ARTICLE

Dispersal of wild and escapee farmed Atlantic cod (Gadus morhua) inNewfoundlandEmily W. Zimmermann, Craig F. Purchase, Ian A. Fleming, and John Brattey

Abstract: Inherent trait differences and changes that arise through domestication could be maladaptive and lead to negativeecological consequences when non-native individuals escape from aquaculture cages and interact with wild populations. Weused acoustic telemetry to map the spatiotemporal distribution of local wild (n = 29) and “escapee” farmed Atlantic cod (Gadusmorhua) (n = 52) through experimental releases off eastern Newfoundland to determine the potential for interaction. Dispersalfrom the cage (>600 m) was rapid (50% dispersal: 12 h for farmed; 5 h for wild) and nonrandom. Most cod (85% farmed, 55% wild)moved northward, remaining close to shore. Although recaptures of escapees during small-scale recreational and commercialfisheries was high (11% farmed; 10% wild), our results suggest that directed efforts to recapture escapees would be logisticallychallenging. Codmigrated a considerable distance (maximum256 km for wild; 157 km for farmed), and some returned to the baythe following year. The similarity of the distribution of escapee farmed and wild cod suggests the potential for interactionsbetween farmed and wild fish, highlighting the importance of minimizing escapes.

Résumé : Des différences de caractères inhérentes et des modifications de caractères découlant de la domestication peuventconstituer de mauvaises adaptations et avoir des conséquences écologiques néfastes quand des individus non indigèness'échappent de cages d'aquaculture et interagissent avec des populations sauvages. Nous avons utilisé la télémétrie acoustiquepour cartographier la répartition spatiotemporelle de morues franche (Gadus morhua) sauvages locales (n = 29) et de moruesd'élevage « échappées » (n = 52) dans le cadre de lâchers expérimentaux sur la côte est de Terre-Neuve afin de déterminer leurpotentiel d'interaction. La dispersion a partir de la cage (>600m) était rapide (50 % de dispersion : 12 h pour les morues d'élevage,5 h pour les morues sauvages) et non aléatoire. La plupart des morues (85 % des morues d'élevage et 55 % des morues sauvages)se déplaçait vers le nord, demeurant près de la côte. Bien que la reprise de morues échappées dans le cadre de pêches récréativeset commerciales a petite échelle fut élevée (11 % des morues d'élevage, 10 % des morues sauvages), nos résultats suggèrent que lareprise intentionnelle de poissons échappés serait difficile d'un point de vue logistique. Les morues ont migré sur des distancesconsidérables (distances maximums de 256 km pour les poissons sauvages et de 157 km pour les poissons d'élevage) et certainessont revenues dans la baie l'année suivante. La similitude de la répartition des morues d'élevage échappées et des moruessauvages laisse indique qu'il y a un potentiel d'interactions de poissons d'élevage et de poissons sauvage, soulignant ainsil'importance de minimiser le nombre de poissons échappés. [Traduit par la Rédaction]

IntroductionLocal adaptation plays a key role in maintaining ecosystem pro-

ductivity and the genetic diversity of wild populations (Kaweckiand Ebert 2004). Domestication combined with environmentalresponses to captive rearing lead to changes inmorphology, phys-iology, and behaviour of farmed relative to wild conspecifics(Einum and Fleming 2001; Huntingford 2004). These potentiallymaladaptive changes for success in the wild could lead to negativeecological consequences if escapees interact and ultimately breedwith wild populations (Fleming et al. 2000; Huntingford 2004;Bekkevold et al. 2006). Interbreeding with wild fish could poten-tially reduce the fitness of native populations through outbreed-ing depression (McGinnity et al. 2003; Hindar et al. 2006). Non-native farmed fish may lack local adaptations and have reducedgenetic variation owing to small founder populations (Fergusonet al. 2007), and crosses with wild fish could result in hybrid off-spring with intermediate or poorer performance than the paren-

tal populations (Einum and Fleming 1997; McGinnity et al. 2003;Fraser et al. 2010).

Fish escaping from net pens is a persistent problem (Jensenet al. 2010) that occurs almost everywhere net pen aquaculture ispracticed (Naylor et al. 2005). In addition to interbreeding, otherpotentially negative impacts of escapes include disease transmis-sion and resource competition (reviewed in Naylor et al. 2005;Diana 2009). Most knowledge of the effects of escapees comesfrom studies on salmon; however, some recent work has beendone on escapee Atlantic cod (Gadus morhua) in Europe (Moe et al.2007; Uglem et al. 2008, 2010; Meager et al. 2009). In the NorthAtlantic, the slow recovery of wild cod stocks following their de-cline (Northeast Atlantic) and collapse (Northwest Atlantic) in thelate 20th century has led to increased farming incentive (Myerset al. 1997; Brown et al. 2003; DFO 2012). Although interest infarming cod has waned recently because of numerous factors,including economics and a resurgence of some wild cod stocks,upon increased demand for cod there is potential to further de-

Received 1 October 2012. Accepted 8 March 2013.

Paper handled by Associate Editor Tara Marshall.

E.W. Zimmermann. Fish Evolutionary Ecology Research Group and Cognitive and Behavioural Ecology Graduate Program, Memorial University of Newfoundland, St. John's, NL A1C 5S7,Canada.C.F. Purchase. Fish Evolutionary Ecology Research Group and Cognitive and Behavioural Ecology Graduate Program, Memorial University of Newfoundland, St. John's, NL A1C 5S7, Canada;Department of Biology, Memorial University of Newfoundland, St. John's, Newfoundland, A1C 5S7, Canada.I.A. Fleming. Fish Evolutionary Ecology Research Group and Cognitive and Behavioural Ecology Graduate Program, Memorial University of Newfoundland, St. John's, NL A1C 5S7, Canada;Department of Ocean Sciences, Ocean Sciences Centre, Memorial University of Newfoundland, St. John's, NL A1C 5S7, Canada.J. Brattey. Science Branch, Fisheries and Oceans Canada, St. John's, NL A1C 5X1, Canada.

Corresponding author: Emily W. Zimmermann (e-mail: ezimmermann@mun.ca).

747

Can. J. Fish. Aquat. Sci. 70: 747–755 (2013) dx.doi.org/10.1139/cjfas-2012-0428 Published at www.nrcresearchpress.com/cjfas on 13 March 2013.

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y T

EX

AS

STA

TE

UN

IV o

n 09

/17/

13Fo

r pe

rson

al u

se o

nly.

velop cod aquaculture (Bolton-Warberg and Fitzgerald 2012). Codare particularly prone to escape from sea cages because of netbiting (Moe et al. 2007; Hansen et al. 2009; Damsgård et al. 2012;Zimmermann et al. 2012), with escape rates of 1.0%, much higherthan the rates of 0.2% for salmon (Jensen et al. 2010). Interbreedingis likely as escapees have been found at wild spawning areas inNorway (Uglem et al. 2008; Meager et al. 2009). Wild cod in theNorthwest Atlantic may be particularly susceptible to hybridiza-tion with escapees because of the low abundance and localizedgenetic structure of coastal populations (Ruzzante et al. 2001;Bekkevold et al. 2006; COSEWIC 2010), especially when escapeesoriginate from non-native populations (Diana 2009).

