Rapid morphological divergence of cultured cod of the northwest Atlantic from their source population

The performance of aquaculture escapees in the wild depends in part on how their morphology differs from that of wild fish. We compared farmed Atlantic cod Gadus morhua morphology to that of wild cod from the same ancestral population. Traditional and geometric morphometrics showed that farmed cod had relatively smaller fins, heads, eyes, and jaws than wild cod for a given size. Conversely, drumming muscle size and metrics of body and liver condition were greater in farmed fish. As the observed differences are likely due to phenotypic plasticity, their fitness consequences for escaped farmed fish may be transient.


INTRODUCTION
Fish exposed to culture develop phenotypes that differ from those of their wild counterparts (Fleming & Gross 1994, Araki et al. 2008, Bailey et al. 2010, Chittenden et al. 2010).The phenotypes they develop may be beneficial under culture but may reduce the fitness of an individual when exposed to another environment (e.g. the wild environment following escape).These cultured phenotypes can be the product of a plastic response whereby different phenotypes can be expressed by a single genotype in response to different environmental conditions (Imre et al. 2002, Skjae raasen et al. 2008, Mayer et al. 2011, Vehanen & Huusko 2011), or these phenotypes may be the result of genetic changes brought about through both intentional and unintentional selection (Fleming et al. 1994, Einum & Fleming 2001, Fleming & Petersson 2001, Hutchings & Fraser 2008, Solberg et al. 2013).The degree of phenotypic change, and its permanence, are both a function of the time an individual has spent in captive conditions (Pakkasmaa et al. 1998, von Cramon-Taubadel et al. 2005), as well as the degree of genetic change from the ancestral lineage due to captivity (Fleming et al. 1994, Blanchet et al. 2008, reviewed by Hutchings & Fraser 2008, Fraser et al. 2010).Thus, if it is presumed that the phenotypes of wild fish are the product of adaptation to their local environment, then the degree to which the phenotype of cultured fish diverges from the wild type is likely a reflection of how maladaptive the cultured phenotype may be if exposed to the wild environment.Furthermore, the 'permanence' of the cultured fish's phenotype, or the degree to which phenotypic plasticity allows it to (re)converge on a wild-type phenotype over time at liberty, may result in a lifetime fitness difference between the 2 groups that is lower than would be predicted based on morphological differences at the time of escape.
Through programmes that sought to diversify the Canadian aquaculture industry, experimental At lantic cod Gadus morhua broodstocks were created from wild-caught fish, and their offspring were stocked to commercial cage aquaculture farms.These first-generation farmed cod afforded us the unique opportunity to study the morphological ef fects of exposure to the aquaculture environment on fish that had not experienced the intensive selection regimes common in more established species (e.g.Atlantic salmon Salmo salar).We compared the morphology of wild cod to farmed individuals created from wild-caught parents that were genetically similar to our wild fish.We then discuss the differences in morphology in terms of potential fitness effects on escapees in the wild.

Data collection
Farmed cod were the progeny of wild-caught fish from Bay Bulls, Newfoundland, Canada (47°18' N, 52°48' W; Northwest Atlantic Fisheries Organization [NAFO] division 3L; Fig. 1), which were spawned between December 2006 and March 2007.The farmed cod were reared in tanks at Memorial University from fertilization until they were transferred en masse to Sapphire Sea Farms' net-pen facility in Bay Bulls on 30 November 2008.Some of them (n = 112) were sampled between 4 and 9 November 2009 during the annual harvest.
Wild cod were captured using baited cod pots on 10 and 20 November 2009 (n = 38 and n = 19, respectively) in Smith Sound, Newfoundland (48°9' N, 53°44' W; NAFO di vi sion 3L; Fig. 1).Cod of Smith Sound and Bay Bulls are thought to be of the same stock, being genetically similar (Beacham et al. 2002, Bradbury et al. 2010, Rose et al. 2011).The wild fish were held in a tank and measured 2−3 wk after collection.The farmed and wild cod were held without feeding prior to measuring to ensure that gut contents did not bias weight or shape measure, and only fish free of obvious skeletal defect were included in the analysis.
After being killed, fish were kept on ice before being arranged left side up, with their median and caudal fins extended and pinned in place, and photographed with a digital camera (Nikon D300) mounted on a tripod.A ruler was included in each photograph to allow for size calibration.
After photographing, the right and left pelvic fin lengths (distance from the origin of the fin to tip of the longest fin ray) were measured (± 0.01 cm) with digital callipers because they could not be measured from the photographs.Fish were weighed whole (± 0.01 g), sexed when the internal organs were removed, and the liver was weighed separately (± 0.01 g).Following the protocol of Rowe & Hutchings (2004), both the right and left drumming muscles were removed and frozen, before being dried to constant mass and weighed together (± 0.001 g).
Eighteen landmarks were recorded as x−y coordinates from the photographs using ImageJ (Schneider et al. 2012; http://rsb.info.nih.gov/ij/download.html;Fig. 2).Standard lengths were measured as the distance between the anterior-most point of the premaxilla and the posterior-most edge of the hypural plate (points 1 and 8 respectively on Fig. 2).The dorsal and anal fin lengths and widths were measured as the distance from the fin origin to the tip of the second fin ray, which was the longest, and as the distance along the fin base from its origin to its distal insertion, respectively (Fig. 2).Unforeseen variation in fin attitude and extension prevented measurement of the size of the left pectoral fin from the digital photographs.A small number of farmed fish (10 of 112) showed malformed fins, and measurements of these fins were excluded from the analysis.

