Fisheries Research 170 (2015) 82–88

Contents lists available at ScienceDirect

Fisheries Research

j ournal homepage: www.elsevier.com/locate/fishres

Fish traceability: Guessing the origin of fish from a seafood market

using fish scale shape

∗

Ana L. Ibánez˜

Universidad Autónoma Metropolitana-Iztapalapa Departamento de Hidrobiología, Av. San Rafael Atlixco 186, Col. Vicentina. México, D.F. 09340, México

a r t i c l e i n f o a b s t r a c t

Article history: Fish traceability is an important tool in fish food safety and a recognition tool in the assessment of

Received 6 December 2014

biodiversity and fisheries. Geometric morphometric methods were used to establish whether scale mor-

Received in revised form 9 May 2015

phology may determine the origin of specimens from a fish market. In order to trace the origin of fish, the

Accepted 19 May 2015

fish scale shape of two mugilids species Mugil cephalus and Mugil curema specimens from three different

Handling Editor B. Morales-Nin

trade premises at the Mexico City central fish market was analyzed and compared with the scale shape of

Available online 6 June 2015

previously collected samples from known areas along the Gulf of Mexico and Pacific coasts. The origin of

the fish market specimens was kept in a closed envelope and was revealed at the end of the study. Scale

Keywords:

shape was described using seven landmarks, the coordinates of which were subjected to a generalised

Fish traceability

Procrustes analysis, followed by a principal components analysis and discriminant analysis. Discriminant

Fish scale shape

Stock identification classification was used as the main indicator to identify the source of the fish, where the percentage of

Fisheries management discrimination traced the origin of the specimens, respectively. A leave-one-out procedure was applied

Geometric morphometrics to validate the results of discriminant functions on shape, and relative warp in size-shape space (shape

Gulf of Mexico

plus size). Results showed that the discriminant analysis correctly classified 83.3% of the cases, with 100%

Mugil curema

for adjacent geographic areas. The relative warp in size-shape space (shape plus size) of fish scales could

Mugil cephalus

be used to reliably trace fish populations by applying geometric morphometric methods to one fish scale

per specimen. In order to effectively use such an approach in fisheries management, it is necessary to

have previous patterns of fish scale shapes for each locality. Since the scale shape was described using

seven landmarks, it may be possible to improve the tracing rate by increasing the number of landmarks

used. Fish scale shape thus offers a straightforward, accessible, quick and inexpensive method to trace

fish stocks.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction when a product comes from a contaminated area and it is fun-

damental to limit damages. In this case, a rapid identification is

European Union law defines “Traceability” as the ability to track required to enable a fast retrieval of the product. It is also required

any food, feed, food-producing animal or substance that will be in population analyses in order to understand population dynamics

used for consumption, through all stages of production, processing and to carry out fisheries assessments.

and distribution (Official Journal of the European Communities, Until now, tracing methods have used molecular biotechnolo-

2002). Traceability is relevant because responds to food security gies (Lockely and Bardsley, 2000), accuracy testing of fish products

threats, documents custody chains and production practices, meets has been carried out using protein-based techniques (Rehbein,

regulatory compliances and analyses logistics and production costs 1990) and nucleic acid methods have become accepted for species

(Storøy et al., 2013). Providing the identity of a product is important identification due to their precision (Carrera et al., 1998; Jerome

as it gives citizens security regarding what they eat. Fish traceabil- et al., 2008). With the range of Polymerase Chain Reactions, inter-

ity information may improve the sale of products by identifying est in AFLPs (Amplified Fragment Length Polymorphisms) and in

appropriate information on the origin and source of fish, including DNA barcodes (normally based on sequences of cytochrome oxi-

whether they are wild or cultured. Fish traceability is also useful dase subunit I) is increasing among animal geneticists (Watanabe

et al., 2004) since these procedures generate hundreds of informa-

tive genetic markers that increase the probability of detection of

∗ species-specific and population-specific polymorphisms (Maldini

Tel.: +52 55 58046585; fax: +52 55 58044738.

E-mail address: [email protected] et al., 2006; Papa et al., 2003; Heras et al., 2006; Durand et al.,

http://dx.doi.org/10.1016/j.fishres.2015.05.016

0165-7836/© 2015 Elsevier B.V. All rights reserved.