The susceptibility of wild cod to interactions with escapees de-pends on themigratory patterns of the two groups; however, littleis known about the migration patterns of wild cod in the North-west Atlantic. The stock structure of Newfoundland cod is com-plex, with some discrete groups (Ruzzante et al. 1996) spawning inspecific habitats (offshore, Lear 1984; inshore off headlands,Templeman 1966; or in bays, Hutchings et al. 1993; Smedbol andWroblewski 1997; Brattey et al. 2008), as well as some mixing ofindividuals between groups (Taggart 1997). In southern NorthwestAtlantic Fisheries Organization (NAFO) Division 3L, cod stocksmay be a mix of fish from the summer feeding groups from thenortheast and south coasts of Newfoundland (Lawson and Rose2000; Robichaud and Rose 2004; Brattey et al 2008).

The degree of interaction between escapee and wild cod maydepend on themovement patterns of the escapees relative to wildindividuals. Previous studies have observed escapee farmed codremaining in the local area in Norwegian fjords; however, popu-lation differences in dispersal and migration may occur betweenthe east and west North Atlantic, as well as between bay systems(Uglem et al. 2008, 2010). For example, wild juvenile cod reared ina farm setting in Newfoundland for 3 years prior to release trav-eled up to 170 km from the release site (Wroblewski and Hiscock2002), overlapping the movements of local wild cod. The move-

ments of escapee hatchery-reared cod in the Northwest Atlantichave to our knowledge not been studied previously.

The risk of negative impacts due to escaping fish has been dem-onstrated in salmon; however, few studies have focused on escap-ing cod (see Moe et al. 2007; Uglem et al. 2008; Meager et al. 2009).To determine the potential for interaction between escapeefarmed cod and local wild cod, we quantified the spatiotemporaldistribution of both farmed and wild cod following simulatedescape events. We tested the hypotheses that (i) escapee farmedcod remain near the cage for some time following escape, facili-tating recapture; (ii) the local-scale dispersal pattern of escapeefarmed cod overlaps with that of local wild cod, facilitating inter-actions near the release site; and (iii) farmed cod dispersal outsidethe release site bay is localized relative to wild cod.

Materials and methods

Study site and cod farm characteristicsThis study was focused in Bay Bulls (47°18=N, 52°48=W) in east-

ern Newfoundland (NL), Canada (Fig. 1). Bay Bulls is 4.5 km longwith an average width of 1.1 km, an area of 5.5 km2, and a maxi-mum depth of 90 m. The cod farm in Bay Bulls, run by SapphireSea Farms, had eight pens (average water depth 15 m), three ofwhich each contained 5000 Atlantic cod 39–54 cm long. Theseadult fish (3+ years of age) came from 26 families that originatedfrom the Cod Genome Project (CGP; Trippel et al. 2011). The CGPwas designed to develop a breeding program and genomics toolsto supply the developing Atlantic cod aquaculture industry inCanada with improved broodstock. Under the auspices of CGP,wild broodstock from Smith Sound, NL (NAFO, Division 3L) werebrought to Memorial University's Ocean Sciences Centre (OSC;St. John's, NL) in July 2007. The offspring used in the present studynormally spawn during late winter and spring, but were inducedto spawn in September–October 2008, hatched October–November2008, and were placed in the cages in July 2009.

Fig. 1. Location of the study area in eastern Canada, with Bay Bulls in detail. Light grey areas represent the detection range of the receivers(500–600 m). C, H1–H5, and M1–M2 refer to the receivers (see Table 2 for details).

748 Can. J. Fish. Aquat. Sci. Vol. 70, 2013

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y T

EX

AS

STA

TE

UN

IV o

n 09

/17/

13Fo

r pe

rson

al u

se o

nly.

For comparison with the farmed cod, adult wild cod werecaught in baited cod pots (poly netting on a metal frame, 1.83 m ×1.83 m × 1.04 m) within Bay Bulls a maximum of 24 h prior totagging and by hand line at the southern headland of Bay Bulls(47°17=14==N, 52°45=49==W), approximately 4 km from the farm, 6 hprior to tagging, with care taken to minimize damage to the fishduring retrieval of the gear. Thesewild cod, believed to be amix ofmigrants from northern (NAFO divisions 2J3KL) and southern(NAFO Subdivision 3Ps) cod stocks, come to the Bay Bulls area inthe summer to feed, before migrating elsewhere to spawn in latewinter and spring. The farmed cod, which derive from a wildNewfoundland source, spawn at a similar time.

Acoustic taggingTo determine possible influences of season, two releases of

acoustically tagged farmed and wild cod were performed at thesame site in Bay Bulls in 20 m of water. The first occurred on9 August 2011 when 21 farmed and 14 wild cod were tagged andreleased simultaneously (Table 1). The second occurred on 10 Oc-tober 2011 when 17 farmed codwere tagged and released, followedan hour later by a release of 14 farmed and 15 wild tagged cod(Table 1). The delayed release was undertaken to determinewhether farmed cod followed the wild cod after release. Wild andfarmed cod were size-matched by length as closely as possible;however, wild cod were on average 13% longer than farmed cod inthe first release (generalized linear model (GLM): F[1,33] = 5.47,p = 0.02), but not in the second (GLM: F[1,27] = 0.61, p = 0.44). Therewere no significant differences in length between releases (GLM:F[2,76] = 0.09, p = 0.92), nor was there an interaction between re-lease and cod type (GLM: F[1,76] = 1.01, p = 0.31). Farmed cod hadgreater girth than wild cod as well as differences in other keymorphometric features such as neck curvature, as has been pre-viously observed (see Uglem et al. 2011).

To tag the cod, individuals were placed on a V-shaped surgicaltable, ventral side up to induce tonic immobility. The eyes werecovered with a wet dark cloth, and seawater was continuouslypoured over the gills. Each fish was surgically implanted with anindividually coded transmitter (VEMCO, Halifax, Nova Scotia,Canada) and externally tagged with a T-bar tag (Floy Tag Manufac-turing, Seattle, Washington, USA) following the methods ofUglem et al. (2008). Cod less than 60 cm were tagged with modelV13-1H transmitters (13 mm × 45mm,mass 6 g, frequency 69 kHz,120 ± 60 s ping rate), while cod longer than 60 cmwere taggedwithmodel V16-5H transmitters (16 mm × 93mm,mass 16 g, frequency69 kHz, 120 ± 60 s ping rate). Surgery time was <2min. Tagged fishwere allowed to recover for up to 60 min after surgery (resump-tion of normal, upright swimming behaviour at the bottom of thetank) prior to release. Transmitter function was checked with amobile receiver (VEMCO VR100) prior to release. Cod were thenreleased by lowering them into thewaterwith dip nets adjacent tothe cage site. Cod were tracked for up to 1 year. All handling andtagging were conducted according to the Canadian Council onAnimal Care regulations for the treatment and welfare of animalsand was approved by the Memorial University Animal Care Com-mittee (protocols 11-17-IF and 12-17-IF). In case of recapture by local

fishermen, a monetary reward was offered upon receipt of thetags to encourage reporting of capture location, gear type, andfish size.

Acoustic arrayThe movements and distribution of the tagged cod were re-

corded using nine individual receivers (VEMCO model VR2W) de-ployed on anchored ropes throughout Bay Bulls (Fig. 1; formooring apparatus details see Brattey et al. 2008). All receiversrecorded the transmitter identification code, date, and time ofdetection when a tagged cod was within the receiver range. Re-ceivers were retrieved as in Brattey et al. (2008). The detectionrange of the transmitters was determined by carrying out a seriesof range tests in which a transmitter was towed slowly away fromthe receiver (as in Uglem et al. 2008). The average detection rangeof the receivers was a radius of 500–600 m.