Size standardization and calculation of condition indices
Size standardization was employed so that only relative differences in trait size between the 2 origins (i.e.wild or farmed) were considered.The lengths and widths of the dorsal and anal fins, the lengths of the pelvic fins, and the weight of the drumming muscles were log 10 -transformed and then standardized using the method of Reist (1986).Each of these traits was standardized for each fish using the formula M std = M obs (Sz mean /Sz obs ) b , where: M is the trait measure, Sz is the size measure to which samples are standardized, superscript b is the trait-specific common within-groups slope, and the subscripts mean, obs, and std refer to the mean, observed (raw), and the size-standardized measurements, respectively.The weight of the drumming muscles was standardized to a common body weight, while the length and width measurements were standardized to a common centroid size.The centroid size, the square root of the sum of the squared distances of each peripheral landmark (i.e.excluding points 13, 14, 17, and 18 in Fig. 2) to the centroid, was calculated in R (R Devel-opment Core Team 2011) using the function gpagen (geomorph package; Adams & Otárola-Castillo 2013).
Condition indices (CI) were calculated for each fish by taking the standardized residuals of the regression of log 10 -transformed standard length on the log 10 -transformed total weight.The liver indices (LI) were calculated similarly from the regression of the log 10 -transformed weight of the liver on the log 10transformed total weight.The standardized residuals convey the condition status of each fish.Positive residuals indicate that the fish is heavier, or possesses a heavier liver for their size than the average, while negative residuals indicate the opposite.

Traditional morphometric, geometric morphometric, and statistical analyses
All statistical and geometric morphometric analyses were conducted in R (R Development Core Team 2011).The traditional morphometric analyses consisted of testing for differences in size-standardized drumming muscle mass, dorsal and anal fin lengths and widths, pelvic fin lengths, as well as CI, and LI individually between fish origins (i.e.wild or farmed) using a linear model with permutation (lmp function, lmPerm package; Wheeler 2010) and type III sumsof-squares (Anova function, car package; Fox & Weisberg 2011) with sex and origin as fixed effects.Using permutation removes the necessity that the data satisfy the assumptions of traditional parametric tests, and allows for the calculation of exact significance levels.The issue of multiple hypothesis testing was addressed by the use of adjusted p-values, with the false discovery rate set to α = 0.05 (Benjamini & Hochberg 1995).
Principal component analysis (PCA), with varimax rotation (prcomp function, stats package; R Development Core Team 2011), was also conducted as part of the traditional morphometric analysis to reduce the number of parameters, using all morphometric measures listed in Table 1, with the exception of standard length, total weight, and drumming muscle mass.Standard length and total weight were excluded because they represent differences in fish size rather than shape (size-standardized).Drumming muscle mass was also excluded because it had missing values, which caused the sample size to drop appreciably.All principal components (PCs) with eigenvalues greater than the mean eigenvalue were considered significant (Jackson 1993).
Geometric morphometric analyses were conducted using the R packages shapes (Dryden 2013) and geo- morph (Adams & Otárola-Castillo 2013).The x−y coordinates collected from the photographs of the fish were first converted to shape coordinates using generalized Procrustes analysis (GPA; Adams et al. 2004).
GPA removes the non-shape aspects of size, (scaling), orientation, and location from the raw x−y coordinates, and also standardizes each individual to a common unit centroid size (Rohlf 1999, Adams et al. 2004).The amount of shape variation attributable to the different origins of the fish (controlling for sex) was quantified using Procrustes ANOVA with permutation, which compares the observed sum-of-squared Procrustes distances to an expected distribution which is calculated through permutation (Goodall 1991).PCA was also conducted on the configuration of the specimens into principal warp space to detect the major features of the shape variation.Differences in PC scores between origins were tested using linear models with sex and origin as fixed effects.

