Implications of the Circumpolar Genetic Structure of Polar Bears for Their Conservation in a Rapidly Warming Arctic
Implications of the Circumpolar Genetic Structure of Polar Bears for Their Conservation in a Rapidly Warming Arctic
Elizabeth Peacock * 0 1 2 3 4 12 13
Sarah A. Sonsthagen 0 1 2 3 12
Martyn E. Obbard 0 1 2 5 14
Andrei Boltunov 0 1 2 6 15
Eric V. Regehr 0 1 2 7 16
Nikita Ovsyanikov 0 1 2 8
Jon Aars 0 1 2 9
Stephen N. Atkinson 0 1 2 4 13
George K. Sage 0 1 2 3 12
Andrew G. Hope 0 1 2 3 12
Eve Zeyl 0 1 2 10
Lutz Bachmann 0 1 2 10
Dorothee Ehrich 0 1 2 10
Kim T. Scribner 0 1 2 11
Steven C. Amstrup 0 1 2
Stanislav Belikov 0 1 2 6 15
Erik W. Born 0 1 2
Andrew E. Derocher 0 1 2
Ian Stirling 0 1 2
Mitchell K. Taylor 0 1 2
ystein Wiig 0 1 2 10
David Paetkau 0 1 2
Sandra L. Talbot 0 1 2 3 12
0 Competing Interests: The author D. Paetkau is an owner of the Canadian company, Wildlife Genetics International. This fact does not alter the authors' adherence to all PLOS ONE policies on sharing data and materials
1 Funding: Funding for collection and analysis of new samples was provided by the U.S. Geological Survey's Changing Arctic Ecosystem Initiative and is supported by funding from the Wildlife Program of the USGS Ecosystem Mission Area, US Fish and Wildlife Service, Ontario Ministry of Natural Resources, Toronto Zoo Endangered Species Fund, Government of Nunavut, Nunavut Wildlife Management Board, Makivik Corporation, Polar Continental Shelf Program, Norwegian Polar Institute, World Wildlife Fund, and Natural Sciences and Engineering Research Council of Canada. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. The author D. Paetkau is an owner of the Canadian company, Wildlife Genetics International. Wildlife Genetics International pro- vided support in the form of salary for author DP, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific role of this author is articulated in the 'author contributions' section
2 Editor: Michael Knapp, Bangor University , United Kingdom
3 Alaska Science Center, US Geological Survey , Anchorage, Alaska , United States of America,
4 Department of Environment, Government of Nunavut , Igloolik, Nunavut , Canada,
5 Ontario Ministry of Natural Resources and Forestry , Peterborough, Ontario , Canada,
6 All-Russian Research Institute for Nature Protection , Moscow, Russian Federation,
7 US Fish and Wildlife Service , Marine Mammals Management, Anchorage, Alaska , United States of America,
8 Wrangel Island State Nature Reserve , Moscow, Russian Federation,
9 Norwegian Polar Institute , Troms , Norway,
10 Natural History Museum, University of Oslo , Oslo , Norway,
11 Department of Zoology, Michigan State University , East Lansing, Michigan , United States of America, 10. Polar Bears International, Bozeman, Montana, United States of America , 1
12 Greenland Institute of Natural Resources, Copenhagen, Denmark , 1
13 Department of Biological Sciences, University of Alberta , Edmonton, Alberta , Canada , 1
14 Science & Technology Branch, Environment Canada , Edmonton, Alberta , Canada , 1
15 Faculty of Science and Environmental Studies, Lakehead University , Thunder Bay, Ontario , Canada , 1
16 Wildlife Genetics International , Nelson, British Columbia , Canada
We provide an expansive analysis of polar bear (Ursus maritimus) circumpolar genetic variation during the last two decades of decline in their sea-ice habitat. We sought to evaluate whether their genetic diversity and structure have changed over this period of habitat decline, how their current genetic patterns compare with past patterns, and how genetic demography changed with ancient fluctuations in climate. Characterizing their circumpolar genetic structure using microsatellite data, we defined four clusters that largely correspond to current ecological and oceanographic factors: Eastern Polar Basin, Western Polar Basin, Canadian Archipelago and Southern Canada. We document evidence for recent (ca. last 1-3 generations) directional gene flow from Southern Canada and the Eastern Polar Basin towards the Canadian Archipelago, an area hypothesized to be a future refugium for polar bears as climate-induced habitat decline continues. Our data provide empirical evidence in support of this hypothesis. The direction of current gene flow differs from earlier patterns of gene flow in the Holocene. From analyses of mitochondrial DNA, the Canadian Archipelago cluster and the Barents Sea
subpopulation within the Eastern Polar Basin cluster did not show signals of
population expansion, suggesting these areas may have served also as past
interglacial refugia. Mismatch analyses of mitochondrial DNA data from polar and
the paraphyletic brown bear (U. arctos) uncovered offset signals in timing of
population expansion between the two species, that are attributed to differential
demographic responses to past climate cycling. Mitogenomic structure of polar
bears was shallow and developed recently, in contrast to the multiple clades of
brown bears. We found no genetic signatures of recent hybridization between the
species in our large, circumpolar sample, suggesting that recently observed hybrids
represent localized events. Documenting changes in subpopulation connectivity will
allow polar nations to proactively adjust conservation actions to continuing decline
in sea-ice habitat.
The distribution and viability of animal populations are dependent on the
quantity, quality and connectivity of habitat. As habitat undergoes change,
populations expand or become increasingly isolated. Because habitat change due
to climate warming has been and is predicted to be most dramatic at the poles and
especially for sea-ice habitat , Arctic species are expected to undergo dramatic
shifts in distribution that may affect their viability . Since 1979, the spatial
extent of Arctic sea-ice in autumn has declined by over 9% per decade through
2010 . Recent modeling predicts that nearly ice-free summers will characterize
the Arctic before mid-century. However, as sea-ice loss is occurring faster than
forecasted , the first nearly ice-free Arctic summer could occur as soon as
2016 . The polar bear (Ursus maritimus), a specialist carnivore whose survival
depends on sea-ice for foraging, migration and mating, is expected to be
increasingly at risk due to changes in sea-ice habitat . One modeling effort that
included projections of future sea-ice conditions forecasted that two-thirds of the
circumpolar population might be extirpated within half a century  unless
greenhouse gas emissions are reduced and climate warming slowed . These
observations and expectations prompted the listing of the polar bear as a
vulnerable species on the IUCN Red List in 2006 and as a threatened species in the
United States in 2008. Subsequently, the 5 nations where polar bears occur are
now developing a circumpolar management plan  outlining conservation
actions and research needed to understand how the species will respond to
changing habitat. Initial research has shown that decreasing metrics of sea-ice
extent have been associated with declining polar bear body condition ,
survival  and population size , although effects of sea-ice decline on
polar bears are variable [15, 16]. Using global circulation models (GCMs) and
polar bear habitat-use data, Durner et al.  predicted that mean summer polar
bear optimal habitat in the Polar Basin (approximately 50% of the most northerly
portion of polar bear range) will decrease by 68% by the end of the 21st century.
Although with faster than forecasted observations of the rate of sea-ice
reduction [3, 4], this timeline will likely be compressed. At the scale of the polar
bears entire circumpolar range, changes in sea-ice phenology and quality are
predicted to influence the demographics [6, 18], connectivity [19, 20], degree of
genetic isolation , and therefore viability, of the 19 global semi-discrete
subpopulations recognized by the IUCN/Polar Bear Specialist Group .
Though movement of individual polar bears across present-day subpopulation
boundaries is evident from capture-recapture data and hunter tag returns ,
satellite telemetry  and genetic surveys [21, 27, 28], the IUCN
subpopulation boundaries were originally developed based on patterns of sea-ice
formation and melt, and observations of polar bear seasonal fidelity. Though
polar bears are currently thought to constitute a panmictic, single evolutionary
unit , reduced range and habitat connectivity may increasingly fragment
subpopulations resulting in transient refugia and meta-population dynamics.
Some subpopulations may become more productive, whereas others may become
less viable . There is already evidence of change in the contemporary
distribution of polar bears. For example, polar bears, once common in
Newfoundland , are now seen there only infrequently and in small numbers.
Similarly, polar bears once regularly summered on St. Lawrence and St. Matthew
islands in the Bering Sea . Now they are irregularly observed in the Bering
Sea and do not spend summers on St. Matthew Island. Although these changes in
polar bear distribution may also have been related to overharvest, the recent
reductions in the extent of sea-ice due would prevent current and regular use of
Future responses of polar bears to declining habitat may mirror their responses
during past fluctuations in climate. Paleodemographic genetic estimates using
mitochondrial DNA (mtDNA) have suggested that polar bear evolution and
effective population size have tracked key climate events, similar to the
paraphyletic brown bear (U. arctos). Both species experienced fluctuations in
effective population size throughout their history, including prolonged declines
over the past 500,000 years, although this decline is more pronounced in polar
bears . Population fluctuations of brown and polar bears appear to be offset,
with population size in brown bears decreasing, and in polar bears increasing,
during Pleistocene cooling events . This offset is consistent with observations
from other paired boreal and Arctic species, including voles (Microtus spp.) and
shrews (Sorex spp.; ). Periodic hybridization between the two species of bears
has resulted in conflicting estimates of divergence time, presence of polar bear-like
mtDNA in certain extant  and extinct  brown bear populations,
and long stretches of polar bear nuclear DNA in the brown bear genome [33, 41].