A.L. Ibá˜nez / Fisheries Research 170 (2015) 82–88 83

2012). Unfortunately, these methods are not only expensive but were collected from commercial fisheries in Madre Lagoon (MA)

also require specialized human resources for their application. and Tamiahua Lagoon (TA) along the Gulf of Mexico, and in July

Novel technologies are thus needed to assess species and identify 2009 from Cuyutlán Lagoon (CU) along the Pacific coast (Fig. 1,

stocks (Fischer, 2013). Table 1). All specimens were adults with an average total length

The shape of fish scales is to a significant extent species-specific of 356 ± 42 mm.

and helps determine stock membership. In recent times, Ibánez˜ M. curema scales were collected on similar days of each of two

et al. (2007) applied geometric morphometric methods (GMM) to consecutive years, 2009 and 2010, from specimens caught by com-

scales in order to identify genera, species and local populations mercial fisheries in Madre Lagoon (MA), Tamiahua Lagoon (TA),

among the Mugilidae. Garduno-Paz˜ et al. (2010) also used fish Cazones Estuary (CA), Alvarado Lagoon (AL) and Mecoacán Lagoon

scale morphometrics and were able to discriminate between sym- (ME) along the Gulf of Mexico. Scales were also collected in July

patric phenotypes of the Arctic charr (Salvelinus alpinus). The use 2009 from specimens from Cuyutlán Lagoon (CU) and the Balsas

of scale shape seems valuable in population discrimination and is River (BA) along the Pacific coast of Mexico (Fig. 1, Table 2). All

an important and practical fisheries management tool. Moreover, specimens were adults with average total lengths of 287 ± 39 and

fish scale shape was used to identify geographic variants among 298 ± 21 mm, for the first and second year respectively.

the Lutjanidae (Lutjanus argentiventris, Lutjanus guttatus and Lut- In all cases, samples of approximately 50 specimens per geo-

janus peru) of three geographic areas along the Pacific coast (Ibánez˜ graphic area and species were obtained, with the exception of M.

et al., 2012a), where results showed that specimens of each species cephalus from San Antonio Bay which had only 20 specimens.

from the three geographic areas formed two local populations. When adults, both species could be clearly distinguishable by

The consistency of the two populations for the three species indi- a trained eye, however, since mullets show extremely conserved

cated that these events do not happen by chance or accidently. morphological characters the specimens were examined using the

Variability in hydro-geomorphology, water productivity and abun- taxonomic keys from Harrison (2002).

dance of fish stocks likely accounted for the observed differences

in scale morphology and explained why scale shape may be used 2.2. Collection of fish scales from Mexico City market samples

to discriminate between stocks (De Pontual and Prouzet, 1987;

Watkinson and Gillis, 2005; Ibánez˜ et al., 2007). Consistency in fish M. curema and M. cephalus specimens were bought at the Mexico

scale shape through time was also tested considering that time may City fish market during the November–December 2012 fishery sea-

affect different locations in a similar way (Ibánez,˜ 2014). However, son. The trade premises for M. curema were stalls 49, 1 and 47, on D

no significant differences in scale shape were detected from year corridor, and the samples were labeled 49D, 1D and 47D respec-

to year, whereas differences were recorded among the geographic tively, while those for M. cephalus were stalls 20, 8 and 49, on

locations. Thus, scale shape differences may be useful to trace fish corridor D, and the samples were labeled 20D, 8D and 49D. A sample

when a previously recorded pattern for each locality exists. Con- of around 50 fish was obtained without recording its origin. The ori-

sequently if it is possible to use fish scales shape to discriminate gin of the specimens was unknown to us and we asked the seller to

species or stocks and trace the origin of fish this will be a relevant write the source of the sample in a closed envelope to be revealed

novel finding that could be useful in fisheries research and also for at the end of the study. All specimens were adults with average

economic purposes. The difference between the previous studies total lengths of 296 ± 28 mm and 361 ± 38 mm, for M. curema and

and this one is that in the aforementioned the origin of the spec- M. cephalus respectively.

imens was established, whereas in the case of the present study, One scale per fish was extracted for analysis from the left side of

the localities where the fish were collected are unknown. Mugil the body, from the dorsal area between the first and second dorsal

species were selected because of the importance of mullets fish- fins. A total of 753 scales was analyzed for M. curema (599 biological

eries, according to FAO (2009), Mexico is one of the 10 counties samples and 154 from the fish market) and of 358 for M. cephalus

with more production of fish mullets. (225 biological samples and 133 from the fish market). One single

In order to trace the origin of the fish under study, Mugil cephalus scale was used with the purpose of evaluating a fast population

(Linnaeus, 1758) and Mugil curema (Valenciennes, 1836) specimens method that does not require repetitive sampling of individuals.

were bought at different trade premises in the Mexico City seafood The scales were dried and stored in paper envelopes. Once in the

central fish market. Fish scale shape was analysed and compared laboratory, the scales were cleaned with soft soap and tap water

with scale shapes of samples (called biological samples) collected with the help of a small toothbrush. They were then rinsed, dried

previously from known areas along the Gulf of Mexico and Pacific and preserved between two glass slides. A digital image of each

coasts. The null hypothesis stated that the differences described scale was taken with a Zeiss stereomicroscope (Stemi 2000-C) and

by the fish scale morphology landmarks between the fish from the an integrated 4.0 MP Canon digital camera. Regenerated scales were

market and the previously known biological samples are insuffi- discarded. Seven landmarks per scale were recorded with the TPS-

cient to allow the identification of the market samples. In other dig software (Rohlf, 2006) following the protocol of Ibánez˜ et al.