Traditional morphometrics
No interactions were detected between sex and origin.Within origin, the size-adjusted dried mass of the drumming muscles was greater in males than in females (Table 1).However, females were bigger and their LIs were greater than those of the males (Table 1).All size-adjusted morphometric measures, with the exception of the width of the first dorsal fin, differed significantly between wild and farmed cod (Table 1).
The first 4 PCs all had eigenvalues greater than the mean eigenvalue, and cumulatively explained 74.3% of the variation in traditional morphometric variables (Table 2).The loadings of wild and farmed fish on PCs 1 and 2 differed significantly (t-test, p < 0.001), while there was no significant difference on PCs 3 and 4 (t-test, both p > 0.05; Fig. 3; PC4 not shown).
The first PC, which explains 44.3% of the variation, was characterized by negative loading of the fin measures, particularly fin lengths (Table 2).PC2 explained 12.6% of the variation, and for the most part is described by positive loadings from CI, LI, and fin widths.Interestingly, on PC2, the fin widths showed moderate to strong positive loadings, while their lengths showed near-zero to moderately negative loadings (Table 2).

Geometric morphometrics
ANOVA with permutation on the Procrustesaligned coordinates of the wild and farmed cod revealed that there was a significant interaction between sex and origin (F 1,140 = 6.112, p < 0.001).Within-origin analysis showed that the shape of the wild males differed from that of the wild females, and the same was true for farmed males and females (both p < 0.05).Testing within sexes, the shape of both farmed females and males was different from that of their wild counterparts (both p < 0.001).
PCA of the configuration of the wild and farmed specimens into the principal warp space revealed 7 PCs with eigenvalues greater than the mean eigenvalue, and cumulatively ex plained 81.90% of the variance.Like the ANOVA above, the scores on PC1 and PC2 showed a significant interaction be tween sex and origin (both p < 0.05; Fig. 4).That said, Fig. 4 shows a clear separation between wild and farmed fish along PC2.PC1 ex plained 30.17% of the variance, and PC2 18.52%.PC1 was, however, significantly correlated with centroid size (Spearman's rho: −0.259, p < 0.01), indicating that the shape differences de scribed by the first PC were mainly related to size.There were no significant differences in shape between origins, sexes, or any interaction between the two for PCs 3−7 (all p > 0.05).
Fig. 5 depicts the difference in shape between farmed females relative to farmed males (Fig. 5a), wild females relative to wild males (Fig. 5b), farmed females relative to wild females (Fig. 5c), and farmed males relative to wild males (Fig. 5d), and is illustrative of the significant sex × origin interaction.Despite detecting significant statistical difference in shape be tween the farmed males and females, their consensus shapes ap pear to be quite congruent even when differences are magnified 3× (Fig. 5a).Wild females appear to be shallower in the abdominal region than the wild males, as indicated by the magnitude of the ventral displacements of points 2, 3, and 4 relative to point 12 (refer to Fig. 2 for description of points and Fig. 5b for relative displacement of points).This difference in body depth seems to be confined to the abdominal region because the displacement of the points on the dorsal surface is offset by the displacement of the points opposite them on the ventral surface in the head (points 1, 13, 15, 16, and 18), and in the caudal regions (points 5, 6, 7, 9, 10, and 11; Fig. 5b).Farmed males and females both show a reduction in head size and  Gadus morhua.DF, AF, and PF refer to the dorsal, anal, and pelvic fins respectively, and their corresponding numbering begins with the most anterior fin.Condition index (CI) and liver index (LI) are the standardized residuals of the regression of standard length, and liver weight on total weight respectively.Fin sizes were standardized to a common centroid size, while the calculation of CI and LI includes an inherent standardization caudal peduncle length relative to their wild counterparts (females: Fig. 5c; males: Fig. 5d).The smaller head size is evidenced by the posterior displacement of points 1, 16, 17, and 18, the anterior displacement of points 13 and 14, and the anteriodorsal displacement of point 15 (Fig. 5c,d).However, farmed males show a greater reduction in jaw length relative to wild males than farmed females do to wild fe males, (point 15; Fig. 5c,d).The posterior displacement of points 6, 7 (females), 9, and 10, while the midlateral portion of the hypural plate (point 8) re mains rela- tively un changed along the anteroposterior axis, is indicative of a truncation of the caudal peduncle.Of particular note, the difference in ab dominal region body depth between the farmed and wild females appears to be greater than the difference between the farmed and wild males (points 3, 4, and 12; Fig. 5c,d).It is worth noting that the dorsal rotation of point 8 in Fig. 5b,d appears most likely to be the result of subtle differences in the overall rotation, or curvature of the wild male specimens, and likely should be taken as spurious.

