A1

FVa

sb

a

ARRA

KPSMCGMM

1

rwssitoMsflc

tes

0d

Fisheries Research 107 (2011) 169–176

Contents lists available at ScienceDirect

Fisheries Research

journa l homepage: www.e lsev ier .com/ locate / f i shres

study of the population structure of the Pacific sardine Sardinops sagax (Jenyns,842) in Mexico based on morphometric and genetic analyses

rancisco Javier García-Rodrígueza,∗, Silvia Alejandra García-Gascab, José De La Cruz-Agüeroa,íctor Manuel Cota-Gómeza

Centro Interdisciplinario de Ciencias Marinas, Instituto Politécnico Nacional, Departamento de Biología Marina y Pesquerías, Colección Ictiológica, Av. Instituto Politécnico Nacional/n, Col. Playa Palo de Santa Rita 23096, Apdo, Postal 592, La Paz, Baja California Sur, MexicoCentro de Investigación en Alimentación y Desarrollo, A.C. Av. Sábalo – Cerritos s/n, Estero del Yugo, Apdo, Postal P 711, Mazatlán, Sinaloa, Mexico

r t i c l e i n f o

rticle history:eceived 25 April 2010eceived in revised form 4 November 2010ccepted 4 November 2010

eywords:acific sardinesardinopsitochondrial DNA

a b s t r a c t

Several studies on the Pacific sardine Sardinops sagax have focused on the identification of stock compo-sition and boundaries, using morphometric and genetic analysis. In this study, geometric morphometricbody landmarks and control region mtDNA sequences were used to examine the population structureof sardines along the Pacific coast of the Baja California Peninsula. Samples from commercial landingsin Ensenada (ENS), Baja California, and Bahia Magdalena (BM), Baja California Sur, were obtained during2006–2007. The population hypotheses tested were based on the distribution of sea surface temperature(SST) along the coast, which was previously used to define stocks. A total of 275 sardines from ENS and119 from BM were used in morphometric analysis. Fifty-three sequences from ENS and 106 from BM were

ontrol regionenetic differentiationorphometric analysisexico

used for genetic comparisons. Morphometric results showed differences among the three groups basedon SST, suggesting the existence of different morphotypes. Percentage of molecular variance explainedby the differences among three groups was significantly different from zero. However, the distribution ofhaplotypes in the groups did not show a clear phylogeographic pattern. Additionally, mismatch distribu-tions supported relatively similar historical demographic events in the three groups. Although evidenceof phenotypic groups along the Pacific coast of the Peninsula was found, current molecular data did not

nce o

clearly support the existe

. Introduction

Biological and ecological knowledge about natural resources iselevant for devising management strategies, especially in speciesith important conservation or commercial status. The Pacific

ardine Sardinops sagax (Jenyns, 1842) is distributed from theoutheastern coast of Alaska to the northwestern coast of Mexico,ncluding the Gulf of California (Kramer and Smith, 1971). It is one ofhe most important schooling pelagic species along the west coastf North America and is captured in Mexico, near Ensenada, Bahiaagdalena, and Guaymas (Lluch-Belda et al., 1986). The Pacific

ardine is a commercially valuable species. Capture records showuctuations over time, with a near collapse during the mid 20thentury (D.F.O., 2004).

Several studies on the Pacific sardine have focused on the iden-ification of stocks. This is relevant because specific biological andcological information of each stock is used to implement harvesttrategies aimed at achieving sustainable exploitation. Stock struc-

∗ Corresponding author. Tel.: +52 612 12 25344; fax: +52 612 12 25322.E-mail address: fjgarciar@ipn.mx (F.J. García-Rodríguez).

165-7836/$ – see front matter © 2010 Elsevier B.V. All rights reserved.oi:10.1016/j.fishres.2010.11.002

f a phylogeographically structured population.© 2010 Elsevier B.V. All rights reserved.

ture studies on the Pacific sardine in Mexican and California havebeen based on various kinds of information, including tagging infor-mation (Clark, 1945) vertebral counts (Clark, 1947; Wisner, 1960),spawning areas (Marr, 1960), blood groups (Sprague and Vrooman,1962; Vrooman, 1964), size-at-age (Wolf and Daugherty, 1964),morphometric data (De La Cruz-Agüero and García-Rodríguez,2004), genetic analysis (Hedgecock et al., 1989; Lecomte et al.,2004; Gutiérrez-Flores, 2007) and cohort analysis (Félix-Uragaet al., 1996).

Relevant information about the movements of fish and abun-dances of the Pacific sardine populations was obtained from theintensive tagging program carried out between 1936 and 1944(Clark, 1945). This study found that fish tagged north of Bahia Sebas-tian Viscaino, Mexico, were caught at northern sites, supportingthe existence of only one stock with a distribution from BritishColumbia to northern and central Baja California. No evidence wasfound of movement toward the north of individuals tagged in Bahia

Magdalena (Clark, 1945). Studies carried out on the seasonal andgeographic distributions of larvae show that the principal spawningareas are centered in three areas (Marr, 1960): off Central Californiain April (Lynn, 2003); near Bahia Magdalena, Baja California Sur, insummer; and in the Gulf of California in fall and winter (Aceves-

170 F.J. García-Rodríguez et al. / Fisheries Research 107 (2011) 169–176

USA

PacificOcean

MEXICO

Gulf of

California

1. Ensenada (ENS)2. Bahia Magdalena (BM)

1Northernzone

Southernzone 2

M(

mtat(iTcaiwtcL

mworTea

2

2

llazo(

Table 1Sampling sites. Data sets are grouped according to sampling site and sea surfacetemperature (SST). Upper data were used for morphometric analysis. Lower datawere used for genetic analysis. Cold Ensenada (CEN), Temperate Bahia Magdalena(TBM) and Warm Bahia Magdalena (WBM).

