351

Biological Journal of the Linnean Society

, 2006,

88

, 351–365. With 8 figures

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London,

Biological Journal of the Linnean Society,

2006,

88

, 351–365

Blackwell Publishing LtdOxford, UKBIJBiological Journal of the Linnean Society0024-4066The Linnean Society of London, 2006*** 2006883351365Original Article

PHENOTYPIC VARIATION IN BODY SHAPE AMONG POPULATIONS OF ATLANTIC CODJ. MARCIL

ET AL

.

*Corresponding author. E-mail: swaind@dfo-mpo.gc.ca

Genetic and environmental components of phenotypic variation in body shape among populations of Atlantic cod (

Gadus morhua

L.)

JULIE MARCIL

1

, DOUGLAS P. SWAIN

2

* and JEFFREY A. HUTCHINGS

1

1

Department of Biology, Dalhousie University, Halifax, Nova Scotia B3H 4J1, Canada

2

Fisheries and Oceans Canada, Gulf Fisheries Centre, PO Box 5030, Moncton, New Brunswick E1C 9B6, Canada

Received 15 October 2004; accepted for publication 12 September 2005

A common-garden experiment was conducted on larvae to test for genetic differences in body shape among popula-tions of Atlantic cod (

Gadus morhua

). Offspring from four north-west Atlantic regions were reared from hatching topostmetamorphosis at two temperatures (7

±

1

°

C and 11

±

1

°

C) and two food levels (1500 and 4500 prey L

−

1

). Bodyshape differed between populations and treatments. Population differences were greatest between south-westScotian Shelf cod and those further north; the former were characterized by a deeper body, larger head, and longercaudal peduncle than cod from the other populations. Significant differences were also observed between two puta-tive populations on the south-west Scotian Shelf, suggesting genetic divergence between spawning aggregations atsmall spatial scales (

<

100 km). Temperature and food supply also influenced body shape, with the effect of theformer being more pronounced. Individuals reared at the higher temperature or food level had a deeper body and alarger head than those reared at the lower temperature or food supply. Phenotypic responses to changes in the rear-ing environment also differed among populations, indicating genetic differences in phenotypic plasticity. Differencesbetween populations in morphology and in phenotypic plasticity suggest genetic divergence at both large (

>

1000 km)and small (

<

100 km) spatial scales. The genetic differences at large spatial scales counteracted the expected effectsof temperature differences in the wild, suggesting countergradient variation in morphology among thesepopulations. © 2006 Her Majesty the Queen in Right of Canada. Journal compilation © 2006 The Linnean Societyof London,

Biological Journal of the Linnean Society

, 2006,

88

, 351–365.

ADDITIONAL KEYWORDS:

allometry – Atlantic cod larvae – countergradient variation – fish morphology –genotype by environment interaction – geometric morphometrics – local adaptation – phenotypic plasticity –

population genetics – reaction norm.

INTRODUCTION

Most documented cases of genetic divergence betweenfish populations originate from studies on freshwaterand anadromous fishes (Hendry & Stearns, 2004). Formarine fishes, there are generally fewer geographicalbarriers to gene flow and most species have high dis-persal capability, either as planktonic eggs and larvae(which can be widely dispersed by ocean currents) oras migratory adults. As a consequence, marine fish areoften assumed to be more genetically homogeneousthan freshwater and anadromous fishes, an assump-

tion that earlier genetic studies appeared to support(Gyllensten, 1985; Ward, Woodwark & Skibinski,1994). However, an absence of geographical barriersdoes not exclude the possibility of local adaptation.

Genetic divergence in response to natural selectioncan be rapid (Reznick

et al

., 1997; Conover & Munch,2002; Koskinen, Haugen & Primmer, 2002) and canoccur despite a level of gene flow that would prevent oreliminate neutral genetic differentiation (Allendorf,1983, 1995; Carvalho, 1993). Consequently, geneticdivergence can develop more rapidly in adaptive thanin selectively neutral characters and may persist inadaptive characters despite gene flow (Hard, 1995;Conover, 1998). Given the prominence of genetic

352

J. MARCIL

ET AL

.

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London,

Biological Journal of the Linnean Society,

2006,

88

, 351–365

examinations of gene flow based on molecular markerspresumed to be neutral to selection, it is possible thatgenetic differentiation for adaptive traits may havebeen under-estimated (Karhu

et al

., 1996; Reed &Frankham, 2001; Koskinen

et al

., 2002). The few stud-ies that have examined adaptive characters in marinefishes have indeed found evidence of genetic differen-tiation between populations in terms of growth rate(Conover & Present, 1990; Schultz, Reynolds &Conover, 1996; Puvanendran & Brown, 1998), foodconversion efficiency (Purchase & Brown, 2000), andvertebral number (Billerbeck, Orti & Conover, 1997).

Fishes that exploit different resources in termsof food, space, and habitat tend to evolve differentmorphologies, reflecting adaptation to these ecologicaldifferences (Alexander, 1974). Evidence of local adap-tation in morphology due to differences in selectionpressure has been documented in a number of fish spe-cies. For example, trophic specialization betweenbenthic and pelagic habitats has been observed amongpopulations of Arctic charr,

Salvelinus alpinus

(Snor-rason

et al

., 1994) and between sympatric groups ofstickleback,

Gasterosteus

spp. (McPhail, 1984). Simi-larly, differences in body shape attributable to con-trasting selection pressures related to swimmingperformance have been documented for northernredbelly dace,

Phoxinus eos

(Toline & Baker, 1993)and coho salmon,

Onchorhynchus kisutch

(Taylor &McPhail, 1985a, b; Swain & Holtby, 1989).

Phenotypic plasticity arises when the same geno-type produces different phenotypes in different envi-ronments (Hutchings, 2004). The set of phenotypicresponses across different levels of an environmentalfactor is termed the reaction norm (Via

et al

., 1995)and is specific to each genotype. Body shape in fisheshas been demonstrated to be influenced by environ-mental factors such as temperature (Martin, 1949;Beacham, 1990), food ration (Currens

et al

., 1989),and type of food or feeding mode (Day, Pritchard &Schluter, 1994; Robinson & Wilson, 1996). Plasticity infish body shape has been hypothesized to be adaptive(Robinson & Parsons, 2002). Adaptive phenotypicplasticity in fish morphology has been demonstratedin crucian carp,

Carassius carassius

, in response tothe presence or absence of a predator (Brönmark &Miner, 1992), and in pumpkinseed sunfish,

Lepomisgibbosus

(Robinson & Wilson, 1996) and stickleback,

Gasterosteus

spp. (Day

et al

., 1994), in response to dif-ferences between benthic and pelagic habitats.

Morphological differences observed in the wild thatappear to be adaptive are often assumed to reflectgenetic divergence between groups when in fact theymay be simply cases of adaptive phenotypic plasticity.Conversely, a lack of phenotypic differences in the wildmay not always reflect genetic similarity betweengroups. When genotypic influences oppose environ-

mental influences, a pattern termed countergradientvariation (Conover & Schultz, 1995), phenotypic sim-ilarity in the wild may reflect genetic divergenceamong groups (Conover, 1998). Therefore, when study-ing genetic divergence using phenotypic characterssuch as body shape, it is necessary to separate thegenetic and environmental components of variation.Common-garden experiments have proven to be a use-ful tool for determining the genetic basis of phenotypicdifferences between putative groups (Conover &Schultz, 1995; Conover, 1998; Swain & Foote, 1999).They consist of rearing the different groups underidentical conditions, thus eliminating the environmen-tal differences between groups. The observed differ-ences can then be attributed to genetic divergence.