In addition, detection rate of stationary reference transmittersmoored near the farm (VEMCO model V16 4-H, 16 mm × 68 mm,mass 24 g, frequency 68 kHz, 700 ± 60 s ping rate) was comparedwith weather conditions (air temperature, water temperature at10 m, wind direction, and wind speed). During June and July, datawere binned in 4 h blocks. Average weather conditions when thedetection rate was poor (<50%) were as follows: air temperature of8.5 ± 2.3 °C (mean ± SD), water temperature of 5.5 ± 1.1 °C, winddirection of 155 ± 79° (south–southeast), and wind speed of 16.2 ±11.7 km·h–1. In comparison, when the detection rate was good(>50%), the air temperature was significantly higher (12.5 ± 4.8 °C;F[1,331] = 7.57, p < 0.01); however, water temperature (5.6 ± 2.1 °C;F[1,331] = 0.01, p = 0.97), wind direction (193 ± 83° south–southwest;F[1,331] = 2.26, p = 0.13), and wind speed (19.1 ± 9.2 km·h–1; F[1,331] =1.00, p = 0.32) were similar to periods of poor detection. It isunclear how air temperature would impact detection rates; how-ever, we concluded that weather was not hindering the detectionrates of the transmitters, and therefore any differences inweatherconditions between the releases of tagged fish would not impactour results.

Within the bay, a portable VR100 hydrophonewas used to listenfor 5min at a grid of stations located 500m apart to determine thegeneral location and transmitter serial number of tagged cod on22 August, 2 September, 13 September, and 16 November 2011.Data acquired with the VR100 were pooled with detections fromthe VR2W arrays. An extensive inshore network of receivers de-ployed by Fisheries and Oceans Canada (DFO), comprising 30 arraysof two to ten receivers (VEMCOVR2Wreceivers) across 250kmof thenortheast coast of Newfoundland, including known cod spawningareas, was used to detect the cod after leaving Bay Bulls (Brattey et al.2008). In addition, temperature was recorded by data loggers (OnsetHobo U22 Water Temp Pro v2, Bourne, Massachusetts, USA) co-moored with each hydrophone and by conductivity–temperature–depth (SeabirdElectronics SeaCATProfiler 19, Bellevue,Washington,USA) casts in the middle of Bay Bulls.

Data analysisFish that showed no movement after release over a period of

several months were assumed to have died and were excludedfrom the analyses (n = 1 farmed and 1 wild cod; 97.5% survival rate,as in Brattey et al. 2008). Data were pooled into three receiverarrays (C, H1–H2, and H3–H5) within Bay Bulls (Fig. 1). Detection ofa fish on at least one of the receivers within the array was definedas presence within that array's detection area. Acoustic noise andsignal collisions caused some false signals, so single ping detec-tions separated bymore than 1 hwere considered to be erroneous,unless validated by detection at one of the nearby receivers. Fishwere defined as departed from the cage when there were no de-tections for 3 h. Proportion of time within Bay Bulls spent withinthe various receiver arrays was determined by the number ofhours during which individual cod were detected at the array

Table 1. Mean length and number of acoustically tagged farmed andwild cod released in three simulated escape events in 2011 in Bay Bulls,Newfoundland (47°18.663=N, 52°48.259=W).

Release Fish type Release dateNo. offish

Mean (±SD)length (cm)

1 Farmed 9 August 2011 21 46.3±2.4Wild 14 51.9±10.5

2.1 Farmed 10 October 2011 17 48.8±2.82.2 Farmed 10 October 2011 14 46.9±3.3

Wild 15 48.9±3.2

Zimmermann et al. 749

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y T

EX

AS

STA

TE

UN

IV o

n 09

/17/

13Fo

r pe

rson

al u

se o

nly.

relative to the total number of hours each individual was detectedwithin the bay.

All datawere analyzed using R 2.15.1 (R Development Core Team2012). Null hypotheses were rejected when p < 0.05. Residuals ofall models were checked for normality and homogeneity. Datawere analyzed with a GLMwith Poisson error distribution and loglink, unless stated otherwise. To determine the role of release dateand fish type on departure time from the release site, as well asarrival time at receiver array detection zones, we used a GLMwitha response variable of departure time and fixed effects of releasedates (n = 3), type of fish (n = 2), and the interaction between thetwo. Subsets of data including only farmed cod were used to de-termine differences in departure time of farmed cod betweenreleases. To determine the proportion of time spent at the re-ceiver arrays within Bay Bulls, we used a GLM with a responsevariable of percent time detected at each array (out of the totaldetections within the bay for each individual) and fixed effects offish type (n = 2), receiver array (n = 3), and release (n = 3), and allinteractions. To determine the role of detection location onmove-ment speed of cod (calculated from the minimum straight linedistance and assuming that fish are at the receiver location whenthe signal was received), we used a GLM with response variable ofswimming speed and a fixed effect of receiver array location.

Results

Distribution and dispersal within the bayFarmed cod remained near the cage longer (Fig. 2; F[1,74] = 114.2,

p < 0.01) and reached the inner fence later than wild cod (F[1,76] =51.89, p < 0.01). Fish from the second release dispersed away fromthe cage quicker than those from the first release (Fig. 2; F[2,74] =78.38, p < 0.01) and reached the inner fence sooner (F[2,76] = 75.33,p < 0.01). Within 3–8 h (among the three releases) after release,50% of the wild cod had dispersed >600 m from the cage; forfarmed cod this took 5–21 h. Time to 90% dispersal from the cagewas 7–15 h for wild cod and 20–42 h for farmed cod. There was nosignificant interaction between release and fish type at the cage(F[1,74] = 0.17, p = 0.68) or at the inner fence (F[1,76] = 0.85, p = 0.36).Farmed cod released independently (Release 2.1) dispersed moreslowly from the cage (F[1,28] = 15.89, p < 0.01) and arrived at theinner fencemore slowly (F[1,29] = 47.71, p< 0.01) than those releasedsimultaneously with wild cod (Release 2.2). All the tagged codwere detected at the inner fence, and the southern receiver (H1)detected more pings than the northern receiver (H2; Table 2).

Wild cod arrived at the outer fence on average 16 h earlier thanfarmed cod (F[1,73] = 77.14, p < 0.01). In addition, cod from theOctober releases took on average 22 h longer to reach the outerfence than cod released in August (F[2,73] = 117.96, p < 0.01). Therewas a significant interaction between release and fish type (F[1,73] =79.53, p < 0.01), with farmed cod released independently in Re-lease 2.1 taking the longest (57 h) to arrive at the outer fence.However, there was no significant difference in arrival time be-tween farmed cod from Release 2.1 and 2.2 (F[1,27] = 3.35, p = 0.07).All but three of the tagged cod (all from Release 2) were detectedwithin the outer fence array; however, the northernmost receiver(H5) has not been retrieved since September 2011 (i.e., before Re-lease 2). The three cod that left the bay undetected by the rest ofthe outer fence arraymay have been detected by H5. Each receiverin the outer fence array detected between 80%–88.6% (Release 1)and 78.3%–94% (Release 2) of the cod. The southernmost receiver(H3) detected the most total pings, but more cod (particularly offarmed origin) were detected by the middle receiver (H4; Table 2).