Differences between wild and farmed fish
Farmed Atlantic cod experience an environment markedly different from that of wild cod.Differences include diet, water temperature and current, fish density, visual and structural complexity, and structure, all of which have been shown to plastically affect the growth, development, and morphology of fishes (Currens et al. 1989, Adams & Huntingford 2002, Marcil et al. 2006, Ambrosio et al. 2008).Not surprisingly, the vast majority of morphological characters we measured differed significantly between wild and farmed individuals, as did their overall shape as evidenced by geometric morphometric analysis.Both traditional and geometric morphometrics indicated that farmed cod had relatively smaller head, jaw, and fin measures, while their body depth, CI, and LI measures were larger than those of the wild cod.
The presence in cultured cod of greater CI and LI than wild cod has been widely documented (e.g.Lie et al. 1986, Svåsand et al. 1996, Grant et al. 1998, Purchase & Brown 2001) and is corroborated by our results.Given that the main site of lipid sequestration in cod is the liver, and liver size and lipid content are directly influenced by the lipid content of the feed, the observed differences in LI are likely reflective of the different diet and physical environment experienced by the wild and farmed cod (Lie et al. 1986, Lambert & Dutil 1997, Morais et al. 2001).Similarly, the greater CI and the greater body depth of the farmed relative to the wild fish in this study are both related to the farmed cod having a higher LI (liver, and as a consequence, visceral mass).
As seen for body depth and LI, the different head morphology in the farmed and wild cod was also likely the result of differences in diet, and perhaps to a lesser extent, physical environment.The jaw and head morphology of fishes have been shown to be highly phenotypically plastic, and this plastic response is related to and influenced by the fish's diet.While studies on the phenotypic effects of different diets are lacking for cod, studies of other species have indicated that smaller heads and jaws are seen in fish which are fed non-elusive, prepared diets (Meyer 1987, Wintzer & Motta 2005), as well as in fish fed a greater ration (Currens et al. 1989).These features are characteristic of the pellet diet and feeding regime of farmed cod, and relatively smaller heads and jaws have been previously observed in cultured cod (Uglem et al. 2011).
Among the head features that were found to be relatively smaller in the farmed than the wild fish was eye size.Apart from simply being proportional to the head size, Devlin et al. (2012) have suggested that the eye development of rapidly growing fish becomes decoupled from their somatic growth, resulting in a negative allometry.
The most consistently observed differences between multiple species of wild and cultured fish are that cultured fish tend to develop relatively smaller fins of all types (e.g.Lund et al. 1989, Swain et al. 1991, Rogdakis et al. 2011, Patiyal et al. 2014).In some cases, this difference in size is the result of the fins of the cultured fish being either damaged or malformed (Bosakowski & Wagner 1994, Latre mouille 2003, Hatlen et al. 2006, Blanchet et al. 2008, Chittenden et al. 2010).However, it is unlikely that contemporary fin damage or malformation affected the results of the present study.The fins of both the wild and farmed fish were checked for signs of damage (e.g.clubbing, or abrasion of fin margin, etc.) or deformity, and measurements from any deformed fins were excluded from the analysis.Whether past damage or abrasion may have resulted in stunting of the size of the farmed cod's fins is also unclear, given the behaviour of cod (decreased wounding with fish size; Hatlen et al. 2006), as well as the great capacity for organ and tissue regeneration present in fish (Azevedo et al. 2011, Shao et al. 2011).It is possible that the smaller fins of the cultured cod resulted in part from a plastic response to water current.Studies in salmonids have shown that lower current velocity and variability experienced in culture can lead to relatively smaller fins (Pakkasmaa & Piironen 2000, Wessel et al. 2006, Keeley et al. 2007).Similarly, when compared to wild fish, farmed cod likely experience similar reductions in water velocity, and hence similar plastic effects on fin size could be expected in our study.
Considering all the observed differences between the farmed and wild cod in our study, the congruence between our results and those of Uglem et al. (2011), the only other study of differences in adult morphology between wild and farmed cod in which sufficient information is reported to allow comparison, is impressive.This is especially true given that the populations examined are thought to have been isolated for at least 100 000 yr (Bigg et al. 2008).This suggests that the observed differences may represent a stereotypical plastic response of Atlantic cod to culture.