Atlantic cod,

Gadus morhua

, is a widespread, com-mercially important marine fish. Adults from manypopulations undergo seasonal migrations betweensummer/spawning grounds and over-winteringgrounds (ICES, 1994; Sinclair & Currie, 1994; Lawson& Rose, 2000a; Comeau, Campana & Chouinard,2002). During those migrations, fish from one stocksometimes mix with those from other adjacent stocks(Lawson & Rose, 2000a). The egg and larval stages arepelagic and are subject to dispersal by water currents.Most Atlantic cod populations in the north-west Atlan-tic have been depleted and remain at a low level ofabundance (Hutchings & Reynolds, 2004). A betterunderstanding of the spatial scale of populationstructure and local adaptation would contribute tothe effective management and conservation of thisspecies.

In the present study, a common-garden experimentwas used to test for genetic differences in juvenilebody shape between four spatially separated popula-tions of Atlantic cod. The effects of temperature andfood density on body shape were also examined totest for genetic differences in phenotypic plasticitybetween the populations. This study tests for geneticdivergence between populations in a marine fish withhigh potential for dispersal and may serve as an exam-ple for other widespread marine fish species that haveevolved over a range of environments and have beensubjected to similar levels of gene flow betweengroups.

MATERIAL AND METHODS

S

TUDY

POPULATIONS

Atlantic cod larvae were compared from four areas(Fig. 1): (1) south-western Scotian Shelf (NorthwestAtlantic Fisheries Organization [NAFO] division 4X);(2) southern Gulf of St Lawrence (NAFO division 4T);(3) Placentia Bay, Newfoundland (NAFO subdivision3Ps); and (4) Bonavista Bay on the north-easterncoast of Newfoundland (NAFO division 3L). The

PHENOTYPIC VARIATION IN BODY SHAPE AMONG POPULATIONS OF ATLANTIC COD

353

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London,

Biological Journal of the Linnean Society,

2006,

88

, 351–365

spawning period for the 4T, 3Ps, and 3L populationsoccurs in spring and early summer, with peak spawn-ing in May and June (Lett, 1980; Myers, Mertz &Bishop, 1993; Hutchings & Myers, 1994; Lawson &Rose, 2000b). Spawning for the 4X population occursduring the winter from February to March (ICES,1994). Because of this temporal difference in spawn-ing, 4X larvae are expected to experience colder tem-peratures than those from the other three populations(Marcil, 2004). Adult cod from the 4T populationundertake extensive seasonal migrations betweenspawning and feeding grounds inside the Gulf of StLawrence and overwintering grounds outside the Gulf(Sinclair & Currie, 1994; Comeau

et al

., 2002). Adultsin the other populations exhibit seasonal movementsat a smaller scale (ICES, 1994; Lawson & Rose,2000a).

C

OMMON

-

GARDEN

EXPERIMENT

Mature, prespawning cod (54–77 individuals per pop-ulation) were collected from each of the four popula-tions (Fig. 1) and allowed to spawn in the laboratory.Cod were collected from the 4X area on 7 January2002 and 16 January 2003, from the 3Ps area on 12April 2002, from the 4T area on 15 May 2003, andfrom the 3L area on 12 June 2003. A number of spawn-ing components are believed to occur in the 4X area

(ICES, 1994). The date of peak spawning differed bymore than 1 month between the two collections from4X, suggesting that they may have been from differentspawning components.

Broodstock from the 4X and 4T populations wereheld in the Aquatron facility at Dalhousie Universitywhereas broodstock from the 3Ps and 3L populationswere held at the Ocean Sciences Centre in St John’s,Newfoundland. Four batches of fertilized eggs werecollected from each of the populations. Each batch con-sisted of eggs from two consecutive days of spawning.All the eggs were incubated at the Ocean SciencesCentre in 250-litre flow-through tanks held between5

°

C and 8

°

C until hatching. When almost all the eggsin a batch had hatched, samples of larvae were trans-ferred to 30-litre aquaria for rearing.

Larvae were stocked in 30-litre aquaria at a densityof 40 larvae per litre. Samples from each batch werestocked in four aquaria, one for each of the followingtreatments: (1) high temperature–high food (HTHF);(2) high temperature–low food (HTLF); (3) lowtemperature–high food (LTHF); and (4) low tempera-ture–low food (LTLF). The high and low temperaturetreatments were 11

±

1

°

C and 7

±

1

°

C, respectively.Temperature was recorded twice daily. The high foodtreatment was 4500 prey per litre; the low food treat-ment was 1000 prey per litre in 2002 and 1500 prey perlitre in 2003. Larvae were fed three times daily:morning, afternoon, and evening. Larvae were fed Iso-chrisis-enriched rotifers from days 2–10 and Algomat-enriched rotifers from days 11–31. The larvae werethen weaned to a diet of

Artemia

. Larvae were fed amixture comprised of 50% rotifers and 50%

Artemia

from days 32–39 and then

Artemia

only from day 40until the end of the experiment. Larvae were rearedunder continuous light at an intensity of 2000 lux.Other physical variables were kept constant amongpopulations throughout the experiment.

M

ORPHOMETRIC

DATA

COLLECTION

Larvae were reared until they had reached an averagestandard length of 25 mm. A photograph of each indi-vidual was taken using a 1.2 megapixel digital cameraand the programme VCS (Visual CommunicationSuite) version 2.1.1. Homologous landmarks, corre-sponding to 12 anatomical structures, were digitizedalong the profile of the larvae (Fig. 2) using the TpsDigprogramme (Rohlf, 2003a). Larvae were not fed for atleast 12 h before sampling to ensure that variation instomach fullness would not influence the shapeanalysis.

A generalized least-squares Procrustes superimpo-sition (Rohlf & Slice, 1990) was performed on all thedigitized sets of coordinates using the TpsRelw pro-gramme (Rohlf, 2003b). This procedure aligned each

Figure 1.

Location of the Atlantic cod populations in thisstudy with their NAFO (Northwest Atlantic FisheriesOrganization) division names. Shading indicates bathyme-try: less than 100 m (lightest), 100–200 m, more than200 m (darkest). Circles indicate broodstock collectionsites.

52º

50º

48º

46º

44º

42º68º 66º 64º 62º 60º 58º 56º 54º 52º

4T

4X

3Ps

3L

354

J. MARCIL

ET AL

.

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London,

Biological Journal of the Linnean Society,

2006,

88

, 351–365

set of coordinates by translation, scaling, and rotation,so that corresponding homologous landmarks lay asclose as possible to each other. A consensus configura-tion was then computed on the set of scaled andaligned specimens. This configuration served as a ref-erence to compute the partial warps and uniform com-ponents that serve as shape variables in geometricmorphometric analyses. Uniform components repre-sent the shape change associated with uniformstretching or compression of all landmarks along their

x

and

y

axes (Rohlf & Bookstein, 2003). Partial warpsare parameters describing the non-uniform deforma-tions in shape between the consensus configurationand the specimens (Rohlf, 1990). The 12 pairs of

x

and

y

coordinates were transformed into 18 partial warpsand two uniform components for each specimen.

