global population structure and demographic history of the grey seal · 2015. 1. 26. · a global...

Global population structure and demographic history ofthe grey seal

A. KLIMOVA,* C. D. PHILLIPS ,† K. FIETZ,‡ § M. T. OLSEN,§ J . HARWOOD,¶ W. AMOS** and

J . I . HOFFMAN*

*Department of Animal Behaviour, University of Bielefeld, Postfach 100131, 33501, Bielefeld, Germany, †Department of

Biological Sciences and the Museum, Texas Tech University, Lubbock, TX 79409, USA, ‡Marine Evolution and Conservation

group, Centre for Ecological and Evolutionary Studies, University of Groningen, Nijenborgh 7, NL 9747 AG, Groningen, The

Netherlands, §Section for Evolutionary Genomics, Centre for Geogenetics, Natural History Museum of Denmark, University of

Copenhagen, Øster Voldgade 5-7, DK-1350, Copenhagen K, Denmark, ¶Sea Mammal Research Unit, Scottish Oceans Institute,

University of St Andrews, St Andrews, Fife, KY16 8LB, UK, **Department of Zoology, University of Cambridge, Downing

Street, Cambridge, CB2 3EJ, UK

Abstract

Although the grey seal Halichoerus grypus is one of the most familiar and intensively

studied of all pinniped species, its global population structure remains to be eluci-

dated. Little is also known about how the species as a whole may have historically

responded to climate-driven changes in habitat availability and anthropogenic exploi-

tation. We therefore analysed samples from over 1500 individuals collected from 22

colonies spanning the Western and Eastern Atlantic and the Baltic Sea regions, repre-

sented by 350 bp of the mitochondrial hypervariable region and up to nine microsatel-

lites. Strong population structure was observed at both types of marker, and highly

asymmetrical patterns of gene flow were also inferred, with the Orkney Islands being

identified as a source of emigrants to other areas in the Eastern Atlantic. The Baltic

and Eastern Atlantic regions were estimated to have diverged a little over 10 000 years

ago, consistent with the last proposed isolation of the Baltic Sea. Approximate Bayes-

ian computation also identified genetic signals consistent with postglacial population

expansion across much of the species range, suggesting that grey seals are highly

responsive to changes in habitat availability.

Keywords: approximate Bayesian computation, asymmetrical gene flow, demographic history,

marine mammal, pinniped, population structure

Received 1 March 2014; revision received 4 June 2014; accepted 25 June 2014

Introduction

A species’ demography is determined by an interaction

between the extrinsic forces generated by its ecology,

including anthropogenic influences, and its intrinsic life

history characteristics. In turn, demographic change is

often reflected in genetic change, both in terms of

genetic diversity and levels of divergence between pop-

ulations or geographic regions. By combining powerful

new analytical approaches with increasingly large

genetic data sets, this link can now be exploited to infer

historical demographic events. Because past demo-

graphic trends may inform future response to environ-

mental perturbations, such as anthropogenically driven

climate change, such knowledge is at a particular pre-

mium given the predictions of current climate models.

Colonially breeding pinnipeds are interesting in the

context of historical demography for several reasons.

First, they have strong dispersal capabilities yet also

tend to be highly philopatric and site faithful (Pomeroy

et al. 1994, 2000; Hoffman et al. 2006a,b; Hoffman & For-

cada 2012). The relationship between these opposing

traits should contribute greatly to population structure,

which will in turn affect the demographic coupling

between breeding colonies and, hence, populationCorrespondence: Joseph I. Hoffman, Fax: +49 (0)521 1062711;

E-mail: [email protected]

© 2014 John Wiley & Sons Ltd

Molecular Ecology (2014) 23, 3999–4017 doi: 10.1111/mec.12850

persistence (Olsen et al. 2014). Second, several pinniped

species show evidence of strong demographic responses

to historical changes in the marine environment, either

through rapid population expansion during periods of

increased food or habitat availability (Matthee et al.

2006; Curtis et al. 2009; Dickerson et al. 2010; Phillips

et al. 2011) or via altered migration dynamics (De Bruyn

et al. 2009). Third, many pinnipeds have been heavily

exploited for their meat, oil and/or fur (Bowen & Lid-

gard 2012). Consequently, contemporary patterns of

genetic diversity may have been shaped by a combina-

tion of relatively ancient events, like the last glacial

maximum (Matthee et al. 2006), more recent anthropo-

genic challenges (Hoffman et al. 2011) and innate life

history characteristics.

The grey seal (Halichoerus grypus) provides an inter-

esting case in point, being widely distributed across

much of the North Atlantic continental shelf, exhibiting

considerable dietary flexibility and breeding in diverse

habitats from pack-ice through land-fast ice to terres-

trial colonies (Kovacs & Lydersen 2008). Three main

geographically isolated grey seal populations are cur-

rently recognized (Bonner 1981). The Western Atlantic

population is distributed along the northeastern sea-

board of the United States and southern Canada. The

Eastern Atlantic population is concentrated around the

coast of the United Kingdom and Ireland but also

includes breeding colonies in Iceland, the Faroe Islands

and along the mainland coast of northern Europe. The

Baltic population is geographically confined to the Bal-

tic Sea (Bonner 1981; Hall 2002; Thompson & H€ark€onen

2008). Population size has probably changed a great

deal over time, being estimated at ~20 000 during the

last glacial maximum when continental shelf habitat

was much reduced (Boehme et al. 2012), before expand-

ing around 12 000 years ago as the ice sheets retreated.

More recently, many breeding colonies were extirpated

or heavily reduced by anthropogenic exploitation and

pollution (Lesage & Hammill 2001; H€ark€onen et al.

2007; Hiby et al. 2007). However, with protection, popu-

lations have rebounded strongly to a global size of

approximately half a million and numbers are still

increasing.

Although all three major grey seal populations prob-

ably experienced broadly similar histories, the timing

and severity of the component events may have dif-

fered considerably. For example, Western Atlantic grey

seals were hunted to population collapse in the 1800s,

later recovering from a few thousand individuals in

the 1960s (Lesage & Hammill 2001) to around 350 000

today. Archaeological evidence indicates that Eastern

Atlantic grey seals were common around 8000–

5500 years ago but gradually declined during the late

Middle Ages when many colonies in the Baltic

disappeared (H€ark€onen et al. 2007). Following protec-

tion under the Grey Seals (Protection) Act in 1914,

numbers in the UK increased from around 9000 in the

mid-1930s to between 80 and 90 thousand individuals

in 2008 (Lonergan et al. 2011). The Baltic population is

somewhat smaller but has experienced significant

growth in recent years and currently numbers around

40 000 individuals (HELCOM 2013). Grey seals are

thought to have entered the Baltic Sea around 6300–

7000 years ago (Schmolke 2008) and declined by over

90% between 1900 and the 1970s due to a combination

of unrestricted hunting and disease susceptibility

caused by the environmental contaminants polychlori-

nated biphenyl and dichlorodiphenyl trichloroethane

(Harding & H€ark€onen 1999). Following a ban on hunt-

ing in 1974 and changes in environmental politics,

the Baltic population has been slowly recovering

(HELCOM 2013).

Previous genetic studies have explored several

aspects of grey seal population structure and historical

demography. Strong philopatry is reflected in the

genetic differentiation between pairs of colonies even

within a geographic region (Allen et al. 1995), although

the pattern appears fluid and may be strongly influ-

enced by directed migration between colonies that have

reached carrying capacity and newly founded or grow-

ing colonies (Gaggiotti et al. 2002). On a broader scale,

the Western and Eastern Atlantic are strongly differenti-

ated from one another, while the Eastern Atlantic and

Baltic appear less so, consistent with the latter having

diverged more recently (Boskovic et al. 1996; Graves

et al. 2009; Cammen et al. 2011). The timing of the split

between the Baltic and the Eastern Atlantic grey seal

populations has been estimated variously at 0.35 million

years ago (Boskovic et al. 1996) based on mitochondrial

DNA, and between 10 000 and 1 million years ago

based on microsatellite data (Graves et al. 2009), the

wide range being due to uncertainty over mutation

rates. Both Boskovic et al. (1996) and Graves et al. (2009)

reported unexpectedly high levels of genetic diversity

that argue against an intense bottleneck.

Until recently, inferences about historical demogra-

phy required a range of simplifying assumptions such

as constant population size or zero migration. The latest

Bayesian approaches now allow more flexible models to

be fitted in which multiple parameters are estimated

simultaneously (Beaumont & Rannala 2004). However,

most real populations still carry more complicated his-

tories than can be accommodated without simplifica-

tion. Consequently, it is of interest to compare

alternative algorithms based on both nuclear and unipa-

rentally inherited markers in a species known to have

experienced dynamic demographic population histories.

Applying this rationale to the grey seal, and to provide


4000 A. KLIMOVA ET AL.

a global view of this species’ population structure

and demographic history, we sequenced 350 bp of the

mitochondrial hypervariable region 1 (HVR1) to aug-

ment published genotype data for nine microsatellites

(Worthington Wilmer et al. 1999; Gaggiotti et al. 2002) in

over 1500 grey seal samples collected from 19 colonies

spanning the Western and Eastern Atlantic Oceans, and

added these to data from three colonies in the Baltic

Sea (Graves et al. 2009).

Materials and methods

Tissue sample collection and DNA extraction

Skin biopsy samples were collected from live grey

seal pups at 19 breeding colonies (see Table 1, Fig. 1

and Fig. S1, Supporting information for details) fol-

lowing Bean et al. (2004). Skin samples were stored

individually in the preservative buffer 20% dimethyl

sulphoxide (DMSO) saturated with salt (Amos &

Hoelzel 1991) and stored at �20 °C. Total genomic

DNA was initially extracted using an adapted Chelex

100 protocol (Walsh et al. 1991), but to improve

sequence quality, this DNA was later purified using a

standard phenol–chloroform procedure (Sambrook

et al. 1989).

HVR1 sequencing and microsatellite genotyping ofnon-Baltic samples

A 489-bp region of the HVR1 of the mitochondrial

control region was PCR amplified using primers

(L16245 50-CACCACAGGCACCCAAAG-30 and H54 50-TCCAAATGGCAATGACACC-30) designed by Graves

et al. (2009) from the complete grey seal mitochondrial

genome (Arnason et al. 1993). Sequencing was con-

ducted by Alpha Biolabs (http://www.nucleics.com).

