DataPDF Available

The Pillars of Hercules as a bathymetric barrier to gene flow promoting isolation in a global deep-sea shark (Centroscymnus coelolepis)

January 2016

January 2016

Authors:

Diana Catarino

University of the Azores

Halvor Knutsen

Institute of Marine Research in Norway

A. Veríssimo

CIBIO Research Center in Biodiversity and Genetic Resources

Esben Moland Olsen

Institute of Marine Research in Norway, Flødevigen

Show all 14 authorsHide

List of sampling sites with number of Centroscymnus coelolepis specimens collected for genetic analyses (N), and genetic diversity indices for the microsatellites and mtDNA control region (CR) markers

…

Pairwise F-statistics (F ST and Φ ST ) and Jost's D est values between sampling sites for microsatellites (below the diagonals) and mtDNA control region (above the diagonals)

…

Median-joining network from 497bp mtDNA control region sequences. Size of circles (haplotypes) is proportional to the relative frequencies in the sample and colour coded according to sampling locality; abbreviations are given in Table 1. Each branch indicates single nucleotide substitutions, except when noted. Bootstrapping support of each branch is 100%, except when noted. *indicates an indel.

…

Barrier-inferred gene flow patterns based on (a) mtDNA and (b) 11 microsatellite loci. Thickness of the barriers (red lines) indicates their rank of importance and numbers indicate the percentage of support. Only barriers with support >60% are shown. Sampled sites abbreviations are given in Table 1.

…

Principal component analysis (PCA) based on 11 microsatellite loci. Circles around the sampling localities highlight the separation according to PCA 1 (P < 0.001). Sampled sites abbreviations are given in Table 1.

…

Figures - uploaded by Francis Neat

Content may be subject to copyright.

Content uploaded by Francis Neat

Content may be subject to copyright.

The Pillars of Hercules as a bathymetric barrier to gene

ﬂow promoting isolation in a global deep-sea shark

(Centroscymnus coelolepis)

DIANA CATARINO,*†HALVOR KNUTSEN,‡§¶ ANA VER



ISSIMO,**†† ESBEN MOLAND

OLSEN,‡§¶ PER ERIK JORDE,¶GUI MENEZES,*†HANNE SANNÆ S,‡DAVID STANKOVI



C,‡‡

JOAN BAPTISTA COMPANY,§§ FRANCIS NEAT,¶¶ ROBERTO DANOVARO,***††† ANTONIO

DELL’ANNO,*** BASTIEN ROCHOWSKI‡‡‡ and SE R GIO STE F ANN I †§§§

*MARE—Marine and Environmental Sciences Centre, Department of Oceanography and Fisheries, University of the Azores,

Rua Prof. Dr. Frederico Machado, Horta, Azores, Portugal, †IMAR-Institute of Marine Research, Department of Oceanography

and Fisheries, University of the Azores, Rua Prof. Dr. Frederico Machado, Horta, Azores, Portugal, ‡Institute of Marine

Research (IMR), Flødevigen N-4817 His, Norway, §University of Agder, Kristiansand N-4604, Norway, ¶Department of

Biosciences, University of Oslo, Centre for Ecological and Evolutionary Synthesis (CEES), PO Box 1066 Blindern, Oslo N-0316,

Norway, **CIBIO-U.P., Centro de Investigacß

ao em Biodiversidade e Recursos Gen

eticos, Campus Agr

ario de Vair~

ao, Rua Padre

Armando Quintas, Vair~

ao 4485-661, Portugal, ††College of William and Mary, Virginia Institute of Marine Science, Route

1208, Greate Road, Gloucester Point 23062, VA, USA, ‡‡Department of Life Sciences, University of Trieste, Via Licio Giorgieri

5, Trieste 34127, Italy, §§Institut de Ci

encies del Mar, CSIC, Passeig Mar

ıtim de la Barceloneta 37-49, Barcelona 08003, Spain,

¶¶Marine Laboratory, Marine Scotland-Science, PO Box 101, 375 Victoria Road, Aberdeen AB11 9DB, UK, ***Department of

Life and Environmental Sciences, Universit

a Politecnica delle Marche, Ancona 60131, Italy, †††Stazione Zoologica Anton

Dohrn, Villa Comunale, Naples, Italy, ‡‡‡School of BioSciences, The University of Melbourne, Parkville, Vic. 3010, Australia,

§§§CNR-ISSIA, Via de Marini 6, Genova 16149, Italy

Abstract

Knowledge of the mechanisms limiting connectivity and gene ﬂow in deep-sea ecosys-

tems is scarce, especially for deep-sea sharks. The Portuguese dogﬁsh (Centroscymnus

coelolepis) is a globally distributed and near threatened deep-sea shark. C. coelolepis pop-

ulation structure was studied using 11 nuclear microsatellite markers and a 497-bp frag-

ment from the mtDNA control region. High levels of genetic homogeneity across the

Atlantic (Φ

=0.0091, F

=0.0024, P>0.05) were found suggesting one large popula-

tion unit at this basin. The low levels of genetic divergence between Atlantic and Aus-

tralia (Φ

=0.0744, P<0.01; F

=0.0015, P>0.05) further suggested that this species

may be able to maintain some degree of genetic connectivity even across ocean basins. In

contrast, sharks from the Mediterranean Sea exhibited marked genetic differentiation

from all other localities studied (Φ

=0.3808, F

=0.1149, P<0.001). This ﬁnding sug-

gests that the shallow depth of the Strait of Gibraltar acts as a barrier to dispersal and that

isolation and genetic drift may have had an important role shaping the Mediterranean

shark population over time. Analyses of life history traits allowed the direct comparison

among regions providing a complete characterization of this shark’s populations. Sharks

from the Mediterranean had markedly smaller adult body size and size at maturity com-

pared to Atlantic and Paciﬁc individuals. Together, these results suggest the existence of

an isolated and unique population of C. coelolepis inhabiting the Mediterranean that

most likely became separated from the Atlantic in the late Pleistocene.

Keywords: barriers to dispersal, deep-sea shark, isolation, Mediterranean, population structure

Received 10 April 2014; revision received 29 October 2015; accepted 2 November 2015

Correspondence: Diana Catarino, Fax: +351 292200411;

E-mail: dianacatarino@uac.pt

Molecular Ecology (2015) 24, 6061–6079 doi: 10.1111/mec.13453

Introduction

The initial idea of large marine populations approximat-

ing panmixia (Caley et al. 1996), showing low genetic

differentiation (Palumbi 1992; Ward et al. 1994; Waples

1998), has recently been challenged (reviewed by Hau-

ser & Carvalho 2008; Salmenkova 2011). Ocean currents,

bathymetry, habitat fragmentation, limited dispersal

and migration have been shown to limit gene ﬂow and

connectivity for many species across their distribution,

promoting genetic divergence. Furthermore, past cli-

matic events such as ice ages have led to the strength-

ening of barriers and promoted vicariance (e.g.

Chevolot et al. 2006; Robalo et al. 2012). Despite some

progress made towards understanding the mechanisms

behind genetic structure in marine organisms, our

knowledge remains scarce regarding deep-sea species

(those living below 200 m depth). The logistic difﬁcul-

ties in sample collection, the limited number of speci-

mens available and taxonomic uncertainty, have

hampered population genetic studies in deep-sea taxa.

Spatial genetic divergence in deep-sea ﬁshes is

thought to be generally low or nearly absent (e.g. Ball

et al. 2000; White et al. 2009; Longmore et al. 2014).

However, there are a few notable exceptions of genetic

differentiation, as in the roundnose grenadier (Knutsen

et al. 2012), the bluemouth rockﬁsh (Aboim et al. 2005)

and the blackspot sea bream (Stockley et al. 2005). These

species differ from each other in their depth distribu-

tion and general life history, and thus, the barriers to

connectivity and mechanisms of differentiation are

likely to be different.

While the number of genetic studies performed in

coastal and pelagic sharks has increased (reviewed by

Dudgeon et al. 2012), few studies have focused on pat-

terns of genetic structure in deep-sea sharks (Straube

et al. 2011; Ver

ıssimo et al. 2011a, 2012; Cunha et al.

2012). These studies indicate that deep-sea sharks main-

tain high levels of gene ﬂow within ocean and some-

times across oceans basins, despite lacking a dispersive

pelagic early life stage (Conrath & Musick 2012). As dis-

tance seems less important in reducing genetic connec-

tivity, deep-sea sharks are an interesting group to test

hypotheses on other factors limiting connectivity.

Just as bathymetric features, such as troughs and

deep basins, can act as barriers to shallow water species

(Knutsen et al. 2009), shallow water passages, sills or

plateaus, are potential factors that may act as barriers

for deep-sea ﬁshes. For instance, the Mediterranean Sea

(hereafter referred to as Mediterranean) is the youngest

deep-sea basin in the world (Emig & Geistdoerfer 2004)

and is connected to the Atlantic Ocean only by the nar-

row and shallow Strait of Gibraltar (12.9 km wide,

284 m deep). The Strait of Gibraltar was also known in

the classical antiquity as the ‘Pillars of Hercules’, due to

the promontories that ﬂank its entrance. The average

depth of the Mediterranean is 1500 m, with the western

basin reaching 3400 m and the eastern reaching 5267 m.

The Mediterranean has been shaped by turbulent geo-

logical events, such as the Messinian salinity crisis (i.e.

the reduction in the Atlantic water inﬂow resulted in

Mediterranean desiccation c. 5.6 million years ago; Hs€

et al. 1973; Krijgsman et al. 1999), which led to massive

extinctions. After this event, the Atlantic marine biota

recolonized the Mediterranean arriving in different and

separated demographic waves depending on their bio-

geographic afﬁnities (warmer or colder) mostly during

the drastic climatic ﬂuctuations occurring in the Pleis-

tocene (reviewed by Emig & Geistdoerfer 2004). The

Strait of Gibraltar acts as a physical barrier isolating the

deep-sea Mediterranean from the adjacent Atlantic

Ocean (Smith & Gale 2009), resulting in impoverished

deep-sea communities in the Mediterranean compared

to the Atlantic (Emig & Geistdoerfer 2004). The Strait of

Gibraltar has been reported as a genetic break for shal-

low water and pelagic marine taxa, as some species dis-

play marked genetic differences between the eastern

North Atlantic and the Mediterranean. However, this is

still controversial as other species do not exhibit such

pattern (reviewed by Patarnello et al. 2007). Genetic

studies on exclusively deep-sea species occurring on

either side of the Strait of Gibraltar are currently lack-

ing. Therefore, the Atlantic/Mediterranean transition is

a very interesting area to study genetic connectivity in

deep-sea species.

The Portuguese dogﬁsh (Centroscymnus coelolepis Bar-

bosa du Bocage & de Brito Capello 1864) has been eco-

nomically exploited and is one of the most widespread

deep-sea sharks, reported all over the Atlantic Ocean,

in the western Mediterranean, in the Indian and Wes-

tern Paciﬁc Oceans (Compagno et al. 2005). This deep-

water squaloid shark (order Squaliformes) inhabits

continental and island slopes, seamounts and ocean

ridges (Ebert & Stehmann 2013), and its vertical distri-

bution ranges from 128 to 3675 m depth (Ebert &

Stehmann 2013). In the Mediterranean, the deepest

specimen was collected at 2863 m (Grey 1956). The Por-

tuguese dogﬁsh is a yolk-sac viviparous shark that

gives birth to 12 pups per litter on average and has a

long reproductive cycle with a gestation period of two

or more years (Yano & Tanaka 1988; Girard & Du Buit

1999; Ver

ıssimo et al. 2003; Figueiredo et al. 2008). Spe-

cies longevity is still unknown (Irvine 2004; Cotton

2010), but it is likely to be as high as the maximum esti-

mated age for a close relative, Centroselachus crepidater

(54 years old; Irvine et al. 2006).

6062 D. CATARINO ET AL.

The species is of commercial interest in the eastern

North Atlantic and in Australian waters (Daley et al.

2002; ICES 2012), where intense ﬁshing mortality led to

the stock’s decline and resulted in strict conservation

policies. High ﬁshing pressure together with the low

fecundity and expected long longevity of C. coelolepis

led the International Union for Conservation of Nature

(IUCN) to recently classify the species as globally near

threatened (Stevens & Correia 2003). The European

Union has implemented a zero catch policy for several

deep-water sharks since 2010 including C. coelolepis, and

no by-catch landings are allowed since 2012 (Council

Regulation (EU) No 1225/2010, Council Regulation

(EU) No 1359/2008). In the Mediterranean, commercial

ﬁsheries usually operate down to 800 m depth, and

ﬁsheries below 1000 m depth were banned since 2005

(FAO 2006). Thus, the species is presently not targeted

commercially, being classiﬁed in the Mediterranean as

Least Concerned (IUCN; Bradai et al. 2012).

A recent genetic study (Ver

ıssimo et al. 2011a) showed

that the species is genetically undifferentiated and proba-

bly consists in a single panmictic unit along the eastern

Atlantic from Ireland to South Africa. Despite the

observed genetic homogeneity over such a large spatial

scale, the species exhibits differences in life history traits

when considering the Mediterranean. In the Atlantic,

C. coelolepis pups are born around 30 cm total length (TL)

and grow to a maximum of 122 cm TL (Girard & Du Buit

1999). Females attain larger sizes than males, with males

maturing around 70–86 cm TL and females around 95–

102 cm TL in the Atlantic and Paciﬁc Oceans (Yano &

Tanaka 1988; Girard & Du Buit 1999; Clarke et al. 2001;

Ver

ıssimo et al. 2003). In the Mediterranean, body size is

considerably smaller: 22 cm TL at birth (Torchio &

Michelangeli 1971) and reaching a maximum size of

65 cm TL (Cl

oet al. 2002). Size at maturity is different in

the Mediterranean compared to the Atlantic and Paciﬁc

Oceans, and fecundity is also expected to be lower in the

Mediterranean (Yano & Tanaka 1988; Cl

oet al. 2002).