Differences between sexes
Cod drumming-muscle weight (Engen & Folstad 1999, Rowe & Hutchings 2004, Skjaeraasen et al. 2006, 2008) and the length of the pelvic fins (Skjaeraasen et al. 2006(Skjaeraasen et al. , 2008(Skjaeraasen et al. , 2012) ) have been shown to be sexually dimorphic in other studies, and our results found this to be true of drumming muscle weight, but marginally not so for pelvic fin length.Both traits are suspected to play important roles in mate choice (Skjaeraasen et al. 2006, 2012, Rowe & Hutchings 2008), and in the case of the pelvic fins, in maintaining ventral alignment during gamete release (Skjaeraasen et al. 2008).
Sampling time and differences in the maturation schedule of male and female cod likely account for the observed differences in body depth, body mass, and LI, and perhaps to some extent drumming muscle mass.Seasonal gonad ripening in cod from this population generally begins at about the same time these fish were sampled (Rideout & Burton 2000).Male Atlantic cod (cultured and wild) generally begin to mature and have functionally mature gonads earlier in the season than females.During maturation, males cease feeding and exhibit a concomitant decrease in body mass and marked hypertrophy of the testes and drumming muscles, while maintaining an LI lower than that of females throughout their reproductive cycle (Fordham & Trippel 1999, Rideout & Burton 2000, Rowe & Hutchings 2004, Solberg & Willumsen 2008).

Implications
When cultured cod escape from net-pens, they interact with wild cod, and are subjected to the selective pressures of the natural environment (Moe et al. 2007, Damsgård et al. 2012, Zimmermann et al. 2012).It is likely that the morphology developed by the cod in culture will be to some degree maladaptive in the wild, and thus any escapees will experi-ence lower fitness than their wild counterparts, as has been seen in other species (Fleming et al. 2000, McGinnity et al. 2003, Meager et al. 2010, Skaala et al. 2012).
The differences in fin size and body condition we documented may result in different swimming performance.However, the relationship between them in cod and other species is not always clear (Rose et al. 1995, Reidy et al. 2000).Fitness effects of the fins may also extend to reproduction, with the relatively smaller fins of the farmed cod imparting a competitive disadvantage during both male−male agonistic interaction and courtship display.Extension of the median fins is a component of male Atlantic cod's 'flaunting display' (shown to both males and females; Brawn 1961), and pelvic fins are used both for display (Skjaeraasen et al. 2010) and to grasp the female and maintain alignment of their urogenital openings during ventral mount (Brawn 1961, Rowe et al. 2008).Moreover, some evidence suggests pelvic fin size may be related to spawning success (Rowe et al. 2008).Such effects may, however, be mitigated to some extent by transience in the differences in fin sizes resulting from convergence through plasticity towards the wild phenotype following escape, as noted in gilthead sea bream Sparus aurata (Arechavala-Lopez et al. 2013), and the same is likely true of condition (CI and LI;Nordeide et al. 1994, Jacobsen & Hansen 2001).
It is worth reiterating that the fish in this study are first-generation offspring of wild-caught parents, and while a single generation in captivity has been shown to affect the fitness of cultured fish (Fleming et al. 1997, Milot et al. 2013), increased generations under selection in a cultured environment can lead to genetic changes (reviewed by Hutchings & Fraser 2008, Nguyen 2015).Such genetic changes could result in permanent phenotypic changes relative to the wild fish, even if they are exposed to the same environment (i.e. after escape; Araki et al. 2008, Christie et al. 2012, Milot et al. 2013).Therefore, any differences in fitness caused by the morphological differentiation between wild and cod ob served in this study would likely be inflated by genotypic and consequent phenotypic changes that accumulate over time through both deliberate and inadvertent selection.