Sites Date SST Groups n Total

Morphometric dataNorthern zone Jan 2007 15.3 CEN 104

Feb 2007 15.0 CEN 31Mar 2007 15.1 CEN 82Apr 2007 15.0 CEN 58 275

Southern zone Jan 2007 22.1 WBM 30Feb 2007 20.2 TBM 12Mar 2007 19.5 TBM 33Apr 2007 18.7 TBM 28May 2007 18.0 TBM 16 119

394Genetic dataNorthern zone – – 6

Jan 2007 15.3 CEN 24Feb 2007 15.0 CEN 15Mar 2007 15.1 CEN 4Apr 2007 15.0 CEN 4 53

Southern zone Jun 2006 18.9 TBM 27Aug 2006 26.5 WBM 19Oct 2006 28.0 WBM 11Nov 2006 26.8 WBM 15Dec 2006 24.3 WBM 26Jan 2007 22.1 WBM 2Feb 2007 20.2 TBM 2

The left side of each individual sardine was photographed bythe same person using a digital camera. A ruler was placed nextto each specimen to obtain scaling information. Since the land-marks alone were insufficient to achieve a good representation of

118° 116° 114° 112° 110°

Fig. 1. Sampling sites: 1 – Ensenada (ENS) and 2 – Bahia Magdalena (BM).

edina et al., 2004). This supports the existence of three stocksSmith, 2005).

In Mexico, De La Cruz-Agüero and García-Rodríguez (2004) usedultivariate morphometric analysis on sardines collected from

wo sites on the western coast of the Baja California Peninsuland found significant differences among the samples. Based onemperature-at-catch and otolith morphometry, Félix-Uraga et al.2004, 2005) found evidence of three stocks of Pacific sardines rang-ng from Bahia Magdalena, Mexico, to San Pedro, California, USA.hese authors found evidence of three stocks: one stock was asso-iated with cold waters (13–17 ◦C, distributed mainly in California),nother was related to temperate waters (17–22 ◦C, largely inhabit-ng the west coast of the Baja California Peninsula) and a third stock

as associated with warm waters (>22 ◦C, concentrated mainly inhe Gulf of California). In contrast, various molecular markers indi-ate a lack of population differentiation (Hedgecock et al., 1989;ecomte et al., 2004; Gutiérrez-Flores, 2007).

The identification of stock structure is relevant to stock assess-ents and harvest management (Emmett et al., 2005). In this studye compared sardines from two sites located along the west coast

f the Baja California Peninsula using two approaches: geomet-ic morphometric and mitochondrial DNA (mtDNA) sequencing.wo hypotheses were tested: (1) there are no significant differ-nces between sampling zones, and (2) there are no differencesssociated with sea surface temperature (SST).

. Material and methods

.1. Sampling

The Pacific sardines used in the present study were col-ected between June 2006 and May 2007 from commercialandings in Ensenada (ENS), Baja California (northern zone),

nd Bahia Magdalena (BM), Baja California Sur (southernone), Mexico (Fig. 1, Table 1). Monthly SST data werebtained from the NOAA OceanWatch-Central Pacific websitehttp://oceanwatch.pifsc.noaa.gov:8080/thredds/dodsC/pfgac/)

Mar 2007 19.5 TBM 2Apr 2007 18.7 TBM 2 106

159

for areas near ENS and BM (Fig. 2). As suggested for stock dis-crimination by Félix-Uraga et al. (2004, 2005), the organizationof morphometric and genetic data sets was based on the limits ofSST: CEN (sardines from Ensenada associated with cold water),TBM (sardines from Bahia Magdalena associated with temperatewater) and WBM (sardines from Bahia Magdalena associated withwarm water) (Table 1, Fig. 3). Although the sampling was designedto obtain a representative number of individuals from each zoneand SST stock, logistical problems made obtaining the samplesoccasionally difficult.

2.2. Morphometric analysis

Fig. 2. Variation in the sea surface temperature (SST) in Ensenada (line with dia-mond) and Bahia Magdalena (line with square) from June 2006 to May 2007.Horizontal lines represent the stock limits according to SST previously suggested(Félix-Uraga et al., 2005).