These striking differences in life history traits between

the Mediterranean and other regions of C. coelolepis range

strongly suggest that the former may harbour a distinct

population of Portuguese dogﬁsh; however, this hypoth-

esis has never been tested.

The aim of the present study was to investigate the

population genetic structure of C. coelolepis, with special

emphasis on the Atlantic/Mediterranean transition, and

to assess to what extent the Strait of Gibraltar may have

acted as a barrier to gene ﬂow, considering the species

bathymetric distribution. As far as we know, this study

is the ﬁrst to analyse the genetic connectivity in the

Atlantic/Mediterranean transition in a bathyal species.

We further investigate large- (Atlantic/Paciﬁc) and

small-scale (within the Mediterranean) connectivity. Life

history traits, such as body size and size at maturity,

were collected from different locations along the species

range to allow for direct comparison among regions

and to provide a complete characterization of this

species’ populations from the studied areas.

Material and Methods

Sampling

Specimens of Centroscymnus coelolepis were collected by

research vessels or commercial ﬁshing vessels (before

ﬁshery closure) from nine main locations (Fig. 1; Tables 1

and S1, Supporting information): Azores (AZ), South of

Azores (SAZ; Irving and Great Meteor seamounts com-

bined), Madeira (MAD), Portugal mainland (PT), Ireland

(IRE), United Kingdom (UK), Mauritania (MAU), west-

ern Mediterranean (MED) and Australia (AUS). Samples

from IRE and MAU and part of the samples from MAD

and PT were previously used in the study by Ver

ıssimo

et al. (2011a; see Table 1 for details). Samples from the

AZ and SAZ were collected using a ‘stone-buoy’ bottom

long-line gear (Menezes et al. 2009). Sharks from the UK

(Campbell et al. 2011) and AUS were collected using

deep-water bottom trawl nets. Sampling in the Mediter-

ranean was performed using baited traps or deep-water

bottom trawl nets (Tosti et al. 2006; Tecchio et al. 2011).

Biological data (length, weight, sex, maturation) were col-

lected on board from fresh specimens or in the laboratory

from specimens frozen on board. Small portions of white

muscle or ﬁn clips were sampled for genetic analyses and

preserved in 95% ethanol or DMSO buffer saturated with

NaCl (Seutin et al. 1991) prior to DNA extraction.

DNA extraction and PCR ampliﬁcation

Mitochondrial DNA (mtDNA). Genomic DNA was

extracted using the E.Z.N.A.

(EaZy Nucleic Acid Isola-

tion) Mag-Bind Tissue DNA kit (OMEGA Bio-Tek,

USA) according to the manufacturer’s instructions with

the KingFisher mL magnetic particle processor

(Thermo–Electron Corporation, USA). The mitochon-

drial control region (CR) was partially ampliﬁed for 155

individuals (this study) by polymerase chain reaction

(PCR) using the primers Pro-L and CoeCRH (Keeney

et al. 2003; Ver

ıssimo et al. 2011a). PCR ampliﬁcation fol-

lowed published protocols (Ver

ıssimo et al. 2011a) with

slight modiﬁcations (Appendix S1, Supporting informa-

tion). No template controls were included to check for

possible contamination. Electrophoresis of PCR prod-

ucts was performed on a 1% agarose gel to evaluate the

integrity of the products. Finally, all ampliﬁed products

were puriﬁed using ExoSAP-IT (USB Corporation,

USA) and sequenced commercially at BMR Genomics

POPULATION STRUCTURE OF THE PORTUGUESE DOGFISH 6063

(www.bmr-genomics.it, Padua, Italy). All haplotypes

were conﬁrmed by sequencing both the forward and

reverse strands for at least one individual per haplotype.

Nuclear DNA (microsatellites). Genomic DNA was

extracted using the E.Z.N.A.

Tissue DNA Kit (OMEGA

Bio-Tek) according to the manufacturer’s instructions. A

total of 503 sharks (Table 1) were genotyped for 11

nuclear microsatellite loci described by Ver

ıssimo et al.

(2010, 2011b). Primers were combined in three different

sets (Table S2, Supporting information) after optimiza-

tion. Sets 1 and 3 were ampliﬁed in multiplexes of four

loci, whereas set 2 was ampliﬁed as a duplex for the loci

Ccoe13 and Ccoe55 and separately for locus Ccoe61.

Ampliﬁcation followed published protocols (Ver

ıssimo

et al. 2011b) with some modiﬁcations (Appendix S1,

Supporting information). PCR products were run for

fragment analyses on an ABI Prism 3130xl (Life Tech-

nologies, USA). All genotypes were scored indepen-

dently by three persons using GENEMAPPER software

(version 4.0; Life Technologies), and inconsistencies

were resolved by replicate runs. Genotypes that could

not be resolved after replicate runs were omitted from

further analyses.

Combined samples

To increase sample sizes per geographical region, some

of the original sample collections were combined based

upon geographical proximity and absence of genetic

divergence between them (exact G-test, see Results

section), resulting in nine (eight for microsatellite data)

Fig. 1 Portuguese dogﬁsh sampled col-

lection sites. Stars indicate samples used

only for mtDNA. Sampled sites abbrevia-

tions are given in Table 1.

6064 D. CATARINO ET AL.

composite collections for the analyses of spatial genetic

structure (see Table 1 for details). Thus, samples from

the Azores, collected from 2003 to 2011, in several geo-

graphically adjacent collection sites around the islands

and seamounts, were combined in one collection (AZ or

AZ/MAR: cf. Table 1). Mediterranean samples were

obtained in several years (2001, 2009 and 2012) at three

main locations: northeast of Barcelona (MED1), south-

east of Ibiza (MED2) and southwest of Sardinia (MED3;

see Fig. 1 and Table 1), and were combined into a sin-

gle MED collection. Likewise, IRE and UK samples

were combined (UK/IRE) based upon the same criteria.

Atlantic results (ATL) refer to all sampled localities

within the Atlantic Ocean (excluding MED), including

South Africa (SAF) in the case of mtDNA.

Statistical analysis

mtDNA sequence analyses. All sequences available for

this species at the National Center for Biotechnology

Information (NCBI) database (Accession nos HQ664432–

HQ664449, Ver

ıssimo et al. 2011a) were also included in

the mtDNA CR alignment, with the respective haplo-

type frequencies. This added 192 additional samples to

the data set, for a total of 347 (see Table 1 for details).

All sequences were aligned using SEAVIEW (Galtier et al.

1996) and CLUSTALX (version 1.8.3.; Thompson et al.

1997).

Comparisons among sampling localities and relation-

ships among CR haplotypes were analysed using the

median-joining network method (Bandelt et al. 1999)

and the maximum parsimony approach (Polzin &

Daneschmand 2003), estimated with the NETWORK soft-

ware version 4.6.1.0. (ﬂuxus-engineering.com) using

default parameters and a nucleotide mutation weight of

10 for transitions/transversions, and 30 for insertions/

deletions, as suggested in the manual. A transition-

to-transversion ratio of 3:1 was estimated from the data

and used as a parameter (DNASP version 5.1; Rozas et al.

2003).

Table 1 List of sampling sites with number of Centroscymnus coelolepis specimens collected for genetic analyses (N), and genetic

diversity indices for the microsatellites and mtDNA control region (CR) markers

Sampling area Code Sampling date (year)

Nuclear microsatellites Mitochondrial CR

AZ and MAR combined AZ/MAR —— — — — 68 11 (2) 0.706 0.0020

Azores AZ 2003–2011 115 0.821 0.014 16.5 (11) 8.9 28 9 (2) 0.778 0.0024

Mid-Atlantic Ridge MAR 2004 —— — — — 40* 7 0.660 0.0018

South of the Azores SAZ 2007 25 0.795 0.036 10.5 (4) 8.3 10 3 0.600 0.0015

UK and IRE combined UK/IRE 68 0.818 0.018 14.2 (6) 8.7 48 8 (2) 0.597 0.0017

Ireland IRE 2006, 2007 35

†

0.817 0.030 12.0 (4) 8.6 38

‡

8 (2) 0.636 0.0019

United Kingdom UK 2011 33 0.820 0.007 11.9 (2) 8.6 10 2 0.467 0.0009

Mauritania MAU 2007 46

†

0.828 0.065 14.2 (9) 9.2 40

‡

7 0.630 0.0017

Mediterranean combined MED 151 0.657 0.041 10.0 (3) 5.6 55 7 (5) 0.579 0.0015

Northeast of Barcelona MED1 2009, 2012 105 0.665 0.024 9.5 (1) 5.7 39 5 (3) 0.566 0.0014

Southeast of Ibiza MED2 2009 15 0.651 0.021 6.0 5.8 —— — —

Southwest of

Sardinia

MED3 2001, 2009 31 0.628 0.114 6.6 (1) 5.3 16 4 (1) 0.642 0.0018

Portugal mainland PT 2005, 2006, 2011 37

†

0.812 0.002 12.3 (4) 8.7 52

‡

9 0.614 0.0015

Madeira MAD 2003 22 0.820 0.044 10.1 (1) 8.6 22

‡

7 (3) 0.671 0.0018

Australia AUS 2009, 2010 39 0.809 0.009 12.5 8.5 27 7 (3) 0.712 0.0023

South Africa SAF 2008 —— — — — 25*6 0.693 0.0018

Southeast of Ibiza samples were not sequenced because the tissue was prioritized for microsatellite analyses, as these markers are

more variable.

, estimated mean gene diversity; F

, inbreeding coefﬁcient; A, mean number of alleles (number of unique alleles); R

, mean allelic

richness (minimum sample size of 13 diploid individuals); H

, number of haplotypes (unique haplotypes); H

, haplotype diversity;

p, nucleotide diversity.

*Samples from MAR and SAF were only available for mtDNA and were sequenced by Ver

ıssimo et al. (2011a), all available at the NCBI.

†

Samples used before by Ver

ıssimo et al. (2011a) and genotyped again for the 11 microsatellite markers used in this study. For PT

sampling, only 25 of the 37 samples used in the microsatellites genotyping were from Ver

ıssimo et al. (2011a), as the DNA was

degraded. The remaining 12 samples came from recent collected specimens at this location.

‡

Samples sequenced by Ver

ıssimo et al. (2011a) and available at NCBI. To this previous data set, 10 new sequences were add for PT

sampling and seven for MAD sampling.

In bold are the main sampled locations analysed.

POPULATION STRUCTURE OF THE PORTUGUESE DOGFISH 6065

Genetic diversity indices, including number of haplo-

types (H

), haplotype diversity (H

) and nucleotide

diversity (p), were calculated for each of the nine sam-

pling localities. Genetic distance (Φ

) between localities

was estimated based on the mean number of pairwise

differences among sequences. All analyses were per-

formed in ARLEQUIN (version 3.5.1.2.; Excofﬁer & Lischer

2010), using 10 000 permutations of the data for esti-

mating P-values when applicable. The false discovery

rate approach (FDR; Benjamini & Hochberg 1995) was

used to correct for multiple comparisons.

Microsatellite analyses. Only samples in which at least

seven loci were successfully ampliﬁed were considered

for downstream analysis, to minimize biases caused by

poor DNA quality and missing genotypes. Levels of

genetic variability were estimated by gene diversity (H

within samples, H

in the total: Nei & Chesser 1983),

observed number of alleles and rareﬁed allelic richness

at each locus and for each sampling locality using FSTAT

(version 2.9.3; Goudet 2001; Tables 1 and S2, Supporting

information). Each locus/sample combination was anal-

ysed for deviations from the Hardy–Weinberg equilib-

rium (HWE) expectations by estimating the inbreeding

coefﬁcient F

within loci (Weir & Cockerham 1984;

Table S2, Supporting information) and tested for signiﬁ-

cance using two-tailed probability exact tests in the GEN-

EPOP software (version 4.0; Raymond & Rousset 1995;

Rousset 2008). The FDR approach was applied when

interpreting the resulting P-values. MICRO-CHECKER soft-

ware (version 2.2.3; van Oosterhout et al. 2004) was

used to estimate the presence of null alleles, stutter and

large allele dropout in all loci.

Genetic differences among localities were quantiﬁed

by F

, using Weir & Cockerham’s (1984) estimator,

over all sampling localities, as well as for pairs of sam-

ples and subsamples. Statistical signiﬁcance of pairwise

tests was assessed by G-tests for allele frequency

differences in GENEPOP using 10 000 dememorizations

and batches, and 10 000 iterations per batch. P-values

in multiple tests situations were evaluated for signiﬁ-

cance using the FDR approach.

Temporal replicates for the Atlantic were represented

by the IRE (2006/2007, n=35) vs.UK (2011, n=33) sam-

ples, and AZ2003 (n=26) vs.AZ2011 (n=50). Within

the MED basin, temporal replicates were collected in

three different years: 2001 (n=30), 2009 (n=71) and

2012 (n=50). Furthermore for MED, juveniles (n=75)

were compared with adults (n=23). The genetic struc-

ture for the temporal replicates was quantiﬁed by F

and tested for genetic divergence as described above.