Sequences were aligned using BIOEDIT 7.0 (Hall 1999)

and all positions differing from the consensus sequence

were inspected by one of us (WA) to verify base calls

were high quality. All sequences were then trimmed to

the length of the shortest sequence (350 bp). Microsatel-

lite data derive from two published studies (Worthing-

ton Wilmer et al. 1999; Gaggiotti et al. 2002), the

samples being genotyped following Allen et al. (1995) at

nine unlinked microsatellite loci: Hg3.6, Hg4.2, Hg6.1,

Hg6.3, Hg8.9, Hg8.10, Hgd.2 (Allen et al. 1995), Pv9 and

Pv11 (Goodman 1997).

Table 1 Numbers of grey seal (Halichoerus grypus) tissue samples for which HVR1 and microsatellite data were analysed. Colonies

are classified according to three ‘regions’ and seven ‘subregions’ as described in the Materials and methods. Data for seals from the

Baltic are from Graves et al. (2009)

Region Subregion Colony

Number of samples genotyped at

Mitochondrial DNA Microsatellites

Western Atlantic Nova Scotia Sable Island 67 28

Eastern Atlantic Orkney Islands Holm of Huip 144 146

Holm of Spurness 144 145

Calf of Eday 130 130

Muckle Greenholm 129 131

Rusk Holm 118 119

Stroma 115 115

Links Ness 89 89

Swona 101 102

Faray 72 73

Copinsay 57 57

Corn Holm 58 59

Calf of Flotta 47 47

Switha 41 41

North West Scotland North Rona 48 52

Monachs 28 32

Eastern Scotland Isle of May 54 54

North East England Farne Islands 49 49

North Atlantic Faroe Islands 31 33

Baltic Baltic Stockholm Archipelago 39 47

Bay of Bothnia 34 34

Estonia 40 50

1635 1633


GREY SEAL POPULATION GENETICS 4001

Inclusion of data from the Baltic

Additional mitochondrial and microsatellite data were

available for three grey seal breeding colonies from the

Baltic Sea (Graves et al. 2009). The mitochondrial trace

files were scrutinized by two of us (KF and MTO) lead-

ing to a number of errors being identified and corrected

(Fietz et al. 2013). These samples were also genotyped

by Graves et al. (2009) for six microsatellite loci, five of

which overlap with our main data set (Hg4.2, Hg6.1,

Hg8.9, Hg8.10 and Pv9). Consequently, all analyses

including samples from the Baltic were restricted to five

microsatellites. Allele frequencies at these loci were har-

monized between the two data sets by Graves et al.

(2009), who independently regenotyped 50 samples and

found the two sets of scores to be identical.

Generation of summary statistics

The number of mitochondrial haplotypes, number of

polymorphic sites, haplotype diversity (h) and nucleo-

tide diversity (p) were calculated using DNASP version

5.0 (Rozas & Rozas 1995). Haplotype frequencies were

calculated using ARLEQUIN version 2.0 (Schneider et al.

2000). Using the Bayesian information criterion (BIC) for

model selection implemented in MEGA 5 (Tamura et al.

2011), the best-fit mutational model for our mtDNA data

was determined to be K2 + G(0.55) + I(0.48) and this,

or the most similar model available, was used in subse-

quent analysis where a substitution model needed to be

specified. GENEPOP (Raymond & Rousset 1995) was used

to test each microsatellite locus for deviations from

Hardy–Weinberg equilibrium, to calculate observed and

expected heterozygosities and to test for linkage disequi-

librium. For each test, we set the dememorization num-

ber to 10 000, the number of batches to 1000 and the

number of iterations per batch to 10 000.

Analysis of population structure

Population structure was assessed using hierarchical

analyses of molecular variance (AMOVA) conducted

within ARLEQUIN. Analyses were carried out at two dif-

ferent hierarchical levels: among the three main geo-

graphic regions (Western Atlantic, Eastern Atlantic and

Baltic) and among the seven subregions (Orkney

Islands, North West Scotland, Eastern Scotland, North

S w e d e nS w e d e n

N o r w a yN o r w a y

F r a n c eF r a n c e

P o l a n dP o l a n d

F i n l a n dF i n l a n d

G e r m a n yG e r m a n y

I t a l yI t a l y

U k r a i n eU k r a i n e

R o m a n i aR o m a n i a

L a t v i aL a t v i a

B e l a r u sB e l a r u sI r e l a n dI r e l a n d

A u s t r i aA u s t r i aH u n g a r yH u n g a r y

L i t h u a n i aL i t h u a n i a

E s t o n i aE s t o n i a

U n i t e d K i n g d o mU n i t e d K i n g d o m

S l o v a k i aS l o v a k i a

C z e c h R e p u b l i cC z e c h R e p u b l i c

D e n m a r kD e n m a r k

C r o a t i aC r o a t i a

B e l g i u mB e l g i u m

S w i t z e r l a n dS w i t z e r l a n d

N e t h e r l a n d sN e t h e r l a n d s

S e r b i aS e r b i a

R u s s i aR u s s i a

S l o v e n i aS l o v e n i a

L u x e m b o u r gL u x e m b o u r g

M o l d o v aM o l d o v a

I s l e o f M a nI s l e o f M a n

F a e r o e I s l a n d sF a e r o e I s l a n d s

J e r s e yJ e r s e y

L i e c h t e n s t e i nL i e c h t e n s t e i n

20°0’0"E

20°0’0"E

10°0’0"E

10°0’0"E

0°0’0"

0°0’0"

10°0’0"W

10°0’0"W

20°0’0"W

20°0’0"W

30°0’0"W

30°0’0"W

40°0’0"W

40°0’0"W

60°0

’0"N

60°0

’0"N

50°0

’0"N

50°0

’0"N

20°0’0"E

20°0’0"E

10°0’0"E

10°0’0"E

0°0’0"

0°0’0"

10°0’0"W

10°0’0"W

20°0’0"W

20°0’0"W

30°0’0"W

30°0’0"W

40°0’0"W

40°0’0"W

50°0’0"W60°0’0"W

60°0’0"W

70°0

’0"N

60°0

’0"N

60°0

’0"N

50°0

’0"N

50°0

’0"N

Sable Island

North Rona

Monachs

Orkney Islands

Isle of May

Farne Islands

Stockholm Archipelago

Bay of Bothnia

FaroeIslands

Estonia

0 195 390 780 Miles

BALTIC SEA

A T L A N T I C O C E A N

0 1,600800 Miles

50°0’0"W

Fig. 1 Map showing the sampling locations of grey seal breeding colonies. The sampling locations of colonies within the Orkney

Islands are shown in Fig. S1 (Supporting information).



East England, North Atlantic, Nova Scotia and Baltic

Sea). For mitochondrial DNA, pairwise population dif-

ferences were assessed using Fst (Weir & Cockerham

1984) and Φst (Kimura 1980), the former looking only at

haplotype frequency differences (Excoffier et al. 1992)

while the latter also incorporates haplotype sequence

similarity. For microsatellite data, pairwise population

differences were assessed using Fst (Weir & Cockerham

1984) and Rst (Slatkin 1995), the latter being a microsat-

ellite-specific measure that takes account of the step-

wise-mutation process. Statistical significance was

assessed using 10 000 permutations of the data. The

program POPULATIONS version 1.2.31 (Langella 1999) was

used to construct neighbour-joining (NJ) trees for the

mitochondrial data using Φst and for the microsatellite

data using Cavalli-Sforza and Edwards chord distance,

Dc (Cavalli-Sforza & Edwards 1967). The latter measure

has been shown to be most effective in recovering the

correct tree topology from microsatellite data under a

variety of evolutionary scenarios (Takezaki & Nei 1996).

Haplotype network

Haplotype networks are often better suited than phylo-

genetic trees to depict relationships among populations

because ancestral haplotypes are often extant and

uncertainty over ancestor–descendent relationships can

be indicated by network reticulations. We therefore

visualized relationships among the three geographic

regions based on the mitochondrial DNA sequence data

by constructing a median joining network within

NETWORK version 4.6.1.1. (Bandelt et al. 1999). This pro-

gram calculates all possible minimum-spanning trees

for the data set and then combines these into a single

minimum spanning network following an algorithm

analogous to that proposed by Excoffier & Smouse

(1994). Inferred intermediate haplotypes are then added

to the network in order to minimize its overall length.

Estimation of migration rates

Migration rates can be estimated at two different time-

scales. In the short term, immigrants from a different,

genetically distinct, population and their offspring will

tend to carry genotypes that are unlikely in a system

where gene flow is negligible and the population is in

Hardy–Weinberg equilibrium (Faubet et al. 2007). Such

individuals can be identified and used to infer the rates

and directions of contemporary gene flow using the

Bayesian program BAYESASS version 3 (Wilson & Rannala

2003). Parameter values examined are migration rates

(DM), allele frequencies (DA) and inbreeding coefficients

(DF). The user chooses parameter value size changes to

achieve the optimal acceptance rate of between 20%

and 60% (Wilson & Rannala 2003). After exploring a

range of values, we analysed the seven subregions

(Orkney Islands, North West Scotland, Eastern Scotland,

North East England, North Atlantic, Nova Scotia and

Baltic Sea) using 3 9 107 iterations following a burn-in

of 3 9 106 iterations, DM = 0.1, DA = 0.4 and DF = 0.8.

Each analysis was repeated five times with different ini-

tialization seeds. The posterior probability distributions

of the parameters of interest were inferred from sam-

ples harvested every 1000 iterations. TRACER version 1.5

(Rambaut & Drummond 2007) was used to monitor

chain mixing and convergence. As high levels of consis-

tency were found among the replicate analyses, data

are presented for the final replicate only.

To explore patterns of historical gene flow, we used

the Bayesian coalescent-based framework implemented

in MIGRATE version 3.5.1 (Beerli & Felsenstein 1999;

Beerli 2006). MIGRATE allows the simultaneous estima-

tion of gene flow (M = m/l, where m = migration rate

and l = mutation rate) between populations and h, a

measure of effective population size (h = 4Nel; where

Ne = effective population size and l = mutation rate).