A key step when studying shape is to account for thesize differences among specimens. There are two typesof size effects: (1) isometric effects

−

a proportionalincrease in all morphometric measurements as sizeincreases and (2) allometric effects – a change inshape as size increases. Size and shape can be stronglyrelated because of allometric growth, differential ratesof growth for different body parts (Gould, 1966). In thepresent study, the aim was to test for shape differencesbetween populations and experimental treatmentsthat could not be attributed to differences in sizebetween individuals from the different populations ortreatments. Thus, there was the need to adjust forboth isometric and allometric effects of size.

Centroid size, the square root of the sum of thesquared distances between the landmarks and theircentroid, is used in the Procrustes superimposition toscale each specimen to a common unit size by dividingeach coordinate of a specimen by its centroid size(Bookstein, 1986), thus removing isometric effects ofsize. To account for allometric effects of size, each mor-phometric variable was adjusted to a common sizebased on its regression on centroid size. Within-groupslopes were estimated from an analysis of covarianceof the model:

D

ijklm

=

µ

+

P

i

+

B

ij

+

F

k

+

T

l

+

β

X

ijklm

+

e

ijlkm

where

P

i

is the effect of population

i (

4X

-

02, 4X-03, 4T,3Ps, or 3L),

B

ij

is the effect of batch

j

nested within

population

i

,

F

k

is the effect of food treatment

k

,

T

l

isthe effect of temperature treatment

l

,

D

ijklm

and

X

ijklm

are the partial warps or uniform components and log

10

centroid size of fish

m

in treatment

kl

of batch

j

nestedwithin population

i

,

β

is the covariate slope, and

e

ijklm

isa random normal deviation. Batch nested within pop-ulation was treated as a random effect and all otherterms were treated as fixed effects. Terms for interac-tions between the main effects (population, food treat-ment, and temperature treatment) were also includedin the model if significant (P < 0.05). Heterogeneity ofslopes between populations and between temperatureand food treatments (and their combinations) wastested based on type I sums of squares. Adjustmentwas based on the common within-group slope if therewas no significant (P < 0.05) heterogeneity of slopesbetween groups, and on separate within-group slopesotherwise. Analyses used the SAS procedures GLMand REG (SAS, 1999–2001).

The partial warps and uniform components of theProcrustes analysis were adjusted to a common sizeusing:

where aij is the adjusted partial warp or uniform com-ponent for specimen i in group j, dij is the original par-tial warp or uniform component, bj is the estimatedregression slope for group j (or the common within-group slope), is the grand mean log10 centroid size,and CSij is the specimen’s log10 centroid size.

STATISTICAL ANALYSIS OF SHAPE VARIATION

Morphometric variation was summarized by extract-ing principal components (PCs) from the covariancematrix of the shape variables (i.e. the adjusted partialwarps and uniform components). The componentscores of the first three principal components wereused as dependent variables in a nested analysis ofvariance (ANOVA) to test for significant shape differ-ences among populations and among treatmentgroups. Type III sums of squares were used because ofunequal sample size among groups. Batch nestedwithin population was treated as a random effectwhereas all other terms were treated as fixed effects.Because sample sizes were unequal, Satterthwaite’sapproximation (Sokal & Rohlf, 1981) was used to con-struct the denominator mean square (and to determineits degrees of freedom) for tests of population effects.

Very few 4X larvae survived the low temperaturetreatment; none survived beyond metamorphosis inthis treatment in both years. Rearing larvae underhigh light intensity may have been a possible cause ofthe low survival of the 4X larvae (Puvanendran &Brown, 1998). As a consequence, two sets of analyseswere performed. The first analysis comprised all four

a d b CS CSiij j j ij= + -( )

CS

Figure 2. Location of the 12 homologous landmarks usedto describe the shape of juvenile cod.

1

2 34

56

7

910

1112

8

Head Abdomen Mid body Posteriorbody

Caudal peduncle

PHENOTYPIC VARIATION IN BODY SHAPE AMONG POPULATIONS OF ATLANTIC COD 355

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

populations (the two 4X year-groups and the groupsfrom 4T, 3Ps, and 3L) compared at the two food levelsof the high temperature treatment. The second anal-ysis included the 4T, 3Ps, and 3L larvae compared atall levels of both the food and temperature treatments.

The assumption of normality of the component scoreswas tested within each treatment group for the firstthree principal components using the Shapiro–Wilkstatistic. Statistical significance was assessed follow-ing sequential Bonferroni adjustments (Rice, 1989). Inmost cases (64 out of 66 cases), the distributions did notdepart significantly from normality. Homogeneity ofvariances was tested using an F-max test for the firstthree principal components of each analysis section.Variance was homogeneous in all cases.

Because slopes were heterogeneous between groupsin many cases, differences between populations orexperimental treatments will vary depending on thesize to which morphometric variables are adjusted.Analyses were repeated adjusting the shape variablesto a larger or smaller centroid size (the largest andsmallest centroid sizes common to all groups in theanalysis). Results, both in terms of the differences inPC score between groups and their statistical signifi-cance, were very similar regardless of the centroid sizeto which shape variables were adjusted. Thus, over therange of sizes produced in our experiments, the differ-ences in shape between groups are not sensitive to thesize to which specimens are adjusted (Marcil, 2004).

In addition to the analysis reported here usinggeometric morphometrics, parallel analyses wereconducted using truss morphometrics (Strauss &Bookstein, 1982). Both approaches revealed very sim-ilar shape differences between populations or experi-mental treatments (Marcil, 2004), and only the resultsusing geometric morphometrics are reported here.

RESULTS

VARIATION IN ALLOMETRY AMONG POPULATIONS AND ENVIRONMENTAL TREATMENTS

Significant allometric differences were found betweenpopulations and treatment groups. In many instances,allometric size adjustment of the partial warps usedseparate within-group slopes between treatmentgroups, populations, or both. The highest percentageof significant allometric differences was foundbetween populations when examining shape variationamong all populations reared at high temperature(Table 1: Analysis 1). In the analysis of the three popu-lations (4T, 3Ps, and 3L) reared at two temperatureand two food levels, significant allometric differencesmost frequently occurred in the temperature (C × T),food (C × F), or population × temperature (C × P × T)model terms (Table 1, analysis 2).

VARIATION IN BODY SHAPE AMONG FOUR POPULATIONS (4X, 4T, 3PS, AND 3L) REARED AT HIGH TEMPERATURE

The first three principal components accounted for50% of the variation in shape (Table 2). They summa-rized shape differences related to both population andfood level, although the effect of population was themost significant.

The first PC, accounting for 25.3% of the variation inbody shape, summarized variation in body depth, cau-dal peduncle length, and head size (Fig. 3A). Fish hav-ing a high PC1 score were characterized by a deepbody, large head, short posterior body, and long caudalpeduncle (Fig. 3A). This component showed a strong

Table 1. Heterogeneity of slopes between populations (P),temperature treatments (T), food treatments (F), and theircombinations

Model term Analysis 1 Analysis 2

C × P 65% 25%C × T – 35%C × F 15% 30%C × P × T – 30%C × P × F 25% 15%C × T × F – 20%C × P × T × F – 10%

The percent of morphometric variables (partial warps oruniform components) with heterogeneous slopes in theregression on log centroid size (C) is shown for each modeleffect (e.g. C × P gives the percent of cases in which slopeswere heterogeneous between populations and C × P × Fthe percent of cases in which slopes differed between eachcombination of population and food level).