Genetic differentiation patterns among samples were

visualized by applying principal component analysis

(PCA) on allele frequencies, using PCA-GEN (version

1.2.1; Goudet 1999). Signiﬁcance of total inertia in each

axis was tested using 10 000 data randomizations.

BAYESCAN (Foll & Gaggiotti 2008) and FDIST2 (imple-

mented in the LOSITAN software: Antao et al. 2008) were

used to check for outlier loci putatively under selection

with emphasis on the Atlantic vs.Mediterranean com-

parisons. Runs were based on the default parameters

for BAYESCAN and based on the recommended options

(on for ‘neutral’ mean F

and force mean F

), the step-

wise-mutation model (SMM) and 50 000 simulations for

LOSITAN. The FDIST2 and BAYESCAN were recently recom-

mended for testing divergent selection due to lower

type I and II error (i.e. detecting false outliers and

rejecting true outliers; Narum & Hess 2011). However,

as both methods have beneﬁts and pitfalls (Hansen

et al. 2010; Narum & Hess 2011), they were used to look

for common trends.

The BOTTLENECK software (version 1.2.02; Piry et al.

1999) was applied to investigate population declines,

using the TPM model with 70–90% stepwise mutations

and 10–30 variance and based on 10 000 interactions.

The Wilcoxon test was used to check for signiﬁcant

heterozygosity excess. Past reduction in effective popu-

lation size was also investigated using the M-ratio test

(Garza & Williamson 2001) and comparing it with the

critical value (M

), following the conservative parame-

ters suggested by the authors (i.e. h=10, 90% one-step

mutation and 3.5 mean size of non-one-step mutation).

Mitochondrial and nuclear markers analyses. In addition to

and Φ

(above), Jost’s Dstatistics (Jost 2008) were

also calculated for both marker types. Pairwise D

est

val-

ues were calculated with the ‘diveRsity’ package (Keenan

et al. 2013) in R(version 3.1.1; R Core Team 2014), using

1000 bootstraps. P-values in multiple tests situations

were evaluated for signiﬁcance using the FDR approach.

Spatial patterns of genetic divergence were visualized

using BARRIER software (version 2.2; Manni et al. 2004)

using sample coordinates and pairwise F

or Φ

estimates as input. The robustness of each barrier

was accessed by using 1000 bootstrapped matrices

calculated with customized Rcodes followed by the

‘POPGENOME’ package (Pfeifer et al. 2014) for the mtDNA

and using the ‘diveRsity’ package for nuclear genes,

both in R.

Approximate Bayesian computation (ABC, Beaumont

et al. 2002), implemented in DIYABC software (version 2.0;

Cornuet et al. 2014), was used to estimate the time since

divergence (t) between Atlantic and Mediterranean and

long-term effective population size (N

) of both popula-

tions. For estimation of the demographic parameters,

we tested a total of four different scenarios (see Fig. S1,

Supporting information) to choose among those that

better ﬁts the present data set: Scenario 1—constant effec-

6066 D. CATARINO ET AL.

tive population size for ATL and MED after divergence;

Scenario 2—MED population size is allowed to change

after divergence while ATL population size is constant

over time; Scenario 3—ATL population size is allowed to

change after divergence while MED population size is

constant over time; and Scenario 4—both ATL and MED

population sizes are allowed to change after divergence.

The reference table (the base of the parameter estimation)

consisted of 1 000 000 simulated data sets for each sce-

nario. Uniform priors were used when building the refer-

ence table and a range of values were allowed for the

historical parameters under consideration: Atlantic

N1 =100–2 000 000; Mediterranean N2 =10–500 000;

t=1–20 000; t1 =1–15 000, t1 <t; Atlantic N1b =10–

80 000; Mediterranean N2b =10–20 000. Several trial

runs were performed to choose the range of values that

best ﬁt the observed data, as suggested in Cornuet et al.

(2010). Sharks have low mutation rates (Martin et al.

1992; Martin 1995) and therefore the range for the mean

mutation rate and the individual locus mutation rate was

set to 1.0 910

6

to 3.0 910

4

(Karl et al. 2011) for the 11

microsatellites markers, leaving the other parameters as

default. Regarding the mtDNA CR, the most appropriate

nucleotide substitution model was selected from the hier-

archical series of likelihood ratio test, implemented in

MEGA (version 5.0; Tamura et al. 2011). From the models

available at the DIYABC, the Hasegawa–Kishino–Yano

(HKY 1985) model (I=0 and G=0.5) had the lowest

Bayesian information criterion (Schwarz 1978) value and

was used. The range for the mtDNA mutation rate was

set for 6.5 910

9

to 1.0 910

7

(Karl et al. 2011), and the

rest of the parameters were left as default. Gaps were

removed from the sequence data previous to the analy-

ses, because DIYABC does not include them in the current

version. The summary statistics used in the analyses

(Table S3, Supporting information) were the ones recom-

mended by Cornuet et al. (2008, 2010), because they are

sensitive to demographic events and divergence time (i.e.

mean genetic diversity and mean Garza–Williamson’s M

are sensitive to demographic changes and F

to diver-

gence time). The posterior probability of each scenario

was estimated using the logistic regression (Cornuet et al.

2008; Cornuet et al. 2010) on the 1% of simulated data sets

closest to the observed data set and using 10 intermedi-

ated values. Type error I and Type error II were esti-

mated for the chosen scenario computing 500 data sets

under each competing scenarios. The posterior distribu-

tion of the parameters was estimated by the logit transfor-

mation of the parameters and using 10 000 selected data.

A ‘model check’ was performed at the end of the analy-

ses using all the other summary statistics not used when

building the reference table (Cornuet et al. 2010; Table S3,

Supporting information). The discrepancy between the

model and the real data was visualized by PCA, as

implemented in DIYABC using 10 000 simulated data sets.

Time of divergence in years was assessed assuming a

generation time of 17 years, an average estimate based

on data from FISHBASE (Froese & Pauly 2013) and for a clo-

sely related species (Irvine et al. 2006).

In order to determine the amount of gene ﬂow

between Atlantic and Mediterranean populations and to

obtain an independent estimate of tand N

, the isola-

tion-with-migration model (IM, as implemented in the

software IMa2; Hey & Nielsen 2007) was used. To mini-

mize the computation time, only a subset of samples

was analysed. The subset consisted in 53 samples

drawn randomly from both Atlantic and Mediterranean

(altogether 106 samples) and only samples with avail-

able microsatellite genotypes and mtDNA sequences

were considered. Mitochondrial CR sequences and

seven of the eleven microsatellite loci were included in

the model. Ccoe9, Saca3853 and SacaGA11 were

excluded, as they contain short interrupted motifs

(Ver

ıssimo et al. 2011b) and Ccoe75 was excluded due

to alleles with single bp repeats. As for the DIYABC anal-

ysis, the HKY model of sequence evolution was applied

to mitochondrial sequences, and a SMM was assumed

for microsatellite loci. Altogether ﬁve parameters were

estimated: current and ancestral population sizes

(hATL, hMED and hANC, respectively), time since

divergence (t) and a single migration parameter (m).

Upper bounds for parameter priors were estimated

from the previous DIYABC analysis and during consecu-

tive preliminary runs of the program, based on initial

estimates of theta as advised in the manual of IMa2.

The ﬁnal prior values used for population size, migra-

tion rate and divergence time were as follows: q 100,

m1,t 30. We ran 150 Markov chains in parallel

under a geometric heating scheme. Several run time set-

tings with different heating schemes and durations

were explored to assess the mixing of the MCMC simu-

lation and chosen accordingly (ha =0.999, hb =0.4).

For the ﬁnal simulation, the manual’s suggestion for

large data sets was followed. First, two independent

jobs were run until a suitable burn-in was reached

(1 051 048 and 1 528 526 steps). Next, a new set of runs

were started by reloading the Markov chain state ﬁle

with an additional short burn-in period of 100 000 and

afterwards 20 000 genealogies were sampled every 25

steps from a total 500 000 steps. Both Markov chain

state ﬁles generated in burn-in runs were used three

times. Finally, all six replicates were combined in L-

mode run with identical parameter settings.

Life history analyses. As not all individuals used in the

genetic analyses had maturity data, a subsample of those

was analysed and summarized for average size, weight,

sex and capture depth for the three main regions anal-

POPULATION STRUCTURE OF THE PORTUGUESE DOGFISH 6067

ysed: Mediterranean, Atlantic and Australia (Paciﬁc). To

increase the sample size and the analyses’ power, a set of

samples collected at the end of 2012 and 2013 (not

screened for genetic data) was added to the Atlantic

(n=32) and Mediterranean (n=43) areas. The biological

data available from our Australian samples were very

limited; thus, a data set previously published (Daley et al.

2002) was included in the analyses (courtesy of K. Gra-

ham). The maturity data (immature vs. mature) were

analysed in detail using generalized linear models

(McCullagh & Nelder 1989) with maturity state as a bin-

ary response variable (juvenile or mature) and individual

body length or body weight as explanatory variables in

two separate models. In both models, effects of area

(Mediterranean vs.Atlantic vs.Paciﬁc) and sex on the

probability of being mature were included in the model.

Akaike’s information criterion (AIC) was used to com-

pare the performance of alternative candidate models

(Burnham & Anderson 1998). The model with the small-

est AIC value is expected to represent the best compro-

mise between lack of precision (including too many

parameters) and bias (too few parameters). Maturity

ogives were constructed based on the selected models,

showing the probability of being mature at a given length

and weight, for each area and sex independently. The

size and weight at which 50% of the sharks were mature

p50

and W

p50

) were extracted from the maturity ogives.

Results

Genetic variability

mtDNA. The CR primers ampliﬁed a 497-bp fragment,

and the 347 successfully obtained sequences contained 19

polymorphic sites (13 transitions, 4 transversions and 3

indels). Two indels occurred only in MED sequences,

where 23 of 55 individuals shared one or more indels.

The 19 polymorphic sites resulted in 27 haplotypes, 15 of

which were exclusive to a single locality (AZ: n=2, IRE:

n=2, MED: n=5, MAD: n=3 and AUS: n=3; Table 1),

while 12 were shared by two or more localities. The esti-

mated haplotype diversity varied between 0.467 and

0.778, whereas the nucleotide diversity ranged from

0.0009 to 0.0024, for UK and AZ, respectively. The lower

estimates refer to the very small UK sample (n=10) and

may represent a sampling artefact: for larger, regional

collections (n=48–68), the lowest haplotype and nucleo-

tide diversities were observed in the MED (Table 1)

although differences among regions were small.

Microsatellites. In total, 503 sharks were genotyped.

Genetic diversity in the total sample (H

) varied among

loci from 0.639 for locus Ccoe2 to 0.954 for Ccoe55

(mean over 11 loci =0.816), with a total number of alle-

les per locus ranging from 5 (Saca3853) to 56 (Ccoe55;

mean =20; Table S2, Supporting information). The

exact G-tests did not ﬁnd any deviations to HWE in the

ﬁnal data set (see Appendix S1, Supporting information

for details). Results from MICRO-CHECKER did not report

any consistent pattern of null alleles within loci, stutter

or large allele dropout. The levels of genetic diversity

) were similar among Atlantic and AUS, ranging

from 0.795 (SAZ) to 0.828 (MAU), but were signiﬁcantly

lower (about 19%) in the MED (H

=0.657; Mann–

Whitney U=23, P<0.05 two-tailed). Allelic richness

) displayed a similar pattern, with reduced levels

inside the MED (5.6) compared to elsewhere

(mean =8.7; Table 1; Fig. S2, Supporting information);

however, the differences were not signiﬁcant (Mann–

Whitney U=33, P>0.05 two-tailed).

Spatial and temporal structure

mtDNA. Pairwise Φ

and D

est

estimates were not sig-

niﬁcant within the Atlantic (Table 2, above diagonals).

However, for Φ

comparisons, differences were appar-

ent between Atlantic and Australian samples, where

three comparisons remained signiﬁcant even after FDR

correction at the 5% level. Regarding D

est

comparisons,

the pairwise values including AUS (D

est

0.0473–0.2775)

were higher than those among Atlantic collections (D

est

0.0000–0.0628); however, none of the ATL/AUS compar-

isons were signiﬁcant after FDR correction. Overall, the

ATL/AUS pairwise comparison was signiﬁcant and

indicative of genetic structure for Φ

(0.0744, P<0.01),

but not for D

est

(0.1543, P>0.5). In contrast, sharks from

the MED were genetically different from all other locali-

ties, with very high and signiﬁcant estimates of genetic

differentiation (mean Φ

=0.3808, P<0.001). As men-

tioned above, the gaps present in several of the Mediter-

ranean sequences were not found in any of the ATL/

AUS samples. Deleting these sites and rerunning the

data (as suggested by Pearce 2006) did not change the

main patterns found. Indeed, pairwise Φ

increased

slightly when the gaps were excluded (results not

shown). Pairwise D

est

comparisons including MED also

showed high and signiﬁcant values (D

est

0.2785–0.7281;

Table 2). The haplotype network for the mtDNA CR

(Fig. 2) consisted in two central, abundant haplotypes

and several other low-frequency-derived haplotypes.

The two central haplotypes, H1 (n=121) and H2

(n=144), occurred in similar frequencies across locali-

ties except for MED, where H2 was not observed.

Instead, the MED had H1 and a unique haplotype (H5)

occurring in high frequency (55% and 36%, respectively),

differing by one or two indels (position 41 and 53 in the

alignment) from the central ones. In the BARRIER analysis,

the strongest genetic barrier (100% support) was found

6068 D. CATARINO ET AL.

between the MED and all other localities (Fig. 3a). Other

secondary barriers are located around the AZ/MAR,

with statistical support between 68% and 98%.