We used the full migration model, which allows bidi-

rectional migration between populations. Following a

burn-in of 10 000 iterations and ten replicates of a single

long chain, 50 000 genealogies were recorded at a sam-

pling increment of 50. The prior distributions for the

parameters were uniform with h priors bounded

between 0 and 24 and M priors bounded between 0

and 1000. A heating scheme was applied based on four

changes with temperatures 1, 1.5, 3 and 1 9 106 as

recommended by authors. Three additional runs were

performed using results from earlier runs as starting

conditions. All of these runs generated acceptable (i.e.

>200) effective sample sizes (ESSs) for all parameters of

interest. A microsatellite mutation rate of 5 9 10�4 per

gamete per generation (Weber & Wong 1993) was used

to scale the resulting values of M to m. Modal posterior

parameter estimates are presented as recommended by

the developers (Beerli 2009). The BAYESASS and MIGRATE

analyses were performed on the full data set, as well as

on a reduced data set in which we assessed potential

biases caused by a large Orkney sample size by restrict-

ing the Orkney Island sample to 134 individuals, com-

prising 10–11 samples randomly selected from each of

the breeding colonies.

Bayesian analysis of population divergence

We next used the program IMA2 (Hey 2010) to estimate

divergence dates at the regional level based on mito-

chondrial data. IMA2 is based on an isolation-with-

migration model and estimates the posterior probability

densities using MCMC simulation of six demographic



parameters including population sizes of the extant as

well as the ancestral population (given as q = 4Nel,where Ne is the effective population size), migration

rates (m = M/l, where M is the migration rate per

generation per gene copy) in both directions, and time

since divergence (t = tl, where t is the time since popu-

lation splitting). All parameter estimates are scaled by

the mutation rate (l), which was given as 2.75 9 10�7

substitutions per site per year (Phillips et al. 2009).

IMA2 assumes stable population sizes and a model of

isolation with gene flow, in which a single panmictic

population split into two at some time in the past but

has since remained connected by some degree of gene

flow (Hey & Nielsen 2004, 2007). Initial runs were per-

formed using wide priors (q ≤ 500, m ≤ 50, t ≤ 50) in

order to identify appropriate values for subsequent

parameter estimation. Simulations were then performed

with q = 200, m = 10 and t = 20 using 20 Markov

chains, the recommended heating scheme and 1 9 107

steps with 1 9 106 steps discarded as ‘burn-in’. Geneal-

ogies were sampled every 100 steps. Estimates were

generated under the HKY mutation model using an

inheritance scalar of 0.25 for mitochondrial sequence

data. Although IMA2 allows for analysis of more than

two populations, this model assumes that a bifurcating

phylogenetic tree can adequately represent the evolu-

tionary history of the focal taxa. Three independent

runs were performed for each comparison in order to

assess convergence.

Bayesian analysis of large sequence data sets can be

computationally prohibitive, but the coalescent process

is relatively robust with respect to the effect of sam-

ple size on tree topology (Felsenstein 2006; Kuhner

2009). Consequently, to expedite analysis of the East-

ern Atlantic population, we restricted the number of

individuals used to 213, comprising 11–12 samples

randomly selected from each of the 18 breeding

colonies.

Modelling alternative demographic histories

Finally, we used approximate Bayesian computation

(ABC; Beaumont et al. 2002; Bertorelle et al. 2010) to

explore likely historical changes in population size.

ABC allows the user to broadly define alternative

demographic scenarios, expressed as a stepwise series

of population size changes, then uses summary statis-

tics from the observed and simulated data sets to esti-

mate parameter values and to assess the relative

support for each scenario. ABC was implemented using

the program DIYABC version 1.0.4.46 (Cornuet et al.

2008, 2010). We developed five different demographic

models describing different possible patterns of effec-

tive population size change over time (see Table S1 and

Fig. S2, Supporting information for details). Priors for

the timing of events and the magnitude of changes of

Ne were based on known events in the demographic

history of this species, including massive habitat loss

during last glacial maximum (Boehme et al. 2012), the

entry of grey seals into the Baltic Sea (Sommer &

Benecke 2003) and recent anthropogenic exploitation

(Lesage & Hammill 2001; H€ark€onen et al. 2007; Hiby

et al. 2007). The first scenario that we evaluated (i) rep-

resented the null hypothesis of constant effective popu-

lation size over time. The alternative scenarios invoked

(ii) population expansion; (iii) population reduction; (iv)

population bottleneck; and (v) population expansion

followed by reduction. The mutation rate for microsatel-

lites was bound between 5 9 10�4 and 5 9 10�3 substi-

tutions/generation. For the HVR1 data, we used the

K2P substitution model as determined through model

selection. The substitution rate was uniformly distrib-

uted between 8.12 9 10�7 and 3.8 9 10�6 substitutions/

site/generation (Phillips et al. 2009; Dickerson et al.

2010). A generation time of 14 years was assumed

(Thompson & H€ark€onen 2008). We used four summary

statistics for microsatellites (mean number of alleles per

locus, mean expected heterozygosity, mean allele size

variance and mean MGW index across loci) and four

summary statistics for the HVR1 (number of haplo-

types, number of segregating sites, mean pairwise dif-

ference and Tajima’s D). These statistics were chosen on

the basis of their sensitivity to demographic change.

After simulating one million data sets for each sce-

nario, we used a polychotomous logistic regression pro-

cedure (Fagundes et al. 2007) to estimate the posterior

probability of each scenario, based on the 1% of simu-

lated data sets for each model producing summary sta-

tistics closest to the observed values. The type I error

rate for model selection was estimated as the proportion

of instances where the selected scenario did not exhibit

the highest posterior probability compared with the

competing scenarios for 500 simulated data sets gener-

ated under the selected scenario. In a similar way, the

type II error rate was estimated by simulating 500 data

sets for each of four alternative scenarios and calculat-

ing the proportion of instances in which the best-sup-

ported model was incorrectly selected as the most

probable one (Cornuet et al. 2010). Using the most prob-

able scenario, local linear regression was used to esti-

mate the posterior distributions of parameters.

Specifically, a logit transformation of parameter values

was performed and the 1% closest simulated data sets

to the observed were used for regression and posterior

parameter estimation (Beaumont et al. 2002). Due to

computational limitations, the ABC analysis of the East-

ern Atlantic population was run on the reduced subset

of 213 samples described above.



Results

We sequenced a 350-bp region of the mitochondrial

HVR1 in over 1500 grey seals from 19 different colonies,

which had previously been genotyped for nine highly

polymorphic microsatellites. Additional published data

were then incorporated from three colonies in the Baltic

(Graves et al. 2009), allowing comparisons to be made

at the levels of three major geographic regions (Western

Atlantic, Eastern Atlantic and Baltic), seven subregions

(Nova Scotia, Orkney Islands, North West Scotland,

Eastern Scotland, North East England, North Atlantic

and Baltic) and 22 colonies (Table 1 and Fig. 1).

The mitochondrial HVR1 was highly informative, con-

taining 94 variable sites of which 57 were parsimony

informative. A total of 167 unique haplotypes were

identified in 1625 grey seal individuals and overall hap-

lotype diversity was high, at 0.941 (Table S2, Supporting

information). Although the most common haplotype

was found in 181 individuals from 18 colonies, there

was little sharing of haplotypes among the three major

geographic regions, none being common to the Western

and Eastern Atlantic, and the latter both sharing a single

haplotype with the Baltic. These patterns are summa-

rized as a minimum spanning network in Fig. 2.

The microsatellite loci were also highly variable, car-

rying on average 9.8 alleles. Pooling all of the data, six

loci were found to deviate significantly from Hardy–

Weinberg equilibrium, three of which remained signifi-

cant after table-wide Bonferroni correction for multiple

statistical tests (Table S3, Supporting information). Several

significant deviations were observed after partitioning

the data into three geographic regions, but these were

not consistent across loci, as would be expected if null

alleles were responsible. Colony-level analyses revealed

only six significant deviations following Bonferroni cor-

rection, five of which involved different loci (data not

shown). No evidence was found for linkage disequilib-

rium among the loci.

Population structure

Strong genetic structure was detected at both the mito-

chondrial HVR1 and microsatellites. Pairwise compari-

sons at the level of three geographic regions yielded

highly significant Fst values for both types of marker

(Table 2, 0.10–0.28 for mtDNA and 0.11–0.24 for micro-

satellites, all significant at P < 0.0001). The equivalent

Φst (0.06–0.08) and Rst (0.08–0.13) values were somewhat

smaller but still highly significant (P < 0.0001 for all

comparisons). At the level of seven subregions, the

HVR1 yielded highly significant Fst and Φst values for

all but a handful of pairwise comparisons, with fewer

significant tests obtained using microsatellites (Table 3).

These patterns of differentiation are summarized by

two neighbour-joining trees, one for each marker class

(Fig. 3). Both trees reveal three distinct clades corre-

sponding to colonies from the UK plus the Faroe

Islands, the Baltic and Sable Island, respectively. How-

ever, the two trees disagree about which of the regions

are most dissimilar, the mitochondrial DNA favouring

the Western and Eastern Atlantic and nuclear data

favouring the Western Atlantic and Baltic. Logically, the

nuclear picture is probably closer to the truth, given

140

50

5

Fig. 2 A median joining network illus-

trating phylogenetic relationships among

the grey seal mitochondrial HVR1

haplotypes. Each line joining two circles

corresponds to a single nucleotide substi-

tution, with red circles representing

hypothetical haplotypes that were not

observed in the sample of individuals.

Circle size reflects the relative frequency

of each of the observed haplotypes.

Regions are colour coded as follows:

white = Western Atlantic, grey = Eastern

Atlantic, black = Baltic.



that East Atlantic seals lie between the Baltic and the

West Atlantic. However, one cannot discount unex-

pected patterns that may have arisen during the demo-

graphic upheaval that likely accompanied the retreat of

the ice sheet at the end of the last glacial maximum.

Next, we used AMOVA to assess the proportion of

genetic variation attributable to each level of population

substructure (Table 4). As expected, a greater proportion

of the null variance was explained by among-region as

opposed to among-subregion components, consistent

with population structure being strongest at the regional

level. Among-region and among-subregion variance

components were higher for Φst than Fst at the HVR1

and for Rst than Fst at microsatellites, again consistent

with the fact that the former measures are considered to

be more appropriately matched to the mutation models

of the two marker classes respectively. Finally, mito-

chondrial Φst explained a marginally greater proportion

of variation both among regions and among subregions

than microsatellite Rst, supporting the impression given

above that levels of paternal gene flow tend to be higher

than maternal gene flow.