Signatures of admixture are consistent with overlapping distributional ranges
(and habitat) of both species in some areas, and viable hybrids recently have been
documented in wild populations . The rate and periodicity of hybridization
events through time may result from shifting ranges of one or both species in
response to climatic change . Hence, connectivity among subpopulations of
polar bears, and interactions between polar bears and brown bears, will continue
to change during the current period of sea-ice decline.
In this paper, we use mtDNA and neutral microsatellite DNA data to establish
the historical and current genetic structure and gene flow dynamics in polar bears
across their entire circumpolar range, and specifically among the subpopulation
units that are used for polar bear management. We employ clustering techniques
[43, 44] to evaluate underlying circumpolar structure, which may be cryptic and
not necessarily related to subpopulation boundaries . By using coalescent
theory , we examine historical (mtDNA) and modern (microsatellite)
asymmetries in gene flow among clusters of subpopulations to determine the
underlying connectivity. We assess symmetry in gene flow during the period of
very recent climate warming and ice habitat decline with Bayesian methods .
We also directly compare polar bear samples from the 1980s through the 2010s
(over ,23 generations), to test for temporal shifts in genetic structure within
subpopulations. We further determine the extent of sex-bias in gene flow as a
mechanism underlying connectivity of subpopulations of polar bears. We
compare the polar bear mtDNA data to similar data in the brown bear, to provide
context to the analyses of phylogenetic structure, historical population
fluctuations and gene flow in polar bears during historical climate fluctuations.
We also address recent hypotheses about the relationship between modern polar
bears and brown bears [33, 40, 41]. Our aim is to provide a comprehensive
baseline of historical and contemporary genetic structure for polar bears,
throughout their circumpolar range, which will complement ecological and
evolutionary research, provide a framework for effective conservation actions
[48, 49], and provide a geographically comprehensive baseline for future
assessment and management.
Materials and Methods
Data were used from samples collected in previously permitted research studies
between 1973 and 2006 (n5869; S11 and S12 Tables; [27, 28]). We also newly
extracted DNA from polar bear tissue collected for ecological research studies
permitted by various jurisdictions (S11 Table; details of permits in S12 Table).
Shed polar bear hair samples from the Laptev Sea (LP) and Kara Sea (KS)
subpopulations were collected from the ground by Russian biologists, and did not
require permitting for collection. Other polar bear samples were collected from
the legal harvest in the territory of Nunavut, Canada (S12 Table). No polar bears
were harvested or captured for the purpose of this study.
We analyzed DNA from 2,748 polar bear samples from 18 subpopulations at 16
21 microsatellite loci (depending on subpopulation), and 411 samples from 15
subpopulations at the mtDNA control region (S1 Fig. a,b). Some samples were
used for both mtDNA and microsatellite analyses. We collected tissue samples of
polar bears year-round from sport or subsistence harvest (n5198), or capture
operations (n52410) in all countries that have polar bears: Canada, Greenland,
Norway, Russia and the United States. Remotely-collected biopsies from Canada
(n565) and shed hair from Russia (n536; S11 Table) were also used. Samples
were collected from 18 of the 19 recognized subpopulations of polar bears 
(S1a,b Fig.): Baffin Bay (BB); Barents Sea (BS); Chukchi Sea (CS); Davis Strait
(DS); East Greenland (EG); Foxe Basin (FB); Gulf of Boothia (GB); Kane Basin
(KB); KS; LP; Lancaster Sound (LS); MClintock Channel (MC); Northern
Beaufort Sea (NB); Norwegian Bay (NW); Southern Beaufort Sea (SB); Southern
Hudson Bay (SH); Viscount Melville (VM); and Western Hudson Bay (WH). The
19th subpopulation the Arctic Basin is an inaccessible, unmanaged
subpopulation, and is thought not to currently function as a year-round habitat.
Only samples from independent polar bears (i.e., .2 years) were used, unless the
accompanying mother and/or sibling were not already in the sample set.
Ninetyseven percent of the samples had known latitude and longitude (S1a,b Fig.), but
analyses were conducted on all samples. For additional details, see S1 Supporting
Information: Details of materials and methods, Sampling.
DNA extraction, PCR amplification, and genotyping or sequencing protocols for
microsatellite and mtDNA data of ursids are described in the literature
[27, 28, 50]. For additional details, see S1 Supporting Information: Details of
materials and methods, Laboratory Techniques.
We quantified genetic variation and tested for neutrality in both microsatellite
and mtDNA data using a variety of computer programs routinely used to analyze
genetic data . For additional details, see S1 Supporting Information:
Details of materials and methods, Genetic Diversity.
We used microsatellite data to test for differences in the distribution of allele
frequencies  between decadal groups for each of nine polar bear
subpopulations for which we had data for multiple decades (S1a Fig.): SB (n583
(1980s), 45 (1990s), 64 (2000s) and 41 (2010s)); CS (n530 (1980s), 27 (1990s), 71
(2000s) and 138 (2010s)); FB (n530 (1990s), 89 (2000s)); BB (n548 (1990s) and
126 (2000s)); WH (n524 (1980s), 22 (2008)); GB (n530 (1990s), 15 (2008)); LS
(n530 (1980s90s), 31 (2000s); the Labrador portion of DS (n527 (1990s) and
217 (2000s)); and the Svalbard region of BS (n5192 (1990s) and 249 (2000s)). We
also used Bayesian clustering  to infer the occurrence of population structure
among groups sampled during different decades. For additional details, see S1
Supporting Information: Details of materials and methods, Decadal Comparisons.
We assessed genetic differentiation among the subpopulations of polar bears using
both microsatellite and mtDNA data. Overall standardized estimates of FST
variance based on microsatellite loci were calculated [55, 61], as were estimates of
inter-subpopulation variance ; significance was based on random
permutation tests. We also addressed previous hypotheses about within-subpopulation
structuring within DS  and SH , using the same tests. For mtDNA data,
we applied the evolutionary model that best fit mtDNA  to calculate W ,
and tested for significance . We used hierarchical analyses of molecular
variance (AMOVA; ) with both marker types to test for significance of
geographic partitioning of hypothesized genetic units. For additional details, see
S1 Supporting Information: Details of materials and methods, Genetic
To assess sex-specific philopatry and gene flow of polar bears, we plotted pairwise
FST values from microsatellite (biparental inheritance) and mtDNA (matrilineal
inheritance) data of each subpopulation, calculated after accounting for
differences in the effective size between the two genomes (FST(nu)51e0.25*ln[1
FST(mt)]) . For interspecific comparison, we plotted these values with pairwise
values generated from populations of brown bears from southeast Alaska .
Modern Circumpolar Structure
We examined the modern structuring of the circumpolar polar bear population
based on microsatellite data using Bayesian clustering methods  implemented
in STRUCTURE version 2.0 . We considered advice from Evanno et al.  and
Pritchard et al.  to determine the most likely number of genetic clusters of
polar bears, in addition to biological rationale. For significance testing, all
avalues were set at 0.05 and, where appropriate, adjusted using Bonferroni
We estimated modern (microsatellite) and historical (mtDNA) gene flow between
clusters that we had previously identified with microsatellite DNA. Estimates of
gene flow among the clusters of polar bears were calculated for microsatellite loci
using BAYESASS, version 3.0.1 [43, 65]. The number of migrants per generation
(Nem) for nuclear microsatellite and number of female migrants per generation
(Nf m) for mtDNA were calculated using MIGRATE version 3.0.3 [46, 65, 68]. For
additional details, see S1 Supporting Information: Details of materials and
methods, Gene Flow.
Phylogenetic Analyses of MtDNA Sequences
Using mtDNA data, we made phylogenetic comparisons among haplotypes found
in 15 subpopulations of polar bears. To root the phylogenetic tree, we included 37
haplotypes from 144 individuals representing the three Alaskan brown bear clades
. Phylogenetic analyses of the polar and brown bear control region sequences
were conducted using PAUP*4.0b8 , using maximum parsimony (MP),
maximum likelihood (ML) and distance (minimum evolution, ME) approaches.
For additional details, see S1 Supporting Information: Details of materials and
methods, Phylogenetic Analyses of MtDNA Sequences.
Changes in Historical Population Size
Using the mtDNA dataset, we assessed historical signatures of population growth/
expansion within subpopulations and/or within larger regional groupings of polar
bears using: extended Bayesian skyline plots ; mismatch distributions  and
their raggedness (rg ; ); the shape of phylogenetic trees; neutrality tests
sensitive to population fluctuations [73, 74]; comparison of diversity indices
(haplotype diversity [h] and nucleotide diversity [p]); and coalescent-based
simulation methods . Significance of expansion measures was tested via
coalescent simulations . To estimate the time since population expansion, we
used mismatch distributions and the nonlinear least-squares approach. For
estimation of mutation rate we used a coalescent Bayesian framework and
included control region haplotype sequences of representative brown and polar
bears . Parameters were set in BEAUti, part of the v1.6.1 software package .
For additional details, see S1 Supporting Information: Details of materials and
methods, Changes in Historical Population Size.