Microsatellites. Signiﬁcant genetic differentiation was

found over all sampling localities both across loci (com-

bining temporal samples; F

=0.0581, P<0.001;

Table S2, Supporting information) and for each locus

separately (F

: 0.0107-0.1275, P<0.001; Table S2,

Supporting information). However, no divergence was

found among sampling localities across loci after

removing MED from the analyses (F

=0.0016,

P> 0.05; Table S2, Supporting information). Thus, the over-

all genetic divergence seems to be caused by divergence

between the MED and the other localities (mean

= 0.1149, P< 0.001; Table 2). Within the MED, no evi-

dence for genetic structure among the three replicate sites was

found (see Table S4, Supporting information).

No signiﬁcant genetic structure was found within the

Atlantic (mean F

=0.0024; P>0.05; Table 2) and the

Table 2 Pairwise F-statistics (F

and Φ

) and Jost’s D

est

values between sampling sites for microsatellites (below the diagonals) and

mtDNA control region (above the diagonals)

AZ SAZ UK/IRE MAU MED PT MAD AUS SAF

F-statistics

AZ/MAR —0.0244 0.0267 0.0011 0.3127*** 0.0168 0.0402 0.0140 0.0166

SAZ 0.0096 —0.0200 0.0426 0.3802*** 0.0344 0.0413 0.0110 0.0412

UK/IRE 0.0005 0.0035 —0.0064 0.4437*** 0.0145 0.0128 0.0997* 0.0118

MAU 0.0007 0.0096 0.0014 —0.3970*** 0.0160 0.0024 0.0413 0.0170

MED 0.1098*** 0.1138*** 0.1149*** 0.1154*** —0.4330*** 0.4495*** 0.2722*** 0.3578***

PT 0.0004 0.0064 0.0006 0.0004 0.1143*** —0.0122 0.0825* 0.0003

MAD 0.0013 0.0078 0.0004 0.0007 0.1051*** 0.0009 —0.0944* 0.0165

AUS 0.0020 0.0108* 0.0005 0.0018 0.1311*** 0.0002 0.0029 —0.0089

Jost D

est

AZ/MAR —0.0183 0.0569 0.0028 0.4914*** 0.0182 0.0362 0.0785 0.0000

SAZ 0.0123 —0.0000 0.0000 0.6142** 0.0000 0.0000 0.1788 0.0089

UK/IRE 0.0012 0.0000 —0.0141 0.7281*** 0.0006 0.0000 0.2775 0.0628

MAU 0.0000 0.0040 0.0000 —0.5594*** 0.0000 0.0039 0.1284 0.0000

MED 0.2681*** 0.2632*** 0.2738*** 0.2681*** —0.6154*** 0.6764*** 0.2785*0.4747***

PT 0.0008 0.0013 0.0000 0.0000 0.2769*** —0.0000 0.1786 0.0143

MAD 0.0008 0.0044 0.0000 0.0000 0.2610*** 0.0001 —0.2230 0.0327

AUS 0.0008 0.0071 0.0003 0.0000 0.3293*** 0.0002 0.0088 —0.0473

Sampled sites abbreviations are given in Table 1.

Signiﬁcant at alpha =*0.05, **0.01, ***0.001 after false discovery rate approach.

Bold indicates signiﬁcant values.

H8 H13

H12

H15

H23

H24

H25

H26

H27

H21

H19

H20

H22

H16

H17

H18

H14

H10

H11

*14%

86% *

86%

71%

AZ/MA

SAZ

UK/IRE

MAU

MED

MAD

AUS

SAF

Fig. 2 Median-joining network from 497-

bp mtDNA control region sequences.

Size of circles (haplotypes) is propor-

tional to the relative frequencies in the

sample and colour coded according to

sampling locality; abbreviations are given

in Table 1. Each branch indicates single

nucleotide substitutions, except when

noted. Bootstrapping support of each

branch is 100%, except when noted.

*indicates an indel.

POPULATION STRUCTURE OF THE PORTUGUESE DOGFISH 6069

overall ATL/AUS comparison showed also no signiﬁ-

cant differences (F

=0.0015; P>0.05), despite the

existence of one ATL/AUS signiﬁcant pairwise compar-

isons (Table 2). D

est

values showed the same patterns as

found with F

values, with no genetic structure

detected within Atlantic, or between ATL/AUS locali-

ties. Regarding comparisons including MED, high and

signiﬁcant pairwise F

and D

est

values were found

(P<0.001 in all cases; Table 2).

Allelic frequencies (Fig. S3, Supporting information)

were very similar between ATL and AUS across all loci;

however, MED exhibited very different allelic frequen-

cies. A shift in the most common alleles was observed

between Atlantic/Australia localities and Mediterranean

ones, where less frequent alleles at the former locations

became more common in the Mediterranean. This

pattern was found also for mtDNA haplotypes. No

temporal heterogeneity among years was detected either

within the Atlantic (8 years period) or the Mediterranean

(11 years period; Table S4, Supporting information).

The BARRIER analysis revealed that the strongest

genetic break was between the MED and all other sam-

ples, with 100% support (Fig. 3b). Other weaker barriers

separated SAZ from the other samples with a support

above 80%. A similar pattern is suggested by the PCA

analyses (Fig. 4), where the MED was clearly separated

from all other samples (PC1: F

=0.0476, 87% of total

inertia, P=0.0001). The PCA further suggested some

separation between SAZ and other samples by PC2;

however, this axis was not signiﬁcant (PC2:

=0.0019, 4% of total inertia, P=1.00). Results from

100

UK/IRE

AZ PT MED

MAD

SAZ

MAU

100

UK/IRE

MED

AZ/MAR

SAZ

MAU

SAF

MAD

mtDNA microsatellites

(a) (b)

Fig. 3 Barrier-inferred gene ﬂow patterns based on (a) mtDNA and (b) 11 microsatellite loci. Thickness of the barriers (red lines)

indicates their rank of importance and numbers indicate the percentage of support. Only barriers with support >60% are shown.

Sampled sites abbreviations are given in Table 1.

–0.6

–0.4

–0.2

0.2

0.4

0.6

–1 –0.6 –0.2 0.2 0.6 1

SAZ

MED

UK/IRE

MAD

AUS

MAU

PC2 (4%inertia)

PC1 (87% inertia)

Fig. 4 Principal component analysis (PCA) based on 11

microsatellite loci. Circles around the sampling localities high-

light the separation according to PCA 1 (P<0.001). Sampled

sites abbreviations are given in Table 1.

6070 D. CATARINO ET AL.

the outlier loci tests showed no candidate loci under

directional selection when comparing ATL/MED.

Demographic analyses

For microsatellite data, no bottlenecks were detected

using both the Wilcoxon tests (P>0.05) and the M-ratio

tests (mean observed M≥0.749, M

0.675 for seven

loci and M

0.725 for 15 loci).

In the DIYABC analysis, scenario 4 was the best sup-

ported with relatively stronger posterior probability

(mean P=0.99, 95% CI =0.98–1.00; Table 3; Fig. S4,

Supporting information). All other scenarios received

much lower support (between P=0.00 and 0.01; Table 3;

Fig. S4, Supporting information). Based on the logistic

estimate, the type error I (probability of being the true

model, but not selected) estimated for scenario 4 was 0.2

and the type error II (probability of not being the true

model, but it is selected) was 0.06. Under scenario 4, the

effective population size estimation for ATL (N1) did not

produce a distinct mode regardless the large priors

given. All the other estimated parameters showed dis-

tinct modes (Fig. S5, Supporting information). The MED

N2 was estimated in 90 900 (95% CI: 36 500–479 900;

Table 3). The simulations showed that both populations’

sizes after divergence (N1b and N2b) were not very small

(Table 3; Fig. S5, Supporting information) with modal

values of 40 500 (95% CI: 16 300–77 400) and 2050 (408–

15 300), respectively. Time since divergence was esti-

mated in 2750 generations (95% CI: 1450–18 100) for sce-

nario 4. Therefore, under this scenario and according to

the generation time used for the species (17 years), the

divergence time between ATL and MED population can

be put around 47 000 years before present (yBP; 95% CI:

24 650–307 700). The posterior distributions of the muta-

tion rates suggested that the prior range selected was

appropriate for the analysed data set (Fig. S6, Supporting

information). Model check showed that the observed

and the simulated data were similar (Fig. S6, Supporting

information), suggesting a good model ﬁt.

A summary of the estimated parameters using IMa2

analyses can be found in Table S5 (Supporting informa-

tion). Replicate runs in IMa2 analyses revealed low

levels of migration between both populations (0.756; a

single sharp peak was observed). Clear peaks were also

produced when estimating divergence time (0.495) and

the two population size parameters (Θ

ATL

=15.55 and

MED

=4.05), while the estimated size of the ancestral

population was on the upper border of the priors

(slightly higher than 100). When converting migration

parameter into per-generation population migration

rates (M=Θ9m/2), peak locations corresponded to

1.530 (M

ATL?MED

) and 5.878 (M

MED?ATL

) migration

events per generation, which means 0.090 and 0.346

Table 3 Posterior parameters values for each scenario estimated with DIYABC

Scenario Parameter Mean Median Mode

95% CI

P(95% CI)Lower Upper

S1 N1 3.77 910

3.34 910

2.81 910

1.12 910

8.86 910

0.00 (0.00–0.00)

N2 4.61 910

2.62 910

9.45 910

4.26 910

2.34 910

t8.91 910

8.18 910

5.69 910

1.78 910

1.87 910

S2 N1 3.65 910

3.30 910

2.57 910

1.26 910

8.09 910

0.00 (0.00–0.00)

N2 3.18 910

3.32 910

4.80 910

7.78 910

4.93 910

t8.79 910

8.04 910

6.05 910

2.01 910

1.84 910

t1 4.66 910

3.68 910

1.92 910

3.30 910

1.32 910

N2b 7.97 910

7.14 910

4.77 910

9.33 910

1.85 910

S3 N1 1.34 910

1.40 910

1.92 910

4.37 910

1.98 910

0.01 (0.00–0.02)

N2 4.08 910

2.69 910

1.76 910

7.60 910

1.93 910

t7.99 910

6.95 910

2.93 910

1.63 910

1.85 910

t1 6.43 910

6.04 910

4.15 910

1.2 910

1.37 910

N1b 4.32 910

4.36 910

4.46 910

7.93 910

7.62 910

S4 N1 1.49 910

1.57 910

1.97 910

6.25 910

1.99 910

0.99 (0.98–1.00)

N2 2.17 910

1.92 910

9.09 910

3.65 910

4.79 910

t7.10 910

5.85 910

2.75 910

1.45 910

1.81 910

t1 4.69 910

3.93 910

2.43 910

9.42 910

1.23 910

N1b 4.78 910

4.77 910

4.05 910

1.63 910

7.74 910

N2b 4.52 910

3.36 910

2.05 910

4.08 910

1.53 910

For each estimated parameter are presented the mean, median and modal value, the lower and upper 95% credible intervals and the

probability P(95% CI) of each scenario under consideration. In bold is shown the best supported scenario.

POPULATION STRUCTURE OF THE PORTUGUESE DOGFISH 6071

migrations events per year taking into account the

C. coelolepis generation time (17 years). In order to date

the divergence time in years and to estimate the effec-

tive population size, we applied the geometric mean of

the mutation rates estimated by the DIYABC approach

considering Scenario 1 with constant Ne, namely

5.3 910

6

(2.75 910

6

–6.98 910

5

) and 1.81 910

8

(9.2 910

9

–8.71 910

8

) for microsatellites and

mtDNA, respectively (Fig. S6, Supporting information).

Applying these rates, IM indicates that the divergence

between Atlantic and Mediterranean populations

occurred at 92 063 yBP (95% CI: 36 267–209 233), while

per generation was estimated as 42 530 (95% CI:

27 487–74 257) and 11 077 (95% CI: 6427–17 641) for

Atlantic and Mediterranean, respectively.

Life history analyses

For both sexes, the largest sharks were found in the

Atlantic. Here, the average body length exceeded the

maximum body length seen in the Mediterranean,

where much smaller sharks dominated the samples

(Table 4). The data supported an interaction effect

between body length and sex, and also an additive effect

of area, on the probability of being mature (Table S6,

Supporting information). In a separate model, the data

supported an interaction effect between body weight

and sex, as well as an additive effect of area, on the

probability of being mature (Table S6, Supporting infor-

mation). These models could not be simpliﬁed

(removing terms) without increasing the AIC value. The

smallest mature sharks were found in the Mediterranean

(Table 4; Fig. 5), where females reached L

p50

at 56 cm

and W

p50

at 2.2 kg, while males reached L

p50

at 46 cm

and W

p50

at 0.5 kg. In the Atlantic and Paciﬁc Oceans,

maturity ogives were similar and shifted towards larger

sizes and weights for both females (>100 cm and

>10 kg) and males (>80 cm and >3 kg; Table 4; Fig. 5).

Discussion

Genetic barrier across Gibraltar

The main ﬁnding of this study is the existence of a

major genetic discontinuity between the Mediterranean

population of Portuguese dogﬁsh and those inhabiting

the Atlantic/Paciﬁc Oceans. This conclusion was sup-

ported by all genetic markers (11 nuclear and 1 mito-

chondrial). Signiﬁcant high levels of pairwise F

(>0.1),

(>0.3) and D

est

(>0.2) between Mediterranean and

all the other localities suggest that gene ﬂow across the

shallow waters of Gibraltar are nearly absent for this

deep-sea species. The low estimated migration rates (1.5

and 5.9 migration events per generation) corroborate

this hypothesis.