Migration rates and directions

Contemporary and historical migration rates among the

seven geographic regions estimated using BAYESASS

(Wilson & Rannala 2003) and MIGRATE (Beerli & Felsen-

stein 1999; Beerli 2006), respectively, revealed interest-

ing patterns. Estimated rates of gene flow out of the

Orkney Islands into other subregions within the East

Atlantic are almost an order of magnitude higher than

all other rates and are the only ones that are statistically

different from zero (Table 5). Although much less

pronounced, there is also a suggestion that the Baltic is

relatively isolated, experiencing low rates of immigra-

tion and emigration, and that Nova Scotia behaves in

the opposite way compared with the Orkney Islands,

with low rates of emigration but moderate levels of

immigration.

Table 2 Estimates of genetic differentiation among the three geographic regions; (a) pairwise Fst and Φst values (above and below

the diagonal, respectively) for the mitochondrial HVR1; (b) pairwise Fst and Rst values (above and below the diagonal, respectively)

based on five microsatellite loci

Region Western Atlantic Eastern Atlantic Baltic

(a) Mitochondrial HVR1 Western Atlantic – 0.28*** 0.18***

Eastern Atlantic 0.08*** – 0.10***

Baltic 0.06*** 0.06*** –

(b) Microsatellites Western Atlantic – 0.11*** 0.17***

Eastern Atlantic 0.08*** – 0.24***

Baltic 0.13*** 0.09*** –

*P < 0.05; **P < 0.01; ***P < 0.001.

Table 3 Estimates of genetic differentiation among the seven subregions; (a) pairwise Fst and Φst values (above and below the diago-

nal, respectively) for the mitochondrial HVR1; (b) pairwise Fst and Rst values (above and below the diagonal, respectively) based on

five microsatellite loci

Subregion

Nova

Scotia

Orkney

Islands

North West

Scotland

Eastern

Scotland

North East

England

North

Atlantic Baltic

(a) Mitochondrial

HVR1

Nova Scotia – 0.28*** 0.28*** 0.41*** 0.35*** 0.26*** 0.19***

Orkney Islands 0.08*** – 0.01 0.05*** 0.02** 0.01 0.10***

North West Scotland 0.07*** 0.01* – 0.06*** 0.03** �0.01 0.07***

Eastern Scotland 0.13*** 0.04*** 0.05*** – �0.01 0.13*** 0.15***

North East England 0.12*** 0.04*** 0.05*** �0.01 – 0.07*** 0.11***

North Atlantic 0.09*** 0.03** 0.02* 0.10*** 0.09*** – 0.06***

Baltic 0.06*** 0.06*** 0.06*** 0.10*** 0.09*** 0.07*** –

(b) Microsatellites Nova Scotia – 0.08*** 0.07*** 0.07*** 0.06*** 0.09*** 0.13***

Orkney Islands 0.11*** – �0.01 0.01 �0.01 0.01 0.09***

North West Scotland 0.10*** �0.01 – �0.01 �0.01 0.01 0.07***

Eastern Scotland 0.07*** �0.01 �0.01 – �0.02 0.01 0.08***

North East England 0.09*** �0.01 �0.02 �0.03 – �0.01 0.05***

North Atlantic 0.13*** �0.02 �0.01 �0.01 �0.04 – 0.08***

Baltic 0.24*** 0.17*** 0.16*** 0.17*** 0.09*** 0.15*** –

*P < 0.05; **P < 0.01; ***P < 0.001.



Longer-term estimates of gene flow tend to be some-

what smaller than the contemporary estimates (Table 6),

but in general are strongly correlated with them (Man-

tel’s r = 0.49, n = 42, P = 0.004) indicating a broadly

consistent signal over time. As with contemporary rates,

the Orkney Islands appear to be the largest source of

emigration, while the Baltic appears relatively isolated.

Interestingly, BAYESASS and MIGRATE reveal subtle but

plausible differences as well as similarities. While con-

temporary emigration from the Orkney Islands is both

very high and largely dominated by movement to other

colonies in the East Atlantic, the historical pattern sug-

gests that local emigration is exceeded by that to the

West Atlantic and Baltic. Similarly, the strongly asym-

metrical pattern inferred for recent gene flow out of

and into Nova Scotia is not apparent in the deeper

history.

Elsewhere, it has been shown that Bayesian migration

rate estimates can be sensitive to differences in

sample size (Faubet et al. 2007). To test whether the

(b) Microsatellites

Nor

th R

onaC

alf of Flo

tta

Sable Island

Faroes FarnesIsle of May

Monachs

Rusk holm

Calf

of E

day

Fara

y Corn Holm

Ho

lm o

f Spu

rness

Links Ness

Holm of HuipStroma

Stockholm

Estonia

Bay of Bothnia

StromaSw

itha

Muckle G

reenholm

Co

pin

say

(a) Mitochondrial DNA

Sabl

e Isl

and

Bay of Bothnia

Estonia

Stockholm

Isle of May

Farnes

Calf of Eday

North Rona

Holm

of SpurnessC

orn

Ho

lmFa

ray

Copi

nsay

Swona

Muckle Greenholm

Rusk holm

StromaHolm of Huip

Links Ness

Calf of Flotta

Switha

hca

no

Ms

Faro

es0.03

0.04

Fig. 3 Neighbour-joining trees con-

structed for (a) the mitochondrial HVR1

and (b) microsatellites using the genetic

distance measures Φst and Dc, respec-

tively (See ‘Materials and methods’ for

details).



disproportionately large sample size of the Orkney

Islands could have affected our results, we therefore

repeated the BAYESASS and MIGRATE analyses after restrict-

ing the Orkney Island sample to a total of 134 individu-

als selected at random from within each of the breeding

colonies. The overall pattern obtained was very similar,

with estimated migration rates out of the Orkney Islands

being somewhat lower than for the full data set but still

significantly different from zero (Tables S4 and S5,

Supporting information). This suggests that the strongly

asymmetric pattern of gene flow out of the Orkney

Islands is not an artefact of unequal sample sizes.

Bayesian analysis of divergence times and effectivepopulation sizes

We next used the program IMA2 (Hey 2010) to estimate

divergence dates and effective population sizes from

our mitochondrial data. Although we were primarily

interested in deriving parameters relating to the diver-

gence of the Baltic region from the Eastern Atlantic, we

also analysed divergence between the Western and

Eastern Atlantic. The resulting marginal posterior prob-

ability distributions of divergence time and effective

population size parameters show clear peaks bounded

within the prior distribution in all pairwise comparisons

(Fig. 4). Parameter estimates together with 95% highest

probability density intervals are also given in Table S6

(Supporting information). Figure 4a shows clear sup-

port for the Baltic having diverged from the Eastern

Atlantic a little over 10 000 years ago, which is consis-

tent with geological records pertaining to the opening

of the Baltic Sea. The posterior probability distribution

for the divergence time of the Western and Eastern

Atlantic is somewhat flatter but indicates a more

ancient divergence, which probably took place around

30 000 years ago. We also obtained smaller Ne estimates

for ancestral than contemporary populations (Fig. 4b),

Table 4 Results of analyses of molecular variance (AMOVA) based on (a, b) the mitochondrial HVR1 sequence data; and (c, d) micro-

satellites, with the data set being partitioned into 3 geographic regions and 7 subregions respectively

Partition Source of variation d.f. Variance

% of total

variance F P

(a) Mitochondrial

DNA (using Fst)

3 regions Among regions 2 0.03 6.08 0.061 <0.001Among colonies within regions 19 0.01 0.72 0.008 <0.001Within colonies 1603 0.46 93.2 0.068 <0.001

7 subregions Among subregions 6 0.02 4.66 0.047 <0.001Among colonies within subregions 15 0 0.02 0.000 0.45

Within colonies 1603 0.46 95.32 0.047 <0.001(b) Mitochondrial

DNA (using Φst)3 regions Among regions 2 0.48 17.9 0.179 <0.001

Among colonies within regions 19 0.02 0.7 0.008 <0.001Within colonies 1603 2.17 81.4 0.186 <0.001

7 subregions Among subregions 6 0.26 10.6 0.106 <0.001Among colonies within subregions 15 0.01 0.25 0.003 0.14

Within colonies 1603 2.22 89.1 0.109 <0.001

Partition Source of variation Sum of squares Variance

% of total

variance F P

(c) Microsatellites

(using Fst)

3 regions Among regions 102.8 0.17 8.5 0.085 <0.001Among colonies within regions 58.0 0.01 0.4 0.004 <0.001Among individuals within colonies 2872.3 0.04 1.9 0.021 <0.001Within individuals 2792.5 1.80 89.2 0.108 <0.001

7 subregions Among subregions 117.1 0.08 4.2 0.042 <0.001Among colonies within subregions 43.8 0.01 0.3 0.003 <0.001Among individuals within colonies 2872.5 0.04 2.0 0.021 <0.001Within individuals 2792.7 1.80 93.4 0.066 <0.001

(d) Microsatellites

(using Rst)

3 regions Among regions 8908.4 15.4 15.8 0.158 <0.001Among colonies within regions 2069.6 1.19 0.2 0.002 0.025

Among individuals within colonies 123965.9 �0.31 �0.3 �0.004 0.627

Within individuals 126709.5 81.8 84.3 0.157 <0.0017 subregions Among subregions 9138.1 6.80 7.7 0.077 <0.001

Among colonies within subregions 1839.9 0.25 0.3 0.003 0.008

Among individuals within colonies 123965.9 �0.31 �0.4 �0.003 0.619

Within individuals 126709.5 81.8 92.4 0.076 0.003



Table

5Rates

ofrecentgen

eflow

amongseven

subregionscalculated

usingfivemicrosatellites

within

theprogram

BAYESASS(W

ilson&

Ran

nala2003).Resultsaregiven

asm,

theproportionofmigrants

per

gen

eration,from

each

ofthesu

bregionsontheleft

(row

headings)

into

thesu

bregionsontheright(columnheadings).95%

confiden

ceintervals

aregiven

inparen

theses

Orkney

Islands

NorthWestScotlan

dEastern

Scotlan

dNorthEastEngland

NorthAtlan

tic

NovaScotia

Baltic

Orkney

Islands

–0.290(0.257–0.322)