words, fish scale shape cannot be used as a barcode for fish trace- (2007).

ability. This was assessed by the GMM, with discriminant analysis The landmarks were located on key features of the scales that are

as the main indicator to identify the source of the fish, where the common to all scales of the two species under study. The landmarks

percentage of discrimination revealed or traced the origin of the considered appropriate were (Fig. 2): landmarks 1 and 3 the ventro-

specimens. The origin of the market samples was written in a closed and dorso-lateral tips of the anterior portion of the scale, landmark

envelope and revealed at the end of the study. 2 the centre of the anterior edge of the scale, landmarks 4 and 6

the boundary between the anterior portion with circuli and the

posterior area covered by cteni, landmark 5 the focus of the scale,

2. Material and methods and landmark 7 the tip of the posterior portion of the scale.

2.1. Collection of fish scales from biological samples 2.3. Morphometrics analyses of biological and city marked

samples

M. cephalus scales were collected in 2009 from specimens caught

in Sabine Lake (SL) and San Antonio Bay (SA), Texas by the staff The configurations of landmark coordinates for the sampled

of the Texas Parks & Wildlife Department, USA. Mexican samples scales were scaled, translated and rotated using generalized

84 A.L. Ibá˜nez / Fisheries Research 170 (2015) 82–88

Fig. 1. Sampling locations in the Gulf of Mexico: Sabine Lake, Texas (SL), San Antonio Bay, Texas (SA), Madre Lagoon, Tamaulipas (MA), Tamiahua Lagoon, Veracruz (TA),

Cazones Estuary, Veracruz (CA), Alvarado Lagoon, Veracruz (AL) and Mecoacán Lagoon, Tabasco (ME). Locations in the Pacific coast: Cuyutlán Lagoon, Colima (CU) and Balsas

River, Michoacán (BA).

Procrustes analysis (GPA) (Rohlf and Slice, 1990). They were then functions, and to assess the efficacy of the latter in classifica-

subjected to tangent projection (Dryden and Mardia, 1998) and tion. This was carried out using cross-validation, in which multiple

subsequently to principal components analysis (PCA; Dryden and repeated analyses were carried out leaving out one individual in

Mardia, 1998; Kent, 1994). For plotting purposes, the principal the construction of the discriminant function before classifying this

component (PC) scores were labelled by species to describe the dis- individual according to the function. This reduces the likelihood of

tribution of the scales. Predicted extreme shapes were computed overestimating the efficacy of discriminant functions by using them

for each principal component. The differences in shape between to classify specimens employed in their construction. Percentage

the extremes of the PCs of interest were visualised using trans- correct classification rates were recorded. Cross validated discrim-

formation grids (Bookstein, 1989; Marcus et al., 1996; Dryden and inant analysis was used to assess and compare the efficacy of shape

Mardia, 1998) computed by Morphologika2 (O’Higgins and Jones, and relative warp in size-shape space (ln centroid size plus shape

2007). variables).

To examine the potential for differences in shape in clas- Discriminant percentage of discrimination of each population

sifying unknown specimens, the scores of specimens on all was used as the main indicator to identify the origin of samples

nonzero PCs were submitted to discriminant analysis (SPSS ver. from the fish market, where the higher percentage of discrimina-

13.0) to compute generalized Mahalanobis’ distances, discriminant tion traced the origin of market specimens.

Table 1

Sample size, origin and site of collection of fish scales and length characteristics for Mugil cephalus specimens.

No. fish sampled Total length Total length

Mean ± SD (mm) Range (mm)

Origin of samples Location Code Year of collection

2009 2012 2009 2012 2009 2012

Biological samples:

Gulf of Mexico

Sabine Lake SL 50 341 ± 43.0 305–500

San Antonio Bay SA 20 344 ± 31.9 310–405

Madre Lagoon MA 55 354 ± 22.8 309–427

Tamiahua Lagoon TA 48 327 ± 25.1 281–400

Pacific Coast

Cuyutlán Lagoon CU 52 405 ± 32.8 221–464

Market samples:

Market site 20D 20D 51 359 ± 20.9 315–421

Market site 8D 8D 51 352 ± 22.6 285–392

Market site 49D 49D 31 372 ± 56.4 291–475

A.L. Ibá˜nez / Fisheries Research 170 (2015) 82–88 85

Table 2

Sample size, origin and site of collection of fish scales and length characteristics for Mugil curema specimens.