Table 2. Nested analysis of variance on size-adjusted par-tial warps among four populations reared at high temper-ature and two food levels

Population Bpopn FoodPopulation× Food

Principal component 1P-value < 0.0001 < 0.0001 0.0034 0.1049% SS 59.5% 7.3% 1.1% 1.0%

Principal component 2P-value 0.0034 0.3039 0.8820 0.1533% SS 35.6% 22.8% 0.0% 9.4%

Principal component 3P-value < 0.0001 0.0004 0.7766 < 0.0001% SS 49.1% 14.7% 0.0% 23.1%

P-values and the percentage of the total sums of squares(%SS) accounted for by each model term are shown. Bpopn,batch nested within population.

356 J. MARCIL ET AL.

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

Figure 3. Mean phenotypic values of the principal components of the size-adjusted partial warps of Atlantic cod from fourpopulations reared at high temperature and two food levels. Symbols show the mean principal component value with itsstandard error for each population. Thin-plate spline deformations corresponding to the extreme positive and negativephenotypic values are depicted on the right hand side of each graph. Deformations are exaggerated three-fold.

Low High

-0.01

0.00

0.01

Low High

-0.01

0.00

0.01

Low High

-0.01

0.00

0.01

A PC1

B PC2

C PC3

eulavC

Pnae

M

Food level

4X-034X-024T3Ps3L

PHENOTYPIC VARIATION IN BODY SHAPE AMONG POPULATIONS OF ATLANTIC COD 357

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

and highly significant population effect and a weakbut significant food effect (Table 2). Population differ-ences were greatest between the two 4X groups andthe three other populations (Figs 3A, 4). The two 4Xgroups tended to have a deeper body, larger head, andlonger caudal peduncle than the other populations(Figs 3A, 5A, B). Fish fed high food densities alsotended to have a slightly deeper body, larger head andlonger caudal peduncle than those fed a low food den-sity, although this effect was slight compared with thepopulation effect.

The second PC, accounting for 13.4% of the variationin body shape, summarized variation in bending alongthe lateral profile of the body (Fig. 3B); this bendingwas an artifact of how a fish was positioned when itspicture was taken. The geometric morphometrics anal-ysis, which accounts for the relative position of thelandmarks, was able to incorporate this small varia-tion in body position even though the bending effectwas reduced by careful positioning of the larvae andthe selection of only the best photographs. In additionto this artifact of fish positioning, variation in bodydepth contributed somewhat to this PC (Fig. 3B). Fishhaving a high PC2 score were characterized by anupward bend and shallow body (Fig. 3B). None of theeffects on PC2 were statistically significant, except fora relatively weak population effect. This reflected ashallower (and upward bent) body in 3Ps fish (Fig. 3B).

The third PC, accounting for 11.5% of the variationin shape, summarized shape variation in head, poste-rior body and caudal peduncle length, and in abdomensize (Fig. 3C). Fish having a high PC3 score were char-acterized by a short head, extended abdomen, and a

short caudal region (Fig. 3C). There was a significantpopulation effect along PC3 with a marked differencebetween the two 4X groups (Fig. 4). At both food levels,the 4X-02 group had higher mean PC3 scores thanthe 4X-03 group (Fig. 3C). There was also a strongpopulation × food interaction for PC3 (Table 2). PC3score increased with an increase in food density for the4T group, decreased with an increase in food densityfor 3Ps and 4X-02 cod, and showed little changebetween food densities for 3L and 4X-03 cod (Fig. 3C).

VARIATION IN BODY SHAPE AMONG THREE POPULATIONS (4T, 3PS, AND 3L) REARED AT HIGH

AND LOW TEMPERATURE

The first three principal components accounted for52% of the variation in the adjusted partial warpscores (Table 3). They summarized shape differencesdue to all terms in the model but primarily differencesrelated to temperature level.

The first and second PCs, accounting for 22.3% and16.6% of the variation in body shape, respectively,summarized variation in body depth, and posteriorbody and caudal peduncle length (Fig. 6A, B). Theyalso included shape variation corresponding to a bend-ing of the body. Fish having a high PC1 score werecharacterized by a shallow body, elongated posteriorbody and caudal peduncle regions, and a slightupward bending of the body (Fig. 6A). Fish having ahigh PC2 score were also characterized by a shallowbody and elongated posterior body and caudal pedun-cle regions, but with the bending of the body directeddownward (Fig. 6B).

Figure 4. Principal component scores from the analysis of size-adjusted partial warps of Atlantic cod from four populationsreared at high temperature and two food levels. Symbols show batch means within each treatment. Error bars indicatethe standard error.

PC 1

-0.020 -0.015 -0.010 -0.005 0.000 0.005 0.010 0.015 0.020

3C

P

-0.010

-0.005

0.000

0.005

0.010

0.015

4X-03 LF4X-03 HF4X-02 LF4X-02 HF4T LF4T HF3Ps LF3Ps HF3L LF3L HF

358 J. MARCIL ET AL.

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

Figure 5. Representative specimens of juvenile Atlantic cod showing the mean shape differences between (A) 4X-03 and(B) 3Ps populations and between 4T cod reared at (C) high temperature and (D) low temperature. Differences in pigmen-tation between specimens are not indicative of population or treatment effects on pigmentation. SL, standard length.

A

B

C

D

25.4 mm SL

25.5 mm SL

23.9 mm SL

23.8 mm SL

PHENOTYPIC VARIATION IN BODY SHAPE AMONG POPULATIONS OF ATLANTIC COD 359

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

There was a strong and highly significant effect oftemperature on the aspects of body shape summarizedby PC1 and PC2 (Table 3), with the difference betweentemperatures particularly strong along PC1. Fishreared at high temperature tended to have a deeperbody, a larger head, and shorter posterior body andcaudal peduncle regions (lower PC1 and PC2 scores)than those reared at low temperature (Figs 5C, D, 6A,B). The only exception was in the PC2 scores of 3Psfish reared at low food density. Significant populationeffects also occurred along PC1 and PC2, although thedifferences between populations were complicated byinteractions with temperature and food treatments(Table 3). Along PC1, 3Ps fish tended to have higherscores (shallow body, elongated caudal region) thandid 4T and 3L fish, except in the HTLF treatment(Fig. 6A). Along PC2, population differences variedbetween both temperature and food treatments.Unlike PC1, there was also a significant effect of fooddensity on PC2 (Table 3). Fish reared at high fooddensity tended to have deeper bodies and shorterposterior body regions, except in the case of 3Ps atlow temperature (Fig. 6B). Significant interactionsbetween population and temperature and populationand food on PC2 (Table 3) reflected the tendency forpopulation differences to be greater at low tempera-ture and low food density (Fig. 6B).

The third PC, accounting for 13.4% of the variationin body shape, summarized variation in body depthand mid-body, posterior body, and caudal pedunclelengths. Fish having a high PC3 score were character-ized by a deep body, a short mid-body region, and longposterior body and caudal peduncle regions (Fig. 6C).As in the first two PCs, there was a significant tem-perature effect on the aspect of shape described byPC3 (Table 3). In this case, however, there was anincrease in body depth with a decrease in temperature

(Fig. 6C). Unlike the patterns observed in the first twoPCs, an increased body depth along PC3 was also asso-ciated with a compression of the mid-body region andan elongation of the caudal peduncle region (Fig. 6C).There was also a slight but significant food effectreflecting the tendency for PC3 score to increase withfood density for 4T and 3L fish reared at hightemperature.