F.J. García-Rodríguez et al. / Fisheries Research 107 (2011) 169–176 171

F ace teD 5).

ttahpbstollfit(ctSrpStl

wtgyGhC

Fprb

ig. 3. Sampling zones along of the year and groups defined by zone and sea surfotted lines represent the stock limits previously suggested (Félix-Uraga et al., 200

he shape, two templates of the digital image were constructedo provide guidelines of equal angular spacing to identify pointslong the body curves using the program MakeFan (H.D. Sheets,ttp://www2.canisius.edu/∼sheets/morphsoft.html). First, a tem-late was constructed based on the landmarks at the end of theranchiostegals rays, on the snout, and at the origin of the dor-al fin. A second template was based on landmarks located athe origin of the anal fin, the end of the dorsal fin, and therigin of the upper lobe of the caudal fin. Additional points (semi-andmarks) were digitized at the intersection of the curve and theines of the templates. Thus, we constructed the digitized con-gurations using 18 points, 16 along the contour and two onhe side of the fish (Fig. 4) using the program TpsDig Ver 1.4Rohlf, 2004). A superimposition method based on generalized Pro-rustes analysis (GPA) was used to remove differences attributedo the position, orientation, and scale between configurations.emi-landmarks were submitted to the alignment algorithm toeduce effects of the arbitrary selection of a limited number ofoints to represent the curves, using the program SemiLand6 (H.D.heets, http://www2.canisius.edu/∼sheets/morphsoft.html). Oncehe semi-landmarks were aligned, they were treated as points inandmark data sets.

Partial warp scores (the contribution that each partialarp makes to the total deformation) were obtained from

he Thin Plate Spline interpolation function using IMP pro-rams. They were subjected to a Principal Component Anal-

sis (PCA) and Canonical Variates Analysis (CVA) using PCA-en6n and CVAGen6m software, respectively (IIMP, H.D. Sheets,ttp://www2.canisius.edu/∼sheets/morphsoft.html). We used thehi square procedure in the PCAGen6n program to test whether

1

23

4 5 67

89

1011

121314

1516

17

18

ig. 4. Schematic representation of the Pacific sardine showing points used for mor-hometric geometric analysis. Points 2, 3, 4, 5, 8, 9 and 10 were found using twoeference systems. The first was based on points 1, 6 and 16, and the second wasased on points 7, 11 and 14.

mperature (SST) in Ensenada and Bahia Magdalena from June 2006 to May 2007.

the principal components (PC) had significantly different vari-ances (Anderson, 1958). PC significant scores were used to comparegroups using ANOVA. A matrix of the assignments was constructedto complement previous analysis by assigning each specimen toone of the known groups (based on the Mahalanobis distance fromthe specimen to the mean value of the nearest group).

Since the CVA suggested significant differences among groups,partial Procrustes distance means (PPDMs) were calculated toperform paired comparisons. The significance of the test wasbased on bootstrapping to determine whether the observedF-value could have been produced by chance, taking intoaccount the distribution of bootstrapped F-values. This analy-sis was carried out using the TwoGroup6 software (IMP, H.D.Sheets, http://www2.canisius.edu/∼sheets/morphsoft.html). Dis-tances obtained were used to construct an unrooted tree basedon Neighbor-Joining (NJ) using Phylip Ver 3.6 (Felsenstein, 2005).Finally, the thin-plate spline interpolating functions were used tovisualize shape changes.

2.3. Genetic analysis

Caudal fin samples were collected in 1.5 mL microtubes contain-ing absolute ethanol and stored at −20 ◦C until laboratory analysis.Total DNA was isolated by taking ∼0.5 g of caudal fin and usingthe “salting out” method (Miller et al., 1988). Isolated DNA was re-suspended in 100 �L deTE (Tris–EDTA pH 8.0). A fragment of thecontrol region (D-loop) of mtDNA was amplified using the primersreported by Bowen and Grant (1997). PCR amplification was car-ried out in 12.5 �L reactions containing 1× PCR buffer with 1.5 mMMgCl2 (Clontech), 131.25 �M of each dNTP, 0.4 �M of each primer,0.5 U Advantage Taq DNA polymerase (Clontech), and 1 �L of DNA.The PCR setup consisted of an initial denaturation step at 94 ◦Cfor 2 min, followed by 35 cycles at 94 ◦C for 1 min, 55 ◦C for 1 min,and 68 ◦C for 2 min. Amplification products were purified using theQiagen MiniElute kit following the instructions suggested by themanufacturer, and sequenced using an automated DNA sequencer(LICOR IR2).

Sequences were aligned and edited using the BioEdit soft-

ware (Hall, 1999), which uses the ClustalW algorithm. Haplotypeand nucleotide diversity were calculated for each data set usingArlequin 3.0 (Excoffier et al., 2005). Population genetic struc-ture was analyzed using the Analysis of Molecular Variance(AMOVA), which estimates the proportion of genetic variation

1 heries Research 107 (2011) 169–176

wbouAmsm(fdd3ppomum

3

3

i8gPACndtewbvLCcT

bPPf

Fig. 5. Distribution of scored frequencies obtained from the CVA in three groups.Black circles represent sardines from the CEN group; gray circles represent sardinesfrom the TBM group; and white circles represent sardines from the WBM group.

Table 2Percent population assignment based on Mahalanobis distance. Original groups arefound along rows. Cold Ensenada (CEN), Temperate Bahia Magdalena (TBM) andWarm Bahia Magdalena (WBM).