We collected multi-locus microsatellite genotypes from samples of 2,748 polar
bears from 18 of 19 circumpolar polar bear subpopulations (data are deposited at
datadryad.org, doi:10.5061/dryad.v2j1r). Both allelic richness (range, 5.06 to 5.94)
and observed heterozygosity (Ho; range, 0.64 to 0.73) were similar across
subpopulations (S1 Table). Overall Ho was 0.70. Significant departures from
Hardy-Weinberg Equilibrium (HWE) were detected in 5 of 338 cases (locus by
subpopulation), with no individual locus or subpopulation accounting for a
disproportionate number of significant test results. Accordingly, no loci were
dropped from further analyses due to deviation from HWE. Overall, significant
(P,0.05) associations of alleles between markers (linkage disequilibrium, LD)
were identified for 47 of 210 pairs of loci (10.5 positive tests would be expected
due to Type I error), suggesting that the global population has significant
admixture or substructuring. LD patterns were absent with the exception of one
pair of loci, suggesting that the observed LD was due to population structure
rather than physical linkage on the chromosome. LD within the DS
subpopulation, especially in association with, but not isolated to, locus G10X, produced 30
of 67 significant (P,0.002) test results. That the LD was relatively restricted to
one subpopulation is suggestive of significant population substructuring within
Analysis of 411 polar bears from 15 subpopulations (Fig. 1a,b; S1 Table)
identified 63 mtDNA control region haplotypes characterized by 35 polymorphic
sites, including 31 transitions, one transversion and three insertion/deletions (S2
Table). Haplotype diversity (h) ranged from 0.00 to 0.95 and nucleotide diversity
(p) ranged from 0.000 to 0.007 (S1 Table). Private haplotypes were observed in
the four clusters identified with microsatellite DNA (below in Genetic
Differentiation of Subpopulations), though most private haplotypes were only
represented by a single or few individuals (Fig. 1b). Haplotype sequences have
been submitted to GenBank (S2 Table). There was no signal of deviation from
selective neutrality for any of the 15 subpopulations (En50.0910.444, P50.303
0.990; S1 Table).
We evaluated genetic differentiation at microsatellite markers within nine
subpopulations between samples collected in early and late time periods (S3
Table). We found no significant levels of genic and genotypic differentiation and
substructure between early and late time periods (S3 Table). Thus, all data were
pooled within each subpopulation for subsequent analyses.
Genetic Differentiation of Subpopulations: Microsatellite Data
The circumpolar estimate of population genetic structure across the 18
subpopulations was significant (FST50.034; 95% CI: 0.0270.043). The upper
limit of FST for this data set, taking into account the level of genetic diversity, is
0.295, therefore our overall estimate accounts for 11.5% of the maximum possible
level of genetic structure. Thirty-one (20%) of 153 possible pairwise comparisons
among subpopulations showed a signal of significant differentiation (adjusted for
multiple-comparisons) based on at least one of four metrics (mean FST, RST, genic
and/or genotypic differentiation; S4 and S5 Tables; genotypic differentiation was
assessed for those subpopulations out of HWE). For example, WH exhibited
genetic differentiation from most other subpopulations, with 11 of 17 possible
comparisons showing significant differentiation with all metrics. In contrast, BS
was genetically similar to most other subpopulations, and was typically grouped
with EG, KS and LP. Significant genetic structure was observed within DS and SH,
Fig. 1. Relationships between mitochondrial haplotypes of polar bears from the circumpolar range (15 subpopulations). a. Minimum evolution tree
showing the relationships between 63 mitochondrial DNA control region haplotypes for polar bears from these subpopulations, the ancient Poolepynten
(GenBank Accession No. GU573488) polar bear and haplotypes found within the three clades of Alaskan brown bears (GenBank Accession
No. KM821364KM821401). Numbers represent distances between deeper nodes, under the Tamura-Nei distance (I+G0.69) model. Filled circles indicate
nodes with.70% bootstrap support, and arrows at nodes indicate 5069% bootstrap support. b. Unrooted 95% parsimony network showing relationships of
the 64 haplotypes. The size of the node corresponds to the frequency of each haplotype (numbered) with black squares representing unsampled
indicating further regional structuring within these subpopulations (S6 Table).
Using the Fischers Exact test and a50.05 for comparison to Paetkau et al. 
who found 98% of the 120 pairwise comparisons significant, 150 of our 153
possible tests (98%) were significant. Using this latter test, the three pairs of
subpopulations that did not show differentiation from each other were MC and
LS, KS and LP, and KS and BS.
Using the polar bear microsatellite data and the program STRUCTURE , DK
was maximized when K (i.e., number of likely clusters)52 (DK5577.7; S2 Fig.).
According to this model, polar bears residing in the Polar Basin were assigned to
one cluster and all other polar bears were assigned to another. The second highest
DK was with K53 (DK5263.2), which partitioned polar bears residing in the
Canadian Archipelago and Southern Canada into their own clusters, with the
third cluster in the Polar Basin (Fig. 2a,b); K53 was identified as the most likely
clustering pattern using Pritchard et al.s  criteria. Regional sub-structuring
also was uncovered within higher hierarchical clusters (in the K53 scenario), but
only significant within the Polar Basin. Within the Polar Basin, polar bears were
further separated into Eastern and Western Polar Basin clusters (K52, DK5271.2;
S3 Fig. c). The analyses within Canadian Archipelago Cluster (K52, DK5255.7)
and Southern Canada Cluster (K52, DK5350.5) did not result in a discrete
geographical pattern similar to that observed within the Polar Basin (S3a,b Figs.).
The split found within the Polar Basin group reflects some known ice patterns (see
Discussion) and thus we decided to group polar bears into four clusters for
subsequent analyses: Eastern Polar Basin, Western Polar Basin, Canadian
Archipelago and Southern Canada.
Using AMOVA, we rejected all but one hypothesis of among-group variance in
microsatellite allele frequencies, which placed polar bear subpopulations into
three clusters (Polar Basin, Canadian Archipelago, and Southern Canada; S7
Genetic Differentiation of Subpopulations: MtDNA data
We observed significant global variance in mtDNA haplotypic diversity
(WST50.210; P,0.001). Population pairwise WST values ranged from 20.046 to
0.805, and 77 of 105 pairwise comparisons were significant (S5 Table). We
detected significant differences in the distribution of haplotypes (x2 df53513.21
, P,0.002, a50.05) in all comparisons that involved subpopulations
characterized by three or fewer mtDNA samples (MC, VM and NW). In 33 of 39
Fig. 2. Assignment of individual polar bears (S11 Table) from their circumpolar range (19
subpopulations) to regional genetic clusters. a. STRUCTURE  assignment plot for microsatellite
signatures (n52,899) of polar bears. Y-axis represents proportional membership each of three most-likely
groups identified by program STRUCTURE (Southern Canada [red dots], Canadian Archipelago [blue dots] and
the Polar Basin [yellow dots]). Note, based on subsequent analysis (S2c Fig., S6 Table) we discuss the Polar
Basin cluster as two groups: the Eastern Polar Basin Western Polar Basin clusters. Individuals are organized
(each represented by a single vertical line) along the X-axis according to subpopulation: East Greenland (EG),
Barents Sea (BS); Kara Sea (KS); Laptev Sea (LP); Chukchi Sea (CS); Southern Beaufort Sea (SB); Northern
Beaufort Sea (NB); Viscount Melville (VM); MClintock Channel (MC); Gulf of Boothia (GB); Lancaster Sound
(LS); Norwegian Bay (NW); Kane Basin (KB); Baffin Bay (BB); Davis Strait (DS); Foxe Basin (FB); Western
Hudson Bay (WH) and Southern Hudson Bay (SH). Individuals within each subpopulation are arranged
according membership to one of the three clusters. b. Geographical locations of (n52,650) samples in the
three genetic clusters.
comparisons involving these three subpopulations, we rejected the null hypothesis
of equal distribution of haplotypes (S5 Table).
Significant mean WCT values were observed for all AMOVA hypotheses based
on mtDNA data (i.e., historical structure and sex specific dispersal). Among these,
the hypothesis that grouped subpopulations into two broad geographic regions
(the Polar Basin, and Canadas eastern Arctic and Subarctic) yielded the highest
mean WCT value (S7 Table).
Mitogenomic structure for polar bears was less than the partitioning observed
among brown bear populations separated by similar geographic distances (S4
Fig.). Pairwise comparisons among subpopulations revealed that genetic structure
at nuclear markers was typically less than expected given the genetic structure
observed in mtDNA (S4 Fig.).
Asymmetrical gene flow was inferred among the four clusters (Easter Polar Basin,
Western Polar Basin, Canadian Archipelago, Southern Canada) across both
marker types and all analyses. In the microsatellite data set, biases in the
directionality of gene flow estimated using the coalescent were not as strong as
those estimated using allelic frequency (Table 1; Fig. 3). Microsatellite data
analyzed using allelic frequencies (i.e., past 13 generations) indicated directional
gene flow into the Canadian Archipelago cluster from the other clusters (except
from the Western Polar Basin), as well as from the Eastern Polar Basin into the
Western Polar Basin (Table 1; Fig. 3). The Eastern Polar Basin and Southern
Canada clusters were represented by a high proportion of non-migrant
individuals, suggesting these regions are sources of dispersal (S8 Table).
Coalescent-based estimates of the microsatellite data revealed asymmetrical gene
flow from Western Polar Basin cluster into the Canadian Archipelago cluster, with
all other comparisons suggesting symmetrical gene flow as indicated by
overlapping 95% confidence limits (Table 1).