DISCUSSION

GENETIC DIVERGENCE AMONG POPULATIONS OF ATLANTIC COD

The present study provides evidence of genetic differ-ences in morphology and allometry between popu-lations of Atlantic cod. Significant morphometricdifferences were found between the 4T, 3Ps, and 3Lpopulations when reared in a common environment.However, the greatest morphological differences werefound between these populations and the two 4Xgroups. Fish from both 4X groups tended to have adeeper body shape than did those from the other pop-ulations. Just as morphological differences were great-est for the 4X population, ecological differences areexpected to be greatest between this group and theother three populations. Cod from the south-westScotian Shelf spawn during the winter, whereas theother three populations spawn in the spring and earlysummer. It can therefore be expected that environ-mental conditions experienced by the 4X cod duringthe larval and early juvenile stages are much differentfrom the environmental conditions experienced by theother cod populations examined in the present study.

Significant morphometric differences were alsofound between the two 4X groups when reared in acommon environment. This suggests the presence ofvery fine-scale genetic differentiation within the same

Table 3. Nested analysis of variance on size-adjusted partial warps among three populations (P) reared at two tempera-ture (T) and two food (F) levels

Population Bpopn Temperature Food P × T P × F T × F P × T × F

Principal component 1P-value 0.0011 0.0002 < 0.0001 0.3577 < 0.0001 0.0376 0.7051 < 0.0001% SS 9.5% 3.0% 53.1% 0.1% 3.0% 0.7% 0.0% 1.9%

Principal component 2P-value 0.0267 < 0.0001 < 0.0001 0.0099 0.0002 < 0.0001 0.0735 0.0822% SS 21.6% 19.4% 15.9% 3.1% 7.9% 9.0% 1.5% 2.4%

Principal component 3P-value 0.1377 < 0.0001 < 0.0001 0.0127 0.5258 0.6827 0.7062 0.0464% SS 19.2% 39.0% 19.6% 2.3% 0.5% 0.3% 0.1% 2.3%

P-values and the percentage of the total sums of squares (%SS) accounted for by each model term are shown. Bpopn, batchnested within population.

360 J. MARCIL ET AL.

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

Figure 6. Mean phenotypic values of the principal components (PC) of the size-adjusted partial warps of Atlantic codfrom three populations reared at two temperature and two food levels. Symbols show the mean PC value with its standarderror for each population. Symbol shading indicates temperature treatment. Thin-plate spline deformations correspondingto the extreme positive and negative phenotypic values are depicted on the right hand side of each graph. Deformationsare exaggerated three-fold.

Low High

-0.01

0.00

0.01

Low High

-0.01

0.00

0.01

Low High

-0.01

0.00

0.01

A PC1

B PC2

C PC3

eulavC

Pnae

M

Food level

4T LT4T HT3Ps LT3Ps HT3L LT3L HT

PHENOTYPIC VARIATION IN BODY SHAPE AMONG POPULATIONS OF ATLANTIC COD 361

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

stock. These fine-scale genetic differences are a possi-ble consequence of the different spawning componentssuspected to exist in this division (ICES, 1994).

The genetic differences observed among the fourpopulations of Atlantic cod in this study suggest thepresence of local adaptation in morphology amongthese populations, as well as between different spawn-ing components of the same population. For a commer-cially important species, such as the Atlantic cod,evidence of local adaptation among and within popu-lations has important implications for managementand conservation. A population adapted to its localenvironment will have reduced fitness in contrastingenvironments. In this case, a depleted or extirpatedpopulation might not be readily replaced by adjacentstocks, which would be poorly adapted to its environ-ment. The existence of genetic subcomponents withinpopulations, as suggested by the differences betweenthe 4X groups in the present study, also complicatesresource management. Management of stock com-plexes as a single unit may lead to the extinction ofsubunits, reducing overall productivity of the stockcomplex (Frank & Brickman, 2000).

Differences in body shape and allometry betweenpopulations of Atlantic cod reared in a common envi-ronment are interpreted here as evidence for geneticdivergence among the populations. However, these dif-ferences could also reflect maternal effects or theeffects of variation in the extent of selective mortality.As expected in marine fish larvae, mortality was quitehigh for all populations and treatments. If selectivemortality among phenotypes was the major cause ofthe shape differences among groups, the frequencydistribution of PC scores for high-mortality groupswould be expected to be a subset of the distributionsfor the low-mortality groups. This was clearly not thecase in the present study: when the phenotypic distri-bution of a high- and a low-mortality group were plot-ted against each other, the first group had phenotypesmore extreme than any observed among the lattergroup (Fig. 7). Furthermore, the extent of phenotypicdifferences between groups was unrelated to the dif-ference in mortality between them (Marcil, 2004).Maternal effects could also contribute to the morpho-logical differences between populations. However,maternal effects have a greater influence early indevelopment and are less likely to be substantial inpostmetamorphosis juveniles (Heath, Fox & Heath,1999). The differences documented in the presentstudy between populations reared in a common envi-ronment most likely reflect genetic differences ratherthan differences in selective mortality or maternaleffects.

Evidence of genetic divergence in adaptive quanti-tative and neutral molecular traits has been docu-mented between populations of marine fishes in a

number of cases. Conover & Present (1990) reporteddifferences in growth capacity of Atlantic silversides,Menidia menidia, from Nova Scotia, New York, andSouth Carolina when reared in a common environ-ment. Purchase & Brown (2000) also observed differ-ences in the growth rate of larvae and the foodconversion efficiency of juveniles between Atlantic codfrom the Grand Banks and Gulf of Maine when rearedunder identical conditions. Genetic studies using mic-rosatellite DNA have reported differences at selec-tively neutral loci among Atlantic cod stocks at thelarge scale examined in the present study (NovaScotia to Newfoundland) (Bentzen et al., 1996; Ruz-zante, Taggart & Cooke, 1999) but not at the finerscale examined in the present study on the south-westScotian Shelf. Taken together, the inferences of geneticdifferentiation in the present study coupled with pre-vious evidence of spatially variable restrictions ingene flow are consistent with the hypothesis thatAtlantic cod can differ genetically at comparativelylarge and small geographical scales.

ENVIRONMENTAL EFFECTS ON FISH MORPHOLOGY

Body shape in fishes is known to be affected by anumber of environmental factors, including rearingtemperature (Martin, 1949; Beacham, 1990; presentstudy), water velocity (Imre, McLaughlin & Noakes,2002), quantity of food (Currens et al., 1989; presentstudy), and type of food or feeding mode (Meyer, 1987,1990; Wainwright, Woodwark & Skibinski, 1991; Wim-berger, 1991, 1992; Day et al., 1994; Robinson & Wil-son, 1995; Day & McPhail, 1996). In the present study,rearing temperature had a strong effect on the bodyshape of juvenile Atlantic cod. Fish reared at high

Figure 7. Principal component 1 scores for two popula-tions with different survival rates. Survival rate of the 3Psgroup was approximately twice that of the 4X-03 group.

PC 1-0.03 -0.02 -0.01 0.00 0.01 0.02 0.03

ycneuqerF

0

5

10

15

20

25 4X-03 HT/HF3Ps HT/HF

362 J. MARCIL ET AL.

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

temperature were characterized by a deeper body, alarger head, and shorter caudal region than fishreared at low temperature. This temperature effect onbody shape was similar for all three populationstested, although the effect was slightly stronger forPlacentia Bay (3Ps) cod. Similar effects of rearing tem-perature on body shape have been reported in studieson juvenile chum salmon, Onchorhynchus keta (Bea-cham, 1990), and on juvenile carp, Cyprinus carpio(Loy, Cataudella & Corti, 1996). In all three cases,juveniles reared at warmer temperatures tended tohave deeper bodies, suggesting that this may be a gen-eral tendency in fishes.