Table 4 Biological data sampled from three main regions, showing the sample size (N) of females (F) and males (M), mean and

(range) of capture depth, body length (total length, TL) and body weight (total weight, TW), and the size (L

p50

) and weight (W

p50

)at

ﬁrst maturation

Sampled region Sex NCapture depth (m) L

p50

(cm) W

p50

(cm)TL (cm) TW (kg)

Mediterranean F 106 2136 (1500–2853) 41 (20–65) 0.6 (0.04–1.9) 56 2.2

M 78 2081 (1500–2853) 42 (21–58) 0.5 (0.05–1.4) 46 0.5

Atlantic F 73 1454 (1075–2475) 105 (64–119) 9.2 (0.7–16.2) 109 10.2

M 134 1333 (1025–2475) 92 (69–104) 5.7 (1.9–8.3) 83 3.7

Australia F 22 1062 (900–1070) 80 (55–110) 4.7 (1.0–13.0) 111* 11.2*

M 91 1009 (390–1070) 79 (48–97) 3.9 (0.7–8.0) 84 4.1

Part of the Australian data is from the literature (Daley et al. 2002), courtesy of K. Graham.

*Only two of the 22 females were mature.

20 40 60 80 100 120

0.04.0 8.

(a)

0 2 4 6 8 10121416

0.0 4.08.0

(b)

)

Body length (cm)

)erutam

Body weight (kg)

Fig. 5 Maturity ogives for Portuguese dogﬁsh, showing the

probability of being mature within the observed ranges of

body length (a) and body weight (b) for females (solid lines)

and males (broken lines) collected from the Mediterranean

(black), Atlantic (red) and Paciﬁc (blue).

6072 D. CATARINO ET AL.

Although the reported bathymetric distribution of

this species (between 128 and 3675 m; Ebert & Steh-

mann 2013) would not rule out transition across the

shallows of Gibraltar, it is highly unlikely as most stud-

ies reported a considerably deeper minimum depth of

occurrence, depending on the region, that is below

500 m off Ireland (Clarke et al. 2001) and below 900 m

in the Azores (Menezes et al. 2009; Table S1, Supporting

information). The average capture depth in this study

was 2109 m within the Mediterranean, with no individ-

uals being caught shallower than 1500 m depth. Such

ﬁndings are in accordance with previous studies in the

Mediterranean (Carrass

on et al. 1992; Tecchio et al.

2011), where the existence of a deeper-dwelling commu-

nity of this species seems to prevail, possibly to avoid

interspeciﬁc competition with shallower-dwelling spe-

cies of similar trophic level (Carrass

on et al. 1992).

Therefore, the very deep habits of the Portuguese dog-

ﬁsh together with the shallow bathymetry of the Strait

of Gibraltar (284 m) may act as a physical barrier to the

movement of individuals, leading to genetic isolation of

the Mediterranean population. In other species where

an Atlantic/Mediterranean genetic barrier has been

found (all epipelagic species with planktonic larvae),

the Almeria-Oran oceanographic density front has been

identiﬁed as the barrier to gene ﬂow, rather than the

Strait of Gibraltar (Patarnello et al. 2007 and references

therein). The Almeria-Oran front (Tintore et al. 1988) is

located at the eastern end of the Alboran Sea and can

extend to 300 m deep, limiting extensively the mixing

of the epipelagic eggs and larvae from the Atlantic with

the rest of the Mediterranean and vice versa. The exis-

tence of pelagic larval stages in many species promotes

passive dispersal and gene ﬂow, while the absence of

such larval stages in sharks makes gene ﬂow dependent

on active individual dispersal (Conrath & Musick 2012).

Therefore, in the case of the viviparous deep-sea shark

Centroscymnus coelolepis, the shallow depth of the Strait

of Gibraltar, instead of the Almeria-Oran front, is a bio-

logically more realistic barrier to gene ﬂow. The

hypothesis of bathymetry as a barrier to genetic connec-

tivity is supported by studies in other deep-sea taxa,

such as the roundnose grenadier (Knutsen et al. 2012)

and tusk (Knutsen et al. 2009). A physical barrier at the

Strait of Gibraltar seems also relevant for several

deep-sea species and especially sharks, because only a

fraction of the Atlantic deep-sea species occur inside

the Mediterranean (Emig & Geistdoerfer 2004; Ebert &

Stehmann 2013).

Genetic pattern across the Atlantic and Paciﬁc

In contrast to the marked genetic discontinuity

across the Strait of Gibraltar, the apparent genetic

homogeneity across the Atlantic Ocean suggests a single

Atlantic panmictic population, consistent with the ﬁnd-

ing of Ver

ıssimo et al. (2011a). The genetic homogeneity

across distant locations (Ver

ıssimo et al. 2011a; this

study) is probably due to the high swimming capacity

and long lifespan of the species (Irvine et al. 2006; Maia

et al. 2012). Therefore, active dispersal is likely to be a

major source of connectivity among areas without major

physical barriers, as has also been implied for other

deep-sea sharks (Straube et al. 2011; Cunha et al. 2012;

Ver

ıssimo et al. 2012).

Active dispersal over a long generation time may also

explain the low level of differentiation between Atlantic

and Australia. Furthermore, females are known to store

sperm after mating (up to several months or years;

Moura et al. 2011) and may mate whenever they get the

opportunity, saving the sperm until suitable conditions

arise to fertilize and produce the young. This strategy

may also contribute to population mixing during long

distance foraging movements.

Demographic events and population divergence

The low levels of genetic variability, as those found

within the Mediterranean, are sometimes associated

with past bottlenecks. However, no evidence of a bottle-

neck event was detected by the tests employed, perhaps

due to limitations of such tests (see Peery et al. 2012 for

details). The best supported scenario chosen by DIYABC,

where both populations N

were allowed to change

after divergence, showed high modal values of N

after

divergence (N1b and N2b, Fig. S5, Supporting informa-

tion, Table 3), suggesting that no severe bottleneck has

occurred. However, in the Mediterranean, a possible

stronger reduction in the population cannot be entirely

ruled out because the lower credible interval for N2b is

as low as 500. Nevertheless, the reduction in heterozy-

gosity in the Mediterranean was not very large (~19%),

so it is possible that the divergence between the Atlan-

tic and Mediterranean populations may have occurred

simply by postisolation genetic drift over time. Results

from the ABC analyses also suggest that historical N

after divergence in the Mediterranean was at least one

order of magnitude smaller than in the Atlantic, which

may had facilitated the action of drift in the Mediter-

ranean. Results from IM approach also support that N

is signiﬁcantly smaller (no overlapping CI) within the

Mediterranean compared with the Atlantic. The shift in

the common allele frequencies across the Strait of

Gibraltar (Fig. S3, Supporting information) and the loss

of the high-frequency H2 Atlantic haplotype in the

Mediterranean are in agreement with what to expect in

an isolated population subjected to the action of genetic

drift (Hedrick 2011) and are consistent with previous

POPULATION STRUCTURE OF THE PORTUGUESE DOGFISH 6073

studies for the area (i.e. P

erez-Losada et al. 2002;

Lemaire et al. 2005). This pattern suggests that demo-

graphic events such as isolation followed by genetic

drift have been acting over time and may be an impor-

tant force shaping the gene pool of this Mediterranean

shark population. Mediterranean populations of some

large pelagic predators such as tunas, swordﬁsh and

dolphinﬁsh have been shown to have a similar genetic

pattern (i.e. genetic differentiation and low genetic

diversity comparing with their Atlantic counterparts),

explained as the result of vicariance and demographic

events (Vi~

nas et al. 2004; Alvarado-Bremer et al. 2005;

Nakadate et al. 2005; D

ıaz-Jaimes et al. 2010).

The ABC results from scenario 4 suggested the sepa-

ration between Atlantic and Mediterranean at about

47 000 yBP (95% CI: 24 650–307 700 yBP), at the late

Pleistocene, during the last glacial period (110 000–

12 000 yBP; Kukla 2005). However, the relatively wide

credible intervals suggest that an earlier separation may

also be possible. A similar estimate was also provided

by IM approach suggesting that divergence occurred at

92 063 yBP (95% CI: 36 267–209 233 yBP). The DIYABC

approach has the advantage of statistically choosing the

scenario that best explains the studied data set and esti-

mating the different parameters under that scenario. On

the other hand, the IM approach has the advantage of

considering and estimating migration rates which DIYABC

has not. As DIYABC does not assume migration after

divergence, the estimates of divergence time and effec-

tive population sizes may be biased. However, in the

present study, the estimated migration rates are very

low suggesting that this pitfall may have little effect on

the DIYABC estimates. Regardless of the exact timing of

the Atlantic/Mediterranean split, the models suggest

that this separation has probably occurred during the

last glacial period, which reached its maximum at the

last glacial maximum (LGM; 26 000–20 000 yBP; Clark

et al. 2009). According to the credible intervals, the sep-

aration occurred no later than the LGM, suggesting that

this climatic event had an important role in the separa-

tion of the Atlantic and Mediterranean shark popula-

tions or at least contributed to keep them apart in case

of an earlier split. This ﬁnding is in accordance with

Emig & Geistdoerfer’s (2004) suggesting that the condi-

tions for the bathyal and abyssal species during the last

glacial period were very harsh and therefore entrance

in the Mediterranean by Atlantic deep-sea species was

thought to be limited. The geological history of the

Mediterranean and the several periods of drastic cli-

matic variability are thought to play a major role in the

separation of populations for several species, including

elasmobranchs, resulting in genetic heterogeneity

between the isolated groups (Hewitt 1996; Chevolot

et al. 2006; Grifﬁths et al. 2011). Different thermal

regimes in the water column during different climatic

events may have contributed to the shift in the species

vertical range and thereby promote or prevent the

entrance in the Mediterranean by Atlantic specimens

(Emig & Geistdoerfer 2004).

Life history traits

Portuguese dogﬁsh from the Mediterranean were smal-

ler and reached sexual maturity at smaller sizes com-

pared to those from the Atlantic and Australia. Females

in the Mediterranean reach only about half the maxi-

mum size (65 cm) of that expected for an Atlantic

female shark (122 cm). Furthermore, size at birth seems

to be smaller in the Mediterranean (about 17–22 cm;

Table 4; Torchio & Michelangeli 1971; Carrass

on et al.

1992) compared to the Atlantic (about 30 cm; Girard &

Du Buit 1999). The capture of four recently born speci-

mens (20–22 cm TL, yolk-sac scar was still present)

within Mediterranean during this study and one (32 cm

TL, D. Catarino, personal observation) more recently

within Atlantic seems to conﬁrm this pattern. Smaller

adult sizes in Mediterranean populations compared to

their Atlantic conspeciﬁcs are a common trend for sev-

eral species. This trend may result from a combination

of several factors, in which limited food resources and

higher temperature in the Mediterranean are likely to

play a major role (Stefanescu et al. 1992; Massut

ıet al.

2004). The Mediterranean deep biological communities

live in warmer and saltier waters (~13 °C, >38&;

Tecchio et al. 2011) than those in Atlantic and Paciﬁc

Oceans (Table S1, Supporting information). Also food

availability (downward particle ﬂuxes; biomass and

abundance indices) is lower within the Mediterranean

(Danovaro et al. 1999; Massut

ıet al. 2004) and thereby

deep-sea communities are faced with an oligotrophic

environment. Nevertheless, hotspots of higher produc-

tivity are also present, such as submarine canyons

(Danovaro et al. 2010 and references therein). Therefore,

deep-sea species crossing Gibraltar have not only to

overcome the shallow physical barrier but also to adjust

to the unique environmental conditions of the deep

Mediterranean basin.

As explained above, the genetic break observed at

molecular markers across the Strait of Gibraltar in

C. coelolepis is probably the result of demographic

events (isolation +genetic drift) rather than natural

selection because no signal of selection was detected.

Ambient temperature plays a major role in the meta-

bolic rates of ectothermic organisms, including sharks

(Bernal et al. 2012), and this condition coupled to a

low-food environment as seen in the Mediterranean

deep-sea may set up strong selection pressures on traits

related to feeding activity/strategy and growth.

6074 D. CATARINO ET AL.

Therefore, we suggest that the selection signals should

be further investigated, speciﬁcally regarding genes

controlling metabolic activity and growth.

In conclusion, the Mediterranean Portuguese dogﬁsh

population is genetically distinct and isolated from its

Atlantic and Australian counterparts, and the striking

differences in their respective life history traits suggest

they are following different evolutionary trajectories.

Acknowledgements

The authors are very thankful to the several persons and insti-

tutions that provided data, samples and assistance at different

stages of this work: R. Medeiros, A. Rosa, E. Giacomello

(IMAR/DOP, Portugal); H. Ramos (SeaExpert, Portugal); K.