0.282(0.244–0.319)

0.261(0.212–0.310)

0.261(0.205–0.317)

0.022(0–0.059)

0.007(0–0.017)

NorthWestScotlan

d0.002(0–0.007)

–0.012(0–0.034)

0.014(0–0.039)

0.020(0–0

.056)

0.017(0–0.048)

0.008(0–0.020)

Eastern

Scotlan

d0.002(0–0.005)

0.010(0–0.028)

–0.020(0–0.051)

0.014(0–0

.042)

0.014(0–0.04)

0.005(0–0.014)

NorthEastEngland

0.001(0–0.002)

0.008(0–0.022)

0.008(0–0.022)

–0.010(0–0

.028)

0.013(0–0.037)

0.004(0–0.011)

NorthAtlan

tic

0.001(0–0.002)

0.006(0–0.017)

0.007(0–0.021)

0.009(0–0.026)

–0.013(0–0.037)

0.004(0–0.012)

NovaScotia

0.001(0–0.002)

0.006(0–0.018)

0.008(0–0.023)

0.009(0–0.024)

0.010(0–0

.028)

–0.003(0–0.009)

Baltic

0.000(0–0.001)

0.005(0–0.014)

0.006(0–0.016)

0.008(0–0.023)

0.009(0–0

.026)

0.010(0–0.029)

–

Tab

le6

Modal

estimates

ofhistoricalgen

eflow

among

seven

subregionscalculated

using

fivemicrosatellites

within

theprogram

MIG

RATE(Beerli&

Felsenstein

1999;Beerli

2006).Resultsaregiven

asm,theproportionofmigrants

per

gen

eration,from

each

ofthesu

bregionsontheleft

(row

headings)

into

thesu

bregionsontheright(columnhead-

ings).95%

confiden

ceintervalsaregiven

inparen

theses

Orkney

Islands

NorthWestScotlan

dEastern

Scotlan

dNorthEastEngland

NorthAtlan

tic

NovaScotia

Baltic

Orkney

Islands

–0.025(0.010–0.040)

0.016(0.004–0.030)

0.011(0–0

.020)

0.009(0–0.019)

0.011(0–0.020)

0.019(0.008–0.030)

NorthWestScotlan

d0.006(0–0.015)

–0.005(0–0

.016)

0.005(0–0

.014)

0.005(0–0.016)

0.006(0–0.015)

0.003(0–0.011)

Eastern

Scotlan

d0.005(0–0.013)

0.004(0–0.012)

–0.003(0–0

.010)

0.003(0–0.010)

0.003(0–0.011)

0.003(0–0.011)

NorthEastEngland

0.004(0–0.013)

0.001(0–0.009)

0.004(0–0

.012)

–0.003(0–0.010)

0.005(0–0.013)

0.003(0–0.010)

NorthAtlan

tic

0.006(0–0.017)

0.004(0–0.014)

0.004(0–0

.015)

0.004(0–0

.014)

–0.003(0–0.011)

0.003(0–0.010)

NovaScotia

0.004(0–0.012)

0.006(0–0.016)

0.004(0–0

.012)

0.003(0–0

.011)

0.004(0–0.012)

–0.007(0–0.017)

Baltic

0.005(0–0.013)

0.001(0–0.009)

0.003(0–0

.010)

0.003(0–0

.011)

0.003(0–0.011)

0.007(0–0.017)

–



consistent with range-wide historical population expan-

sion. Interestingly, the numerically smallest contempo-

rary population in the Baltic region recovered the

largest estimate of Ne.

Approximate Bayesian Computation (ABC) analysis

Finally, we analysed the combined mitochondrial and

microsatellite data from each of the three main grey seal

regions within an ABC framework to evaluate relative

statistical support for the following five demographic

scenarios: (i) constant population size; (ii) population

expansion; (iii) population reduction; (iv) population

bottleneck; and (v) population expansion followed by

reduction (see Fig. S2, Supporting information for fur-

ther details). The posterior probability of each compet-

ing scenario was evaluated using a polychotomous

logistic regression on the 1% of simulated data sets clos-

est to the observed. The population expansion model

was clearly the most probable for the Baltic and Eastern

Atlantic, but results obtained for the Western Atlantic

were less clear-cut, with a 68.3% probability being

recovered in favour of the expansion model. We next

evaluated the power of the model choice procedure by

calculating type I and II error rates for the best-sup-

ported model. In all cases, type I error ranged from

16.8% to 19.2% and type II error rates were <10% (Table

S7, Supporting information). For each population, we

estimated the marginal posterior probability density of

each parameter based on expansion scenarios using the

1% closest simulated data sets from the observed data

set. Posterior estimates of expansion time and contem-

porary Ne were similar among regions, although pre-

expansion Ne appears to have been largest for the Baltic

region (Fig. 5 and Table 7).

Discussion

We conducted a comprehensive survey of the popula-

tion structure and historical demography of a widely

distributed pinniped species, the grey seal. Strong

genetic structuring was found in both the mitochondrial

and nuclear data sets, which becomes progressively

weaker over finer geographic scales. Signatures of pop-

ulation expansion were detected for all three regions

using ABC. Migration rate estimates revealed added

complexity, with a possible key role for the Orkney

Islands, both historically and currently, while the Baltic

appears relatively isolated.

The literature contains a wealth of increasingly

sophisticated statistical tools for inferring demographic

parameters from genetic data. We have deployed a

number of these to analyse a large data set from the

grey seal while endeavouring to minimize the overlap.

Thus, we aimed to infer current population substruc-

ture, current and historical migration rates and histori-

cal changes in population size using both nuclear and

mitochondrial data. While a reasonably cohesive picture

emerges, it is also noticeable that many simplifying

assumptions still need to be made because no one pro-

gram yet estimates all parameters simultaneously. Thus,

IMA2 assumes constant population sizes, something that

historical records and other genetic analyses show to be

false. Equally, to estimate recent patterns of gene flow

requires an assumption of approximate equilibrium in

(a)

(b)

Fig. 4 Marginal posterior probability distributions obtained in

pairwise IMa2 analyses between the Eastern Atlantic and Baltic

and between the Western and Eastern Atlantic (see ‘Materials

and methods’ for details). (a) Divergence times in years, (b)

Effective population sizes. ‘Ancestral 1’ corresponds to Ne

prior to the divergence of the Western and Eastern Atlantic,

and ‘Ancestral 2’ corresponds to Ne prior to the divergence of

the Baltic from the Eastern Atlantic. Note that two Ne estimates

are available for the Eastern Atlantic because this region fea-

tures in two pairwise comparisons (‘Eastern Atlantic 1’ and

‘Eastern Atlantic 2’ correspond to analyses involving the Baltic

and Western Atlantic, respectively).



the period immediately before. Moreover, the observed

patterns could potentially be affected by unsampled

‘ghost’ populations in other parts of the Eastern Atlan-

tic range such as the Wadden Sea, Denmark, Norway

and Iceland. For these reasons, although we expect the

overall patterns to be genuine, no one result should be

taken entirely at face value.

Population structure

Bearing in mind the provisos above, there is a strong

consensus that the Baltic, Western Atlantic and Eastern

Atlantic show a high degree of genetic differentiation at

both markers. This is expected given the geographic

and ‘anthropogenic’ barriers that have largely isolated

the Baltic population from the remainder of the species’

distribution and the large geographic separation

between the Western and Eastern Atlantic. However,

there is a slight contradiction. Based on the two pre-

ferred measures, Φst and Rst, the most dissimilar popu-

lation pairs are the Western and Eastern Atlantic

according to mtDNA data, and the Baltic and Western

Atlantic according to nuclear data. It is unclear whether

this discrepancy is genuine or arises out of statistical

noise. If it is genuine, then the contrasting patterns

could arise in either of two ways. If the Atlantic popu-

lations were ancestrally comparatively small, this could

have allowed more rapid divergence within the Atlantic

at mitochondrial markers due to their lower effective

population size. Alternatively, if the Baltic became more

isolated and gene flow reduced across the Atlantic,

the faster mutation rate of microsatellites would allow

the Baltic population to diverge faster, particularly from

the more geographically distant Western Atlantic.

Below the level of the three major regions, we again

find contrasting patterns between nuclear and mito-

chondrial markers. This is seen in Table 3, where pair-

wise comparisons not involving the Baltic or Nova

Scotia in the Western Atlantic never reveal significant

differences using microsatellites but almost invariably

do show differentiation using mitochondrial DNA. This

pattern probably stems from the greater fidelity to pup-

ping sites shown by adult female seals, many of whom

return to breed time and again within a few tens of

metres of where they bred in previous seasons (Pome-

roy et al. 2000). In contrast, males, not being tied to

newborn pups, have the potential to visit more than

one breeding colony during a single breeding season.

Indeed, this is made easier by the way the timing of the

breeding season varies around the UK. How many

males do this remains to be determined, although radio-

and satellite-tagging studies suggest it may be fairly

common (Thompson et al. 1996; McConnell et al. 1999;

Jessop et al. 2013). Having said this, male-mediated

(a)

(b)

(c)

Fig. 5 Posterior density curves of model parameters based

on 10 000 accepted values from 1 9 106 iterations of the popula-

tion expansion model (see ‘Materials and methods’ for details)

shown separately for grey seals from the Western Atlantic (full

lines), Eastern Atlantic (dashed lines) and Baltic (dotted lines).

Results are shown for (a) contemporary population size, (b)

expansion time, and (c) Pre-expansion population size.



gene flow cannot be very high, at least in certain areas

of the species distribution, because previous studies

have identified highly significant microsatellite allele

frequency differences between UK colonies (Allen et al.

1995) and in our current data set, AMOVA identifies a

small but highly significant proportion of the total vari-

ation attributable to between subregion differences.

Migration rates

Our estimates of recent and historical gene flow are

intriguing in that they both identify the Orkney Islands

as playing a pivotal role in regional grey seal demogra-

phy. High levels of recent gene flow out of the Orkney

Islands have already been reported (Gaggiotti et al.