Fig. 4 .Fig. 5 .
Fig. 4. Ordination plot for configurations of specimens into principal warp space for geometric morphometric analysis of Gadus morhua.Individuals are plotted by origin and sex using colour and shape respectively (farmed = black, wild = red, males = triangles, females = circles; farmed males: n = 58, farmed females: n = 50, wild males: n = 13, wild females: n = 23).Ellipses represent 95% CI for the groups.The same colour scheme is used to denote origins, but sexes are distinguished by line type (solid = males, dashed = females)

ture
Research Building and Sapphire Sea Farms for allowing us to collect data during their harvest, and for their support during this time.The wild cod were collected with the help of Dennis Ivany and John Brattey.Ryan Stanley was instrumental in the creation of Fig.1.We also thank the members of the Fleming and Purchase labs for their assistance and friendship.The Natural Sciences and Engineering Research Council of Canada funded this research through a strategic grant to I.A.F. and C.F.P., and the Research and Development Corporation of Newfoundland provided additional support to B.F.W. LITERATURE CITED Adams CE, Huntingford FA (2002) The functional significance of inherited differences in feeding morphology in a sympatric polymorphic population of Arctic charr.Evol Ecol 16: 15−25 Adams DC, Otárola-Castillo E (2013) geomorph: an R package for the collection and analysis of geometric morphometric shape data.Methods Ecol Evol 4: 393−399 Adams DC, Rohlf FJ, Slice DE (2004) Geometric morphometrics: ten years of progress following the 'revolution'.Ital J Zool 71: 5−16 Ambrosio PP, Costa C, Sánchez P, Flos R (2008) Stocking density and its influence on shape of Senegalese sole adults.Aquacult Int 16: 333−343 Araki H, Berejikian BA, Ford MJ, Blouin MS (2008) Fitness of hatchery-reared salmonids in the wild.Evol Appl 1: 342−355 Arechavala-Lopez P, Sanchez-Jerez P, Izquierdo-Gomez D, Toledo-Guedes K, Bayle-Sempere JT (2013) Does fin damage allow discrimination among wild, escaped and farmed Sparus aurata (L.) and Dicentrarchus labrax (L.)?J Appl Ichthyol 29: 352−357 Azevedo AS, Grotek B, Jacinto A, Weidinger G, Saude L (2011) The regenerative capacity of the zebrafish caudal fin is not affected by repeated amputations.PLoS ONE 6: e22820 Bailey MM, Lachapelle KA, Kinnison MT (2010) Ontogenetic selection on hatchery salmon in the wild: natural selection on artificial phenotypes.Evol Appl 3: 340−351 Beacham TD, Brattey J, Miller KM, Le KD, Withler RE (2002) Multiple stock structure of Atlantic cod (Gadus morhua) off Newfoundland and Labrador determined from genetic variation.ICES J Mar Sci 59: 650−665 Benjamini Y, Hochberg Y (1995) Controlling the false discovery rate: a practical and powerful approach to multiple testing.J R Stat Soc B 57: 289−300 Bigg GR, Cunningham CW, Ottersen G, Pogson GH, Wadley MR, Williamson P (2008) Ice-age survival of Atlantic cod: agreement between palaeoecology models and genetics.Proc R Soc B 275: 163−173 Blanchet S, Páez DJ, Bernatchez L, Dodson JJ (2008) An integrated comparison of captive-bred and wild Atlantic salmon (Salmo salar): implications for supportive breeding programs.Biol Conserv 141: 1989−1999 Bosakowski T, Wagner EJ (1994) Assessment of fin erosion by comparison of relative fin length in hatchery and wild trout in Utah.Can J Fish Aquat Sci 51: 636−641 Bradbury IR, Hubert S, Higgins B, Borza T and others (2010) Parallel adaptive evolution of Atlantic cod on both sides of the Atlantic Ocean in response to temperature.Proc R Soc B 277: 3725−3734

Table 1
. Mean (± SD) morphometric measures and analyses by sex and farmed or wild origin Atlantic cod Gadus morhua.Standard length and weight measures are unstandardized; the calculation of condition index (CI) and liver index (LI) includes an inherent standardization.Drumming muscle weight has been standardized to a common weight, while all other measures have been standardized to a common centroid size.DM: combined dried mass of right and left drumming muscles; DF1, DF2, DF3: first through third dorsal fins; AF1 and AF2: first and second anal fins; PF: pelvic fins.There were no significant interactions between sex and origin for any of the measures.Adjusted p-values are shown, and those significant are in bold (α = 0.05)

Table 2 .
Percentage of explained variance, eigenvalues, and the loadings of the measurements included in the principal component analysis (PCA) (with varimax rotation) on the first 4 principal components (PCs), for farmed and wild