The proportion of time spent by codwithin Bay Bulls in range ofvarious receiver arrays differed (F[2,231] = 55.55, p < 0.01), with codspending 38.8% ± 32.1% of their time at the outer fence, comparedwith 31.7% ± 26.3% near the cage and 29.6% ± 22.7% at the innerfence. There was a significant interaction between cod type andreceiver array (Fig. 3; F[1,231] = 528.04, p < 0.01) and between release

and receiver array (F[1,231] = 104.79, p < 0.01); cod from Releases 1and 2.2 spent the most time at the outer fence (45.8% ± 31.3% and42.4% ± 35.2%, respectively), while farmed fish from Release 2.1(released independently) spent the most time at the cage (54.3% ±19.8%). Overall, the median time spent in the bay was relativelyshort, with farmed cod spending 14 days (range of 9 h to 29 days)and wild cod spending 20 days (range of 5 h to 40 days) within thebay. The receiver array at the feeding grounds off of the mouth ofBay Bulls (M1 and M2) was deployed just prior to Release 2, so nocomparison can bemade between Releases 1 and 2 at this location.Cod spent a significant proportion of time at this array, with codfrom Release 2 spending approximately one-quarter of their timewithin and near Bay Bulls (27.0% ± 37.0%) at the feeding grounds.Wild cod spent significantly more time at the feeding groundsthan farmed cod (F[1,44] = 5.76, p = 0.02), spending on average 30%of their time in Bay Bulls at the feeding grounds, compared withonly 26% for the farmed cod.

Dispersal outside the bayOnce the cod left Bay Bulls, we could only detect them along the

coast, where receiver arrays were located. Of the cod detectedoutside Bay Bulls (n = 67, 82.7%), most were detected on arrays tothe north, including Petty Harbour (73.2%), approximately 21.3 kmalong the coast north of Bay Bulls. Cod from Release 1 (August)took the longest to arrive at Petty Harbour (40.1 ± 29.3 days; F[2,57] =59.5, p < 0.01; Fig. 4). Overall, farmed cod arrived more quickly(28.1 ± 22.7 days) than wild cod (41.9 ± 27.6 days; F[1,57] = 37.6,p < 0.01), and specifically farmed cod from Release 2.1 arrivedmore quickly (21.4 ± 7.3 days) than those from Release 2.2 (28.9 ±24.1 days; F[1,34] = 19.99, p < 0.01). However there was no significantinteraction between the release and cod type (F[1,57] = 0.42, p = 0.52). Agreater percentage of the farmed than wild cod were detected atPetty Harbour (Fisher's exact test p < 0.01). As of the end of January2012 (when detections ceased for the winter), 44 of the 52 taggedfarmed cod (84.6%) had been detected at Petty Harbour, compared

Fig. 2. Dispersal of tagged cod away from the cage release site, asmeasured by the percentage of fish detected following release atthe cage site hydrophone, for Release 1 and Release 2. Farmed (�)and wild (●) cod released simultaneously, as well as the farmedcod released independently (an hour earlier) during the secondrelease (e), are shown.

750 Can. J. Fish. Aquat. Sci. Vol. 70, 2013

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y T

EX

AS

STA

TE

UN

IV o

n 09

/17/

13Fo

r pe

rson

al u

se o

nly.

with only 16 of 29 taggedwild cod (55.2%). Arrival in Petty Harbourwas highly variable and took from 4 days (0.13 body lengths(BL)·s−1) to 117 days (0.005 BL·s−1), with cod traveling on average 1.1 ±0.9 km·day−1 (0.03 ± 0.02 BL·s−1). Seventy-one percent (34 farmed,10 wild) of the tagged cod spent 1 day or less in Petty Harbour.Based on the average travel speed, these fish simply passed by thearray, while 27% (12 farmed, 6 wild) remained in the area for2–23 days (mean ± SD: 2.8 ± 4.3 days). The four cod that passedCape Broyle, 27.8 km along the coast south of Bay Bulls, traveledthe same speed as those that went north, traveling on average1.5 ± 2.1 km·day−1 (0.04 ± 0.06 BL·s−1; F[1,64] = 0.03, p = 0.86).

Farther afield, 16 cod (12 farmed, 4 wild) were detected in Con-ception Bay through December 2011, approximately 80 km alongthe coast north from the release site (Fig. 5). Total detectionsdecreased after December, with no detections from January to

April, when water temperatures were significantly colder (–0.28 ±0.95 °C) thanwhen fishwere present (1.82 ± 2.40 °C; F[1,7548] = 2034,p < 0.01). Cod were not detected again until April 2012 (see follow-ing section). None of the cod released herein were detected on theDFO arrays beyond Bellevue in southern Trinity Bay (256 km alongthe coast from Bay Bulls) north to Twillingate (see Brattey et al.2008 for array locations) or on an additional array located atTriton, Notre Dame Bay.

Acoustic detections after JanuaryAs of mid-January 2013, 10 cod had been detected since January

2012. From Release 1, two farmed cod were detected at the mouthof Bay Bulls (one in September, and one in April, whichwas also inPetty Harbour in July), one farmed cod was detected in Bauline(76 km north and west along the coast from Bay Bulls), and onewild cod was detected at Grates Cove between Conception andTrinity bays (160 km away, see Fig. 5). Another wild cod was de-tected heading north past Cape Broyle in May, then past themouth of Bay Bulls and Petty Harbour, into Conception Bay andTrinity Bay as far as Bellevue by June 27, having covered a mini-mum distance of 256 km in 36 days, traveling 0.2 BL·s−1, beforereturning to Conception Bay in July. One farmed cod from Release2.1 was detected in Conception Bay in May. From Release 2.2, one

Table 2. Details of acoustic receiver deployments along the Avalon Peninsula, indicating location, deployment date, depth deployed, and amountof time deployed during the study (9 August 2011 to 19 September 2012).

Receiverarray ID

Location ofarray

Distance tonext array(km)

Deploymentdate Last retrieval date

No. ofreceivers Depth (m)

Deploymenttime (%)

No. of pings(farmed, wild)

C Cage 0 6 June 2011 12 September 2012 1 20 84 43 643 (51f, 28w)H1 Inner fence 1.5 10 June 2011 19 September 2012 1 40 100 24 635 (51f, 29w)H2 Inner fence 10 June 2011 19 September 2012 1 40 100 22 750 (52f, 29w)H3 Outer fence 1.9 10 June 2011 19 September 2012 1 50 100 32 330 (38f, 27w)H4 Outer fence 10 June 2011 19 September 2012 1 70 100 14 402 (47f, 27w)H5 Outer fence 10 June 2011 1 September 2011a 1 70 100b 4 692 (18f, 10w)M1 Mouth 2.1 4 October 2011 19 September 2012 1 40 86 22 092 (15f, 16w)M2 Mouth 4 October 2011 12 August 2012 1 40 76 50 015 (18f, 13w)CB Cape Broyle 25.7 9 June 2011 12 June 2012 3 100–163 100 552 (1f, 3w)PH Petty Harbour 20.2c 29 June 2011 18 June 2012 2 91–115 100 34 325 (46f, 18w)CSF Cape St. Francis 45.7 29 June 2011 18 June 2012 4 50–210 100 7 309 (12f, 2w)BL Bauline 10.2 12 July 2011 19 September 2012 3 30 100 1 923 (6f, 3w)BI Bell Island 10.7 27 June 2011 18 June 2012 4 136–163 100 633 (6f, 1w)SC Salmon Cove 22.4 29 June 2011 18 June 2012 3 43–92 100 69 (3f, 1w)BC Baccalieu 43.7 30 June 2011 27 August 2012 3 105–120 100 227 (4f, 3w)GC Grates Cove 11.2 30 June 2011 28 June 2012 2 65–324 100 17 (0f, 2w)HH Hants Harbour 35.9 1 July 2011 28 June 2012 2 66–286 100 1 (0f, 1w)BV Bellevue 39.3 1 July 2011 28 June 2012 4 74–146 100 39 (0f, 1w)

Note: Minimum distances to next array were calculated using straight line approximations. See Figs. 1 and 2 for locations of receivers. Deployment time wascalculated from 9 August 2011 to 19 September 2012. Total number of pings and the number of individual farmed (f) and wild (w) cod detected at each array aresummarized. See Figs. 1 and 5 for locations.

aReceiver has not been retrieved since the second release of cod.bData was only recovered from 9 August 2011 to 1 Sepember 2011 (5% of the study duration).cDistance measured from mouth of Bay Bulls.