No. Fish sampled Total length Total length

±

Mean SD (mm) Range (mm)

Origin of samples Location Code Year of collection

2009 2010 2012 2009 2010 2012 2009 2010 2012

Biological samples:

Gulf of Mexico

Madre Lagoon MA 51 50 286 ± 12.2 296 ± 13.8 249−312 258−324

Tamiahua Lagoon TA 46 50 299 ± 22.2 295 ± 22.9 230−350 244−332

Cazones Estuary CA 50 50 307 ± 18.8 313 ± 26.7 245−347 235−360

Alvarado Lagoon AL 50 50 276 ± 23.3 296 ± 21.5 224−339 246−339

Mecoacán Lagoon ME 53 49 286 ± 42.3 294 ± 18.3 223−405 256−336

Pacific Coast

± −

Cuyutlán Lagoon CU 50 226 10.2 200 251

Balsas River BA 50 329 ± 22.4 261−372

Market samples:

Market site 49D 49D 52 288 ± 16.7 255−335

Market site 1D 1D 51 326 ± 17.2 299−361

Market site 47D 47D 51 274 ± 17.7 244−323

Total length, centroid size (CS) and ln CS were compared along the dorso-ventral axis (equivalent to elongation along the

between species and across population with a two-way (popula- anterior–posterior axis).

tion \ species) analysis of variance (ANOVA). Heteroscedasticity in Species and also population differences are highly significant for

the ANOVA were detected by means of Levene’s test for all size size (P = 0.001 species; P = 0.001 populations). M. cephalus are larger

measures (P = <0.001) and ANOVA results must, accordingly, be than M. curema specimens in all cases; also intraspecific differences

interpreted with caution. are highly significant.

The cross-validated discriminant analysis using shape and rela-

tive warp in size-shape space (shape plus size) variables (PC scores)

3. Results

from all scales correctly classified 98.2% by species in both cases,

whereas the population analysis by species correctly classified

The GPA/PCA carried out with all fish scales using shape and rela-

(cross-validated) 53.2% and 57.9% and 46.3% and 58.5% for shape

tive warp in size-shape space (shape plus size) explaining 44.6% and

and relative warp in size-shape space (RWSS) for M. cephalus and

98.4% of the total variance in the first PC and the second accounted

M. curema, respectively (Table 3). Classification results improved

for 23.7% and 0.7%, respectively. The two species were clearly sep-

when size information was included in the discriminant analysis,

arated in the plot of the two principal components from the shape

except for species analysis where the cross-validated classification

analysis (Fig. 3). The specimens of M. cephalus and M. curema were

rate remained the same (Table 3).

separated on the first PC.

The cross-validated classification results from the analysis

The general pattern of morphological differences described by

among local populations for M. cephalus and M. curema identi-

these first two PCs was explored using transformation grids (Fig. 3).

fied highly significant discrimination in all cases (Wilks’ lambda

The grid on the left represents the mean shape warped to a PC1

values between 0.055 and 0.204 and P < 0.001, Table 3). In

score of −0.15 for M. cephalus and for M. curema with the grid on

both species higher discrimination percentages were between

the right representing a PC1 score of 0.15, the deformations indi-

35.4% and 96.2% and 36.5% and 78.4% for M. cephalus and M.

cate that with increasing PC1, the focus becomes less central with

curema, respectively (Table 4A and B). Most misclassifications

a little narrowing between landmarks 4 and 6. Also, the exposed

occurred between the neighbor locations or between the fish

portion of the scale (the area among the landmarks 4–7) becomes

market samples. Once these results were recorded, the origin

relatively narrower. Shape variations through the PC2 compressed

of the market samples was revealed. The correctly guessed ori-

gin is marked with an asterisk in Table 4 which in most of the

cases (with exception of ID fish market sample for M. curema)

represents the higher value of the misclassifications. Therefore,

the discriminant analysis carried out on relative warp in size-

shape space (shape plus size) correctly calculated 83.3% of the

market cases with 100% for the adjacent geographic areas. Com-

paratively there is more shape variability among the different

specimens from M. curema geographic areas documented through misclassifications.

4. Discussion

The aim of this study was to trace the origin of fish collected at

the Mexico City central fish market based on the morphometrics of

one fish scale per specimen. In order for the method to be useful

to managers, it needs to be based on simple scale features com-

mon to all scales. We have used only seven landmarks in previous

studies, and they have proved successful in the analysis of species

Fig. 2. Definition of landmarks. and local variants discrimination (Ibánez˜ et al., 2007, 2009, 2011,

86 A.L. Ibá˜nez / Fisheries Research 170 (2015) 82–88

Fig. 3. Principal components analysis (PC1 and PC2) of scale shape of the whole sample where Mugil cephalus and M. curema are labeled with gray and white diamonds,

respectively. The grids are shape variation visualized using transformation grids.