Food ration also had a significant impact on codbody shape, although its effect was weak comparedwith the effects of population and temperature. Fishreared at high food density tended to be deeper bodiedwith larger heads and caudal peduncles. To our knowl-edge, no other study using a common-garden experi-mental protocol has investigated the effect of foodration on fish morphology, except for the extreme com-parison of starved vs. fed fish (Currens et al., 1989).

Most studies demonstrating environmental effectson fish body shape have used freshwater or anadro-mous fishes. Studies on marine fishes are rare, pre-sumably due to the difficulty of rearing these fish.However, the few studies available (Loy et al., 1996;Sarà, Favaloro & Mazzloa, 1999; present study) indi-cate that rearing environment may also have animportant effect on morphology in these fishes.

EVIDENCE OF GENETIC DIFFERENTIATION IN PHENOTYPIC PLASTICITY

Phenotypic responses to a change in environmentalconditions were different among populations of Atlan-tic cod. Phenotypic responses to temperature weresimilar between populations in terms of direction butdiffered in terms of magnitude. There was a generaltendency for fish reared at high temperature to bedeeper bodied with a large head and short caudalregion, but this response was slightly more pro-nounced for the 3Ps fish than for the 4T and 3L pop-ulations. Greater differences in plasticity were evidentbetween populations in their responses to a change infood ration. Fish from the 4T population showedgreater plasticity to a change in food density thendid the other three populations. These instances ofpopulation × temperature and population × food inter-action represent cases of genotype × environmentinteraction. They suggest that there are genetic differ-ences in plasticity among these populations of Atlanticcod.

Phenotypic plasticity is often assumed to be adap-tive. If different phenotypes are optimal in differentenvironments, a single genotype can produce the dif-

ferent optimal phenotypes through phenotypic plastic-ity (Hutchings, 2004). Phenotypic responses to foodtype and feeding mode often appear to be adaptive. Forexample, limnetic morphs feeding in open water tendto have a more streamlined body shape, whereasbenthic morphs feeding in the littoral zone tend tohave a deeper body shape (McPhail, 1992; Toline &Baker, 1993, 1997). In some fishes, limnetic forms fedbenthic food develop a deeper body shape whereasbenthic forms fed limnetic food develop a more stream-lined body shape (Wimberger, 1991; Day et al., 1994;Robinson & Wilson, 1996). In the examples mentionedabove, phenotypic plasticity was shown to be adaptivewith the morphological change induced by a particularenvironment improving performance in that environ-ment. However, the adaptive significance of the effectsof temperature and food ration reported in the presentstudy is unknown.

COGRADIENT VS. COUNTERGRADIENT VARIATION IN MORPHOLOGY OF ATLANTIC COD

Phenotypic variation across environmental gradientscan be the result of both genotypic and environmentalinfluences on the phenotype. Genotypic variation maybe random across environmental gradients, or it maycovary with environmental variation, either reinforc-ing environmental effects and increasing phenotypicvariation across the gradient (cogradient variation) oropposing environmental effects and decreasing pheno-typic variation across the gradient (countergradientvariation; Conover & Schultz, 1995). Cogradient vari-ation is expected to occur when different phenotypesare favoured in the different environments. On theother hand, countergradient variation is expectedwhen the same phenotype is favoured in the differentenvironments.

The strongest genetic difference observed in thepresent study was between 4X cod and cod from thethree other populations. The 4X cod were character-ized by a deeper body than cod from the other popu-lations. In the wild, a strong environmental gradientis also expected between 4X cod and the other popu-lations. Cod in the 4X population spawn in winter,whereas those in the 4T, 3Ps, and 3L populationsspawn in spring and early summer. Consequently, atthe larval and early juvenile stages when they feed onzooplankton in the upper water layers, 4X cod areexpected to experience colder temperatures than docod from the other populations (Fig. 8). In the presentstudy, fish reared at colder temperatures tended tohave a shallower body shape. Thus, the genetic differ-ences between the 4X fish and the other populationsappear to counteract the environmental differencesexpected in the wild. This suggests the presence ofcountergradient variation in body shape across the

PHENOTYPIC VARIATION IN BODY SHAPE AMONG POPULATIONS OF ATLANTIC COD 363

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

temperature gradient between these populations inthe wild. Countergradient variation in body shape hasbeen demonstrated between early and late-spawningstocks of chum salmon, O. keta (Tallman, 1986; Tall-man & Healey, 1991).

Most previous studies of genetic diversity in mor-phology between groups of fish comprise cases wheredifferent body shapes are favoured between groups,reflecting adaptation to ecological differences betweenthe groups (e.g. limnetic vs. benthic groups, McPhail,1984; coastal vs. inland populations, Taylor & McPhail,1985a, b; stream- vs. lake-dwelling groups, Swain &Holtby, 1989). If genetic and environmental effects onmorphology are countergradient between 4X cod andthe other populations in the present study, this wouldbe a case where the difference in morphology observedbetween groups in a common laboratory environmentdoes not reflect a difference in the optimal body shapebetween groups in the wild. Instead, the genetic dif-ferences revealed in common-garden experimentswould be an adaptation to counteract non-adaptiveenvironmental effects on morphology to produce sim-ilar body shapes in the different populations in thewild. This would be consistent with our expectationthat similar body shapes should be favoured in thesepopulations given their similarity in lifestyle. Furtherwork needs to be carried out to test the hypothesis thatdifferences in body shape between these populations inthe wild are minimized by countergradient variation.One consequence would be that adaptive genetic dif-ferences between populations would be overlooked inthe wild because of the phenotypic similarity producedby countergradient variation.

ACKNOWLEDGEMENTS

We would like to thank the Ocean Sciences Centre andits staff for the use of their facility and in helping rear-ing the cod, with a special thanks to Velmurugu Puv-anendran who oversaw broodstock collection and eggand larval rearing. We also thank Dalhousie Univer-sity for the use of the Aquatron facility, with a specialthanks to Jim Eddington for his help with broodstockand egg collection. Many thanks also to all those otherpeople that provided comments and suggestions dur-ing the length of the study and to two anonymousreviewers for their comments on the manuscript.Financial support from the Natural Sciences andEngineering Research Council of Canada, in a Strate-gic Research Grant to J.A.H., is also gratefullyacknowledged. This paper is dedicated to the memoryof Dr Joe Brown, our friend and colleague at the OceanSciences Centre.

REFERENCES

Alexander RM. 1974. Functional design in fishes. London:Hutchinson & Co. Publishers. Ltd.

Allendorf FW. 1983. Isolation, gene flow, and genetic differ-entiation among populations. In: Schonewald-Cox CM,Chambers SM, MacBryde GF, Thomas WL, eds. Geneticsand conservation: a reference for managing wild animal andplant populations. Reading, MA: Benjamin/Cummings, 51–65.

Allendorf FW. 1995. Genetics: defining the units of conserva-tion. American Fisheries Society Symposium 17: 147–148.