Fig. 3. Percentage of time within Bay Bulls spent by tagged codwithin range of each of three receiver arrays: the cage site, the innerfence, and the outer fence. Data are based on total detections withinthe bay. Error bars show standard deviation among individual cod.

Fig. 4. Cumulative arrival of tagged cod at Petty Harbour.

Zimmermann et al. 751

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y T

EX

AS

STA

TE

UN

IV o

n 09

/17/

13Fo

r pe

rson

al u

se o

nly.

farmed codwas detected in Petty Harbour and at themouth of BayBulls in May, as well as one wild cod in Petty Harbour in June, onewild cod in Conception Bay in July, and one wild cod in Concep-tion Bay in August. Based on detections and recaptures, at least80.2% (48 farmed and 17 wild) of the codwent north. Of these, only22.2% (13 farmed and 5 wild) were detected beyond Petty Harbour.In addition, 12.3% (4 farmed, 6 wild) oscillated north and southamong Petty Harbour, Bay Bulls, and Cape Broyle. In contrast,2.5% were detected only south of Bay Bulls (two wild: one detectedon an array, one recaptured).

Recaptures of tagged fishWithin 1 year of release, a total of nine (six farmed, three wild)

of the 81 tagged cod were recaptured by fishermen (no fish wererecaptured later than 1 year after release). Most of these werecaught by local fishermen (one farmed cod in Bay Bulls, one wildand four farmed cod in Petty Harbour) with hand lines (recre-ational) or gillnet (small commercial fishery). Five of these fromRelease 1 (four farmed and onewild cod) were recaptured between31 August and 10 September 2011, and one farmed cod from Re-lease 2.2 was recaptured 3 December 2011. Recaptures in 2012involved one wild cod from Release 1 recaptured on 7 June 2012 inPlacentia Bay (�260 km south along the coast from the releasesite) in an adjacent stock management area (NAFO Subdivision3Ps; Fig. 5), as well as one wild cod from Release 2.2 and onefarmed cod from Release 1 captured during August 2012 in Con-ception Bay near Brigus and Gull Island, respectively. The recap-ture rate for escaped farmed cod was 11.5% and for wild cod was10.3%, but this was not significantly different between groups(�2 = 0.05, df = 1, p = 0.83).

DiscussionEscapes of farmed fish can be both economically and ecologi-

cally detrimental, especially if farmed fish are from a non-nativepopulation. This study showed that escaped farmed cod dispersedaway from the farm site quickly and broadly and overlapped withthe range of local wild cod, indicating the considerable potentialfor interactions. In a worst case scenario, such interactions couldadversely affect recovering wild populations of cod (through in-terbreeding, competition, and pathogen transfer). For example,cod from the Northwest Atlantic show local adaptation betweennorthern and southern groups (Purchase and Brown 2000, 2001;Hutchings et al. 2007; Bradbury et al. 2010), and non-native south-ern cod have been brought to Newfoundland for culture in sea-cages within the habitat of local wild cod.

Intentionally released farmed cod dispersed away from thefarmmore slowly than wild cod, as seen previously in Newfound-land using juvenile wild cod reared in captivity for 3 years prior torelease (Wroblewski and Hiscock 2002), as well as in other speciesthat show a high degree of site fidelity to sea cages (Bridger et al.2001; Arechavala-Lopez et al. 2011, 2012). Farmed cod dispersalrates after release were similar to previous work; however, wildcod dispersal rates were slower (12 h for 90% dispersal comparedwith only 3 h for 100% dispersal observed by Wroblewski andHiscock (2002)). Differences in seasonal activities such as migra-tion and feeding may affect dispersal rate, as cod released inAugust dispersedmore slowly than those released in October, andboth dispersedmore slowly than the cod released in late April andearly May by Wroblewski and Hiscock (2002) (wild juvenilesreared in captivity for 3 years prior to release). Furthermore, wedetected few fish from December until April, suggesting that dur-ing winter the cod were avoiding the cold shallow coastal waters

Fig. 5. Number of cod detected on DFO receiver arrays moored in coastal waters along the northeast coast of Newfoundland. In total, 81 codwere released with acoustic transmitters from Bay Bulls. Bold letters refer to receiver arrays (for details see Table 2). Black stars represent wildcod recaptured by fishermen, while grey stars represent a mix of wild and farmed cod recaptured. Overall size (diameter) of pie charts arescaled to the total number of cod detected at each array, noted by bold number (e.g., maximum detections at Petty Harbour (PH), 64;minimum detections at Bellevue Beach (BV), 1).

752 Can. J. Fish. Aquat. Sci. Vol. 70, 2013

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y T

EX

AS

STA

TE

UN

IV o

n 09

/17/

13Fo

r pe

rson

al u

se o

nly.

where our receivers were located, as seen previously in Concep-tion Bay (Aggett et al. 1987; Lawson and Rose 2000).

In contrast with our research, Norwegian studies have foundfarmed cod to initially disperse more quickly than wild cod(Svåsand et al. 2000; Uglem et al. 2008). The wild cod used in thepresent study were captured �2 km away from the farm, whereasthose used by Uglem et al. (2008) were caught <200 m from thecage site. Conflicting results may also be due to heritable differ-ences in migratory behaviour. The farmed cod used by Uglemet al. (2008) came from a nonlocal stock that was more migratoryand pelagic in behaviour than the local wild cod stocks. In theNorthwest Atlantic, groups of cod inhabit a more seasonally vari-able environment and tend to disperse along the coast, in contrastwith groups of cod in the more stable environment of the North-east Atlantic that tend to be accurate homers and sedentary(Robichaud and Rose 2004).

Escapee farmed cod in the present study left the farm areamoreslowly than wild cod, but most of the escapees had dispersedwithin 48 h (50% of farmed cod and 34% ofwild cod),much quickerthan found previously (Uglem et al. 2008, 2010), indicating thatany attempt at a recapture fishery would need to be undertakenimmediately after an escape incident. Few studies have quantifiedthe recapture rate of escaped farmed cod; however, results froman acoustic telemetry study in Norway (Uglem et al. 2008) as wellas from stock enhancement studies (Svåsand et al. 2000; Skilbreiand Jørgensen 2010) suggest that the incidental recapture ratesmay be high (30%–44%) with even a modest fishing effort. Whilethere was no dedicated recapture fishery near Bay Bulls, rates ofincidental fishery recaptures did not differ between escapeefarmed (11%) and wild cod (10%; similar to Brattey et al. 2008), incontrast with previous studies (Wroblewski and Hiscock 2002;Uglem et al. 2008). The timing and recapture locations in thisstudy reflect the pattern of seasonal (i.e., summer) small-scalefisheries (recreational, sentinel, and commercial) at the mouth ofBay Bulls and off Petty Harbour. In addition, detections and recap-tures at Petty Harbour indicate cod remained in a narrow corridorclose to shore as theymigrated northward, rather than dispersingrandomly. This suggests that by targeting local wild cod aggrega-tions, such as within and near Bay Bulls, the efficiency of a recap-ture fishery could be increased. Given the broad spatial overlap offarmed escapees with wild cod and the similarity in recapturerates in the present study, it is likely that any targeted fishery forescapee recapture will result in substantial incidental catch ofwild cod. Several factors, such as differences in fish size at release,innate behavioural differences due to stock origin, differing geog-raphy near cod farms, and local fishing effort, however, makecomparisons with other studies and areas difficult.