2012b). However, this could be improved by increasing the number Classification results improved when size information was

of landmarks. Based on these seven landmarks, our results imply included in the discriminant analysis. Using only shape (with-

that there is 83.3% possibility of guessing the exact origin of fish out size), identification rates are much better than expected by

market samples and a 100% of guessing the adjacent fishing areas. chance, and by taking size into account, classification is somewhat

Misclassifications were obtained for samples in nearby geographic improved. The size of the fish (total length) has an impact on scale

areas, separated by around 200 or 300 km. In this sense, the dis- size and shape. Even when the specimens of the two species and

criminant analysis is a fine tool to guess the correct locality of the geographic location were of similar length, differences in centroid

sample from unknown origin. Discriminant analysis extracts vec- size of the scales were found amongst them. Thus, when using

tors of discriminant functions from multivariate data, which exploit this approach to discriminate between species and populations,

the differences among a priori-defined groups (Hair et al., 1998; it is important that centroid size is factored into the procedure.

Zelditch et al., 2004) where the first function created maximizes the The lower total classification result between populations is clearly

differences between groups on that function. The number of func- explained by the use of geographic locations of each species as

tions is equal to the number of groups minus one and, unless there encoded groups in the cross-validated analysis.

are fewer predictor variables than groups, they only account for the A possible explanation for the misclassifications of samples

variance which best discriminates groups according to the model from contiguous geographic areas may lie in the life history of the

(Kovarovic et al., 2011). Examples on recent application of discrim- mullets (Ibánez˜ and Gallardo-Cabello, 2004; Ibánez˜ and Gutiérrez-

inant function analyses using otoliths to discriminate species and Benítez, 2004). Mullets spawn beyond the continental shelf, the

fish stocks can be found in Monteiro et al. (2005) and Burke et al. larvae remain in coastal waters for two or three months (Ditty and

(2008), respectively. The use of otoliths for this purpose is well Shaw, 1996) during which time the currents transport them to estu-

established, and while their use is more labour and equipment arine and coastal regions that may not be those of their parents. A

intensive, their efficacy is clear. However is an invasive method. resemblance of specimens caught at neighbouring sites could thus

Likewise, discriminate analysis has been proved in other species, be explained by this exchange of migrants during the larval phase,

this analysis is highly used in paleontology and archaeological stud- as well as by the similar water temperature and other general water

ies in order to investigate different species or categories (Kovarovic characteristics. A future advance can be the use of geometric mor-

et al., 2011; Meloro, 2011). phometrics and spatially-explicit statistical analyses (Fruciano et

Table 3

Results for discriminant analysis using the cross-validation testing procedure. Discrimination between species and populations. RWSS = Relative warp in size-shape space.

p

Discrimination Procedure Wilks’  Discriminant functions Total classification

1st 2nd

results (%)

Between species (M. Shape 0.209 <0.0001 100 98.2

cephalus and M. curema) RWSS 0.175 <0.0001 100 98.2

Between population

Mugil cephalus Shape 0.096 <0.0001 67.1 24.2 53.2

RWSS 0.055 <0.0001 67.3 18.3 57.9

Mugil curema Shape 0.204 <0.0001 49.7 17.5 46.3

RWSS 0.055 <0.0001 68.5 17.7 58.5

A.L. Ibá˜nez / Fisheries Research 170 (2015) 82–88 87

Table 4

Classification results for discriminant analysis (using cross-validation testing procedure) for geographic areas and fish market samples using relative warp in size-shape space

(RWSS). A) Mugil cephalus; B) Mugil curema. Clues: Sabine Lake, Texas (SL), San Antonio Bay, Texas (SA), Madre Lagoon, Tamaulipas (MA), Tamiahua Lagoon, Veracruz (TA),

Cazones Estuary, Veracruz (CA), Alvarado Lagoon, Veracruz (AL) and Mecoacán Lagoon, Tabasco (ME). Locations in the Pacific coast: Cuyutlán Lagoon, Colima (CU) and Balsas

River, Michoacán (BA). In bold = higher discrimination percentage. * = original location of market samples: Mugil curema: 49D from Tamiahua Lagoon (TA), 1D from Cazones

estuary (CA) and 47D from Madre Lagoon (MA). Mugil cephalus: 49D from Tamiahua Lagoon (TA), 8D from Pacific area in Mazatlán port and 20D from Madre Lagoon (MA).