CEN TBM WBM n

TNt

72 F.J. García-Rodríguez et al. / Fis

ithin and among populations. Information on the differencesetween haplotypes for the AMOVA was obtained from a matrixf Euclidean squared distances, and its significance was testedsing non-parametric permutation procedures as implemented inrlequin 3.0. Molecular pairwise ˚ST (analogous to FST) was esti-ated to evaluate genetic differentiation between pairs of data

ets. ˚ST was also carried out using the Arlequin 3.0. A mini-um spanning network was constructed with Network 4.2.0.1

www.fluxusengineering.com/sharenet.htm), based on haplotyperequencies to search for phylogeographic structure. Historicalemographies for each data set were estimated with mismatchistributions (Rogers and Harpending, 1992) using the Arlequin.0. A unimodal distribution suggests rapid growth from a smallopulation size, while a multimodal distribution reflects long-termopulation stability. The expansion model was tested using the sumf square deviations (SSD) between the observed and the expectedismatch. The P value was based on the number of SSD, calculated

nder simulation larger or equal to the observed SSD as imple-ented in Arlequin 3.0.

. Results

.1. Morphometric analysis

We analyzed 394 sardines from the west coast of the Baja Cal-fornia Peninsula. A total of 275 sardines were obtained for CEN,9 for TBM, and 30 for WBM. ANOVA of PCA scores of the threeroups based on SST showed morphometric differences. Scores ofC1 were statistically different among groups (F = 7.641, P = 0.001).Tukey test indicated that WBM was statistically different from

EN (P = 0.0003) and TBM (P = 0.006). CEN and TBM were not sig-ificantly different (P = 0.675). Scores of PC2 also were statisticallyifferent among groups (F = 58.53, P < 0.05). The Tukey test forhe PC2 indicated that the CEN group was significantly differ-nt from the other two groups (P < 0.05). TBM and WBM scoresere not significantly different (P = 0.859). The variance explained

y the CV1 was 72% and 18% was for CV2. The two canonicalariables indicated significant differences between groups (Wilk’sambda = 0.321, P < 0.05 for CV1; Wilk’s Lambda = 0.691, P < 0.05 forV2) (Fig. 5). Assignment based on the Mahalanobis distances indi-ated a high percentage of discrimination among the three groups.he WBM morphotype showed a minor discrimination (Table 2).

Analysis based on the F-test indicated significant differencesetween each paired Procrustes distances mean (PPDM) (F = 13.67,= 0.0011, PPDM = 0.0111, for CEN–TBM; F = 14.64, P = 0.0011,PDM = 0.0178, for CEN–WBM; F = 6.94, P = 0.0011, PPDM = 0.0142,or TBM–WBM). The divergence morphometric information based

able 3ucleotide substitutions for Parsimoniosus sites in Northern and Southern sampling sites

o the site in the 500 bp sequence. Dashes represent similarities to the consensus (cons).

noitisoP

HAPLOTYPE

1 1 1 1 1 1 2 2 2 2 2 2 2 2 3 3 3 3 3 4 4 3 6 6 0 0 0 1 2 6 0 2 2 3 6 6 7 7 2 2 3 6 7 4 5 7 5 8 6 8 9 0 5 9 5 2 9 8 2 5 3 8 6 9 4 5 7 5 2 N O

#cons C C T T C T G T A A G G G G G A T A A G C G A G #H65 - - - - T - - - - - A - - - - - - - - - - - - - #H101 - T C - T - - - G G - - - - - - C - G A T - - - #H116 - - - - - C A - - - - - - - - - - - - A - - - - #H133 - - - - - - A - G G - - A - - - - - G A T - - - #H134 - - C C - - A - - G A A - - - - - - G - T A - - #H135 - - C - - C A - - - A - A - - - C G G A T - - - #H140 - - - - T - - - G G A - - A - - - - G A T - - - #H141 - T - - - C A C - G A - - A - G - - - - - - G - #H142 - - - - - - - C - G - - - - - - C - - A - A - - #H143 - - C - T C A - G - - - A - - - - - G - T A G -

CEN 88.4 4.0 7.6 275TBM 3.3 86.7 10.0 30WBM 13.5 6.7 79.8 89

on PPDM suggested that the WBM morphotype was relatively moredifferent from the other two groups (Fig. 6). The morphologicalchanges based on partial deformation showed that sardines fromthe northern zone (CEN) tended to have a less depressed shape thatthose found toward the southern zone (WBM, Fig. 6).

3.2. Genetic analysis

A 500 pb fragment from the control region of mtDNA wasobtained. Twenty-four variables sites defined 146 haplotypesamong 159 specimens. This high genetic variability translated intolarge values of haplotype diversity (h = 0.999). No haplotype wasshared between the two zones and the number of frequent haplo-types (those occurring in more than one individual) occurred morein the northern zone (8) than in the southern zone (2) (Table 3).

To test congruence of our results with those suggesting threestocks in Mexico (Félix-Uraga et al., 2004, 2005), samples weregrouped in the same manner as in the morphometric analysis,according to both sampling sites and SST (Table 1). Six individ-

based in a fragment of the control region of mtDNA. Position numbers correspond

senoZ

R T H S O U T H 4 2 3 2 2 2 2 2 2 2

F.J. García-Rodríguez et al. / Fisheries Research 107 (2011) 169–176 173

F otype. Morphotype WBM was morphometrically the most different. Sardines from then ern zone (WBM). Vectors represent the direction and magnitude of deformations takinga

uwstsg

tw˚˚fbs

atSthCatT

Table 5Genetic diversity of DNA sequences used in genetic studies of the Pacific sardine.