Enersen (IMR, Norway); U. Fernandez, A. Mecho, S. Tecchio

and E. Ramirez-Llodra (ICM-CSIC, Spain); C. Mytilineou

(HCMR, Greece); K. Graham (CFRC, Australia); S. Tanaka

(Tokai University, Japan); M. Mea, F. Mairano (Italy); to the

crews and scientiﬁc teams of the R/V Arquip

elago, F/V Santo

Onofre, R/V Garcia del Cid, R/V Sarmiento de Gamboa and MRV

Scotia. This study was performed under the framework of

ReDEco (FP6 ERA-NET, MARIN-ERA/MAR/0003/2008) and

Hermione (FP7-ENV-2008-1, nr 226354) projects. Samples were

collected under other scientiﬁc projects: DEECON (ESF, 06-

EuroDEEP-FP-008 & SFRH-EuroDEEP/0002/2007); CONDOR

(EEA Grants, PT0040/2008); SMaRT (M.2.1.2/029/2011) and

DEMERSAIS (Azorean Government); OASIS (FP5, EVK3-CT-

2002-00073-OASIS); PescProf I (INTERREG III B, MAC/4.2/

M12); BIOFUN (ESF, CTM2007-28739-E); DOS MARES

(CTM2010-21810-C03-03); ADIOS (EVK3-CT-2000-00035). DC is

a PhD student funded by FCT (SFRH/BD/65730/2009); HK is

cofunded by the University of Agder and the University of

Oslo (CEES); AV is a postdoctoral fellow funded by FCT

(POPH/FSE; SFRH/BPD/77487/2011); DS is a postdoctoral fel-

low funded by the DIANET programme (FP1527385002); and

SS is a research fellow supported by the Marie Curie grant

cofunded by the EU under the FP7-People-2012-COFUND; Co-

funding of Regional, National and International Programmes,

GA n.600407 and the Bandiera Project RITMARE. MARE is

funded by FCT through the strategic project UID/MAR/

04292/2013. Finally, we would like to thank to the editor and

to the anonymous referees whose comments and suggestions

helped improve the manuscript.

References

Aboim MA, Menezes GM, Schlitt T, Rogers AD (2005) Genetic

structure and history of populations of the deep-sea ﬁsh

Helicolenus dactylopterus (Delaroche, 1809) inferred from

mtDNA sequence analysis. Molecular Ecology,14, 1343–1354.

Alvarado-Bremer JR, Vi~

nas J, Mejuto J, Ely B, Pla C (2005)

Comparative phylogeography of Atlantic blueﬁn tuna and

swordﬁsh; the combined effects of vicariance, secondary con-

tact, introgression and population expansion on the regional

phylogenies of two highly migratory pelagic ﬁshes. Molecular

Phylogenetics and Evolution,36, 169–187.

Antao T, Lopes A, Lopes RJ, Beja-Pereira A, Luikart G (2008)

LOSITAN: a workbench to detect molecular adaptation

based on a F

-outlier method. BMC Bioinformatics,9, 323–

327.

Ball AO, Sedberry GR, Zatcoff MS, Chapman RW, Carlin JL

(2000) Population structure of the wreckﬁsh Polyprion ameri-

canus determined with microsatellite genetic markers. Marine

Biology,137, 1077–1090.

Bandelt HJ, Forster P, R€

ohl A (1999) Median-joining networks

for inferring intraspeciﬁc phylogenies. Molecular Biology and

Evolution,16,37–48.

Beaumont MA, Zhang W, Balding DJ (2002) Approximate

Bayesian Computation in population genetics. Genetics,162,

2025–2035.

Benjamini Y, Hochberg Y (1995) Controlling the false discovery

rate: a practical and powerful approach to multiple testing.

ournal of the Royal Statistical Society. Series B (Methodological),

57, 289–300.

Bernal D, Carlson JK, Goldman KJ, Lowe CG (2012) Energetics,

metabolism and endothermy in sharks and rays. In: Biology of

Sharks and their Relatives, 2nd edn (eds Carrier JC, Musick JA,

Heithaus MR), pp. 211–237. CRC Press, Boca Raton, FL, USA.

Bradai MN, Saidi B, Enajjar S (2012) Elasmobranchs of the

Mediterranean and Black Sea: Status, Ecology and Biology. Stud-

ies and Reviews. General Fisheries Commission for the

Mediterranean. FAO, No. 91. Rome.

Burnham KP, Anderson DR (1998) Model Selection and Inference:

A Practical Information-Theoretic Approach. Springer, New

York, New York, USA.

Caley MJ, Carr MH, Hixon MA, Hughes TP, Jones GP, Menge BA

(1996) Recruitment and the local dynamics of open marine popu-

lations.AnnualReview ofEcologyand Systematics,27, 477–500.

Campbell N, Neat F, Burns F, Kunzlik P (2011) Species rich-

ness, taxonomic diversity, and taxonomic distinctness of the

deep-water demersal ﬁsh community on the Northeast

Atlantic continental slope (ICES Subdivision VIa). ICES Jour-

nal of Marine Science,68, 365–376.

Carrass

on M, Stefanescu C, Cartes JE (1992) Diets and bathy-

metric distributions of two bathyal sharks of the Catalan

deep sea (western Mediterranean). Marine Ecology Progress

Series,82,21–30.

Chevolot M, Hoarau G, Rijnsdorp AD et al. (2006) Phylogeog-

raphy and population structure of thornback rays (Raja cla-

vata L., Rajidae). Molecular Ecology,15, 3693–3705.

Clark PU, Dyke AS, Shakun JD et al. (2009) The last glacial

maximum. Science,325, 710.

Clarke MW, Conolly PL, Bracken JJ (2001) Aspects of the

reproduction of the deep water sharks Centroscymnus coelole-

pis and Centrophorus squamosus from west of Ireland and

Scotland. Journal of the Marine Biological Association of the Uni-

ted Kingdom,81, 1019–1029.

Cl

o S, Dal

u M, Danovaro R, Vacchi M (2002) Segregation of

the Mediterranean Population of Centroscymnus coelolepis

(Chondrichthyes: Squalidae): A Description and Survey.

NAFO SCR Doc. 02/83.

Compagno LJV, Dando M, Fowler S (2005) Sharks of the World.

Princeton University Press, Princeton, New Jersey, USA.

Conrath CL, Musick JA (2012) Reproductive biology of Elasmo-

branchs. In: Biology of Sharks and their Relatives, 2nd edn (eds

Carrier JC, Musick JA, Heithaus MR), pp. 291–311. CRC

Press, Boca Raton, FL, USA.

Cornuet JM, Santos F, Beaumont MA et al. (2008) Inferring

population history with DIY ABC: a user-friendly approach

POPULATION STRUCTURE OF THE PORTUGUESE DOGFISH 6075

to approximate Bayesian computation. Bioinformatics,24,

2713–2719.

Cornuet JM, Ravign

e V, Estoup A (2010) Inference on popula-

tion history and model checking using DNA sequence and

microsatellite data with the software DIYABC (v1.0). BMC

Bioinformatics,11, 401.

Cornuet J-M, Pudlo P, Veyssier J et al. (2014) DIYABC v2.0: a

software to make Approximate Bayesian Computation infer-

ences about population history using Single Nucleotide Poly-

morphism, DNA sequence and microsatellite data.

Bioinformatics,30, 1187–1189.

Cotton C (2010) Age, growth, and reproductive biology of deep-

water Chondrichthyans. PhD thesis, Faculty School of Marine

Science, VIMS, Gloucester Point, Virginia, USA.

Cunha RL, Coscia I, Madeira C et al. (2012) Ancient divergence

in the trans-oceanic deep-sea shark Centroscymnus crepidater.

PLoS One,7, e49196.

Daley RK, Stevens JD, Graham K (2002) Catch analysis and pro-

ductivity of the deepwater dogﬁsh resource in southern Australia.

CSIRO Marine Research, Fisheries Research and Develop-

ment Corporation (FRDC), and NSW Fisheries, Australia,

FRDC Project No. 1998/108.

Danovaro R, Dinet A, Duineveld G, Tselepides A (1999) Ben-

thic response to particulate ﬂuxes in different trophic envi-

ronments: a comparison between the Gulf of Lions-Catalan

Sea (western-Mediterranean) and the Cretan Sea (eastern-

Mediterranean). Progress in Oceanography,44, 287–312.

Danovaro R, Company JB, Corinaldesi C et al. (2010) Deep-sea

biodiversity in the Mediterranean Sea: the known, the

unknown, and the unknowable. PLoS One,5, e11832.

D

ıaz-Jaimes P, Urlbe-Alcocer M, Rocha-Olivares A, Garcia-

de-Leon FJ, Nortmoon P, Durand JD (2010) Global phylo-

geography of the dolphinﬁsh (Coryphaena hippurus): the

inﬂuence of large effective population size and recent disper-

sal on the divergence of a marine pelagic cosmopolitan spe-

cies. Molecular Phylogenetics and Evolution,57, 1209–1218.

Dudgeon CL, Blower DC, Broderick D et al. (2012) A review of

the application of molecular genetics for ﬁsheries manage-

ment and conservation in sharks and rays. Journal of Fish

Biology,80, 1789–1843.

Ebert DA, Stehmann MFW (2013) Sharks, batoids, and chimaeras

of the North Atlantic. FAO Species Catalogue for Fishery Pur-

poses. No. 7. Rome.

Emig CC, Geistdoerfer P (2004) The Mediterranean deep-sea

fauna: historical evolution, bathymetric variations and geographi-

cal changes. Carnets de G

eologie/Notebooks on Geology.

Article 2004/01 (CG2004_A01_CCE-PG), 10 pp.

Excofﬁer L, Lischer HEL (2010) Arlequin suite ver 3.5: a new

series of programs to perform population genetics analyses

under Linux and Windows. Molecular Ecology Resources,10,

564–567.

FAO (2006) General Fisheries Commission for the Mediter-

ranean. Report of the ninth session of the Scientiﬁc Advisory

Committee. FAO Fisheries Report No. 814. Rome.

Figueiredo I, Moura T, Neves A, Gordo LS (2008) Reproductive

strategy of leafscale gulper shark Centrophorus squamosus and

the Portuguese dogﬁsh Centroscymnus coelolepis on the Por-

tuguese continental slope. Journal of Fish Biology,73, 206–225.

Foll M, Gaggiotti O (2008) A genome scan method to identify

selected loci appropriate for both dominant and codominant

markers: a Bayesian perspective. Genetics,180, 977–993.

Froese R, Pauly D (2013) FishBase (version 12/2013). Available

from www.ﬁshbase.org (accessed 7 February 2014].

Galtier N, Gouy M, Gautier C (1996) SEAVIEW and PHYLO_-

WIN: two graphic tools for sequence alignment and molecu-

lar phylogeny. Computer Applications in the Biosciences,12,

543–548.

Garza JC, Williamson EG (2001) Detection of reduction in pop-

ulation size using data from microsatellite loci. Molecular

Ecology,10, 305–318.

Girard M, Du Buit MH (1999) Reproductive biology of two

deep-water sharks from the British Isles, Centroscymnus coelo-

lepis and Centrophorus squamosus (Chondrichthyes: Squali-

dae). Journal of the Marine Biological Association of the UK,79,

923–931.

Goudet J (1999) PCA-GEN (version 1.2.1). Available from http://

www2.unil.ch/popgen/softwares/pcagen.htm (accessed 5

November 2013).

Goudet J (2001) FSTAT, a program to estimate and test gene

diversities and ﬁxation indices version 2.9.3.1. Available

from www2.unil.ch/popgen/softwares/fstat.htm.

Grey M (1956) The distribution of ﬁshes found below a depth

of 2000 m. Fieldiana (Zoology),36,75–183.

Grifﬁths AM, Sims DW, Johnson A et al. (2011) Levels of con-

nectivity between longnose skate (Dipturus oxyrinchus) in the

Mediterranean Sea and the north-eastern Atlantic Ocean.

Conservation Genetics,12, 577–582.

Hansen MM, Meier K, Mensberg KLD (2010) Identifying foot-

prints of selection in stocked brown trout populations: a spa-

tio-temporal approach. Molecular Ecology,19, 1787–1800.

Hasegawa M, Kishino H, Yano T (1985) Dating the human-ape

splitting by a molecular clock of 30 mitochondrial DNA.

Journal of Molecular Evolution,22, 160–174.

Hauser L, Carvalho GR (2008) Paradigm shifts in marine ﬁsh-

eries genetics: ugly hypotheses slain by beautiful facts. Fish

and Fisheries,9, 333–362.

Hedrick PW (2011) Genetics of Populations, 4th edn. Jones and

Bartlett, Burlington, Massachusetts, USA.

Hewitt GM (1996) Some genetic consequences of ice ages, and

their role in divergence and speciation. Biological Journal of

the Linnean Society,58, 247–276.

Hey J, Nielsen R (2007) Integration within the Felsenstein equa-

tion for improved Markov chain Monte Carlo methods in

population genetics. PNAS,104, 2785–2790.

Hs€

u KJ, Ryan WBF, Cita MB (1973) Late Miocene desiccation

of the Mediterranean. Nature,242, 240–244.

ICES (2012) Report of the Working Group on Elasmobranch Fishes

(WGEF),19–26 June 2012, Lisbon, Portugal. ICES CM 2012/

ACOM:19. 547 pp.

Irvine SB (2004) Age, growth and reproduction of deepwater dog-

ﬁshes from southeastern Australia. PhD thesis. Deakin Univer-

sity, Warrnambool, Victoria, Australia.

Irvine SB, Stevens JD, Laurenson LBJ (2006) Surface bands on

deepwater squalid spines: an alternative method of ageing

Centroselachus crepidater.Canadian Journal of Fisheries and

Aquatic Sciences,63, 617–627.

Jost L (2008) G

and its relatives do not measure differentia-

tion. Molecular Ecology,17, 4015.

Karl S, Castro A, Lopez J, Charvet P, Burgess G (2011) Phylo-

geography and conservation of the bull shark (Carcharhinus

leucas) inferred from mitochondrial and microsatellite DNA.

Conservation Genetics,12, 371–382.