During their short stay in the bay, dispersal varied betweenfarmed andwild cod. Many of the wild cod quickly returned to themouth of the bay where they had originally been captured, someremaining in the area for several months. In contrast, farmed coddispersed more slowly and moved around within the bay beforeleaving. The wild cod had previous experience of the local envi-ronment and location of feeding areas at themouth; however, thefarmed codmay have been exploring the bay because of their lackof knowledge of the environment and possibly their inability toevaluate habitat qualities (Uglem et al. 2008; Dempster et al. 2010).In addition, farmed cod released together with wild cod dispersedmore rapidly than farmed cod released independently of wild cod.The high variability in arrival time at Petty Harbour indicates thatalthoughmany wild and farmed cod moved in the same direction(i.e., northward), individual fish were not in close proximity.Farmed cod may have been attracted to aggregations of wild codthrough some unknown sensorymechanism beyond vision. Inter-actions between escaped farmed and wild fish may thus increaseif farms are located near a local wild aggregation. Targeted fish-eries to recover escaped fish may also result in high incidentalcatches of wild fish.

Following dispersal from Bay Bulls, the majority of farmed cod(85%) were detected at Petty Harbour, arriving faster than wildcod, and both farmed and wild cod did not linger there for morethan a day before continuing north. The average travel speed of0.03 BL·s−1 reported here matched that observed by Comeau et al.(2002) for summer migrants. Our study was conducted primarilyduring the summer and early autumn, when cod are inshore andfeeding, possibly resulting in slower migrations along the coastthan the speed of 0.23 BL·s−1 seen by Rose et al. (1995) during theshoreward spring migration. Basal swimming speed in fishes is afunction of body length, with larger individuals tending to showgreater swimming speeds (Santos 2011). Comparisons are there-fore influenced bymany factors, including the size composition ofthe cod being studied.

Few studies have tracked the dispersal of escapee farmed cod,and these studies have been limited mostly to tracking withinfjord systems up to 15 km (but see Uglem et al. 2008). Courtesy oftheDFO acoustic receiver arrays dispersed aroundNewfoundland,we detected farmed cod up to 114 km away from the release site.Similar results were observed previously, with wild juvenile cod(raised in a farm situation for 3 years prior to release) detected upto 170 km away from a release site in Trinity Bay, NL (Wroblewskiand Hiscock 2002). Cod begin long migrations by age 4 or 5 (Rose1993; Lawson and Rose 2000), consistent with the size and age ofcod tagged in this study. Our results indicate that farmed cod areable to disperse great distances and overlap with the range of wildcod.

We observed a wide geographic dispersal of both wild andfarmed cod from Bay Bulls north to Trinity Bay and south toPlacentia Bay, consistent with historical descriptions of the“Avalon stock complex” (Templeman 1962, 1979). Lear (1984)found that cod that overwinter on offshore banks migrate to east-ern Newfoundland for the summer, illustrating the wide overlapof cod from different management divisions (Robichaud and Rose2004). Previous studies found that 10%–30% of the tagged PlacentiaBay cod migrated east and north along the Avalon Peninsula,traveling as far as Trinity Bay, suggesting that some of the codcaptured at Bay Bulls may have originated from Placentia Bay(Lawson and Rose 2000; Robichaud and Rose 2001). Only abouthalf of all tagged cod in our study were detected north of BayBulls, suggesting that some wild cod migrated south toward Pla-centia Bay (including one recaptured there in June 2012, as alsoseen in Brattey 2000) or to the offshore banks (Lear 1984). Previousstudies have shown that wild cod are able to migrate along thesame route, returning to the same site several years in a row (Rose1993; Robichaud and Rose 2001), as indicated by the redetection offive cod in Bay Bulls in the spring and fall of 2012, as well as theoverwinter survival of six other tagged cod. However, owing tothe small sample size, we cannot draw further conclusions on theorigins of the wild cod.

All tagged cod were detected within Bay Bulls; however, 14.8%were never detected along the coast beyond the bay. It is notpossible to distinguish among natural mortality, unreported fish-ingmortality, emigration, or transmitter failure. Transmitters arereliable and were tested prior to release. Some undetected codmay have been caught in lost fishing gear (e.g., “ghost-fishing”), orthe fisher did not claim the reward; however, tagging is well ad-vertised and it is believed that most recaptures are reported(Brattey et al. 2008). It is thereforemore likely that undetected codmoved beyond the range of the receiver arrays, offshore to deeperwater. If feasible, future studies should expand the receiver arrayalong the southern coast of the Avalon Peninsula and into deeperwater.

In conclusion, escapee farmed cod dispersed from the releasesite more slowly than wild cod. However, dispersal of both groupswas rapid (within 24 h), making recapture of escapees logisticallychallenging. Following the initial dispersal period, most codmoved northward, staying close to the shore. This nonrandom

Zimmermann et al. 753

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y T

EX

AS

STA

TE

UN

IV o

n 09

/17/

13Fo

r pe

rson

al u

se o

nly.

dispersal and narrow migration corridor may facilitate an effi-cient recapture fishery. However, the spatiotemporal distribu-tions of escapee farmed cod and wild cod were very similar,suggesting the potential for a high incidental catch of wild cod inrecapture attempts, as well as negative interactions between thedomesticated and often nonlocal farmed cod and the local wildcod. Further studies are therefore needed to minimize the num-ber of escapes from farms, to investigate the potential for aneffective recapture strategy, and to consider means of preventinginteractions, such as interbreeding, with wild fish. These stepswill help with the development of environmentally sound aqua-culture not only for cod, but other marine finfish raised in netpens.

AcknowledgementsWe thank D. Porter for his help with the receiver arrays and

P. Williams for the use of his fishing boat. We also thank SapphireSea Farms for their help with the farmed fish. This work wassupported by funding from grants awarded to I.A.F. and C.F.P.from the Natural Sciences and Engineering Research Council ofCanada by way of a Strategic Project grant and the Canada Foun-dation for Innovation. In addition, we received funding from Fish,Food and Allied Workers to J.B. through the Fisheries ScienceCollaborative Program and graduate funding to E.Z. from theResearch and Development Corporation of Newfoundland andLabrador and Memorial University. Two anonymous reviewersprovided comments that improved an earlier version of themanuscript.

ReferencesAggett, D., Gaskill, H.S., Finlayson, D., May, S., Campbell, C., and Bobbitt, J. 1987.

A study of factors influencing availability of cod in Conception Bay, New-foundland, in 1985. Can. Tech. Rep. Fish. Aquat. Sci. 1562.

Arechavala-Lopez, P., Uglem, I., Fernandez-Jover, D., Bayle-Sempere, J.T., andSanchez-Jerez, P. 2011. Immediate post-escape behaviour of farmed seabass(Dicentrarchus labrax L.) in the Mediterranean Sea. J. Appl. Ichthyol. 27: 1375–1378. doi:10.1111/j.1439-0426.2011.01786.x.

Arechavala-Lopez, P., Uglem, I., Fernandez-Jover, D., Bayle-Sempere, J.T., andSanchez-Jerez, P. 2012. Post-escape dispersion of farmed seabream (Sparusaurata L.) and recaptures by local fisheries in theWesternMediterranean Sea.Fish. Res. 121–122: 126–135.

Bekkevold, D., Hansen, M.M., and Nielsen, E.E. 2006. Genetic impact of gadoidculture on wild fish populations: Predictions, lessons from salmonids, andpossibilities for minimizing adverse effects. ICES J. Mar. Sci. 63(2): 198–208.doi:10.1016/j.icesjms.2005.11.003.