Mitochondrial gene H � Reference

Cytochrome b 0.89 0.0050 Lecomte et al. (2004)

TST

ig. 6. Tree of morphometric divergence and configuration means of each morphorthern zone (CEN) showed a less depressed body shape than those from the souths reference the overall mean configuration (black dots).

als were excluded from the analysis because the sampling dateas unknown. Therefore, this genetic analysis was based on 153

equences (47 from CEN, 33 from TBM, and 73 from WBM). All ofhe excluded specimens had unique haplotypes. Nucleotide diver-ity was relatively larger in the CEN group and smaller in the WBMroup (Table 4).

AMOVA revealed significant genetic differences amonghe three groups (˚ST = 0.02903, P < 0.001). Pairwise FSTas also significantly different between groups (P < 0.001;ST (CEN–TBM) = 0.03907, ˚ST (CEN–WBM) = 0.02632, andST(TBM–WBM) = 0.02615). However, a phylogeographic pattern

rom total haplotypes was not apparent, as the associationsetween clades and particular groups were unclear (data nothown).

Mismatch distributions for the three groups were unimodalnd the sudden expansion model fitted all mismatch distribu-ions (SSD = 0.005, P = 0.058 for CEN; SSD = 0.004, P = 0.281 for TBM;SD = 0.003, P = 0.068 for WBM) (Fig. 7). Taking into account thathe nucleotide diversity of the control region could be 3.6 times

igher than the Cytb (Table 5), and that the divergence rate of theyb can be regarded as 2% per million years (Lecomte et al., 2004),rough estimate of the divergence rate of the control region in

he Pacific sardine could be approximately 7.2% per million years.hus, considering a generational time of 4.4 years for sardines

able 4amples size (n), number of haplotypes (nh), haplotype diversity (h) and nucleotide divemperate Bahia Magdalena (TBM) and Warm Bahia Magdalena (WBM).

Stocks n nh Mean number of pairwise d

CEN 47 38 9.35 ± 4.37TBM 33 30 9.01 ± 4.26WBM 73 73 8.51 ± 3.98

NAD6 0.94 0.0097 Gutiérrez-Flores (2007)NAD5 0.96 0.0101 Gutiérrez-Flores (2007)Control Region 0.99 0.0182 Present study

(Murphy, 1967; Butler et al., 1993) and the tau values (�, muta-tional timescale) estimated in the present study, 9.71, 9.29 and8.64 for CEN, TBM and WBM respectively, the beginning of sud-den expansion could have happened between ∼282,000 (95% CIbetween 243,000 and 311,000), and ∼317,000 (95% CI between260,000 and 362,000), years ago.

4. Discussion

Geometric and genetic analyses have been used as alternative

and robust tools for the discrimination of biological groups (De LaCruz-Agüero and García-Rodríguez, 2004; Bowen and Grant, 1997).A combination of both methods could provide a better understand-ing about the processes affecting that discrimination. In this study,Pacific sardines from the Baja California Peninsula were compared

ersity (�) and standard deviation (SD) for each SST-group. Cold Ensenada (CEN),

ifferences H ± SD � ± SD

0.991 ± 0.0064 0.018709 ± 0.0097100.992 ± 0.0104 0.018015 ± 0.0094651.000 ± 0.0023 0.017029 ± 0.008828

174 F.J. García-Rodríguez et al. / Fisheries

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

200

180

160

140

120

100

80

60

40

20

0

120

100

80

60

40

20

0

450

400

350

300

250

200

150

100

50

0

Number of pairwise differences

Fre

qu

en

cy

CEN

TBM

WBM

FBa

uy

fwhs1Vapd(sna

tGaflpEziVww

ig. 7. Mismatch distribution from the control region for the three morphotypes.ars represent observed distribution and lines represent expected distributionccording to the sudden expansion model.

sing geometric morphometric analysis and mtDNA sequence anal-sis to identify morphological and/or genetic groups.

Morphometric differences were found between individualsrom two sites (ENS–BM). Such morphotypes may be associatedith biologically distinct groups in the Mexican Pacific. This ideaas been discussed by several authors, who based their conclu-ions on data from several methodological tools (Clark, 1945,947; Marr, 1960; Wisner, 1960; Sprague and Vrooman, 1962;rooman, 1964; Wolf and Daugherty, 1964; De La Cruz-Agüerond García-Rodríguez, 2004; Félix-Uraga et al., 2004, 2005). Mor-hometric analysis using corporal distances revealed significantifferences between sardines from the Baja California PeninsulaDe La Cruz-Agüero and García-Rodríguez, 2004). The authorsuggested that these morphometric differences resulted from phe-otypic plasticity in the Pacific sardine, considering the absence ofllozyme-frequency differences (Hedgecock et al., 1989).

Our data also indicated morphometric differences betweenhree groups separated by location and sea surface temperature.enetic analyses based on AMOVA support this result, suggestinggenetically structured population resulting from a limited geneow. These findings may be explained by the heterogeneous dis-ersion of larvae and adults. Several studies suggest that Puntaugenia (28◦N) in the Baja California Peninsula is a transition

one that sets population boundaries for several species, includ-ng the Pacific sardine (Clark, 1947; Hubbs, 1960; Vrooman, 1964;alentine, 1966). This zoogeographical limit has been associatedith oceanographic processes such as the Davidson Current (a pole-ard flowing counter current to California current system) and

Research 107 (2011) 169–176

semi-permanent eddies that limit the distribution of some organ-isms (Hewitt, 1981).