Among gene flow estimates based on mtDNA, there was a signal of effective
dispersal from the Southern Canada cluster into both Eastern and Western Polar
Basin clusters (Table 1).
Nuclear DNA: within Holocene Nem
*Gene-flow estimates are listed as immigration into population A from population B and emigration from population A into population B. For example, gene
flow between WP and CA with microsatellite loci is 4.3 Nem into the CA from WP and 3.2 Nem from CA into WP. Because the 95% CI do not overlap WP is
listed as the source.
Parameter estimates are listed for each cluster pair*, as well as the directionality of gene flow between cluster pairs (source, sink, and symmetrical )
assigned on the basis of 95% confidence intervals (in parentheses). The Eastern Polar Basin cluster (EP) includes polar bears from East Greenland,
Barents Sea, Kara Sea and Laptev Sea subpopulations. The Western Polar Basin cluster (WP) includes polar bears from the Chukchi Sea, southern
Beaufort Sea and northern Beaufort Sea. The Canadian Archipelago cluster (CA) includes Viscount Melville, MClintock Channel, Gulf of Boothia, Lancaster
Sound, Norwegian Bay, Kane Basin, Baffin Bay and the region north of Hudson Strait in Davis Strait (DS). The Southern Canada cluster (SC) includes Foxe
Basin, Southern Hudson Bay, Western Hudson Bay and the region south of Hudson Strait in DS.
Fig 3. Recent directional gene flow (ca. 310 generations) calculated on the basis of allelic frequencies
(number of migrants, m) among polar bear clusters. Data generated using the program BAYESASS ,
examining gene flow relationships between the four clusters of polar bears (Southern Canada (SC; red),
Canadian Archipelago (CA; blue), Eastern Polar Basin (EP; yellow) and Western Polar Basin (WP; green)),
identified by program STRUCTURE analysis of microsatellite data. Arrow widths represent only directional gene
flow values that are significantly different from zero (no migration) and from the value for migration in the
Fluctuations in Historical Effective Population Size
Globally, using mtDNA, polar bears showed a strong signal of historical
population growth (.18 times the SD; g5915.33, SD548.01), and the mismatch
distribution did not deviate significantly from a sudden expansion model
(SSD50.044, P50.103) but did show significant raggedness (rg50.013; S1 Table).
Based on the coalescent, the Western and Eastern Polar Basin and Southern
Canada clusters showed a significant signal of historical growth (S1 Table).
However, significant geographic expansion (and growth) is consistently evident
for only a few subpopulations within these regions: for the Eastern Polar Basin,
only the LP subpopulation; for the Western Polar Basin, the CS and SB; and for
Southern Canada, only SH showed significant historical expansion (S1 Table). CS
and SH each exhibited a particularly strong signal of growth (.6, and.19 times
SD, respectively; S1 Table); both subpopulations were also characterized by
significantly negative Fus Fs and, for SH, also a significantly negative Tajimas D
(S1 Table). Among subpopulations with strong growth signals, mismatch
distributions for the LV, SB and CS did not differ significantly from a sudden
expansion model, although the latter showed a signal of significant raggedness (S1
Table). Similarly, mismatch distributions for SH did not deviate from a sudden
expansion model (S1 Table). The Canadian Archipelago cluster did not show a
signal of growth overall, nor did its component subpopulations (S1 Table).
Extended Bayesian plots did not indicate significant growth within regional
clusters (S5b-e Fig.), although a signature of growth was observed overall (S5a
We estimated time since population expansion, based on mismatch
distributions (S6 Fig.), a mutation rate of 11.0% per million years (based on our
coalescent Bayesian analysis of mtDNA that was internally calibrated by
incorporating data from the entire mitogenome of a 120,000 year old polar bear
fossil ), and based on a generation time of ten years for polar bears and six
years for brown bears (following ), to be approximately 98,260 (t52.512) and
56,410 (t51.405) years ago, respectively, for polar bears of the Western Polar
Basin cluster (SB and CS, pooled) and brown bears of the Eastern Beringian
Clade. Since the algorithm assumes population expansion, the proposed
expansion date for the Western Beringian brown bear clade (31,079 years,
t51.290) is tentative (S9 Table). We note that a lineage-specific mutation rate of
11% for the mtDNA control region is lower than the 30% multiple calibration
rate estimated by Saarma et al.  for tip rates, but greater than their 5% root
As observed in previous studies based on mtDNA gene sequences [35, 37], extant
brown bears are paraphyletic with respect to polar bears (Fig. 1a); brown bears
from Admiralty, Chichagof and Baranof (ABC) islands of the Alexander
Archipelago of southeastern Alaska share a mtDNA lineage more similar to polar
bear than other brown bears in North America, due to recent  or ancient
[33, 36] hybridization, incomplete lineage sorting [37, 80], or both. Further, in
contrast to brown bears of eastern and western Beringia, modern polar bears show
relatively shallow within-species divergence; control region haplotypes in brown
bears of eastern Beringia differ from western Beringian haplotypes by at least
3.0%, whereas polar bear haplotypes differ by less than 0.5% (Fig. 1a,b). The
results of minimum evolution analyses were largely congruent with analyses based
on maximum parsimony and maximum likelihood. All analyses placed
homologous sequence data generated from a ca. 120 ka polar bear fossil from
Poolepynten (Svalbard) see  phylogenetically closer to the ABC brown bear
clade relative to the majority of polar bear haplotypes globally (Fig. 1a). Average
corrected genetic distances between modern polar bear haplotypes and haplotypes
found in ABC brown bears were more than twice as large as average within-polar
and within-ABC brown bear haplotype distances (S10 Table). The average
corrected distance between the modern polar bear group and the Poolepynten
polar bear (TrN+I+G50.012) was the same as the distance between the latter and
the ABC brown bears (S10 Table).
Contemporary response to changing climate
We detected an increase in directional gene flow of polar bears from the Eastern
Polar Basin towards the Canadian Archipelago and Western Polar Basin, and from
Southern Canada to the Canadian Archipelago, within the last 13 generations
(BAYESASS analysis of microsatellite DNA). The directional gene flow from the
Eastern Polar Basin to the Canadian Archipelago and Western Polar Basin is
specifically contemporary (past 13 generations; BAYESASS analyses of
microsatellite DNA), occurring during this current era of sea-ice decline. Estimates of
asymmetry in historical gene flow did not show a similar pattern: coalescent
estimates of asymmetry in gene flow (MIGRATE analyses) did not show
directionality based on maternally-inherited mtDNA (late Pleistocene signature),
or on microsatellite DNA (Holocene signature). The northern parts of the
Canadian Archipelago have been predicted to retain polar bear ice habitat farther
into the future than other Arctic areas [6, 17]. This directional gene flow from the
Eastern Polar Basin cluster towards the Canadian Archipelago and Western Polar
Basin clusters is supported by predominant seasonal ice movement from the
Eastern Polar Basin (e.g., Laptev Sea) towards the northern edge of the Canadian
Archipelago and northern Greenland i.e., part of the Beaufort Gyre . As the
extent of sea-ice continues to decrease in response to climate change, the northern
edge of the Canadian Archipelago and adjacent areas of the Polar Basin are
predicted to retain the last vestiges of summer and autumn sea-ice habitat [6, 17].
Our findings constitute empirical evidence in support of this hypothesis. Note
although the Beaufort Gyre moves ice from the western portion of the Western
Polar Basin towards the Canadian Archipelago, we did not detect gene flow in this
Throughout the Holocene, polar bears in the Canadian Archipelago have had
year-round access to sea-ice from which to hunt for seals. Conversely, in much of
the seasonal-ice ecoregion  in the eastern Arctic and subarctic of Canada (i.e.,
our Southern Canada cluster), polar bears spend several months fasting ashore.
The climate warming of recent decades  has led to earlier break-up of sea ice
and longer ice-free seasons in these subpopulations of the Southern Canada
cluster. Decreased body condition, lower survival, and reduced abundance of
polar bears in some of subpopulations in the Southern Canada cluster have been
linked to these sea ice changes [9, 11, 13], but see ). In contrast, some
researchers hypothesize the Canadian Archipelago will become more productive
habitat for polar bears as annual ice over shallow waters (better conditions for seal
prey) will replace thick, multi-year and less productive ice habitat . Derocher
et al.  and Durner et al.  predicted such a northerly shift in polar bears as
productive habitat changes as a result of climate warming. Our empirical gene
flow data support this prediction. In addition to patterns of ice movement within
the Polar Basin, and changes in availability of preferred polar bear habitat,
evidence of net gene flow into the Canadian Archipelago and Western Polar Basin
may also be influenced by asymmetries in harvest pressure. The Canadian
Archipelago and the Western Polar Basin have at least 2.8 times and 1.5 times,
respectively, higher polar bear harvest than in the Eastern Polar Basin  in
which only the East Greenland subpopulation is harvested.