A)

Mugil cephalus

SL SA MA TA CU 20D 8D 49D Total

SL 44.0 22.0 10.0 15.0 0.0 2.0 0.0 7.0 100.0

SA 8.4 47.6 17.3 0.0 0.0 21.7 1.0 4.0 100.0

MA 9.1 5.5 48.2 23.6 0.0 8.1 1.8 3.6 100.0

TA 0.0 4.2 27.1 43.8 0.0 0.0 2.1 22.9 100.0

CU 0.0 0.0 0.0 0.0 96.2 0.0 3.8 0.0 100.0

20D 2.0 4.0 16.0* 0.0 0.0 78.0 0.0 0.0 100.0

8D 0.0 1.0 0.6 0.0 20.0* 0.0 78.4 0.0 100.0

49D 0.0 0.0 10.8 47.2* 0.8 3.9 1.9 35.4 100.0

B)

Mugil curema

MA TA CA AL ME BA CU 49D 1D 47D Total

MA 78.4 6.0 0.0 0.0 0.0 0.0 0.0 1.9 0.0 13.7 100.0

TA 11.8 60.9 4.0 0.0 2.2 0.0 1.5 19.6 0.0 0.0 100.0

CA 0.0 0.0 68.0 4.0 0.0 0.0 0.0 4.0 16.0 8.0 100.0

AL 4.0 2.0 6.0 58.0 16.0 2.0 0.0 0.0 8.0 4.0 100.0

ME 0.5 1.9 7.9 24.5 55.8 1.9 0.0 0.0 3.8 3.8 100.0

BA 1.9 2.0 2.0 0.0 0.0 72.0 22.1 0.0 0.0 0.0 100.0

CU 0.0 3.8 0.0 0.0 2.0 15.8 78.4 0.0 0.0 0.0 100.0

49D 0.0 30.0* 0.0 0.0 0.0 0.0 0.0 65.0 0.0 5.0 100.0

1D 0.0 0.0 14.0* 4.0 0.0 0.0 0.0 26.0 56.0 0.0 100.0

47D 28.8* 8.0 6.0 0.0 0.0 0.0 0.0 20.7 0.0 36.5 100.0

al., 2011) these methods could improve the correct classification Acknowledgments

rates.

The higher morphological variation in the specimens of M. The study was supported by funding from the Universidad

curema especially in the Atlantic may be due to the presence of Autónoma Metropolitana-Iztapalapa (UAMI) and from the Secre-

more than one population in the Gulf of Mexico or else that there taría de Educación Pública and the Consejo Nacional de Ciencia y

are different phenotypes in the same sample (Ibánez-Aguirre˜ and Tecnología SEP-CONACyT, Ciencia Básica: 2011-01-165569 (Con-

Lleonart, 1996; Ibánez-Aguirre˜ et al., 2006). ferred to UAMI 2012-2015). Many thanks to the Texas Parks

Growth seems to be a key factor related to fish scale shape. and Wildlife Department, USA to Mark Fisher (Science Director),

Changes in fish scale shape were reported for juveniles of the Jerry Mambretti (Sabine Lake Ecosystem Leader), Norman Boyd

cyprinid roach (Rutilus rutilus (L.)) reared in a UK fish farm when (San Antonio Bay Ecosystem Leader) and James Simons (A & M

the fish went through compensatory growth due to the variety of Texas University) who collected the Mugil cephalus samples from

holding facility movements, which are surrogates of food avail- Texas. The author is grateful to Paola Castillo, Elaine Espino, Eddie

ability, density, temperature and other variables (Ibánez˜ et al., Espinosa, Juan Juárez, Eloísa Pacheco and Angel Romero who helped

2012b). Environmental influences on morphology have led to the in Mexican collections. Many thanks to Eddie Espinosa and Eloísa

description of phenotypic populations characterized by pheno- Pacheco for the digitalization of scales. Also many thanks to two

typic differences that may be fully environmentally induced (Swain anonymous referees for very helpful suggestions and comments

and Foote, 1999). These differences could cause morphometric on an earlier version of this paper.

characteristics to vary between groups, allowing a robust morpho-

metric classification (Cronin-Fine et al., 2013). As this method is

References

not invasive, it may prove useful to trace ornamental fish allowing

endangered species to be returned to the wild. Also, it is quick and

Bookstein, F.L., 1989. Principal warps: thin-plate splines and the decomposition of

less costly than other analyses, allowing many individuals from a deformations. IEEE Trans. Pattern. Anal. 11, 567–585.

Burke, N., Brophy, D., King, P.A., 2008. Shape analysis of otolith annuli in Atlantic

population or community to be screened.

herring (Clupea harengus); a new method for tracking fish populations. Fish.

Res. 91, 133–143.

Carrera, E., Garcia, T., Céspedes, A., Gonzáles, I., Sanz, B., Hernández, P.E., Mar-

tin, R., 1998. Identification of Atlantic salmon (Salmo salar) and rainbow trout

5. Conclusion

(Oncorhynchus mykiss) by using Polymerase Chain Reaction amplification and

restriction analysis of the mitochondrial cytochrome b gene. J. Food Prot. 61,

Our results show that fish scale shape may be used to reliably 482–486.