Bolton-Warberg, M., and Fitzgerald, R.D. 2012. Benchmarking growth of farmedAtlantic cod, Gadus morhua: a case study in Ireland. Aquac. Res. 43: 670–678.doi:10.1111/j.1365-2109.2011.02872.x.

Bradbury, I.R., Hubert, S., Higgins, B., Borza, T., Bowman, S., Paterson, I.G.,Snelgrove, P.V.R., Morris, C.J., Gregory, R.S., Hardie, D.C., Hutchings, J.A.,Ruzzante, D.E., Taggart, C.T., and Bentzen, P. 2010. Parallel adaptive evolu-tion of Atlantic cod on both sides of the Atlantic Ocean in response to tem-perature. Proc. R. Soc. B Biol. Sci. 277(1701): 3725–3734.

Brattey, J. 2000. Stock structure and seasonal movements of Atlantic cod (Gadusmorhua) in NAFO Divs. 3KL inferred from recent tagging experiments. Cana-dian Stock Assessment Secretariat Res. Doc. 2000/084.

Brattey, J., Healey, B., and Porter, D. 2008. Northern cod (Gadus morhua) 16 yearsafter themoratorium: new information from tagging and acoustic telemetry.Canadian Stock Assessment Secretariat Res. Doc. 2008/047.

Bridger, C.J., Booth, R.K., McKinley, R.S., Scruton, D.A., and Linstrom, R.T. 2001.Monitoring fish behaviour with a remote combined acoustic/radio biotelem-etry system. J. Appl. Ichthyol. 17: 126–129.

Brown, J.A., Minkoff, G., and Puvanendran, V. 2003. Larviculture of Atlantic cod(Gadus morhua): Progress, protocols and problems. Aquaculture, 227(1–4):357–372. doi:10.1016/S0044-8486(03)00514-3.

Comeau, L.A., Campana, S.E., and Castonguay, M. 2002. Automated monitoringof a large-scale cod (Gadus morhua) migration in the open sea. Can. J. FishAquat. Sci. 59(12): 1845–1850. doi:10.1139/f02-152.

COSEWIC. 2010. COSEWIC assessment and status report on the Atlantic CodGadus morhua in Canada. Committee on the Status of Endangered Wildlife inCanada. Ottawa, Ont.

Damsgård, B., Høy, E., Uglem, I., Hedger, R.D., Izquierdo-Gomez, D., andBjørn, P.A. 2012. Net-biting and escape behaviour in farmed Atlantic codGadus morhua: effects of feed stimulants and net traits. Aquacult. Environ.Interactions, 3: 1–9. doi:10.3354/aei00047.

Dempster, T., Sanchez-Jerez, P., Uglem, I., and Bjørn, P.A. 2010. Species-specific

patterns of aggregation of wild fish around fish farms. Estuar. Coastal ShelfSci. 86(2): 271–275. doi:10.1016/j.ecss.2009.11.007.

DFO. 2012. Stock Assessment Update of Northern (2J3KL) Cod. DFO Can. Sci.Advis. Sec. Sci. Resp. 2012/009.

Diana, J.S. 2009. Aquaculture production and biodiversity conservation. BioSci-ence, 59(1): 27–38.

Einum, S., and Fleming, I.A. 1997. Genetic divergence and interactions in thewild among native, farmed and hybrid Atlantic salmon. J. Fish Biol. 50(3):634–651. doi:10.1111/j.1095-8649.1997.tb01955.x.

Einum, S., and Fleming, I.A. 2001. Implications of stocking: ecological interac-tions between wild and released salmonids. Nord. J. Freshw. Res. 75: 56–70.

Ferguson, A., Fleming, I.A., Hindar, K., Skalla, Ø., McGinnity, P., Cross, T., andProdöhl, P. 2007. Farm escapes, In: The Atlantic salmon: Genetics, Conserva-tion, and Management. Edited by E. Verspoor, L. Stradmeyer, and J. Nieslsen,Blackwell, Oxford. pp. 367–409.

Fleming, I.A., Hindar, K., Mjolnerod, I.B., Jonsson, B., Balstad, T., and Lamberg, A.2000. Lifetime success and interactions of farm salmon invading a nativepopulation. Proc. R. Soc. B Biol. Sci. 267(1452): 1517–1523. doi:10.1098/rspb.2000.1173.

Fraser, D.J., Houde, A.L.S., Debes, P.V., O'Reilly, P., Eddington, J.D., andHutchings, J.A. 2010. Consequences of farmed-wild hybridization across di-vergent wild populations and multiple traits in salmon. Ecol. Appl. 20: 935–953. doi:10.1890/09-0694.1. PMID:20597281.

Hansen, L.A., Dale, T., Damsgård, B., Uglem, I., Aas, K., and Bjørn, P.A. 2009.Escape-related behaviour of Atlantic cod, Gadus morhua, in a simulated farmsituation. Aquac. Res. 40(1): 26–34.

Hindar, K., Fleming, I.A., McGinnity, P., and Diserud, O. 2006. Genetic and eco-logical effects of salmon farming on wild salmon: modelling from experi-mental results. ICES J Mar. Sci. 63: 1234–1247. doi:10.1016/j.icesjms.2006.04.025.

Huntingford, F.A. 2004. Implications of domestication and rearing conditionsfor the behaviour of cultivated fishes. J. Fish Biol. 65(Suppl. A): 122–142.

Hutchings, J.A., Myers, R.A., and Lilly, G.R. 1993. Geographic variation in thespawning of Atlantic cod, Gadus morhua, in the northwest Atlantic. Can. J.Fish. Aquat. Sci. 50(11): 2457–2467. doi:10.1139/f93-270.

Hutchings, J.A., Swain, D.P., Rowe, S., Eddington, J.D., Puvanendran, V., andBrown, J.A. 2007. Genetic variation in life-history reaction norms in marinefish. Proc. R. Soc. B Biol. Sci. 274(1619): 1693–1699.

Jensen, Ø., Dempster, T., Thorstad, E.B., Uglem, I., and Fredheim, A. 2010. Es-capes of fishes from Norwegian sea-cage aquaculture: causes, consequencesand prevention. Aquacult. Environ. Interactions, 1: 71–83.

Kawecki, T.J., and Ebert, D. 2004. Conceptual issues in local adaptation. Ecol.Lett. 7: 1225–1241.

Lawson, G.L., and Rose, G.A. 2000. Seasonal distribution and movements ofcoastal cod (Gadus morhua L.) in Placentia Bay, Newfoundland. Fish. Res. 49(1):61–75. doi:10.1016/S0165-7836(00)00187-9.

Lear, W.H. 1984. Discrimination of the stock complex of Atlantic cod (Gadusmorhua) off Southern Labrador and Eastern Newfoundland, as inferred fromtagging studies. J. Northw. Atl. Fish. Sci. 5: 143–159. doi:10.2960/J.v5.a19.

McGinnity, P., Prodöhl, P., Ferguson, A., Hynes, R., Maoiléidigh, N.Ó., Baker, N.,Cotter, D., O'Hea, B., Cooke, D., Rogan, G., Taggart, J., and Cross, T. 2003.Fitness reduction and potential extinction of wild populations of Atlanticsalmon, Salmo salar, as a result of interactions with escaped farm salmon.Proc. R. Soc. B Biol. Sci. 270(1532): 2443–2450. doi:10.1098/rspb.2003.2520.