Contemporary Genetic Structure
Our examination of asymmetry in gene flow was based on results of Bayesian
clustering which grouped the circumpolar population of polar bears into three or
four regional genetic clusters that represent patterns of breeding within the recent
Holocene (Fig. 2a,b). Generally, individual polar bears from entire
subpopulations were assigned to one of three regional clusters. The exception was the Davis
Strait subpopulation, which was split between two regional clusters: Southern
Canada and the Canadian Archipelago. Targeted Bayesian analyses also supported
the further division of the Polar Basin cluster into the Eastern and Western Polar
Basins (S3c Fig.). Although three regional groupings were also uncovered using
traditional metrics (AMOVA analyses of both microsatellite loci and mtDNA),
only mtDNA AMOVA analyses supported the presence of the four regional
clusters uncovered using Bayesian clustering, which includes the eastern and
western Polar Basin split (S7 Table). The greater level of population and regional
structuring uncovered using data from the maternally-inherited mtDNA relative
to the biparentally-inherited nuclear microsatellite loci is not surprising. Pairwise
FST values for microsatellite loci are generally lower than expected based on FST
values for mtDNA, reflecting hypothesized greater levels of philopatry among
females (S4 Fig.).
Oceanographic patterns of ice formation, ice melt, and the convergence and
divergence of ice from shore create different ecological circumstances for polar
bears  that likely contribute to these clustering patterns. The Canadian
Archipelago cluster is confined by narrow, shallow straits between islands,
insulating the region from the ice dynamics of the Polar Basin. The Canadian
Archipelago was identified by Amstrup et al.  as one of four unique eco-regions
for polar bears. Three subpopulations Kane Basin, Baffin Bay and the more
northerly region of Davis Strait group with the subpopulations in the Canadian
Archipelago proper (Lancaster Sound, Norwegian Bay, MClintock Channel,
Viscount Melville, Gulf of Boothia) into the Canadian Archipelago genetic cluster.
This is consistent with a general northerly seasonal ice movement and the
directional gene flow towards the Canadian Archipelago. The grouping of these
additional subpopulations with the Canadian Archipelago proper supports the
proposed Central Arctic conservation unit of polar bears defined based on
ecological similarities .
The Southern Canada cluster includes subpopulations in the Hudson Bay
ecological complex [48, 82], for which breeding habitat overlaps in central
Hudson Bay . Polar bears sampled south of Hudson Strait in the Davis Strait
subpopulation also clustered with the Hudson Bay complex of polar bears. The
marked genetic differences between southern and northern Davis Strait bears
must reflect differences in southern versus northern breeding areas in the spring.
However, polar bears in Davis Strait were largely sampled in the fall; this suggests
that the fidelity to the two subregions within Davis Strait is year-round. Polar
bears occupying southern and northern Davis Strait also differ in diet  and
rates of survival and recruitment . Though movement has been documented
between these two regions [24, 84, 85], movement across Hudson Strait may be
seasonal or has not resulted in significant evolutionary dispersal. As Hudson Strait
becomes ice-free ever-earlier in spring , with continued climate warming ,
the division between northern and southern Davis Strait may increase. Also within
the Southern Canada cluster, we found similar genetic substructure, albeit
shallow, within the Southern Hudson Bay subpopulation, confirming significant
divergence between polar bears sampled during the autumn on the Ontario Coast
of Hudson Bay and those sampled on Akimiski Island within James Bay . The
restricted genetic interchange within both Davis Strait and Southern Hudson Bay
exacerbates the conservation concern of these two populations, which are at the
southern fringe of current polar bear distribution [15, 87, 88].
Increasingly restricted gene flow as ice extent continues to decline could drive
the current system, which appears to be moderately structured among broad
regions but less structured among subpopulations, toward one characterized by
meta-population dynamics. If periodic immigration into increasingly isolated
subpopulations were not possible, subpopulations could become increasingly
vulnerable to extirpation. Yet, our failure to detect shifts in distributions of alleles
within the nine subpopulations of polar bears that were sampled over the past
three decades (i.e., 1980s-early 2010s), indicate that demographic fluctuations
during the past several generations have been insufficient to influence genetic
partitioning of subpopulations within regions.
Historical Response to Changing Climates
Mismatches between expansions and contractions of polar bears and brown bears
appear to coincide with past climate fluctuations. We timed the expansion of
Western Polar Basin polar bears to the early Wisconsin glacial period
(approximately 98,000 years ago), preceding the expansion of Western Beringian
brown bears before the Last Glacial Maximum (LGM), approximately 31,000
years ago, and Eastern Beringian brown bears during the last glacial at ,56,000
years ago. Deep nuclear and mitogenomic sequencing (of a few individuals)
attributed to differential responses to climate changes and hybridization
[33, 40, 41] corroborate out of phase population trends. Deep nuclear genomic
sequencing suggested that brown bear effective population size apparently
increased during the last interglacial (approximately 130114 kya), and decreased
during the last glacial period. The opposite pattern was uncovered for polar bears,
which showed a marked increase in population size coincident with Pleistocene
cooling, and a decrease coincident with climate warming associated with the last
interglacial . Signatures of evolutionarily recent events (i.e., subsequent to the
LGM) may have low power from whole genomic data . However, similar to
other studies , our analyses, based on multiple expansion and growth metrics,
suggest that polar bear population sizes have fluctuated in response to climate
cycling. Lack of consistency across these metrics, however, may benefit by
simulated tests of directed hypotheses that investigate timing and extent of
effective population size changes [89, 90], and account for among-lineage
variation in comparative demographic analyses .
The phylogenetic relationships and timing of lineage divergence within and
between polar and brown bears remain controversial, due to the difficulty in
discerning between incomplete lineage sorting and ancient and recent (periodic)
hybridization between the two species [41, 92, 93], coupled with uncertainty in
estimates of marker-specific  and genome-wide mutation rates  when
comparing across species. Deep nuclear genomic sequence data places the initial
divergence between ancestors of modern polar bears and brown bears much
earlier (ca. 4 mya; , although see , than estimates based on single
nucleotide polymorphisms (ca. 1.2 mya; ), 14 nuclear introns (ca. 0.230.93
mya; , 13 Y chromosome markers (1.12 mya; ), or on the mitochondrial
genome alone (ca. 150,000 years ago; ). The latter estimate may reflect the
capture of the polar bear mitogenome by brown bears, due to a hybridization
event, rather than the initial divergence of the species. Analysis of mitogenomic
data, including data from an ancient (120,000 year old) fossil polar bear, suggests
an age for the extant polar bear matrilineage of less than 45,000 years, consistent
with recent and rapid growth of modern polar bear populations toward the end of
the last Pleistocene glacial but prior to the LGM . The shallow within-species
structure of modern polar bears relative to brown bears, reflected in our mtDNA
haplotype tree (based on control region sequences from over 400 polar bears),
supports the hypothesis that modern polar bears stem from a single refugial
lineage. Fossil DNA evidence  indicates potential loss of diversity due to
extirpated lineages. Thus, as with other species  analyses of ancient DNA
sequences in polar bears has provided insight into evolutionarily recent events by
facilitating the calibration of molecular clocks and detecting signals of past
Interglacial Refugia and Hybridization
The Svalbard region has been proposed as a previous interglacial refugium,
retaining a source population for range expansion during cooler (glacial) periods
. We observed a signal of stability among polar bears comprising the
Canadian Archipelago and the Barents Sea subpopulations (i.e., Svalbard) within
the Eastern Polar Basin. This indicates that these areas may have served as
previous interglacial refugia and provided leading-edge expansion of polar bears
into other areas during glacial periods. Combined with our evidence of
contemporary gene flow, this supports the hypothesis that these regions may
become a future refugium for polar bears. Although analyses of Y chromosome
loci failed to uncover introgression of the polar bear Y chromosome into brown
bears and vice versa , analyses of both the nuclear autosomal and the
mitochondrial genome suggest that brown and polar bears possess introgressed
alleles [95, 33, 41]. Recently, analysis of mtDNA extracted from fossil bear samples
from two proposed interglacial refugia one in northern Europe (Ireland; ),
and one in the Alexander Archipelago in Alaska  indicates that modern
polar bears stem from one or several hybridization event(s) between polar bears
and brown bears co-occupying periglacial late Pleistocene habitats. In northern
Europe, hybridization is timed to approximately 2832 kya  and in southeast
Alaska is inferred to be between 26 kya years ago and the Pleistocene/Holocene
boundary . Analysis of entire mitogenomes from brown and polar bears ,
however, places their common matrilineal ancestor (from the Alexander
Archipelago) at ca. 152 kya, with radiation of the modern polar bear
mitochondrial crown group at ca. 44 kya. Any hybridization event must therefore
have occurred within this interval, substantially predating the hybridization event
proposed by Cahill et al. . Our mtDNA control region data are also
inconsistent with a recent hybrid origin of modern polar bears, at least in the
proposed Alexander Archipelago refugium (92 kya for coalescence of polar bears
and Alexander Archipelago brown bears; 64 kya for coalescence of the polar bear
crown group; S10 Table), and suggest an earlier (pre-LGM) hybridization event.
Similar to data from the Y chromosome  and earlier analyses of a reduced
suite of microsatellite loci and smaller geographic coverage , we found no
evidence of contemporary admixture between polar bears and brown bears.
Though F1 and F2 hybrids have recently been observed in the Northern Beaufort
Sea and Viscount Melville , our extensive sampling suggests this current
hybridization is thus far localized.