Cronin-Fine, L., Stockwell, J.D., Whitener, Z.T., Labbe, E.M., Willis, T.V., Wilson, K.A.,

trace species and fish populations by applying geometric morpho-

2013. Application of morphometric analysis to identify alewife stock structure

metric methods to fish scales. In order to use such an approach

in the Gulf of Maine. Marine and Coastal Fisheries: dynamics. Manag. Ecosyst.

effectively in fisheries management it is important to have previ- Sci. 5, 11–20, http://dx.doi.org/10.1080/19425120.2012.741558

ous patterns of fish scale shapes for each locality. Therefore, fish De Pontual, H., Prouzet, P., 1987. Atlantic salmon Salmo salar L., stock discrimination

by scale-shape analysis. Aquacult. Fish. Manag. 18, 277–289.

scale shape offers a straightforward, accessible, rapid, reliable and

Ditty, J.G., Shaw, R.F., 1996. Spatial and temporal distribution of larval striped mullet

inexpensive method to trace fish stocks.

(Mugil cephalus) and white mullet (M. curema, family: Mugilidae) in the northern

88 A.L. Ibá˜nez / Fisheries Research 170 (2015) 82–88

Gulf of Mexico, with notes on mountain mullet, Agonostomus monticola. Bull. Ibánez,˜ A.L., 2014. Fish scale shape variation by year and by geographic location,

Mar. Sci. 59, 271–288. could scales be useful to trace fish?: A case study on the Gulf of Mexico and

Dryden, I.L., Mardia, K.V., 1998. Statistical Shape Analysis. John Wiley & Sons, London. Mexican Pacific coast. Fish. Res. 156, 34–38.

Durand, J.D., Shen, K.N., Chen, W.J., Jamandre, B.W., Blel, H., Diop, K., Nirchio, M., Jerome, M., Martinsohn, J.T., Ortega, D., Carreau, P., Verrez-Bagnis, V., Mouchel,

Garcia de Leon, F.J., Whitfield, A.K., Chang, C.W., Borsa, P., 2012. Systematics O., 2008. Species determination in fish products by molecular markers

of the grey mullets (Teleostei: Mugiliformes: Mugilidae): molecular phyloge- and support through a public domain database. J. Agric. Food Chem. 56,

netic evidence challenges two centuries of morphology-based taxonomic. Mol. 3460–3469.

Phylogenet. Evol. 64, 73–92, http://dx.doi.org/10.1016/j. ympev.2012.03.006. Kent, J.T., 1994. The complex Bingham distribution and shape analysis. J.R. Statist.

FAO, 2009. Mugil cephalus. In Cultured aquatic species fact sheets. Text by Saleh, Soc. B 56, 285–299.

M.A. Edited and compiled by Valerio Crespi and Michael New. CD-ROM (multi- Kovarovic, K., Aiello, L.C., Cardini, A., Lockwood, C.A., 2011. Discriminant function

lingual). ftp://ftp.fao.org/fi/document/aquaculture/culturedspecies/file/en/en analyses in archaeology: are classification rates too good to be true? J. Arch. Sci.



flatheadgreymullet.htm (accessed 06.05.15). 38, 3006–3018.

Fischer, J. (Ed.), 2013. Fish identification tools for biodiversity and fisheries assess- Lockely, A.K., Bardsley, R.G., 2000. DNA-based methods for food authentication.

ments: review and guidance for decision-makers. FAO Fisheries and Aquaculture Trends Food Sci. Technol. 11, 67–77.

Technical Paper No. 585. Rome, FAO. 2013. 107 pp. Maldini, M., Marzano, F.N., González-Fortes, G., Papa, R., Gandolfi, G., 2006. Fish and

Fruciano, C., Tigano, C., Ferrito, V., 2011. Geographical and morphological variation seafood traceability based on AFLP markers: elaboration of a species database.

within and between colour phases in Coris julis (L 1758), a protogynous marine Aquaculture 261, 487–494.

fish. Biol. J. Linn. Soc. 104, 148–162. Marcus, L.F., Corti, M., Loy, A., Naylor, G.J.P., Slice, D. (Eds.), 1996. Advances in Mor-

Garduno-Paz,˜ M.V., Demetriou, M., Adams, C.E., 2010. Variation in scale shape among phometrics. Nato ASI series. Plenum Press, New York.

alternative sympatric phenotypes of Arctic charr Salvelinus alpinus from two Meloro, C., 2011. Feeding habits of Plio-Pleistocene large carnivores as revealed by

lakes in Scotland. J. Fish Biol. 76, 1491–1497. the mandibular geometry. J. Vert. Paleontol. 31, 428–446.

Hair, J.F., Anderson, R.E., Tatham, R.L., Black, W.C., 1998. Multivariate Data Analysis, Monteiro, L.R., Di Beneditto, A.P.M., Guillermo, L.H., Rivera, L.A., 2005. Allometric

Fifth ed. Prentice Hall, New Jersey. changes and shape differentiation of Sagitta otoliths in sciaenid fishes. Fish. Res.