Our results could indicate the existence of several genetic pop-ulations and suggest that previous studies based on molecular data(Hedgecock et al., 1989; Lecomte et al., 2004; Gutiérrez-Flores,2007) failed to discern a genetic population structure by usingmtDNA regions having less variability than the Control Region. Thehigh mutation rate of the control region provides more opportu-nity for drift to vary allele frequencies, so a more powerful analysiscan be seen here than with prior mtDNA studies too. The use ofmicrosatellite markers also failed to identify populations based onSST groups (Gutiérrez-Flores, 2007). However, mtDNA has a lowereffective population size than nuclear markers, so it is expected tobe more sensitive to barriers to gene flow than the nuclear mark-ers used previously. Lack of phylogeographical structure could beexplained because the populations are not total isolated from oneanother, so that lineage sorting is not occurring.

Alternatively the differences could be related to causes otherthan the existence of a structured population. According to Waples(1998) the rejection of the null hypothesis (no population differ-entiation) in species with high gene flow can be associated withthe selection of the alpha level (Type I error), biologically insignif-icant differences, or the violation of assumptions about samplinginstead of biologically important differences. Pacific sardine popu-lations, like those of many marine species, are challenging to definebecause of their large sizes and because high levels of dispersalproduce only weak phylogeographic pattern, if at all.

Based on the above, an alternative explanation for our resultsmay be related to mechanisms producing “Chaotic patchiness”(Hedgecock, 1994). This situation is related to the occurrenceof a slight but significant local or microgeographic populationstructure despite a large potential for gene flow between subpop-ulations. It may be explained either by differential survival of fishwith particular genotypes after recruitment, or by variation in thegenetic composition of recruits. Selection along an environmentalgradient may lead to post-recruitment differences among subpop-ulations. Alternatively, a large variance in reproductive successcould lead to pre-recruit genetic heterogeneity. Although manycases of chaotic genetic patchiness are described in invertebratesanimals (Larson and Julian, 1999), instances of chaotic genetic het-erogeneity chaotic have been suggested from fishes, specificallyfrom northern anchovy Enqraulis mordax since a lack geographi-cal pattern within of the central stock of northern anchovy wasevidenced (Hedgecock et al., 1994). Similar processes could occurin Pacific sardine considering that both species show a relativelysimilar population dynamics, associated with large expansions andcontractions of range with changes in abundance in response toclimate change (Lluch-Belda et al., 1989).

In addition, small sample sizes and the high level of geneticdiversity in the control region of mtDNA may have provided onlysmall amounts of statistical power. The high haplotype diversityindicates that a much more intense sampling strategy is needed totest for genetic differences on a small spatial scale. The high diver-sity found in the present study also makes it difficult to undertakea monthly analysis due to the limited sample size. An additionaleffort at increasing the amount of data in order to do a temporaland a more geographically detailed study, and applying a sam-pling design based on the existing knowledge of the resource, couldstrengthen the analysis of the population structure.

Analyses using other molecular markers with different muta-tion rates have been performed and found no evidence of genetic

population structure (Hedgecock et al., 1989; Lecomte et al., 2004;Gutiérrez-Flores, 2007). Those results are relevant since they havebeen based on molecular markers with different degrees of poly-morphism (Koehn et al., 1980; Karl and Avise, 1992; Pogson et al.,1995), as can be seen from their different mutation rates (Table 5).

heries

Tdane(fglf

apnmtsSe

tr2ssganatfitrec

mtnPizoi

A

gSTIClv

R

A

AB

B

F.J. García-Rodríguez et al. / Fis

he weak phylogeographic structure detected in this study may beue to the high migratory ability of adult sardines (Clark, 1945)nd to the highly dynamic meso-scale currents in the Califor-ia Current System (Maluf, 1983; Kessler, 2006), which dispersesggs and larvae. The Pacific sardine is a multiple-batch spawnerTorres-Villegas, 1997), thus the potential for random genetic dif-erentiation from genetic drift may be reduced by overlappingenerations. Retention eddies, on the other hand, may localizearvae long enough to produce both detectable morphological dif-erences among areas and chaotic genetic structure.

The demographic history, inferred from mismatch distribution,lso suggests similar evolutionary events among the three mor-hotypes, although weak demographic expansion gradients wereoticed from North to South (Fig. 7). A right-shifted unimodalismatch distribution found toward the northern zone suggests

hat the northern group represents an older demographic expan-ion than southern groups (see Rogers and Harpending, 1992).imilar northern-to-southern gradients were found by Lecomtet al. (2004).

Our results suggest that at least three morphotypes occur onhe western coast of the Baja California Peninsula, which may cor-espond to the three stocks proposed by Félix-Uraga et al. (2004,005). Moreover, the morphometric data suggest that the mostimilar morphotypes were CEN and TBM. The morphotype pre-umably originating from the Gulf of California (WBM) showed thereatest morphological differentiation. In agreement with previousnalyses, morphological differences may be more related to phe-otypic plasticity than to the genetic variation (De La Cruz-Agüerond García-Rodríguez, 2004). Higher discrimination of morpho-ype WBM, should therefore, be associated with environmentalactors in the Gulf of California. Common events caused by the Cal-fornia Current and other oceanographic processes may promotehe greater morphological similarity between CEN and TBM. Theesults obtained from the partial warps analysis suggest an appar-nt morphological north-to-south pattern; individuals were moreompressed in the Gulf of California than in the Pacific.