Sex Bias in Gene Flow
It is not possible to place radio collars on male polar bears due to the males neck
circumferences exceeding the circumference of the skull, and tracking male polar
bears by other means has had mixed success. Limited data have revealed no
differences between the male and female movement distances (but see
[24, 100, 101]). In contrast to other bear species, it is evident that female polar
bears, even with young, cover extremely large distances on their mobile ice
platforms . In contrast, we also know that movements of female polar bears
with young of the year are often restricted during the spring  and seasonal
fidelity can be high . Also in contrast to other ursids, movements for both
sexes do not appear to be constrained due to territorial behavior. Slight
malebiased dispersal and gene flow, however, was found using molecular techniques
within the Barents Sea subpopulation , suggesting that sex-specific movement
patterns, resulting in gene flow, do occur. Our joint analysis of
biparentallyinherited and maternally-inherited genomes across 18 subpopulations suggest that
gene flow is mediated more by male polar bears, and that females show higher
Future Research Direction
Though we increased the geographic sampling of polar bears, our study was
limited by lack of modern samples from the Kara Sea in the Russian Arctic
(Eastern Polar Basin cluster). This region is of conservation concern due to
hydrocarbon exploration and unquantified levels of poaching , and should be a
target of research, including research on genetic diversity. New samples from the
Northern Beaufort Sea, and a reanalysis for asymmetry in gene flow within the
Western Polar Basin and towards the Canadian Archipelago could test our finding
of gene flow towards this region. We predict gene flow into this subpopulation
because of declining polar bear individual and population status in the Southern
Beaufort Sea [10, 12], the direction of annual ice drift towards the Northern
Beaufort Sea, and the stability or slight increase in abundance of bears in the
Northern Beaufort Sea . In addition, because our data suggest gene flow into
the Canadian Archipelago, and because this region is likely to retain ice habitat
longer into the future [6, 17] than other parts of the Arctic, new sampling should
focus on Norwegian Bay, Nunavut, Canada, for which we did not obtain new
samples, and the Queen Elizabeth Islands, which are specifically in the convergent
ice zone of the Canadian Archipelago . Further, analyses of additional samples
from the Northern Beaufort Sea and neighboring Viscount Melville
subpopulations may yield signals of more contemporary hybridization events . Finally,
with the isolation of functional single amino acid polymorphisms for polar bears
, comparing functional and neutral genetic diversity with variation in
ecological strategies  should prove informative for understanding adaptation to
past, current and future environments.
Our work updates and expands previous circumpolar genetic analyses of polar
bears [27, 45], corroborating and refining those results. We exposed the
asymmetries in gene flow among subpopulations, and also provided a deeper
historical context by combining analyses of microsatellite and mtDNA data. The
signal of novel and recent gene flow towards the Canadian Archipelago, a
potential refugium , supports increasing research, monitoring and proactive
conservation in this region. The relatively high genetic diversity we report for
polar bears provides an expanded baseline for future comparisons as climate
change and harvest continue to impact polar bear distribution, connectivity and
genetic diversity . Given that Arctic habitats are changing rapidly, our analyses
provide evidence of potential future centers of diversity for polar bears, and
insight into differential responses of polar and brown bears (sister species) to
common environmental processes. Our work provides a circumpolar perspective
on how changing habitat is influencing gene flow in a species of worldwide
conservation concern, and illustrates the value of incorporating genetic
information in analyses to understand the response of species to climate change.
S1 Fig. Locations of polar bears, a., sampled at known latitude and longitude
(n52,650) in 18 circumpolar subpopulations of polar bears, recognized by the
IUCN/Polar Bear Specialist Group, and amplified at microsatellite loci: Baffin
Bay (BB); Barents Sea (BS); Chukchi Sea (CS); Davis Strait (DS); East
Greenland (EG); Foxe Basin (FB); Gulf of Boothia (GB); Kane Basin (KB); Kara
Sea (KS); Laptev Sea (LP); Lancaster Sound (LS); MClintock Channel (MC);
Northern Beaufort Sea (NB); Norwegian Bay (NW); Southern Beaufort Sea
(SB); Southern Hudson Bay (SH); Viscount Melville (VM); and Western
Hudson Bay (WH). Circles identify bears sampled at known latitude and
longitude in the 1980s (n5157), 1990s (n5613), 2000s (n51,708) and 2010s
(n5183). b. Locations of 402 polar bears samples in 15 subpopulations with
known latitude and longitude amplified at the mitochondrial DNA control region.
S2 Fig. The average (95% Confidence Intervals) of 5 runs per cluster, of the
negative log likelihood of the probability of the microsatellite data given the
number of clusters of polar bears, K, simulated by the program STRUCTURE (1),
in the circumpolar Arctic.
S3 Fig. Genetic substructure of polar bears within the three broad clusters
identified by program STRUCTURE (1): a. two most likely sub-clusters within the
Canadian Archipelago Cluster [(Viscount Melville (VM), Gulf of Boothia (GB),
Norwegian Bay (NW), Lancaster Sound (LS), Kane Basin (KB)] with Baffin Bay
(BB) and Davis Strait (DS;) b. two most likely sub-clusters within the Southern
Canada Cluster [(Western Hudson Bay (WH), Southern Hudson Bay (SH) and
Foxe Basin (FB)) and Davis Strait (DS; sorted by high to low latitude)] and c.
the two most likely sub-clusters within the Polar Basin: the Eastern Polar Basin
Cluster [(East Greenland (EG), Barents Sea (BS), Kara Sea (KS), Laptev Sea
(LS)] and the Western Polar Basin Cluster [Chukchi Sea (CS), Southern
Beaufort Sea (SB) and Northern Beaufort Sea (NB)]. All individuals are sorted
left to right by high to low latitude.
S4 Fig. Scatter plot of observed pairwise mtDNA WST versus pairwise
microsatellite FST values for 21 microsatellite loci (circles) for 15
subpopulations of polar bears. The line represents the expected microsatellite FST value
given the genetic differentiation observed at mtDNA (2): FST(nu)51e0.25*ln[1
FST(mt)]. Generally the pairwise comparisons are below the expectation line (i.e.,
lower FST derived from microsatellite markers compared with the mtDNA), which
suggests higher female philopatry relative to males (i.e., male biased gene flow).
Stars show similar comparisons of pairwise mtDNA WST and microsatellite FST
values for brown bear populations in Alaska. Black circles show values that
represent polar bear subpopulations that are #900 kilometers between each other
for comparison to the brown bear populations shown, which are 900 km apart
from each other.
S5 Fig. Extended Bayesian skyline plots (EBSPs) and pairwise mismatch
distributions for (a) global-wide (b) Eastern Polar Basin, (c) Western Polar
Basin, (d) Canadian Archipelago, and (e) Southern Canada clusters of polar
bears. EBSPs indicate population growth from past (right) to present (left)
including median population size through time (black line) and 95% highest
probability distribution (grey interval). Log-transformed y-axes represent
population size as a function of effective size (Ne) and generation time (G).
Mismatch distributions indicate the frequency of expected (grey line) and
observed (black) pairwise differences.
S6 Fig. Mismatch distributions of pairwise differences based on mtDNA data of
polar bears from the southern Beaufort (SB) and Chukchi Sea (CS)
subpopulations and brown bears from the Western and Eastern Beringian
Clades. Mismatch signals between the lineages are offset, signifying different
periods of demographic growth.
S1 Table. Estimates of genetic diversity in 18 circumpolar subpopulation of
polar bears, arranged within the four genetic clusters identified in this paper,
including allelic richness (AR), observed heterozygosity (Ho), expected
heterozygosity (He) for the microsatellite data; and the number of haplotypes
(k), haplotype (h) and nucleotide diversity (p), and Ewen-Wattersens
neutrality (En) for the mitochondrial DNA control region. Genetic demographic
statistics include the growth statistic (g) with standard deviation (bold if
significant where g$SD[g]), Fus Fs, Tajimas D, raggedness (rg), and Deviation
from a sudden expansion (SSD). Bold values for microsatellite data signify
microsatellite data from subpopulations that were out of Hardy-Weinberg
Equilibrium at a50.05/number of loci. Bold values for demographic statistics
indicate significance at P#0.05. Abbreviations of subpopulations are as follows:
Baffin Bay (BB); Barents Sea (BS); Chukchi Sea (CS); Davis Strait (DS); East
Greenland (EG); Foxe Basin (FB); Gulf of Boothia (GB); Kane Basin (KB); Kara
Sea (KS); Laptev Sea (LP); Lancaster Sound (LS); MClintock Channel (MC);
Northern Beaufort Sea (NB); Norwegian Bay (NW); Southern Beaufort Sea (SB);
Southern Hudson Bay (SH); Viscount Melville (VM); and Western Hudson Bay
S2 Table. Distribution of haplotypes (and GenBank accession numbers) at the
mitochondrial DNA control region in 18 subpopulations of polar bears: Baffin
Bay (BB); Barents Sea (BS); Chukchi Sea (CS); Davis Strait (DS); East
Greenland (EG); Foxe Basin (FB); Gulf of Boothia (GB); Kane Basin (KB); Kara
Sea (KS); Laptev Sea (LP); Lancaster Sound (LS); MClintock Channel (MC);
Northern Beaufort Sea (NB); Norwegian Bay (NW); Southern Beaufort Sea
(SB); Southern Hudson Bay (SH); Viscount Melville (VM); Western Hudson
S3 Table. Genetic differentiation results of comparisons of microsatellite data
from earlier and later samples within nine global subpopulations and regions
of polar bears: the Svalbard portion of the Barents Sea (BS); Baffin Bay (BB);
Chukchi Sea (CS); Foxe Basin (FB); Gulf of Boothia (GB); the Labrador portion
of Davis Strait (DS); Lancaster Sound (LS); Southern Beaufort Sea (SB) and
Western Hudson Bay (WH). Degrees of freedom are shown in parentheses. The K
metric represents the likely number of clusters for the subpopulation or region
with decadal data pooled as ascertained using the Bayesian clustering program
BAPS. Values in bold show significant differentiation between the groups (a50.05,
Bonferroni corrections applied).