2002) and seem to reflect a general expansion of the

entire grey seal population in this area. Once a colony

reaches its local carrying capacity, groups of emigrants

leave to seek less crowded breeding areas, preferring

nearer sites but certainly being capable of moving much

further away. Overall, therefore, our results suggest that

many colonies in the Orkney Islands may be at, or close

to, carrying capacity and this both encourages emigra-

tion and discourages immigration. Although colonies

elsewhere in the Eastern Atlantic may also be approach-

ing or have reached their carrying capacity, it is inter-

esting that the Orkney Islands are the sole subregion

that reveals a strong signal of emigration, mostly to

other sites within the Eastern Atlantic, although with

weak evidence of perhaps indirect movement as far as

Sable Island via colonies in southern Greenland (Ro-

sing-Asvid et al. 2010). Although not undertaking sea-

sonal migrations such as ringed seals (Pusa hispida),

harp seals (Pagophilus groenlandicus) and hooded seals

(Cystophora cristata), long-distance movements of indi-

vidual grey seals within the Eastern Atlantic have been

documented by tagging studies (Thompson et al. 1996;

McConnell et al. 1999; Jessop et al. 2013) and, over time,

stepping-stone-like movements between the Western

and Eastern Atlantic could potentially occur via colo-

nies in southern Greenland (Rosing-Asvid et al. 2010)

and Iceland (Hauksson 2007a,b).

Over the longer term, the Orkney Islands also appear

to be important with all emigration values from this

area being greater than any other values (Table 6), even

for analyses based on the reduced data set (Table S6,

Supporting information). However, while contemporary

emigration from the Orkney Islands is both very high

and largely dominated by movement to other colonies

in the Eastern Atlantic, the historical pattern suggests

that local emigration was more restricted in the Eastern

Atlantic, mainly involving neighbouring colonies at

North Rona, the Monachs and the Isle of May. As else-

where, we must not forget the possibility that contem-

porary and historic estimates could be slightly distorted

by the much larger sample size of Orkney Island indi-

viduals. Nevertheless, the zooarchaeological record of

seals is particularly rich in the Orkney Islands relative

to other parts of the British Isles (Fairnell 2003), hinting

that the Orkney Islands is probably an area that has

been particularly suited to this species for a long time.

Consequently, it appears that, whenever conditions

allow, seals in this region have been able to fill avail-

able breeding sites and this has generated a reliable

source of emigrants who leave to seed other areas.

Divergence times

It is thought that a transient dispersal corridor connect-

ing the Baltic and North Seas formed around

10 000 years ago and closed during the formation of

Ancylus Lake around 1000 years later (Sommer &

Table 7 Prior uniform distributions, mean, median, mode, quantiles and estimates of bias and precision (MRB = mean relative bias,

RMSE = root mean square error) for the posteriors of parameters calculated from simulations of population expansion. N1, current

effective population size; N2, effective population size before expansion; T1, time from expansion (generations ago). See Fig. S2 and

Table S7 (Supporting information) for details

Region Parameter Prior

Posterior

MRB RMSEMean Median Mode 5% 95%

Western Atlantic N1 1–1.5 9 105 5.15 9 104 3.62 9 104 1.27 9 104 6.37 9 103 1.36 9 105 0.99 3.27

N2 100–10 000 4.84 9 103 4.68 9 103 4.03 9 103 1.23 9 102 4.46 9 103 0.74 3.26

T1 1–5000 1.68 9 103 1.23 9 103 3.67 9 102 9.67 9 102 9.20 9 103 2.77 28.38

Eastern Atlantic N1 1–1.5 9 105 3.92 9 104 2.62 9 104 1.58 9 104 7.71 9 103 1.19 9 105 0.68 2.95

N2 100–10000 4.87 9 103 4.72 9 103 4.25 9 103 7.74 9 102 9.28 9 103 0.64 2.01

T1 1–5000 1.94 9 103 1.69 9 103 8.70 9 101 1.02 9 102 4.59 9 103 1.52 6.93

Baltic N1 1–1.5 9 105 4.68 9 104 2.97 9 104 1.20 9 104 8.10 9 103 1.32 9 105 0.88 3.43

N2 100–10 000 6.10 9 103 6.34 9 103 7.71 9 103 1.73 9 103 9.60 9 103 0.79 2.46

T1 1–5000 2.05 9 103 1.74 9 103 1.77 9 102 1.39 9 102 4.70 9 103 7.39 99.18



Benecke 2003). Previous genetic studies provide rather

broad estimates of the divergence time of the Baltic and

Eastern Atlantic grey seal regions, ranging between

11 000 and 0.35 million years ago depending on data

type, mutation rate assumptions and the method of esti-

mation (Boskovic et al. 1996; Graves et al. 2009). Bosko-

vic et al. (1996) interpreted their upper estimate (0.35

million years ago) as meaning that an ancient, unknown

event may have driven the separation of the Baltic

seals, but also cautioned that their mutation rate

assumption could be incorrect. Our estimate of 7000–

30 000 years ago is far closer to the previous lower esti-

mate (Graves et al. 2009) and includes the narrow time

window during which the dispersal corridor was

known to be open. This provides support for the

hypothesis that the opening and subsequent closing of

the Baltic Sea was the primary factor promoting the

genetic divergence of the Baltic grey seal. Also,

although capable of moving long distances, an addi-

tional factor leading to increased divergence between

Eastern Atlantic and Baltic sea populations could be the

loss of intermediate haul-out sites, and hence dispersal

corridors, caused by anthropogenic disturbance, popu-

lation size reductions and local extinctions of grey seal

populations starting the Middle Ages in the Nether-

lands, Germany, Denmark and western Sweden.

Effective population sizes

Counterintuitively, the numerically smaller Baltic grey

seal region recovered a larger estimate of Ne than either

the Western or Eastern Atlantic regions. This is consis-

tent, however, with previous findings by Graves et al.

(2009) as well as with the mitochondrial network dia-

gram (Fig. 2). Here, one can see the origin of the signal

of population expansion, with a few common haplo-

types surrounded by many closely related but rare satel-

lite haplotypes. It is also clear why the Baltic appears to

be the larger population because the Baltic haplotype

network covers a wider range, at least in this two

dimensional representation, and shows affinities with

both the Western and the Eastern Atlantic. This might

indicate that the Baltic population in some sense carries

a more ancestral signature, although a similar pattern

could plausibly arise through drift alone, with the two

Atlantic populations both having lost one of their two

deepest branches. Differences in social group structure

and breeding behaviour might also potentially contrib-

ute towards the observed differences in effective popu-

lation size. Grey seals in the British Isles tend to breed

in colonies, whereas Baltic grey seals breed on ice or

land, depending on the prevailing environmental condi-

tions (Jussi et al. 2008). The more dynamic nature of the

Baltic breeding habitat may therefore facilitate greater

admixture, which would tend to inflate Ne relative to

the more static, colonial structure of British grey seals.

Historical demographic patterns

Trends in historical population size were studied using

several different methods applied to the mitochondrial,

nuclear and combined data sets. The dominant feature

supported in all cases was a period of recent population

expansion, with historical population sizes consistently

appearing smaller than modern sizes. This is seen most

intuitively in the simple network diagram, in which

small numbers of common haplotypes are surrounded

by haloes of rare haplotypes differing by a single substi-

tution, the classical signal of population expansion. Sim-

ilarly, although no single model is strongly supported

for the Western Atlantic, overall ABC consistently

favours population expansion over both the simple con-

stant population size model and more complicated

models featuring either recent bottlenecks or periods of

expansion and contraction.

Modal estimates of expansion times from the ABC

analysis range from approximately 87–367 generations

ago. To convert generations to time, we assume a gener-

ation time of 14 years (Thompson & H€ark€onen 2008),

implying that expansion occurred around 1218–

5380 years ago. However, generation length is usually

appreciably shorter during population expansion than

when a population is at carrying capacity (Harwood &

Prime 1978). As our estimate was based on an expand-

ing population, it is likely to be an underestimate,

implying that the likely date range for expansion is

appreciably older.

The confidence intervals on all date ranges are inevi-

tably large. We chose HVRI and microsatellites because

their high mutation rates afford good resolution for

recovering the branching pattern of lineages within

each population. However, regardless of the accuracy

by which the actual lineages are reconstructed, the spar-

sity of deeper nodes and the stochasticity inherent in

lineage sorting means that the older regions of any

given gene tree are consistent with a broad range of

demographic scenarios. Consequently, while the expan-

sion times estimated for each of the regions are broadly

consistent, we cannot say with confidence that they do

not differ.

With the above provisos, the observed pattern of

range-wide population expansion is consistent with the

increase in habitat availability that would have occurred

as the ice sheets retreated after the last glacial maximum

(LGM). At the peak of the LGM, the grey seal’s range in

the western North Atlantic appears to have been

restricted to the continental shelf edge between around

40 and 45°N and around the Flemish Cap, whereas on



the eastern side, only the coast of the Bay of Biscay and

the northeastern coast of the Iberian Peninsula would

have been suitable (Boehme et al. 2012). It has been esti-

mated that the total amount of habitat available for grey

seals globally could have been as little as 3% of today’s

value (Boehme et al. 2012). Moreover, regional primary

productivity would probably also have been severely

depleted (Pyenson & Lindberg 2011).

Finally, genetic studies have explored the population

structure of several other Northern Hemisphere pinni-

ped species (e.g. Stanley et al. 1996; Andersen et al.

1998; Goodman 1998; Hoffman et al. 2006a,b, 2009; Mar-

tinez-Bakker et al. 2013), raising the question of whether

other species show evidence of parallel demographic

responses due to changes in their shared environment.

Unfortunately, few of the other studies tested for histor-

ical demographic patterns and, of those that did, even

fewer provide putative dates for historical expansions.

Comparisons among studies are also hampered by dif-

ferences in genetic markers used, the analytical

approaches and the assumptions made about mutation

rate and generation time. Nevertheless, our results are

broadly in line with findings from northern fur seals,

which appear to have undergone population expansion

around 11 000 years ago (Dickerson et al. 2010) and har-

bour (Westlake & O’Corry-Crowe 2002) and hooded

seals (Coltman et al. 2007), which both show signals of

recent population expansion interpreted as being con-

sistent with postglacial expansion.