Beacham TD. 1990. A genetic analysis of meristic and mor-phometric variation in chum salmon (Oncorhynchus keta) atthree different temperatures. Canadian Journal of Zoology68: 225–229.

Bentzen P, Taggart CT, Ruzzante DE, Cook D. 1996.Microsatellite polymorphism and the population structureof Atlantic cod (Gadus morhua) in the northwest Atlantic.Canadian Journal of Fisheries and Aquatic Sciences 53:2706–2721.

Billerbeck JM, Ortí G, Conover DO. 1997. Latitudinal vari-ation in vertebral number has a genetic basis in the Atlanticsilverside, Menidia menidia. Canadian Journal of Fisheriesand Aquatic Sciences 54: 1796–1801.

Bookstein FL. 1986. Size and shape spaces for landmark datain two dimensions. Statistical Science 1: 181–242.

Brönmark C, Miner JG. 1992. Predator-induced phenotypi-cal change in body morphology in Crucian carp. Science 258:1348–1350.

Carvalho GR. 1993. Evolutionary aspects of fish distribution:genetic variability and adaptation. Journal of Fish Biology43 (Suppl. A): 53–73.

Comeau LA, Campana SE, Chouinard GA. 2002. Timing ofAtlantic cod (Gadus morhua L.) seasonal migrations in thesouthern Gulf of St Lawrence: interannual variability andproximate control. ICES Journal of Marine Science 59: 333–351.

Figure 8. Mean temperature in the upper water layer (0–30 m depth) following spawning in the areas occupied byeach cod population in this study. Spawning is assumed tooccur in February for the 4X population and May for theother populations. For details, see Marcil (2004).

Month after spawning

0 1 2 3

)C°(

erutarepmeT

0

2

4

6

8

10

12

14 4X4T3Ps3L

364 J. MARCIL ET AL.

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

Conover DO. 1998. Local adaptation in marine fishes: evi-dence and implication for stock enhancement. Bulletin ofMarine Science 62: 477–493.

Conover DO, Munch SB. 2002. Sustaining fisheries yieldsover evolutionary time scales. Science 297: 94–96.

Conover DO, Present TMC. 1990. Countergradient varia-tion in growth rate: compensation for length of the growingseason among Atlantic silversides from different latitudes.Oecologia 83: 316–324.

Conover DO, Schultz ET. 1995. Phenotypic similarity andthe evolutionary significance of countergradient variation.Trends in Ecology and Evolution 10: 248–252.

Currens KP, Sharpe CS, Hjort R, Schreck CB, Li HW.1989. Effects of different feeding regimes on the morphomet-rics of chinook salmon (Oncorhynchus tshawytscha) andrainbow trout (O. mykiss). Copeia 3: 689–695.

Day T, McPhail JD. 1996. The effect of behavioural and mor-phological plasticity on foraging efficiency in the threespinestickleback (Gasterosteus sp.). Oecologia 108: 380–388.

Day T, Pritchard J, Schluter D. 1994. A comparison of twosticklebacks. Evolution 48: 1723–1734.

Frank KT, Brickman D. 2000. Allee effects and compensa-tory population dynamics within a stock complex. CanadianJournal of Fisheries and Aquatic Sciences 57: 513–517.

Gould SJ. 1966. Allometry and size in ontogeny and phylog-eny. Biology Reviews 41: 587–640.

Gyllensten U. 1985. The genetic structure of fish: differencesin the intraspecific distribution of biochemical genetic vari-ation between marine, anadromous, and freshwater species.Journal of Fish Biology 26: 691–699.

Hard JJ. 1995. A quantitative genetic perspective on the con-servation of intraspecific diversity. American Fisheries Soci-ety Symposium 17: 304–326.

Heath DD, Fox CW, Heath JW. 1999. Maternal effects on off-spring size: variation through early development of chinooksalmon. Evolution 53: 1605–1611.

Hendry AP, Stearns SC, eds. 2004. Evolution illuminated:salmon and their relatives. New York, NY: Oxford UniversityPress.

Hutchings JA. 2004. Norms of reaction and phenotypic plas-ticity in salmonid life histories. In: Hendry AP, Stearns SC,eds. Evolution illuminated; salmon and their relatives. NewYork, NY: Oxford University Press, 154–174.

Hutchings JA, Myers RA. 1994. Timing of cod reproduction:interannual variability and the influence of temperature.Marine Ecology Progress Series 108: 21–31.

Hutchings JA, Reynolds JD. 2004. Marine fish populationcollapses: consequences for recovery and extinction risk.Bioscience 54: 297–309.

ICES. 1994. Spawning and life history information fornorth Atlantic cod stocks. International Council for theExploration of the Sea, Cooperative Research Report 205:1–150.

Imre I, McLaughlin RL, Noakes DLG. 2002. Phenotypicplasticity in brook charr: changes in caudal fin induced bywater flow. Journal of Fish Biology 61: 1171–1181.

Karhu A, Hurme P, Karjalainen M, Karvonen P,Kärkkäinen K, Neale D, Savolainen O. 1996. Do molec-

ular markers reflect patterns of differentiation in adaptivetraits of conifers? Theoretical and Applied Genetics 93: 215–221.

Koskinen MT, Haugen TO, Primmer CR. 2002. Contempo-rary fisherian life-history evolution in small salmonid popu-lations. Nature 419: 826–830.

Lawson GL, Rose GA. 2000a. Seasonal distribution andmovements of coastal cod (Gadus morhua L.) in PlacentiaBay, Newfoundland. Fisheries Research 49: 61–75.

Lawson GL, Rose GA. 2000b. Small-scale spatial and tem-poral patterns in spawning of Atlantic cod (Gadus morhua)in coastal Newfoundland waters. Canadian Journal of Fish-eries and Aquatic Sciences 57: 1011–1024.

Lett PF. 1980. A comparative study of the recruitment mech-anisms of cod and mackerel, their interaction, and its impli-cation for dual stock assessment. Canadian Technical Reportof Fisheries and Aquatic Sciences No. 988. Ottawa: Supplyand Services Canada.

Loy A, Cataudella S, Corti M. 1996. Shape changes duringthe growth of the sea bass, Dicentrarchus labrax (Teleostea:Perciformes), in relation to different rearing conditions. In:Marcus LF, Corti M, Loy A, Naylor GJP, Slice D, eds.Advances in morphometrics. New York, NY: Plenum Press,399–405.

Marcil J. 2004. Genetic and environmental components of phe-notypic variation in body shape among populations of Atlan-tic cod (Gadus morhua). MSc Thesis, Dalhousie University.

Martin WR. 1949. The mechanics of environmental control ofbody form in fishes. Publications of the Ontario FisheriesResearch Laboratory, No. 70. University of Toronto StudiesBiological Series, No. 58. Toronto: University of TorontoPress, 1–91.

McPhail JD. 1984. Ecology and evolution of sympatric stick-lebacks (Gasterosteus): morphological and genetic evidencefor a species pair in Enos Lake, British Columbia. CanadianJournal of Zoology 62: 1402–1408.

McPhail JD. 1992. Ecology and evolution of sympatric stick-lebacks (Gasterosteus): evidence for a species-pair in PaxtonLake, Texada Island, British Columbia. Canadian Journal ofZoology 70: 361–369.

Meyer A. 1987. Phenotypic plasticity and heterochrony inCichlasoma managuense (Pisces, Cichlidae) and their impli-cations for speciation in cichlid fishes. Evolution 41: 1357–1369.