S4 Table. Below the diagonal, significant (+) and non-significant (2)
differences for genic differentiation, FST, and genotypic differentiation (the
latter only for comparisons of subpopulations that were found to be out of
Hardy-Weinberg equilibrium) for pairwise comparisons of 18 subpopulations
of polar bears using microsatellite data. Bonferroni-corrected significance levels
were adjusted for the number of loci compared for each pair (0.05/number of loci
compared). Shaded blocks delineate the four clusters based on our analysis:
Eastern Polar Basin, Western Polar Basin, Canadian Archipelago and Southern
Canada. Above the diagonal: significant (+) and non-significant (2) differences
for FST, hST and haplotypic differentiation using mtDNA data for pairwise
comparisons of 15 subpopulations. Abbreviations for subpopulations are as
follows: Baffin Bay (BB); Barents Sea (BS); Chukchi Sea (CS); Davis Strait (DS);
East Greenland (EG); Foxe Basin (FB); Gulf of Boothia (GB); Kane Basin (KB);
Kara Sea (KS); Laptev Sea (LP); Lancaster Sound (LS); MClintock Channel (MC); Northern Beaufort Sea (NB); Norwegian Bay (NW); Southern Beaufort Sea (SB); Southern Hudson Bay (SH); Viscount Melville (VM); and Western Hudson Bay (WH).
S5 Table. Pairwise estimates of population differentiation among 15
(mitochondrial DNA) or 18 (microsatellite DNA) circumpolar subpopulations
of polar bears: Baffin Bay (BB); Barents Sea (BS); Chukchi Sea (CS); Davis
Strait (DS); East Greenland (EG); Foxe Basin (FB); Gulf of Boothia (GB); Kane
Basin (KB); Kara Sea (KS); Laptev Sea (LP); Lancaster Sound (LS); MClintock
Channel (MC); Northern Beaufort Sea (NB); Norwegian Bay (NW); Southern
Beaufort Sea (SB); Southern Hudson Bay (SH); Viscount Melville (VM);
Western Hudson Bay (WH). Significant values (a50.002 and 0.05 for
microsatellite and mtDNA comparisons, respectively) are in bold text.
S6 Table. Genetic differentiation within the Southern Hudson Bay (SH) and
Davis Strait (DS) subpopulations of polar bears sampled during autumn while
bears are on land. In the two-way comparison within DS: Northern indicates
samples from polar bears north of Hudson Strait; Southern indicates samples
south of Hudson Strait; samples from the Hudson Strait portion of the Foxe Basin
subpopulation are not included in this comparison. In the three-way comparison:
Southern, as above; Northern Baffin includes samples north of Frobisher Bay on
Baffin Island to the border with the Baffin Bay subpopulation; Hudson Strait
includes samples south of Frobisher Bay on Baffin Island and along Hudson Strait,
including adjacent Hudson Strait samples from the Foxe Basin subpopulation.
Akimiski Island is in James Bay, which is in the SH subpopulation; the Hudson
Bay Coast is also within SH. All values are significant at a50.05/number of loci.
S7 Table. Hierarchical analysis molecular variance (AMOVA) among
subpopulations and various groupings of microsatellite (hCT, hSC) and
mitochondrial (WCT,WSC) alleles to test hypotheses of groupings of the global
subpopulations of polar bears: Baffin Bay (BB); Barents Sea (BS); Chukchi Sea
(CS); Davis Strait (DS); East Greenland (EG); Foxe Basin (FB); Gulf of Boothia
(GB); Kane Basin (KB); Kara Sea (KS); Laptev Sea (LP); Lancaster Sound (LS);
MClintock Channel (MC); Northern Beaufort Sea (NB); Norwegian Bay (NW);
Southern Beaufort Sea (SB); Southern Hudson Bay (SH); Viscount Melville
(VM); and Western Hudson Bay (WH). Bold values are significant at a50.05/
number of loci 50.00230.003 (for microsatellite data) or 0.05 (for mtDNA data).
S8 Table. The proportion of non-migrants (95% CI) over the last ca. 13
generations for four genetic clusters of polar bears calculated using program
BAYESASS (1). The Eastern Polar Basin Cluster includes polar bears from East
Greenland, Barents Sea, Kara Sea and Laptev Sea subpopulations. The Western
Polar Basin Cluster includes polar bears from the Chukchi Sea, Southern Beaufort
Sea and Northern Beaufort Sea subpopulations. The Canadian Archipelago
Cluster includes Viscount Melville, MClintock Channel, Gulf of Boothia,
Lancaster Sound, Norwegian Bay, Kane Basin, Baffin Bay MUs, and the region
north of Hudson Strait in the Davis Strait subpopulations. The Southern Canada
Cluster includes Foxe Basin, Southern Hudson Bay, Western Hudson Bay MUs,
and the region south of Hudson Bay in the Davis Strait subpopulations. Theta (h)
for each cluster is calculated from microsatellite (effective population size (Ne),
scaled to mutation rate (m) and mtDNA data (female effective population size
(Nf), scaled to mutation rate) using the program MIGRATE.
S9 Table. Coalescent times (thousands of years ago) to most recent common
ancestor for major mitochondrial lineages of brown and polar bears based on
581 bp of the mitochondrial DNA control region, excluding indels. Provided
are nodal support for lineages based on Bayesian analysis in BEAST and median
age of nodes with 95% confidence intervals (CI). ABC bears refer to brown bears
from the Alexander Archipelago, Alaska, USA.
S10 Table. Average pairwise distances within and among haplotypes from polar
bears sampled from 15 subpopulations, the ancient Poolypenten (GenBank
Accession No. GU573488)* polar bear and haplotypes found within the three
clades of Alaskan brown bears. Values were generated using the Tamura-Nei
(I+G0.69) model of substitution.
S1 Supporting Information. Detailed materials and methods.
Technological support was provided by University of Alaska Life Science
Informatics computer cluster (NIH P20RR016466). We thank all the additional
scientists and technicians who assisted in the collection of samples across the
Arctic. Specifically, we thank the following for field, laboratory and graphics
assistance: A. Coxon, C. Didham, M. Dyck, M. Fowler. Any use of trade, product
or firm names is for descriptive purposes only and does not imply endorsement by
the US Government. The findings and conclusions in this article are those of the
authors and do not necessarily represent the views of the USFWS. This article has
been peer reviewed and approved for publication consistent with USGS
Fundamental Science Practices (http://pubs.usgs.gov/circ/1367).
Conceived and designed the experiments: EP DP SAS SLT AGH GKS. Performed
the experiments: EP SAS AGH SLT GKS. Analyzed the data: EP SAS AGH SLT
GKS. Contributed reagents/materials/analysis tools: EP MEO AB EVR NO JA SNA
EZ LB DE SCA SB EWB AED IS MKT OW. Wrote the paper: EP SLT SAS.
Provided expertise and editing: EP SAS MEO AB EVR JA SNA AGH EZ LB DE
KTS SCA SB EWB AED IS MKT OW DP SLT GKS.
16. Rode KD, Regehr EV, Douglas DC, Durner GD, Derocher AE, et al. (2014) Variation in the response
of an Arctic top predator experiencing habitat loss: feeding and reproductive ecology of two polar bear
populations. Glob Chang Biol 20: 7688.
63. Excoffier L, Smouse PE, Quattro JM (1992) Analysis of molecular variance inferred from metric
distances among DNA haplotypes - application to human mitochondrial DNA restriction data. Genetics
1. Gillett NP , Stone DA , Stott PA , Nozawa T , Yu A , et al. ( 2008 ) Attribution of polar warming to human influence . Nat Geosci 1 : 750 - 754 .
2. Post E , Bhatt US , Bitz CM , Brodie JF , Fulton TL , et al. ( 2013 ) Ecological consequences of sea-ice decline . Science 341 : 519 - 24 .
3. Stroeve J , Holland MM , Meier W , Scambos T , Serreze M ( 2007 ) Arctic sea ice decline: Faster than forecast . Geophysical Research Letters 34 : L09501 .
4. Overland JE , Wang M ( 2013 ) When will the summer Arctic be nearly sea ice free? Geophys Res Lett 40 : 2097 - 2101
5. Stirling I , Derocher AE ( 1993 ) Possible impacts of climatic warming on polar bears . Arctic 46 : 240 - 245 .
6. Amstrup SC , Marcot BG , Douglas DC ( 2008 ) A Bayesian network modeling approach to forecasting in the 21st century worldwide status of polar bears . In: DeWeaver ET , Bitz CM , Tremblay LB , editors. Arctic sea ice decline: observations , projections, mechanisms, and implications: American Geophysical Union , pp. 213 - 268 .
7. Amstrup SC , DeWeaver ET , Douglas DC , Marcot BG , Durner GM , et al. ( 2010 ) Greenhouse gas mitigation can reduce sea-ice loss and increase polar bear persistence . Nature 468 : 955 - 960 .
8. Vongraven D , Aars J , Amstrup SC , Atkinson SN , Belikov S , et al. ( 2012 ) A circumpolar monitoring framework for polar bears . Ursus Monograph Series Number 5 : 66 .