Conclusion

We elucidated the population structure and historical

demography of grey seals globally. The overall pattern

that emerges is one of strong population structure cou-

pled with range-wide historical population expansion.

Highly asymmetrical patterns of gene flow were also

inferred, with the Orkney Islands being identified as a

source of migrants for many other populations. Taken

together, these findings suggest that grey seals are

highly responsive to changes in habitat availability and

that the Orkney Islands may play a pivotal role in local

population dynamics.

Acknowledgements

We are grateful to Phil Harrison for assisting with the collec-

tion of grey seal tissue samples from the Orkney Islands. We

are also indebted to Jeff Graves for sharing genetic data for the

Baltic grey seal populations. Samples were collected and

retained under permits issued by the Department for Environ-

ment, Food and Rural Affairs (DEFRA). JIH was funded by a

Marie Curie FP7-Reintegration-Grant within the 7th European

Community Framework Programme (PCIG-GA-2011-303618).

References

Allen PJ, Amos W, Pomeroy PP, Twiss SD (1995) Microsatellite

variation in grey seals (Halichoerus grypus) shows evidence of

genetic differentiation between two British breeding colonies.

Molecular Ecology, 4, 653–662.Amos W, Hoelzel AR (1991) Long-term preservation of whale

skin for DNA analysis. Report of the International Whaling

Commission, Special Issue 13, 99–103.

Andersen LW, Born EW, Gjertz I et al. (1998) Population structure

and gene flow of the Atlantic walrus (Odobenus rosmarus rosma-

rus) in the eastern Atlantic Arctic based on mitochondrial DNA

and microsatellite variation.Molecular Ecology, 7, 1323–1336.

Arnason U, Gullberg A, Johnsson E, Ledje C (1993) The nucleo-

tide sequence of the mitochondrial-DNA molecule of the

gray seal, Halichoerus grypus, and a comparison with mito-

chondrial sequences of other true seals. Journal of Molecular

Evolution, 37, 323–330.

Bandelt H-J, Forster P, Rohl A (1999) Median-joining networks

for inferring intraspecific phylogenies. Molecular Biology and

Evolution, 16, 37–48.Bean K, Amos W, Pomeroy PP et al. (2004) Patterns of parental

relatedness and pup survival in the grey seal (Halichoerus

grypus). Molecular Ecology, 13, 2365–2370.

Beaumont MA, Rannala B (2004) The Bayesian revolution in

genetics. Nature Reviews Genetics, 5, 251–261.

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

bayesian computation in population genetics. Genetics, 162,

2025–2035.Beerli P (2006) Comparison of Bayesian and maximum- likeli-

hood inference of population genetic parameters. Bioinformat-

ics, 22, 341–345.

Beerli P (2009) How to use migrate or why are markov chain

monte carlo programs difficult to use? In: Population Genetics

for Animal Conservation (eds Bertorelle G, Bruford MW, Haue

HC, Rizzoli A, Vernesi C), pp. 42–79. Cambridge University

Press, Cambridge.

Beerli P, Felsenstein J (1999) Maximum-likelihood estimation of

migration rates and effective population numbers in two pop-

ulations using a coalescent approach. Genetics, 152, 763–773.

Bertorelle G, Benazzo A, Mona S (2010) ABC as a flexible

framework to estimate demography over space and time:

some cons, many pros. Molecular Ecology, 19, 2609–2625.Boehme L, Thompson D, Fedak M et al. (2012) How many

seals were there? The global shelf loss during the last glacial

maximum and its effect on the size and distribution of grey

seal populations. PLoS ONE, 7, e53000.

Bonner WN (1981) Grey seal, Halichoerus grypus. In: Handbook

of Marine Mammals (eds Ridgway SH, Harrison RJ), pp. 111–

144. Academic Press, London.

Boskovic R, Kovacs KM, Hammill MO, White BN (1996) Geo-

graphic distribution of mitochondrial DNA haplotypes in

grey seals (Halichoerus grypus). Canadian Journal of Zoology-

Revue Canadienne De Zoologie, 74, 1787–1796.Bowen WD, Lidgard D (2012) Marine mammal culling pro-

grams: review of effects on predator and prey populations.

Mammal Review, 43, 207–220.

Cammen K, Hoffman JI, Knapp LA, Harwood J, Amos W

(2011) Geographic variation of the major histocompatibility

complex in Eastern Atlantic grey seals (Halichoerus grypus).

Molecular Ecology, 20, 740–752.



Cavalli-Sforza LL, Edwards AWF (1967) Phylogenetic analysis:

models and estimation procedures. American Journal of

Human Genetics, 19, 233–257.

Coltman DW, Stenson G, Hammill MO et al. (2007) Panmictic

population structure in the hooded seal (Cystophora cristata).

Molecular Ecology, 16, 1639–1648.Cornuet JM, Santo F, Beaumont MA et al. (2008) Inferring pop-

ulation history with DIYABC: a user-friendly approach to

Approximate Bayesian Computations. Bioinformatics, 24,

2713–2719.Cornuet JM, Ravigne 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, 104.

Curtis C, Stewart BS, Karl SA (2009) Pleistocene population

expansions of Antarctic seals.Molecular Ecology, 18, 2112–2121.De Bruyn M, Hall BL, Chauke LF et al. (2009) Rapid response

of a marine mammal species to Holocene climate and habitat

change. PLoS Genetics, 5, e10000554.

Dickerson BR, Ream RR, Vignieri SN, Bentzen P (2010) Popula-

tion structure as revealed by mtDNA and microsatellites in

Northern fur seals, Callorhinus ursinus, throughout their

range. PLoS ONE, 5, e10671.

Excoffier L, Smouse PE (1994) Using allele frequencies and geo-

graphic subdivision to reconstruct gene trees within a species:

molecular variance parsimony. Genetics, 136, 343–359.Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecu-

lar variance inferred from metric distances among mtDNA

haplotypes: application to human mitochondrial DNA

restriction data. Genetics, 131, 479–491.

Fagundes NJR, Ray N, Beaumont MA et al. (2007) Statistical

evaluation of alternative models of human evolution. Pro-

ceedings of the National Academy of Sciences of the United States

of America, 104, 17614–17619.

Fairnell EH (2003) The Utilisation of fur-Bearing Animals in the

British Isles - A Zooarchaeological Hunt for Data. MSC in

Zooarchaeology. The University of York, UK.

Faubet P, Waples RS, Gaggiotti OE (2007) Evaluating the per-

formance of a multilocus Bayesian method for the estimation

of migration rates. Molecular Ecology, 16, 1149–1166.

Felsenstein J (2006) Accuracy of coalescent likelihood estimates:

do we need more sites, more sequences, or more loci? Molec-

ular Biology and Evolution, 23, 691–700.Fietz K, Graves JA, Olsen MT (2013) Control control control: a

reassessment and comparison of GenBank and chromato-

gram mtDNA sequence variation in baltic grey seals (Halic-

hoerus grypus). PLoS ONE, 8, e72853.

Gaggiotti OE, Jones FC, Lee WM et al. (2002) Patterns of

colonization in a metapopulation of grey seals. Nature, 416,

424–427.

Goodman SJ (1997) Dinucleotide repeat polymorphisms at seven

anonymous microsatellite loci cloned from the European Har-

bour Seal (Phoca vitulina vitulina). Animal Genetics, 28, 310–311.Goodman SJ (1998) Patterns of extensive genetic differentiation

and variation among European harbor seals (Phoca vitulina

vitulina) revealed using microsatellite DNA polymorphisms.

Molecular Biology and Evolution, 15, 104–118.Graves JA, Helyar A, Biuw M et al. (2009) Microsatellite and

mtDNA analysis of the population structure of grey seals

(Halichoerus grypus) from three breeding areas in the Baltic

Sea. Conservation Genetics, 10, 59–68.

Hall TA (1999) BioEdit: a user-friendly biological sequence

alignment editor and analysis program for windows 95/98/

NT. Nucleic Acids Symposium Series, 41, 95–98.

Hall A (2002) Gray seal Halichoerus grypus. In: Encyclopedia of

Marine Mammals (eds Perrin WF, Wursig J, Thewissen GM),

pp. 522–524. Academic Press, San Diego, California.

Harding KC, H€ark€onen TJ (1999) Development in the Baltic

grey seal (Halichoerus grypus) and ringed seal (Phoca hispida)

populations during the 20th century. Ambio, 28, 619–627.

H€ark€onen T, Brasseur S, Teilmann J et al. (2007) Status of grey

seals along mainland Europe from the Southwestern Baltic

to France. NAMMCO Scientific Publications, 6, 57–68.Harwood J, Prime JH (1978) Some factors affecting the size of

British grey seal populations. Journal of Applied Ecology, 15,

401–411.

Hauksson E (2007a) Abundance of grey seals in Icelandic

waters, based on trends of pup-counts from aerial surveys.

NAMMCO Scientific Publications, 6, 85–97.Hauksson E (2007b) Abundance of grey seals in Icelandic

waters, based on trends of pup-counts from aerial surveys.

NAMMCO Scientific Publications, 6, 85–97.

HELCOM (2013) Baltic Grey Seal Censuses in 2012. Helsinki

Commission, Finland. Available from http://www.helcom.fi/.

Hey J (2010) Isolation with migration models for more than

two populations. Molecular Biology and Evolution, 27, 905–920.

Hey J, Nielsen R (2004) Multilocus methods for estimating

population sizes, migration rates and divergence time, with

applications to the divergence of Drosophila pseudoobscura

and D. persimilis. Genetics, 167, 747–760.Hey J, Nielsen R (2007) Integration within the Felsenstein equa-

tion for improved Markov Chain Monte Carlo methods in

population genetics. Proceedings of the National Academy of

Sciences of the United States of America, 104, 2785–2790.Hiby L, Lundberg T, Karlsson O et al. (2007) Estimates of the

size of the Baltic grey seal population based on photo-identi-

fication data. NAMMCO Scientific Publications, 6, 163–175.