Meyer A. 1990. Ecological and evolutionary consequences ofthe tropic polymorphism in Cichlasoma citrinellum (Pisces:Cichlidae). Biological Journal of the Linnean Society 39:279–299.

Myers RA, Mertz G, Bishop CA. 1993. Cod spawning in rela-tion to physical and biological cycles of the northern North-west Atlantic. Fisheries Oceanography 2: 154–165.

Purchase CF, Brown JA. 2000. Interpopulation difference ingrowth rates and food conversion efficiencies of young GrandBanks and Gulf of Maine Atlantic cod (Gadus morhua).Canadian Journal of Fisheries and Aquatic Sciences 57:2223–2229.

Puvanendran V, Brown JA. 1998. Effect of light intensity onthe foraging and growth of Atlantic cod larvae: interpopula-

PHENOTYPIC VARIATION IN BODY SHAPE AMONG POPULATIONS OF ATLANTIC COD 365

© 2006 Her Majesty the Queen in Right of Canada.Journal compilation © 2006 The Linnean Society of London, Biological Journal of the Linnean Society, 2006, 88, 351–365

tion difference? Marine Ecology Progress Series 167: 207–214.

Reed DH, Frankham R. 2001. How closely correlated aremolecular and quantitative measures of genetic variation? Ameta-analysis. Evolution 55: 1095–1103.

Reznick DN, Shaw FH, Rodd HF, Shaw RG. 1997. Evalu-ation of the rate of evolution in natural populations of gup-pies (Poecilia reticulata). Science 275: 1934–1937.

Rice WR. 1989. Analyzing tables of statistical tests. Evolution43: 223–225.

Robinson BW, Parsons KJ. 2002. Changing times, spaces,and faces: tests and implications of adaptive morphologicalplasticity in the fishes of northern postglacial lakes. Cana-dian Journal of Fisheries and Aquatic Sciences 59: 1819–1833.

Robinson BW, Wilson DS. 1995. Experimentally inducedmorphological diversity in Trinidadian guppies (Poeciliareticulata). Copeia 1995: 294–305.

Robinson BW, Wilson DS. 1996. Genetic variation and phe-notypic plasticity in a trophically polymorphic population ofpumpkinseed sunfish (Lepomis gibbosus). Evolutionary Ecol-ogy 10: 631–652.

Rohlf JF. 1990. Morphometrics. Annual Review in Ecologyand Systematics 21: 299–316.

Rohlf JF. 2003a. Tpsdig, Version 1.39. Stony Brook, NY:Department of Ecology and Evolution, State University ofNew York.

Rohlf JF. 2003b. Tpsrelw: analysis of relative warps, Version1.35. Stony Brook, NY: Department of Ecology and Evolu-tion, State University of New York.

Rohlf JF, Bookstein FL. 2003. Computing the uniform com-ponent of shape variation. Systematic Biology 52: 66–69.

Rohlf JF, Slice D. 1990. Extensions of the Procrustes methodfor the optimal superimposition of landmarks. SystematicZoology 39: 40–59.

Ruzzante DE, Taggart CT, Cook D. 1999. A review of theevidence for genetic structure of cod (Gadus morhua) popu-lations in the NW Atlantic and population affinities of larvalcod of Newfoundland and the Gulf of St. Lawrence. FisheriesResearch 43: 79–97.

Sarà M, Favaloro E, Mazzola A. 1999. Comparative morpho-metrics of sharpsnout seabream (Diplodus puntazzo Cetti,1777), reared in different conditions. Aquacultural Engineer-ing 19: 195–209.

SAS. 1999–2001. SAS user’s guide: statistics, Version 8.02.Cary, NC: SAS Institute.

Schultz ET, Reynolds KE, Conover DO. 1996. Counter-gradient variation in growth among newly hatched Fundu-lun heteroclitus: geographic differences revealed bycommon-environment experiments. Functional Ecology 10:366–374.

Sinclair A, Currie L. 1994. Timing of cod migrations into andout of the Gulf of St. Lawrence based on commercial fisher-ies, 1986–93. DFO Atlantic Fisheries Research Document 94/47: 18p.

Snorrason SS, Skùlason S, Jonsson B, Malmquist HJ,Jonasson PM, Sandlund OT, Lindem T. 1994. Trophicspecialization in Artic charr Salvelinus alpinus (Pisces;

Salmonidae): morphological divergence and ontogeneticniche shifts. Biological Journal of the Linnean Society 52: 1–18.

Sokal RR, Rohlf JF. 1981. Biometry. San Francisco, CA: W.H.Freeman.

Strauss RE, Bookstein FL. 1982. The truss: body formreconstructions in morphometrics. Systematic Zoology 31:113–135.

Swain DP, Foote CJ. 1999. Stocks and chameleons: the useof phenotypic variation in stock identification. FisheriesResearch 43: 113–128.

Swain DP, Holtby LB. 1989. Differences in morphology andbehavior between juvenile coho salmon (Oncorhynchuskisutch) rearing in a lake or in its tributary stream. Cana-dian Journal of Fisheries and Aquatic Sciences 46: 1406–1414.

Tallman RF. 1986. Genetic differentiation among seasonallydistinct spawning populations of chum salmon, Oncorhyn-chus keta. Aquaculture 57: 211–217.

Tallman RF, Healey MC. 1991. Phenotypic differentiation inseasonal ecotypes of chum salmon, Oncorhynchus keta.Canadian Journal of Fisheries and Aquatic Sciences 48: 661–671.

Taylor EB, McPhail JD. 1985a. Variation in body morphol-ogy among British Columbia populations of coho salmon,Oncorhynchus kisutch. Canadian Journal of Fisheries andAquatic Sciences 42: 2020–2028.

Taylor EB, McPhail JD. 1985b. Variation in burst and pro-longed swimming performance among British Columbiapopulations of coho salmon, Oncorhynchus kisutch. Cana-dian Journal of Fisheries and Aquatic Sciences 42: 2029–2033.

Toline AC, Baker AJ. 1993. Foraging tactic as a potentialselection pressure influencing geographic differences in bodyshape among populations of dace (Phoxinus eos). CanadianJournal of Zoology 71: 2178–2184.

Toline AC, Baker AJ. 1997. Evidence of a heritable compo-nent to body shape in the northern redbelly dace (Phoxinuseos). Canadian Journal of Zoology 75: 1247–1253.

Via S, Gomulkiewicz R, De Jong G, Scheiner SM,Schlichting CD, Van Tienderen PH. 1995. Adaptive phe-notypic plasticity: consensus and controversy. Trends inEcology and Evolution 10: 212–217.

Wainwright PC, Osenberg CW, Mittelbach GG. 1991.Trophic polymorphism in the pumpkinseed sunfish (Lepomisgibbosus Linnaeus): effects of environment on ontogeny.Functional Ecology 5: 40–55.

Ward RD, Woodwark M, Skibinski DOF. 1994. A com-parison of genetic diversity levels in marine, freshwater,and anadromous fishes. Journal of Fish Biology 44: 213–232.

Wimberger PH. 1991. Plasticity of jaw and skull morphologyin the neotropical cichlids Geophagus brasiliensis and G.steindachneri. Evolution 45: 1545–1563.

Wimberger PH. 1992. Plasticity of fish body shape. Theeffects of diet, development, family and age in two species ofGeophagus (Pisces: Cichlidae). Biological Journal of the Lin-nean Society 45: 197–218.