Future analyses should be focused in the identification oforphologic criteria to distinguish individuals of different morpho-

ypes. Since phylogeographic structure is not clear, it is presentlyot possible to support the idea of a limited gene flow of theacific sardine along its distribution area. Future genetic compar-sons considering samples from three different zones (northernone, southern zone, and Gulf of California) grouped on the limitsf SST, and larger sample sizes for comparing temporal variationntra SST stock should clarify our results.

cknowledgements

This study was funded by grants from the Secretaría de Investi-ación y Posgrado – Instituto Politécnico Nacional (SIP-20071113,IP-20080573 and SIP-20090333) and with the support of J.R.orres-Villegas. FJGR and JDA thank the grants by EDI-IPN, COFAA-PN, and SNI-CONACYT. We thank the “Colección Ictiológica”,ICIMAR-IPN. We thank Stewart Grant for commenting on an ear-

ier draft of the paper and three anonymous reviewers for theiraluable suggestions and criticism.

eferences

ceves-Medina, G., Jimez-Rosenberg, S.P.A., Hinojosa-Medina, A., Funes-Rodriguez,R., Saldierna-Martinez, R.J., Smith, P.E., 2004. Fish larvae assemblages in the Gulfof California. J. Fish Biol. 65, 832–847.

nderson, T.W., 1958. An Introduction to Multivariate Analysis. Wiley.

owen, B.W., Grant, W.S., 1997. Phylogeography of the sardines (Sardinops

spp.): assessing biogeographic models and population histories in temperateupwelling zones. Evolution 51, 1601–1610.

utler, J.L., Smith, P.E., Lo, N.C.H., 1993. The effect of natural variability of life-history parameters on anchovy and sardine population growth. CalCOFI Rep.34, 104–111.

Research 107 (2011) 169–176 175

Clark, F.N., 1945. Results of tagging experiments in California Waters on Sardine(Sardinops caerulea). Fish. Bull. 61, 93pp.

Clark, F.N., 1947. Analysis of populations of the Pacific sardine on the basis of verte-bral counts. Fish. Bull. 65, 26pp.

D.F.O., 2004. Pacific Sardine. DFO Can. Sci. Advis. Sec. Stock Status Report 2004/037.De La Cruz-Agüero, J., García-Rodríguez, F.J., 2004. Morphometric stock structure

of the Pacific sardine Sardinops sagax (Jenyns, 1842) off Baja California, Mexico.In: Elewa, A.M. (Ed.), Morphometrics: Applications in Biology and Paleontology.Springer-Verlag, New York, NY, pp. 115–124.

Emmett, R.L., Brodeur, R.D., Miller, T.D., Pool, S.S., Krutzikowsky, G.K., Bentley, P.J.,Mccrae, J., 2005. Pacific sardine (Sardinops sagax) abundance, distribution, andecological relationships in the Pacific northwest. CalCOFI Rep. vol. 46, 122–143.

Excoffier, L., Laval, G., Schneider, S., 2005. Arlequin ver 3.0: An integrated softwarepackage for population genetics data analysis. Evol. Bioinform. Online 1, 47–50.

Félix-Uraga, R., Alvarado, R.M., Carmona, R., 1996. The sardine fishery along the westcoast of Baja California, 1981–1994. CalCOFI Rep. 37, 188–192.

Félix-Uraga, R., Gómez-Munoz, V.M., García-Franco, W., 2004. On the existence ofPacific sardine groups off the West coast of Baja California and Southern Cali-fornia. CalCOFI Rep. 45, 146–151p.

Félix-Uraga, R., Gómez-Munoz, V.M., Quinonez-Velázquez, C., Melo-Barrera, F.N.,Hill, K.T., García-Franco, W., 2005. Pacific sardine (Sardinops sagax) stock dis-crimination off the West coast of Baja California and Southern California usingotolith morphometry. CalCOFI Rep. 46, 113–121.

Felsenstein, J., 2005. PHYLIP (Phylogeny Inference Package) version 3.6. Distributedby the author. Department of Genome Sciences, University of Washington, Seat-tle.

Gutiérrez-Flores, C., 2007. Estructura genética poblacional de la sardina del pacíficonororiental Sardinops sagax caeruleus. Tesis de Maestría, CICESE, Ensenada, B.C,p. 112.

Hall, T.A., 1999. BioEdit: a user-friendly biological sequences alignment edit andanalysis program for Windows 95/98/NT. Nucleic Acids Sympos. Ser. 41, 95–98.

Hedgecock, D., 1994. Temporal and spatial genetic structure of marine animal pop-ulations in the California Current. CalCOFI Rep. 35, 73–81.

Hedgecock, D., Hutchinson, E.S., Li, G., Sly, F.L., Nelson, K., 1989. Genetic and morpho-metric variation in the Pacific sardine, Sardinops sagax caerulea: comparisonsand contrasts with historical data and with variability in the northern anchovy,Engraulis mordax. U.S. Fish Bull. 87, 653–671.

Hedgecock, D., Hutchinson, E.S., Li, G., Sly, F.L., Nelson, K., 1994. The central stockof northern anchovy (Engraulis mordax) is not a randomly mating population.CalCOFI Rep. 35, 121–136.