Hoffman JI, Forcada J (2012) Extreme natal philopatry in

female Antarctic fur seals (Arctocephalus gazella). Mammalian

Biology, 77, 71–73.Hoffman JI, Matson C, Amos W et al. (2006a) Deep genetic

subdivision within a continuously distributed and highly

vagile marine mammal, the Steller’s sea lion Eumetopias juba-

tus. Molecular Ecology, 15, 2821–2832.Hoffman JI, Trathan PN, Amos W (2006b) Genetic tracking

reveals extreme site fidelity in territorial male Antarctic fur

seals Arctocephalus gazella. Molecular Ecology, 15, 3841–3847.

Hoffman JI, Dasmahapatra KK, Amos W et al. (2009) Contrast-

ing patterns of genetic diversity at three different genetic

markers in a marine mammal metapopulation. Molecular

Ecology, 18, 2961–2978.

Hoffman JI, Grant SM, Forcada J, Phillips CD (2011) Bayesian

inference of a historical genetic bottleneck in a heavily

exploited marine mammal. Molecular Ecology, 20, 3989–4008.Jessop M, Cronin M, Hart T (2013) Habitat-mediated dive

behavior in free-ranging grey seals. PLoS ONE, 8, e63720.

Jussi M, Harkonen T, Helle E, Jussi I (2008) Decreasing ice cov-

erage will reduce the breeding success of Baltic Grey seal

(Halichoerus grypus) females. Ambio, 37, 80–85.

Kimura M (1980) A simple method for estimating evolutionary

rate of base substitution through comparative studies of nucle-

otide sequences. Journal of Molecular Evolution, 16, 111–120.



Kovacs KM, Lydersen C (2008) Climate change impacts on

seals and whales in the North Atlantic Arctic and adjacent

shelf seas. Scientific Progress, 91, 117–150.

Kuhner MK (2009) Coalescent genealogy samplers: windows

into population history. Trends in Ecology and Evolution, 24,

86–93.Langella O (1999) Populations. CNRS, Gif Sur Yvette, France.

Lesage V, Hammill MO (2001) The status of the grey Seal, Hal-

ichoerus grypus, in the Northwest Atlantic. Canadian Field Nat-

uralist, 115, 653–662.Lonergan M, Duck CD, Thompson D, Moss S, McConnell B

(2011) British grey seal (Halichoerus grypus) abundance in

2008: an assessment based on aerial counts and satellite

telemetry. ICES Journal of Marine Science, 68, 2201–2209.Martinez-Bakker ME, Sell SK, Swanson BJ, Kelly BP, Tallmon

DA (2013) Combined genetic and telemetry data reveal high

rates of gene flow, migration, and long-distance dispersal

potential in Arctic ringed seals (Pusa hispida). PLoS ONE, 10,

e77125.

Matthee CA, Fourie F, Oosthuizen WH, Meyer MA, Tolley KA

(2006) Mitochondrial DNA sequence data of the Cape fur

seal (Arctocephalus pusillus pusillus) suggest that population

numbers may be affected by climatic shifts. Marine Biology,

148, 899–905.McConnell BJ, Fedak MA, Lovell P, Hammond PS (1999)

Movements and foraging areas of grey seals in the North

Sea. Journal of Applied Ecology, 36, 573–590.

Olsen MT, Andersen LW, Dietz R et al. (2014) Integrating

genetic data and population viability analyses for the identi-

fication of harbour seal (Phoca vitulina) populations and man-

agement units. Molecular Ecology, 23, 815–831.Phillips CD, Trujillo RG, Gelatt TS et al. (2009) Assessing

substitution patterns, rates and homoplasy at HVRI of

Steller sea lions, Eumetopias jubatus. Molecular Ecology, 18,

3379–3393.Phillips CD, Gelatt TS, Patton JC, Bickham JW (2011) Phyloge-

ography of Steller sea lions: relationships among climate

change, effective population size, and genetic diversity. Jour-

nal of Mammalogy, 92, 1091–1104.Pomeroy PP, Anderson SS, Twiss SD, McConnell BJ (1994) Dis-

persion and site fidelity of breeding female grey seals (Halic-

hoerus grypus) on North Rona, Scotland. Journal of Zoology

(London), 233, 429–447.Pomeroy PP, Twiss SD, Redman P (2000) Philopatry, site fidel-

ity and local kin associations within grey seal breeding colo-

nies. Ethology, 106, 899–919.

Pyenson ND, Lindberg DR (2011) What happened to gray

whales during the Pleistocene? The ecological impact of sea-

level change on benthic feeding areas in the north Pacific

ocean. PLoS ONE, 6, e21295.

Rambaut A, Drummond AJ (2007) Tracer v1.4. Available from

http://beast.bio.ed.ac.uk/Tracer.

Raymond M, Rousset F (1995) Genepop (Version 1.2) - popula-

tion genetics software for exact tests of ecumenicism. Journal

of Heredity, 86, 248–249.Rosing-Asvid A, Teilmann J, Dietz R, Olsen MT (2010) First

record of Grey seals in Greenland. Arctic, 63, 471–473.Rozas J, Rozas R (1995) DnaSP, DNA sequence polymorphism:

an interactive program for estimating population genetics

parameters from DNA sequence data. Computer Applications

in the Biosciences, 11, 621–625.

Sambrook J, Fritsch EF, Maniatis T (1989) Molecular Cloning: A

Laboratory Manual, 2nd edn. Cold Spring Harbour Laboratory

Press, New York.

Schmolke U (2008) Holocene environmental changes and the

seal (Phocidae) fauna of the Baltic Sea: coming, going and

staying. Mammal Review, 38, 231–246.Schneider S, Roessli D, Excoffier L (2000) ARLEQUIN ver 2.000:

A Software for Population Genetics Data Analysis. Genetics and

Biometry Laboratory, Department of Anthropology and Ecol-

ogy, University of Geneva, Geneva, Switzerland.

Slatkin M (1995) A measure of population subdivision based on

microsatellite allele frequencies. Genetics, 139, 457–462.Sommer R, Benecke N (2003) Post-Glacial history of the Euro-

pean seal fauna on the basis of sub-fossil records. Beitr€age Zu

Arch€aozoologie und Pr€ahistorischer Anthropologie, 6, 16–28.

Stanley HF, Casey S, Carnahan JM et al. (1996) Worldwide

patterns of mitochondrial DNA differentiation in the harbor

seal (Phoca vitulina). Molecular Biology and Evolution, 13,

368–382.

Takezaki N, Nei M (1996) Genetic distances and reconstruction

of phylogenetic trees from microsatellite DNA. Genetics, 144,

389–399.Tamura K, Peterson D, Peterson N et al. (2011) MEGA5:

molecular evolutionary genetics analysis using maximum like-

lihood, evolutionary distance, and maximum parsimony

methods. Molecular Biology and Evolution, 28, 2731–2739.Thompson D, H€ark€onen T (2008) Halichoerus grypus. In: IUCN

Red List of Threatened Species. Version 2012.2. <www.iucnred-

list.org>.

Thompson PM, McConnell BJ, Tollit DJ et al. (1996) Compara-

tive distribution, movements and diet of harbour and grey

seals from the Moray Firth, NE Scotland. Journal of Applied

Ecology, 33, 1572–1584.Walsh PS, Metzger DA, Higuchi R (1991) Chelex100 as a med-

ium for simple extraction of DNA for PCR-based typing

from forensic material. BioTechniques, 10, 506–513.

Weber JL, Wong C (1993) Mutation of human short tandem

repeats. Human Molecular Genetics, 2, 1123–1128.

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

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

Westlake RL, O’Corry-Crowe GM (2002) Macrogeographic

structure and patterns of genetic diversity in harbor seals

(Phoca vitulina) from Alaska to Japan. Journal of Mammalogy,

83, 1111–1126.

Wilson GA, Rannala B (2003) Bayesian inference of recent

migration rates using multilocus genotypes. Genetics, 163,

1177–1191.Worthington Wilmer J, Allen PJ, Pomeroy PP, Twiss SD, Amos

W (1999) Where have all the fathers gone? An extensive mi-

crosatellite analysis of paternity in the grey seal (Halichoerus

grypus). Molecular Ecology, 8, 1417–1429.

J.I.H., W.A. and J.H. designed the research; J.I.H., W.A.

and J.H. performed the research; J.H. contributed

reagents/analytic tools; A.K., J.I.H., C.D.P., K.F. and

M.T.O. analysed the data; A.K., J.I.H., W.A., C.D.P. and

M.T.O. wrote the paper.



Data accessibility

Mitochondrial DNA sequences are available from Gen-

Bank (Accession nos KJ769328–KJ769461). Raw micro-

satellite genotypes and complete data set of the

mitochondrial control region sequences are available via

Dryad Digital Repository (doi:10.5061/dryad.r232 g).

Supporting information

Additional supporting information may be found in the online ver-

sion of this article.

Table S1 Prior distributions for demographic parameters in

the Approximate Bayesian Computation (ABC) analysis.

Table S2 Summary statics for the mitochondrial HVR1 dataset

broken down by region, sub-region and colony.

Table S3 Summary of the microsatellite loci used in this study,

including the number of alleles (A) per locus, allelic richness

(Ar), expected heterozygosity (HE), observed heterozygosity

(HO) and results of an exact test for Hardy-Weinberg equilib-

rium (P-values without Bonferroni correction).

Table S4 Rates of recent gene flow among seven sub-regions

calculated using five microsatellites within the program BAYE-

SASS (Wilson & Rannala 2003;) using a reduced dataset of 134

Orkney Island samples (See Materials and methods for details).

Table S5 Modal estimates of historical gene flow among seven

sub–regions calculated using five microsatellites within the

program MIGRATE (Beerli 2006; Beerli & Felsenstein 1999) using

a reduced dataset of 134 Orkney Island samples (See Materials

and methods for details).

Table S6 IMA2 estimates and 95% highest probability density

(HPD) intervals for various demographic parameters in pair-

wise comparisons between the Baltic and Eastern Atlantic and

between the Western and Eastern Atlantic (see Materials and

methods for details).

Table S7 Posterior probabilities of competing scenarios in the

approximate Bayesian computation (ABC) analysis, including

error rates for the top ranked scenarios for each region.

Fig. S1 Sampling locations of 13 grey seal colonies within the

Orkney Islands.

Fig. S2 Five alternative demographic scenarios evaluated in the

grey seal using Approximate Bayesian Computation (ABC).

See Materials and methods for details.



global population structure and demographic history of the grey seal · 2015. 1. 26. · a global...

Documents