Harrison, I.J., 2002. Mugilidae: mullets. In: Carpenter K, editor. The living marine 74, 288–299.

resources of the Western Central Atlantic. Vol. 2. Bony fishes part 1 (Acipenseri- Official Journal of the European Communities, 2002. Regulation (EC) No.

dae to Grammatidae). FAO Species identification Guide for Fisheries purposes. 178/2002 of the European Parliament and the Council of 28 January

Rome: FAO. pp 1071–1085. 2002.

Heras, S., González-Castro, M., Roldán, M.I., 2006. Mugil curema in Argentinean O’Higgins, P., Jones, N., 2007. Morphologika2 v2.5. Hull York Medical School.

waters: combined morphological and molecular approach. Aquaculture 261, Available from http://sites.google.com/site/hymsfme/downloadmorphologica

473–478. (accessed 04.11.14).

Ibánez-Aguirre,˜ A.L., Lleonart, J., 1996. Relative growth and comparative morpho- Papa, R., Nonnis Marzano, F., Rossi, V., Gandolfi, G., 2003. Genetic diversity and adapt-

metrics of Mugil cephalus L. and M. curema V. in the Gulf of Mexico. Sci. Mar. 60, ability of two species of Mugilidae (Teleostei: Perciformes) of the Po river delta

361–368. coastal lagoons. Oceanol. Acta 26, 121–128.

Ibánez-Aguirre,˜ A.L., Cabral-Solís, E., Gallardo-Cabello, M., Espino-Barr, E., 2006. Rehbein, J., 1990. Electrophoretic techniques for species identification of fishery

Comparative morphometrics of two population of Mugil curema in the Atlantic products. Z. Lebensm.-Unters. Forsch 191, 1–10.

and Pacific coasts. Sci. Mar. 70, 139–145, ISSN 0214-8358. Rohlf, F.J. 2006. Tps Series. Department of Ecology and Evolution, State University,

Ibánez,˜ A.L., Gallardo-Cabello, M., 2004. Reproduction of Mugil cephalus and M. N.Y., Stony Brook. Available from http://life.bio.sunysb.edu/morph (accessed

curema (Pises:Mugilidae) from a coastal lagoon to the northwest of the Gulf 08.08.14).

of Mexico. Bull. Mar. Sci. 75, 37–49. Rohlf, F.J., Slice, D.E., 1990. Extensions of the Procrustes method for the optimal

Ibánez,˜ A.L., Gutiérrez-Benítez, O., 2004. Climate variables and spawning migrations superimposition of landmarks. Syst. Zool. 39, 40–59.

of the striped mullet and white mullet in the Northwestern area of the Gulf of Storøy, J., Thakur, M., Olsen, P., 2013. The TraceFood Framework—principles and

Mexico. J. Fish Biol. 65, 822–831. guidelines for implementing traceability in food value chains. J. Food Engr. 115,

Ibánez,˜ A.L., Cowx, I.G., O’Higgins, P., 2007. Geometric morphometric analysis of fish 41–48.

scales for identifying genera, species and local populations within the Mugilidae. Swain, D.P., Foote, C.J., 1999. Stocks and chameleons: the use of phenotypic variation

Can. J. Fish. Aquat. Sci. 64, 1091–1100. in stock identification. Fish. Res. 43, 113–128.

Ibánez,˜ A.L., Cowx, I.G., O’Higgins, P., 2009. Variation in elasmoid fish scale patterns Watanabe, T., Fujita, H., Yamasaki, K., Seki, S., Taniguchi, N., 2004. Preliminary

is informative with regard to taxon and swimming mode. Zool. J. Linn. Soc. 155, study on linkage mapping based on microsatellite DNA and AFLP markers

834–844. using homozygous clonal fish in ayu (Plecoglossus altivelis). Mar. Biotechnol. 6,

Ibánez,˜ A.L., Espino-Barr, E., Gallardo-Cabello, M., 2012a. Population connectivity 327–334.

among geographic variants within the Lutjanidae through fish scales shape Watkinson, D.A., Gillis, D.M., 2005. Stock identification of Lake Winnipeg walleye

recognition. Sci. Mar. 76, 667–675, http://dx.doi.org/10.3989/scimar.03675.09C based on fourier and wavelet description of scale outline signals. Fish. Res. 72,

(ISSN: 0214-8358). 193–203.

Ibánez,˜ A.L., Pacheco-Almanzar, E., Cowx, I.G., 2012b. Does compensatory growth Zelditch, M.L., Swiderski, L.D., Sheets, H.D., Fink, W.L., 2004. Geometric Morphomet-

modify fish scale shape? Environ. Biol. Fish. 94, 477–482. rics for Biologists. A Primer. Elsevier, St. Louis.