Hewitt, R., 1981. Eddies and speciation in the California Current. CalCOFI Rep. XXII,96–98.

Hubbs, C.L., 1960. The marine vertebrates of the outer coast, Symp: the biogeographyof Baja California and Adjacent Seas. Syst. Zool. 9 (3 and 4), 134–147.

Karl, S., Avise, J.C., 1992. Balancing selection at allozyme loci in oyster: implicationsfrom nuclear RFLPs. Science 256, 100–102.

Kessler, W.S., 2006. The circulation of the eastern tropical Pacific: a review. Prog.Oceanogr. 69, 181–217.

Koehn, R.K., Newell, R.I.E., Immerman, F.W., 1980. Maintenance of an aminopepti-dase allele frequency cline by natural selection. Proc. Natl. Acad. Sci. U.S.A. 77,5385–5389.

Kramer, D., Smith, P.E., 1971. Seasonal and geographic characteristics of fisheryresources, California current. Region VII. Pacific Sardine Com. Fish. Rev. 33 (10),7–11.

Larson, R.J., Julian, R.M., 1999. Spatial and temporal genetic patchiness in marinepopulations and their implications for fisheries management. CalCOFI Rep. 40,94–99.

Lecomte, F., Grant, W.S., Dodson, J.J., Rodríguez-Sánchez, R., Bowen, B., 2004. Liv-ing with uncertainty: genetic imprints of climate shifts in East Pacific anchovy(Engraulis mordax) and sardine (Sardinops sagax). Mol. Ecol. 13, 2169–2182.

Lluch-Belda, D., Crawford, R.J.M., Kawasaki, T., MacCall, A.D., Parrish, R.H., Schwart-zlose, R.A., Smith, P.E., 1989. World-wide fluctuations of sardine and anchovystocks: the regime problem. S. Afr. J. Mar. Sci. 8, 195–205.

Lluch-Belda, D., Magallon, F.J., Schwartzlose, R.A., 1986. Large fluctuations in the sar-dine fishery in the Gulf of California: possible causes. CalCOFI Rep. 27, 136–140.

Lynn, R.J., 2003. Variability in the spawning habitat of Pacific sardine (Sardinopssagax) off southern and central California. Fish. Oceanogr. 12 (3), 1–13.

Maluf, L.Y., 1983. The physical oceanography. In: Case y, T.J., Cody, M.L. (Eds.), IslandBiogeography in the Sea of Cortez. University of California Press, Los Angeles,CA, pp. 26–44.

Marr, J.C., 1960. The causes of major variations in the catch of the Pacific sardine,Sardinops caerulea (Girard). In: Rosa, H., Murphy, G.I. (Eds.), Proceedings of theWorld Scientific meeting on the Biology of Sardines and Related Species, vol. III.Food and Agriculture Organization of the United Nations, pp. 667–791.

Miller, S.A., Dykes, D.D., Polesky, H.F., 1988. A simple salting out procedure forextracting DNA from human nucleated cells. Nucleic Acids Res. 16, 1215.

Murphy, G.I., 1967. Vital statistics of the Pacific sardine (Sardinops caeruea) and the

population consequences. Ecology 48, 731–736.

Pogson, G.H., Mesa, K.A., Boutilier, R.G., 1995. Genetic population structure and geneflow in the Atlantic cod, Gadus morhua: a comparison of allozyme and nuclearRFLP loci. Genetics 139, 375–385.

Rogers, A.R., Harpending, H., 1992. Population growth makes waves in the distribu-tion of pairwise genetic differences. Mol. Biol. Evol. 9, 552–569.

1 heries

R

S

S

T

76 F.J. García-Rodríguez et al. / Fis

ohlf, F.J., 2004. TpsDIG Version 1.40. Department of Ecology and Evolution, StateUniversity of New York at Stony Brook, New York.

mith, P.E., 2005. A History for subpopulation structure in the Pacific sardine(Sardinops sagax) population off Western North America. CalCOFI Rep. 46,75–82p.

prague, L.M., Vrooman, A.M., 1962. A racial analysis of the Pacific sardine (Sardinopscaerulea), based on stidies of erythrocyte antigens. Ann. N. Y. Acad. Sci. 97,131–138.

orres-Villegas, J.R., 1997. La reproducción de la sardina monterrey Sardinopscaerulea (Girard, 1854) en el Noroeste de México y su relación con el ambiente.Ph.D. thesis, Universidad Politécnica de Cataluna.

Research 107 (2011) 169–176

Valentine, J., 1966. Numerical analysis of marine molluscan ranges on the extrat-ropical northeastern Pacific shelf. Limnol. Oceanogr. 11, 198–211.

Vrooman, A.M., 1964. Serologically differentiated subpopulations of the Pacific sar-dine, Sardinops caerulea. J. Fish. Res. Board Can. 21, 691–701.

Waples, R.S., 1998. Separating the wheat from the chaff: patterns of genetic differ-

entiation in high gene flow species. J. Hered. 89, 438–450.

Wisner, R.L., 1960. Evidence of northward movement of stocks of Pacific sardinebased on the number of vertebrae. CalCOFI Rep. 8, 75–82.

Wolf, R., Daugherty, A.E., 1964. Age and length composition of the sardine catch offthe Pacific Coast of the United States and Mexico in 1961 and 1962. Calif. FishGame 50, 241–242.