6076 D. CATARINO ET AL.

Keenan K, McGinnity P, Cross TF, Crozier WW, Prod€

ohl PA

(2013) diveRsity: an R package for the estimation of popula-

tion genetics parameters and their associated errors. Methods

in Ecology and Evolution,4, 782–788.

Keeney DB, Heupel MR, Hueter RE, Heist EJ (2003) Genetic

heterogeneity among blacktip shark, Carcharhinus limbatus,

continental nurseries along the US Atlantic and Gulf of Mex-

ico. Marine Biology,143, 1039–1046.

Knutsen H, Jorde PE, Sannaes H et al. (2009) Bathymetric barri-

ers promoting genetic structure in the deepwater demersal

ﬁsh tusk (Brosme brosme). Molecular Ecology,18, 3151–3162.

Knutsen H, Jorde PE, Bergstad OA, Skogen M (2012) Popula-

tion genetic structure in a deepwater ﬁsh Coryphaenoides

rupestris: patterns and processes. Marine Ecology Progress

Series,460, 233–246.

Krijgsman W, Hilgen FJ, Rafﬁ I, Sierro FJ, Wilson DS (1999)

Chronology, causes and progression of the Messinian salinity

crisis. Nature,400, 652–655.

Kukla G (2005) Saalian supercycle, Mindel/Riss interglacial

and Milankovitch’s dating. Quaternary Science Reviews,24,

1573–1583.

Lemaire C, Versini JJ, Bonhomme F (2005) Maintenance of

genetic differentiation across a transition zone in the sea: dis-

cordance between nuclear and cytoplasmic markers. Journal

of Evolutionary Biology,18,70–80.

Longmore C, Trueman CN, Neat F et al. (2014) Oceanic scale

connectivity and life cycle reconstruction in a deep-sea ﬁsh.

Canadian Journal of Fisheries and Aquatic Sciences,71, 1312–1323.

Maia AMR, Wilga CAD, Lauder GV (2012) Biomechanics of

locomotion in sharks, rays and chimaeras. In: Biology of

Sharks and their Relatives, 2nd edn (eds Carrier JC, Musick JA,

Heithaus MR), pp. 125–152. CRC Press, Boca Raton, FL, USA.

Manni F, Gu

erard E, Heyer E (2004) Geographic patterns of

(genetic, morphologic, linguistic) variation: how barriers can

be detected by “Monmonier’s algorithm”. Human Biology,76,

173–190.

Martin AP (1995) Mitochondrial DNA sequence evolution in

sharks: rates, patterns and phylogenetic inferences. Molecular

Biology and Evolution,12, 1114–1123.

Martin AP, Naylor GJP, Palumbi SR (1992) Rates of mitochon-

drial DNA evolution in sharks are slow compared with

mammals. Nature,357, 153–155.

Massut

ı M, Gordon JDM, Moranta J et al. (2004) Mediterranean

and Atlantic deep-sea ﬁsh assemblages: differences in bio-

mass composition and size-related structure. Scientia Marina,

68, 101–115.

McCullagh P, Nelder JA (1989) Generalized Linear Models. Chap-

man and Hall/CRC, Boca Raton, Florida, USA.

Menezes GM, Rosa A, Melo O, Pinho MR (2009) Demersal ﬁsh

assemblages off the Seine and Sedlo seamounts (northeast

Atlantic). Deep-Sea Research II,56, 2683–2704.

Moura T, Serra-Pereira B, Gordo LS, Figueiredo I (2011) Sperm

storage in males and females of the deepwater shark Por-

tuguese dogﬁsh with notes on oviducal gland microscopic

organization. Journal of Zoology,283, 210–219.

Nakadate M, Vi~

nas J, Corriero A, Clarke S, Suzuki N, Chow S

(2005) Genetic isolation between Atlantic and Mediterranean

albacore populations inferred from mitochondrial and

nuclear DNA markers. Journal of Fish Biology,66, 1545–1557.

Narum SR, Hess JE (2011) Comparison of FST outlier tests for SNP

loci under selection. Molecular Ecology Resources,11, 184–194.

Nei M, Chesser RK (1983) Estimation of ﬁxation indices and

gene diversities. Annals of Human Genetics,47, 253–259.

van Oosterhout C, Hutchinson WF, Wills DPM, Shipley P

(2004) MICRO-CHECKER: software for identifying and cor-

recting genotyping errors in microsatellite data. Molecular

Ecology Notes,4, 535–538.

Palumbi SR (1992) Marine speciation on a small planet. Trends

in Ecology and Evolution,7, 114–118.

Patarnello T, Volckaert FAMJ, Castilho R (2007) Pillars of Her-

cules: is the Atlantic-Mediterranean transition a phylogeo-

graphical break? Molecular Ecology,16, 4426–4444.

Pearce JM (2006) Minding the gap: frequency of indels in

mtDNA control region sequence data and inﬂuence on pop-

ulation genetic analyses. Molecular Ecology,15, 333–341.

Peery MZ, Kirby R, Reid BN et al. (2012) Reliability of genetic

bottleneck tests for detecting recent population declines.

Molecular Ecology,21, 3403–3418.

P

erez-Losada M, Guerra A, Carvalho GR et al. (2002) Extensive

population subdivision of the cuttleﬁsh Sepia ofﬁcinalis (Mol-

lusca: Cephalopoda) around the Iberian Peninsula indicated

by microsatellite DNA variation. Heredity,89, 417–424.

Pfeifer B, Wittelsb€

urger U, Ramos-Onsins SE, Lercher MJ

(2014) PopGenome: an efﬁcient swiss army knife for popula-

tion genomic analyses in R. Molecular Biology Evolution,31,

1929–1936.

Piry S, Lukart G, Cornuet JM (1999) BOTTLENECK: a com-

puter program for detecting recent reductions in the effective

population size using allele frequency data. Journal of Hered-

ity,90, 502–503.

Polzin T, Daneschmand SV (2003) On Steiner trees and mini-

mum spanning trees in hypergraphs. Operations Research Let-

ters,31,12–20.

R Core Team (2014) R: A Language and Environment for Statisti-

cal Computing. R Foundation for Statistical Computing,

Vienna, Austria. Available from http://www.R-project.org.

(accessed 30 July 2015).

Raymond M, Rousset F (1995) GENEPOP (version 1.2): popula-

tion genetics software for exact tests and ecumenicism. Jour-

nal of Heredity,86, 248–249.

Robalo JI, Castilho R, Francisco SM et al. (2012) Northern refu-

gia and recent expansion in the North Sea: the case of the

wrasse Symphodus melops (Linnaeus, 1758). Ecology and Evolu-

tion,2, 153–164.

Rousset F (2008) Genepop’007: a complete re-implementation

of the Genepop software for Windows and Linux. Molecular

Ecology Resources,8, 103–106.

Rozas J, S

anchez-DelBarrio JC, Messeguer X, Rozas R (2003)

DnaSP, DNA polymorphism analyses by the coalescent and

other methods. Bioinformatics,19, 2496–2497.

Salmenkova EA (2011) New view on the population genetic

structure of marine ﬁsh. Russian Journal of Genetics,47,1279–

1287.

Schwarz G (1978) Estimating the dimension of a model. Annals

of Statistics,6, 461–464.

Seutin G, White BN, Boag PT (1991) Preservation of avian

blood and tissue samples for DNA analyses. Canadian Journal

of Zoology,69,82–90.

Smith AB, Gale A (2009) The pre-Messinian deep-sea Neogene

echinoid fauna of the Mediterranean: surface productivity

controls and biogeographical relationships. Palaeogeography,

Palaeoclimatology, Palaeoecology,281, 115–125.

POPULATION STRUCTURE OF THE PORTUGUESE DOGFISH 6077

Stefanescu C, Rucabado J, Lloris D (1992) Depth-size trends in

western Mediterranean demersal deep-sea ﬁshes. Marine

Ecology Progress Series,81, 205–213.

Stevens J, Correia JPS (2003) SSG Australia & Oceania Regional

Workshop, Centroscymnus coelolepis. In: IUCN 2013. IUCN

Red List of Threatened Species (Version 2013.2). Available

from www.iucnredlist.org (accessed 14 January 2014).

Stockley B, Menezes GM, Pinho MR, Rogers AD (2005) Genetic

population structure and evidence of an extreme bottleneck

in the black-spot sea bream (Pagellus bogaraveo) from the NE

Atlantic. Marine Biology,146, 793–804.

Straube N, Kriwet J, Schliewen UK (2011) Cryptic diversity

and species assignment of large lantern sharks of the

Etmopterus spinax clade from the Southern Hemisphere

(Squaliformes, Etmopteridae). Zoologica Scripta,40,61–75.

Tamura K, Peterson D, Peterson N, Stecher G, Nei M, Kumar S

(2011) MEGA5: molecular evolutionary genetics analysis

using maximum likelihood, evolutionary distance, and maxi-

mum parsimony methods. Molecular Biology and Evolution,

28, 2731–2739.

Tecchio S, Ram

ırez-Llodra E, Sard

aFet al. (2011) Drivers of

deep Mediterranean megabenthos communities along longi-

tudinal and bathymetric gradients. Marine Ecology Progress

Series,439, 181–192.

Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins

DG (1997) The ClustalX windows interface: ﬂexible strategies

for multiple sequence alignment aided by quality analysis

tools. Nucleic Acids Research,25, 4876–4882.

Tintore J, Violette PE, Blade I, Cruzado A (1988) A study of an

intense density front in the eastern alboran sea: the Almeria-

Oran front. Journal of Physical Oceanography,18, 1384–1397.

Torchio M, Michelangeli M (1971) Prima segnalazione in acque

italiane di uno squalide del genere Centroscymnus.Natura,

62, 241–245.

Tosti L, Danovaro R, Dell’Anno A et al. (2006) Vitellogenesis in

the deep-sea shark Centroscymnus coelolepis.Chemistry and

Ecology,22, 335–345.

Ver

ıssimo A, Gordo L, Figueiredo I (2003) Reproductive biol-

ogy and embryonic development of Centroscymnus coelolepis

in Portuguese mainland waters. ICES Journal of Marine

Science,60, 1335–1341.

Ver

ıssimo A, McDowell JR, Graves JE (2010) Global population

structure of the spiny dogﬁsh Squalus acanthias, a temperate

shark with an antitropical distribution. Molecular Ecology,19,

1651–1662.

Ver

ıssimo A, McDowell JR, Graves JE (2011a) Population struc-

ture of a deep-water squaloid shark, the Portuguese dogﬁsh

(Centroscymnus coelolepis). ICES Journal of Marine Science,68,

555–563.

Ver

ıssimo A, Moura T, McDowell J, Graves J, Gordo L, Hoelzel

R (2011b) Isolation and characterization of ten nuclear

microsatellite loci for the Portuguese dogﬁsh Centroscymnus

coelolepis.Conservation Genetics Resources,3, 299–301.

Ver

ıssimo A, McDowell JR, Graves JE (2012) Genetic population

structure and connectivity in a commercially exploited and

wide-ranging deepwater shark, the leafscale gulper (Cen-

trophorus squamosus). Marine and Freshwater Research,63,505–

512.

Vi~

nas J, Alvarado-Bremer J, Pla C (2004) Phylogeography of

the Atlantic bonito (Sarda sarda) in the northern Mediter-

ranean: the combined effects of historical vicariance, popula-

tion expansion, secondary invasion, and isolation by

distance. Molecular Phylogenetics and Evolution,33,32–42.

Waples R (1998) Separating the wheat from the chaff: patterns

of genetic differentiation in high gene ﬂow species. Journal of

Heredity,89, 438–450.

Ward RD, Woodwark M, Skibinski DOF (1994) A comparison

of genetic diversity levels in marine, freshwater, and anadro-

mous ﬁshes. Journal of Fish Biology,44, 213–232.

Weir BS, Cockerham CC (1984) Estimating F-statistics for the

analysis of population structure. Evolution,38, 1358–1370.

White TA, Stefanni S, Stamford J, Hoelzel AR (2009) Unex-

pected panmixia in a long-lived, deep-sea ﬁsh with well

deﬁned spawning habitat and relatively low fecundity.

Molecular Ecology,18, 2563–2573.

Yano K, Tanaka S (1988) Size at maturity, reproductive cycle,

fecundity, and depth segregation of the deep sea squaloid

sharks Centroscymnus owstoni and C. coelolepis in Suruga Bay,

Japan. Nippon Suisan Gakkaishi,54, 167–174.

S.S., H.K. and G.M. supervised and coordinated the

work. D.C., S.S. and H.K. conceived the study. D.C.,

S.S., H.K., R.D. and A.D. initiated the work. D.C., S.S.,

A.V., G.M., J.B.C., F.N., R.D., A.D. and B.R. organized

and collected ﬁeld samples. D.C. performed laboratory

work with major contributions of S.S. and H.S. D.C.,

H.S. and H.K. scored the microsatellites genotypes.

D.C., H.K. and E.O. analysed the data and performed

the statistical analyses with major contributions of

P.E.J., A.V., D.S. and S.S. D.C. and H.K. led the writing

of the text to which all authors contributed.

Data accessibility

The mtDNA CR sequences can be accessed in GenBank

by the Accession nos HQ664432–HQ664449 and

KT320896–KT321040 (new sequences from this study).

The ﬁle with all used and aligned sequences, the list of

haplotypes with haplotype frequencies, the microsatel-

lite genotypes, the biological data used in the life his-

tory analyses and the command lines used in IMa2 are

available at DRYAD: doi:10.5061/dryad.ss368.

Supporting information

Additional supporting information may be found in the online ver-

sion of this article.