Meager, J.J., Skjæraasen, J.E., Fernö, A., Karlsen, Ø., Løkkeborg, S., Michalsen, K.,and Utskot, S.O. 2009. Vertical dynamics and reproductive behaviour offarmed and wild Atlantic cod Gadus morhua. Mar. Ecol. Prog. Ser. 389: 233–243. doi:10.3354/meps08156.

Moe, H., Dempster, T., Sunde, L.M., Winther, U., and Fredheim, A. 2007. Tech-nological solutions and operational measures to prevent escapes of Atlanticcod (Gadus morhua) from sea cages. Aquacult. Res. 38(1): 91–99. doi:10.1111/j.1365-2109.2006.01638.x.

Myers, R.A., Hutchings, J.A., and Barrowman, N.J. 1997. Why do fish stocks col-lapse? The example of cod in Atlantic Canada. Ecol. Appl. 7(1): 91–106.

Naylor, R., Hindar, K., Fleming, I.A., Goldburg, R., Williams, S., Volpe, J.,Whoriskey, F., Eagle, J., Kelso, D., and Mangel, M. 2005. Fugitive salmon:assessing the risks of escaped fish from net-pen aquaculture. Bioscience,55(5): 427–437. doi:10.1641/0006-3568(2005)055[0427:FSATRO]2.0.CO;2.

Purchase, C.F., and Brown, J.A. 2000. Interpopulation differences in growth ratesand food conversion efficiencies of young Grand Banks and Gulf of MaineAtlantic Cod (Gadus morhua). Can. J. Fish. Aquat. Sci. 57(11): 2223–2229. doi:10.1139/f00-204.

Purchase, C.F., and Brown, J.A. 2001. Stock-specific changes in growth rates, foodconversion efficiencies, and energy allocation in response to temperaturechange in juvenile Atlantic cod. J. Fish Biol. 58: 36–52. doi:10.1111/j.1095-8649.2001.tb00497.x.

R Development Core Team. 2012. R: a language and environment for statisticalcomputing, R Foundation for Statistical Computing, Vienna, Austria.

Robichaud, D., and Rose, G.A. 2001. Multiyear homing of Atlantic cod to a spawn-ing ground. Can. J. Fish. Aquat. Sci. 58(12): 2325–2329. doi:10.1139/f01-190.

Robichaud, D., and Rose, G.A. 2004. Migratory behaviour and range in Atlanticcod: Inference from a century of tagging. Fish Fish. 5(3): 185–214. doi:10.1111/j.1467-2679.2004.00141.x.

754 Can. J. Fish. Aquat. Sci. Vol. 70, 2013

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y T

EX

AS

STA

TE

UN

IV o

n 09

/17/

13Fo

r pe

rson

al u

se o

nly.

Rose, G.A. 1993. Cod spawning on a migration highway in the north-west Atlan-tic. Nature, 366(6454): 458–461. doi:10.1038/366458a0.

Rose, G.A., deYoung, B., and Colbourne, E.B. 1995. Cod (Gadus morhua L.) migra-tion speeds and transport relative to currents on the north-east Newfound-land Shelf. ICES J. Mar. Sci. 52(6): 903–913. doi:10.1006/jmsc.1995.0087.

Ruzzante, D.E., Taggart, C.T., Cook, D., and Goddard, S. 1996. Genetic differen-tiation between inshore and offshore Atlantic cod (Gadus morhua) off New-foundland: Microsatellite DNA variation and antifreeze level. Can. J. Fish.Aquat. Sci. 53(3): 634–645. doi:10.1139/f95-228.

Ruzzante, D.E., Taggart, C.T., Doyle, R.W., and Cook, D. 2001. Stability in thehistorical pattern of genetic structure of Newfoundland cod (Gadus morhua)despite the catastrophic decline in population size from 1964 to 1994. Con-serv. Genet. 2(3): 257–269. doi:10.1023/A:1012247213644.

Santos, T.C. 2011. Applied aspects of fish swimming performance. In Ency-clopedia of fish physiology: from genome to environment [online]. Edited byA.P. Farrell . Elsevier Science and Technology. Available fromhttp://lib.myilibrary.com.qe2a-proxy.mun.ca?ID=311405 [accessed 30 July2012].

Skilbrei, O., and Jørgensen, T. 2010. Recapture of cultured salmon following alarge-scale escape experiment. Aquacult. Environ. Interactions, 1: 107–115.doi:10.3354/aei00011.

Smedbol, R.K., and Wroblewski, J.S. 1997. Evidence for inshore spawning ofnorthern Atlantic cod (Gadus morhua) in Trinity Bay, Newfoundland, 1991–1993. Can. J. Fish. Aquat. Sci. 54(S1): 177–186. doi:doi:10.1139/f96-146.

Svåsand, T., Kristiansen, T.S., Pedersen, T., Salvanes, A.G.V., Engelsen, R.,Nævdal, G., and Nødtvedt, M. 2000. The enhancement of cod stocks. Fish andFish. 1: 173–205. doi:10.1046/j.1467-2979.2000.00017.x.

Taggart, C.T. 1997. Bank-scale migration patterns in northern cod. NAFO Sci.Coun. Studies, 29: 51–60.

Templeman, W. 1962. Divisions of cod stocks in the northwest Atlantic. ICNAF.Templeman, W. 1966. Marine resources of Newfoundland. Bull. Fish. Res. Board

Can. No. 154. Redbook, 1962, Part III. pp. 79–129.Templeman, W. 1979. Migration and intermingling of stocks of Atlantic cod,

Gadus morhua of the Newfoundland and adjacent areas from tagging in 1962–1966. ICNAF Res. Bull. 14: 5–48.

Trippel, E.A., Rise, M.L., Gamperl, A.K., Johnson, S.C., Robinson, J.A.B., Culver, K.,Rise, M., and Bowman, S. 2011. Development of genomics tools and elitebroodstock for Atlantic cod culture in Canada. World Aquaculture, 42:50–54.

Uglem, I., Bjørn, P.A., Dale, T., Kerwath, S., Økland, F., Nilsen, R., Aas, K.,Fleming, I., and McKinley, R.S. 2008. Movements and spatiotemporal distri-bution of escaped farmed and local wild Atlantic cod (Gadus morhua L.). Aqua-cult. Res. 39(2): 158–170. doi:10.1111/j.1365-2109.2007.01872.x.

Uglem, I., Bjørn, P.A., Mitamura, H., and Nilsen, R. 2010. Spatiotemporal distri-bution of coastal and oceanic Atlantic cod Gadus morhua sub-groups afterescape from a farm. Aquacult. Environ. Interactions, 1: 11–19.

Uglem, I., Berg, M., Varne, R., Nilsen, R., Mork, J., and Bjørn, P.A. 2011. Discrim-ination of wild and farmed Atlantic cod (Gadus morhua) based onmorphologyandscale-circulipattern. ICES J.Mar. Sci.68(9): 1928–1936.doi:10.1093/icesjms/fsr120.

Wroblewski, J.S., and Hiscock, H.W. 2002. Enhancing the reproductive potentialof local populations of coastal Atlantic cod (Gadus morhua). Can. J. Fish. Aquat.Sci. 59(10): 1685–1695. doi:10.1139/f02-136.

Zimmermann, E.W., Purchase, C.F., and Fleming, I.A. 2012. Reducing the inci-dence of net cage biting and the expression of escape-related behaviors inAtlantic cod (Gadus morhua) with feeding and cage enrichment. Appl. Anim.Behav. Sci. 141: 71–78. doi:10.1016/j.applanim.2012.07.009.

Zimmermann et al. 755

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y T

EX

AS

STA

TE

UN

IV o

n 09

/17/

13Fo

r pe

rson

al u

se o

nly.