Appendix S1 Material and methods.

Table S1 Portuguese dogﬁsh sampled collection sites with

averaged coordinates in decimal degrees within sampling area.

Table S2 Nuclear microsatellite loci combined in multiplexes

(Set) and annealing temperature (T

6078 D. CATARINO ET AL.

Table S3 Summary statistics used in the computations in the

DIYABC software.

Table S4 Pairwise F

results for temporal replicates and com-

bined sampled locations using the 11 nuclear loci.

Table S5 IM model estimates: time since divergence (t), effec-

tive population sizes of ancestral and descended populations

(ΘANC, ΘMED, ΘATL) and migration rate (mATL<>MED)

with the lower and upper boundaries of the 95% highest poste-

rior density (HPD) interval.

Table S6 Parameter estimates, with standard error (SE) and P-

values, of the generalized linear models used for describing

Portuguese dogﬁsh maturity ogives.

Fig. S1 DIYABC schematic representation of the four alternative

scenarios used for Approximate Bayesian computation

simulations.

Fig. S2 Genetic diversity patterns across all localities, based on

11 microsatellite loci.

Fig. S3 Graphical representation of the allele frequencies and

haplotype frequencies for the 11 microsatellites markers and

mtDNA CR for the three main regions analysed.

Fig. S4 Relative posterior probability of the demographic sce-

narios estimated using the logistic regression approach.

Fig. S5 DIYABC posterior probability distributions (green lines)

for the six parameters estimated: Atlantic N

(N1), Mediter-

ranean N

(N2), Atlantic N

after divergence (N1b), Mediter-

ranean N

after divergence (N2b), time since divergence (t),

and t1 under the best supported scenarios (S4).

Fig. S6 DIYABC analyses with estimated distributions for the

genetic markers and control quality of each scenario.

POPULATION STRUCTURE OF THE PORTUGUESE DOGFISH 6079

The demersal bathyal fish assemblage of the Central-Western Mediterranean: Depth distribution, sexual maturation and reproduction

Article

Sep 2020
Deep Sea Res

This study compiles information on the size distribution and reproductive patterns of five chondrichthyans and 10 bony species dwelling on the continental slope of the Central-Western Mediterranean (south-eastern Sardinian waters, depth 720-1890 m). The abundance of the relatively less numerous chondrichthyans was found to decrease with depth. Almost all chondrichthyan species had a long reproductive period throughout the year at all depths. In bony fishes, the upper bathymetric stratum (<1100 m) was dominated by species with continuous and seasonal (autumn-winter) spawning cycles. At intermediate and lower depths (>1100 m), continuous reproduction was the most common pattern. Both chondrichthyans and bony fishes exhibited large sizes at maturity and at first maturity (>75% and >60% of their maximum body length, respectively), supporting the hypothesis that most deep-sea fish populations are characterized by a delayed maturity, which makes them highly vulnerable to fishing pressure and over-exploitation. In some sharks, population partitioning was observed, according to life stage as a function of depth: adult/mature females occupied deeper habitats than males and juveniles. Conversely, Dipturus nidarosiensis female and male juveniles (>1100 m) were segregated from the adult/mature population, dwelling between 1000 and 1100 m. In four of the 10 bony fishes examined, we found a clear and significant predominance of females, particularly pronounced in Alepocephalus rostratus, which is clearly related to their K-reproductive strategy. The four key-selected species (Galeus melastomus, A. rostratus, Bathypterois mediterraneus and Trachyrincus scabrus), characterized by different reproductive strategies/modal-ities and feeding behaviours, were differently affected by depth and fluxes of organic matter (Particulate Organic Carbon concentration) from the photic zone towards the sea-bottom as a function of sex and life-stage.

Panmixia in the American eel extends to its tropical range of distribution: Biological implications and policymaking challenges

Article

Full-text available

Nov 2023
EVOL APPL

The American eel (Anguilla rostrata) has long been regarded as a panmictic fish and has been confirmed as such in the northern part of its range. In this paper, we tested for the first time whether panmixia extends to the tropical range of the species. To do so, we first assembled a reference genome (975 Mbp, 19 chromosomes) combining long (PacBio and Nanopore and short (Illumina paired‐end) reads technologies to support both this study and future research. To test for population structure, we estimated genotype likelihoods from low‐coverage whole‐genome sequencing of 460 American eels, collected at 21 sampling sites (in seven geographic regions) ranging from Canada to Trinidad and Tobago. We estimated genetic distance between regions, performed ADMIXTURE‐like clustering analysis and multivariate analysis, and found no evidence of population structure, thus confirming that panmixia extends to the tropical range of the species. In addition, two genomic regions with putative inversions were observed, both geographically widespread and present at similar frequencies in all regions. We discuss the implications of lack of genetic population structure for the species. Our results are key for the future genomic research in the American eel and the implementation of conservation measures throughout its geographic range. Additionally, our results can be applied to fisheries management and aquaculture of the species.

Challenges in avoiding deep-water shark bycatch in Azorean hook-and-line fisheries

Article

Full-text available

Oct 2022

Deep-water sharks are highly diverse, vulnerable, and understudied as a group, despite the increasing pressures on their populations. Twenty-five species of deep-water sharks have been recorded in the Azores, an oceanic archipelago in the mid-North Atlantic, that are regularly caught as bycatch in hook-and-line fisheries. Avoiding the bycatch of deep-water sharks presents multiple challenges due to their high catchability, difficulties in correctly identifying species, and the general lack of data on these species. This review summarizes the findings of recent studies from the region, providing an up-to-date science-based framework for mitigating bycatch effects of Azorean hook-and-line fisheries. Several depth-based, area-based, and gear-based measures have been studied that demonstrate the potential to either avoid or increase the survival of deep-water shark bycatch. However, these measures may have limited efficacy for some species (e.g. highly mobile species) and thus, limited widespread applicability. Convincing fishers to avoid deep-water shark bycatch is also a challenge given the antagonistic interactions with sharks damaging the catch and fishing gear, while simultaneously a market incentive for shark liver oil remains. It highlights the need to proactively engage fishers and incentivize the mitigation of bycatch of deep-water sharks in Azorean waters.

Geographic variation in gene‐flow from a genetically distinct migratory ecotype drives population genetic structure of coastal Atlantic cod (Gadus morhua L.)

Article

Full-text available

Jun 2022
EVOL APPL

Identifying how physical and biotic factors shape genetic connectivity among populations in time and space is essential to our understanding of the evolutionary trajectory as well as the management of marine species. Atlantic cod is a widespread and commercially important marine species displaying several ecotypes with different life history strategies. Using three sets of SNPs: neutral, informative, and genome‐inversion linked, we studied population genetic structure of ~2500 coastal Atlantic cod (CC) from 40 locations along Norway´s 2500km coastline, including nine fjords. We observed: 1) a genetic cline, suggesting a mechanism of isolation by distance, characterized by a declining FST between CC and North East Arctic Cod (NEAC – genetically distinct migratory ecotype) with increasing latitude, 2) that in the north, samples of CC from outer‐fjord areas were genetically more similar to NEAC than were samples of CC from their corresponding inner‐fjord areas, 3) greater population genetic differentiation among CC sampled from outer‐fjord areas along the coast, than among CC sampled from their corresponding inner‐fjord areas, 4) genetic differentiation among samples of CC from both within and among fjords. Collectively, these results permit us to draw two main conclusions. First, that differences in the relative presence of the genetically highly distinct, migratory ecotype NEAC, declining from north to south and from outer to inner fjord, plays the major role in driving population genetic structure of the Norwegian CC. Second, that there is limited connectivity between CC from different fjords. These results suggest that the current management units implemented for this species in Norway should be divided into smaller entities. Furthermore, the situation where introgression from one ecotype drives population genetic structure of another, as is the case here, may exist in other species and geographical regions, thus creating additional challenges for sustainable fisheries management.

Barriers in a sea of elasmobranchs: From fishing for populations to testing hypotheses in population genetics

Article

Full-text available

Sep 2021
GLOBAL ECOL BIOGEOGR

Background: The interplay of animal dispersal and environmental heterogeneity is fundamental for the distribution of biodiversity on earth. In the ocean, the interaction of physical barriers and dispersal has primarily been examined for organisms with planktonic larvae. Animals that lack a planktonic life stage and depend on active dispersal are however likely to produce distinctive patterns. Methods: We used available literature on population genetics and phylogeography of elasmobranchs (sharks, rays and skates) to examine how marine barriers and dispersal ecology shape genetic connectivity in animals with active dispersal. We provide a global geographical overview of barriers extracted from the literature and synthesize the geographical and hydrological factors, spatial and temporal scales to characterize different types of barriers. The three most studied barriers were used to analyse the effect of elasmobranch dispersal potential and barrier type on genetic connectivity. Results: We characterized nine broad types of marine barriers, with the three most common barriers being related to ocean bathymetry. The maximum depth of occurrence, maximum body size and habitat of each species were used as proxies for dispersal potential, and were important predictors of genetic connectivity with varying effect depending on barrier type. Environmental tolerance and reproductive behaviour may also play a crucial role in population connectivity in animals with active dispersal. However, we find that studies commonly lack appropriate study designs based on a priori hypotheses to test the effect of physical barriers while accounting for animal behaviour. Main conclusions: Our synthesis highlights the relative contribution of different barrier types in shaping elasmobranch populations. We provide a new perspective on how barriers and dispersal ecology interact to rearrange genetic variation of marine animals with active dispersal. We illustrate methodological sources that can bias the detection of barriers and provide potential solutions for future research in the field.

Combined COI barcode‐based methods to avoid mislabelling of threatened species of deep‐sea skates

Article

Full-text available

Jun 2021
ANIM CONSERV

Skates are characterised by conservative body morphology which hampers identification and leads to frequent taxonomic confusion and market mislabelling. Accurate specimen classification is crucial for reliable stock assessments and effective conservation plans, otherwise the risk of extinction could be unnoticed. The misclassification issue is evident for the genus Dipturus, distributed worldwide, from the continental shelf and slope to the deep sea. In this study, barcode cytochrome oxidase I gene (COI) sequences were used along with species delimitation and specimen assignment methods to improve taxonomy and zoogeography of species of conservation interest inhabiting the Atlantic Ocean and Mediterranean Sea. In this study, we provided new evidence of the occurence of D. nidarosiensis in the Central‐Western Mediterranean Sea and the lack of Atlantic‐Mediterranean genetic divergence. The Atlantic endangered species D. laevis and D. batis clustered together under the same molecular operational taxonomic unit (MOTU) with any delimitation methods used, while the assignment approach correctly discriminated specimens into the two species. These results provided evidence that the presence of the barcode gap is not an essential predictor of identification success, but the use of different approaches is crucially needed for specimen classification, especially when threshold‐ or tree‐based methods result less powerful. The analyses also showed how different putative, vulnerable, species dwelling across South‐Western Atlantic and South‐Eastern Pacific are frequently misidentified in public sequence repositories. Our study emphasised the limits associated to public databases, highlighting the urgency to verify and implement the information deposited therein in order to guarantee accurate species identification and thus effective conservation measures for deep‐sea skates.

New records and range extension of the Portuguese dogfish Centroscymnus coelolepis in the south‐western Atlantic Ocean, with comments on its morphology

Article

Full-text available

Jan 2020
J FISH BIOL

The Portuguese dogfish Centroscymnus coelolepis is a wide‐ranging deep‐water shark and a common by‐catch component of the catches of several mid‐ to deep‐water fisheries. In the present study, two new records from the south‐western Atlantic Ocean are reported based on specimens caught by bottom‐longline fishing vessels operating in the Argentinean–Uruguayan Common Fishing Zone. Species identification based on morphology and detailed morphometrics, as well as molecular data are presented for one of the specimens. The distribution of the species over the south‐western Atlantic is discussed on the basis of the available bibliography and a thorough revision of museum collections. The records presented here expand the species' previously acknowledged distribution southwards, from around 21° S to at least 38° S, suggesting it occurs continuously along the shelf break of eastern South America. However, given the limited access to specimens of deep‐water sharks in the region, the abundance and real extent of C. coelolepis distribution in the south‐western Atlantic as well as its interaction with deep‐water fisheries remain to be fully assessed.

GENEPOP (Version 1.2): Population Genetics Software for Exact Tests and Ecumenicism

Article

Full-text available

May 1995
J HERED

Biomechanics of Locomotion in Sharks, Rays, and Chimeras

Chapter

Full-text available

Mar 2004

Energetics, Metabolism and Endothermy in Sharks and Rays

Chapter

Full-text available

Apr 2012

Mediterranean and Atlantic deep-sea fish assemblages: Differences in biomass composition and size-related structure

Article

Full-text available

Dec 2004

Data obtained over a period of twenty years from 214 bottom trawls, towed on a single warp at depths between 402 and 1993 m in the western Mediterranean (Algerian and Balearic basins) and eastern North Atlantic (Rockall Trough and Porcupine Seabight), allowed a standardised comparison of density, biomass composition and size-related structure for both the whole fish fauna and for the most common species found within the deep-sea fish assemblages. All four areas are characterised by distinctly different and well-documented oceanographic conditions, biogeographical affinities and fishing exploitation. The results showed clear differences between the Atlantic and the Mediterranean deep-sea fish fauna, not only in density, species richness and composition, but also in the structure of the biomass that constitutes these assemblages. These differences are discussed in relation to environmental conditions and fishing pattern, which have determined the adaptive responses of both individual species and the whole ecosystem.

R: A Language and Environment for Statistical Computing

Book