9. Stirling I , Lunn NJ , Iacozza J ( 1999 ) Long-term trends in the population ecology of polar bears in western Hudson Bay in relation to climatic change . Arctic 52 : 294 - 306 .
10. Rode KD , Amstrup SC , Regehr EV ( 2010 ) Reduced body size and cub recruitment in polar bears associated with sea ice decline . Ecol Appl 20 : 768 - 782 .
11. Rode KD , Peacock E , Taylor MK , Stirling I , Born EW , et al. ( 2012 ) A tale of two polar bear populations (Ursus maritimus): ice habitat, harvest, and body condition . Popul Ecol 54 : 3 - 18 .
12. Regehr EV , Hunter CM , Caswell H , Amstrup SC , Stirling I ( 2010 ) Survival and breeding of polar bears in the southern Beaufort Sea in relation to sea ice . J Anim Ecol 79 : 117 - 127 .
13. Regehr EV , Lunn NJ , Amstrup SC , Stirling I ( 2007 ) Effects of earlier sea ice breakup on survival and population size of polar bears in western Hudson Bay . J Wildl Manage 71 : 2673 - 2683 .
14. Peacock E , Laake J , Laidre KL , Born EW , Atkinson SN ( 2012 ) The utility of harvest recoveries of marked individuals to assess polar bear (Ursus maritimus) survival . Arctic 65 : 391 - 400 .
15. Peacock E , Taylor MK , Laake J , Stirling I ( 2013 ) Population ecology of polar bears in Davis Strait, Canada and Greenland . J Wildl Manage 77 : 463 - 476 .
17. Durner GM , Douglas DC , Nielson RM , Amstrup SC , McDonald T , et al. ( 2009 ) Predicting 21st-century polar bear habitat distribution from global climate models . Ecol Monog 79 : 25 - 58 .
18. Molnar PK , Derocher AE , Theimann GW , Lewis MA ( 2010 ) Predicting survival, reproduction and abundance of polar bears under climate change . Biol Conserv 143 : 1612 - 1622 .
19. Derocher AE , Lunn NJ , Stirling I ( 2004 ) Polar bears in a warming climate . Integr Comp Biol 44 : 163 - 176 .
20. Cherry SG , Derocher AE , Thiemann GW , Lunn NJ ( 2013 ) Migration phenology and seasonal fidelity of an Arctic marine predator in relation to sea ice dynamics . J Anim Ecol 82 : 912 - 921 .
21. Crompton AE , Obbard ME , Petersen SD , Wilson PJ ( 2008 ) Population genetic structure in polar bears (Ursus maritimus) from Hudson Bay, Canada: Implications of future climate change . Biol Conserv 141 : 2528 - 2539 .
22. Obbard ME , Thiemann GW , Peacock E , DeBruyn TD ( 2010 ) Polar Bears: Proceedings of the 15th Working Meeting of the IUCN/SSC Polar Bear Specialist Group , Copenhagen, Denmark 29 June-3 July 2009 . Gland, Switzerland and Cambridge, UK: IUCN , p vii + 235 pp.
23. Taylor MK , Lee J ( 1995 ) Distribution and abundance of Canadian polar bear populations - a management perspective . Arctic 48 : 147 - 154 .
24. Taylor MK , Akeeagok S , Andriashek D , Barbour W , Born EW , et al. ( 2001 ) Delineating Canadian and Greenland polar bear (Ursus maritimus) populations by cluster analysis of movements . Can J Zool 79 : 690 - 709 .
25. Mauritzen M , Derocher AE , Wiig , Belikov S , Boltunov AN , et al. ( 2002 ) Using satellite telemetry to define spatial population structure in polar bears in the Norwegian and western Russian Arctic . J Appl Ecol 39 : 79 - 90 .
26. Amstrup SC , McDonald TL , Durner GM ( 2004 ) Using satellite radiotelemetry data to delineate and manage wildilfe populations . Wildl Soc Bull 32 : 661 - 679 .
27. Paetkau D , Amstrup SC , Born EW , Calvert W , Derocher AE , et al. ( 1999 ) Genetic structure of the world's polar bear populations . Mol Ecol 8 : 1571 - 1584 .
28. Zeyl E , Aars J , Ehrich D , Wiig ( 2009 ) Families in space: relatedness in the Barents Sea population of polar bears (Ursus maritimus) . Mol Ecol 18 : 735 - 749 .
29. Smith P , Stirling I , Jonkel C , Juniper I ( 1975 ) Notes on the present status of the polar bear (Ursus maritimus) in Ungava Bay and northern Labrador . Canadian Wildlife Service, Progress Notes 53 . 8 p.
30. Packard AS ( 1886 ) The former southern limits of the white or polar bear . Am Nat 20 : 655 - 659 .
31. Rausch RL ( 1953 ) On the land mammals of St . Lawrence Island, Alaska. Murrelet 34 : 18 - 26 .
32. Elliot HW ( 1875 ) Polar bears on St . Matthew Island. Harper's Weekly Journal of Civilization, May 1 Issue , pp. 1 - 2 .
33. Miller W , Schuster SC , Welch AJ , Ratan A , Bedoya-Reina OC , et al. ( 2012 ) Polar and brown bear genomes reveal ancient admixture and demographic footprints of past climate change . Proc Natl Acad Sci U S A - Plus . 109 : E2382 - 2390 .
34. Hope AG , Waltari E , Payer DC , Cook JA , Talbot SL ( 2013 ) Future distribution of tundra refugia in northern Alaska . Nature Clim Change doi: 10.1038/nclimate1926.
35. Cronin MA , Amstrup SC , Garner GW , Vyse ER ( 1991 ) Interspecific and intraspecific mitochondrial DNA variation in North American bears (Ursus) . Can J Zool 69 : 2985 - 2992 .
36. Lindqvist C , Schuster SC , Sun Y , Talbot SL , Qi J , et al. ( 2010 ) Complete mitochondrial genome of a Pleistocene jawbone unveils the origin of polar bear . Proc Natl Acad Sci U S A 107 : 5053 - 5057 .
37. Talbot SL , Shields GF ( 1996 ) Phylogeography of Brown Bears (Ursus arctos) of Alaska and paraphyly within the Ursidae . Mol Phylogenet Evol 5 : 477 - 494 .
38. Barnes I , Matheus P , Shapiro B , Jensen D , Cooper A ( 2002 ) Dynamics of Pleistocene population extinctions in Beringian brown bears . Science 295 : 2267 - 2270 .
39. Davison J , Ho SY , Simon YW , Bray SC , Korsten M , et al. ( 2011 ) Late-Quaternary biogeographic scenarios for the brown bear (Ursus arctos), a wild mammal model species . Quaternary Sci Rev 30 : 418 - 430 .
40. Edwards CJ , Suchard MA , Lemey P , Welch JJ , Barnes I , et al. ( 2011 ) Ancient hybridization and an Irish origin for the modern polar bear matriline . Current Biology 21 : 1251 - 1258 .
41. Cahill JA , Green RE , Fulton TL , Ovsyanikov N , Salamzade R , et al. ( 2013 ) Genomic evidence for island population conversion resolves conflicting theories of polar bear evolution . PLOS Genetics 9 : 1 - 8 .
42. Stirling I ( 2011 ) Polar bears: A natural history of a threatened species (Fitzhenry & Whiteside, Markham, Ontario), 300 p.
43. Pritchard JK , Stephens M , Donnelly P ( 2000 ) Inference of population structure using multilocus genotype data . Genetics 155 : 645 - 959 .
44. Corander J , Marttinen P ( 2006 ) Bayesian identification of admixture events using multi-locus molecular markers . Molec Ecol 15 : 2833 - 2843 .
45. Cronin MA , MacNeil MD ( 2012 ) Genetic relationships of extant brown bears (Ursus arctos) and polar bears (Ursus maritimus) . J Hered 103 : 873 - 881 .
46. Beerli P , Felsenstein J ( 1999 ) Maximum likelihood estimation of migration rates and population numbers of two populations using a coalescent approach . Genetics 152 : 763 - 773 .
47. Wilson GA , Rannala B ( 2003 ) Bayesian inference of recent migration rates using multilocus genotypes . Genetics 163 : 1177 - 1191 .
48. Thiemann GW , Derocher AE , Stirling I ( 2008 ) Polar bear Ursus maritimus conservation in Canada: an ecological basis for identifying designatable units . Oryx 42 : 504 - 515 .
49. Peacock E , Derocher AE , Theimann GW , Stirling I ( 2011 ) Conservation and management of Canada's polar bears in a changing Arctic . Can J Zool 89 : 371 - 385 .
50. Jackson JV , Talbot SL , Farley S ( 2008 ) Genetic characterization of Kenai brown bears (Ursus arctos): microsatellite and mitochondrial DNA control region variation in brown bears of the Kenai Peninsula, southcentral Alaska . Can J Zool 86 : 756 - 764 .
51. Swofford DL , Selander RB ( 1981 ) Biosys-1 - A FORTRAN program for the comprehensive analysis of electrophoretic data in population genetics and systematics J Hered 72 : 281 - 283 .
52. Parks SDE ( 2001 ) Trypanotolerance in West African Cattle and the population genetic effects of selection . Ph. D. (University of Dublin).
53. Goudet J ( 1995 ) FSTAT (Version 1.2): A computer program to calculate F-statistics . J Hered 86 : 485 - 486 .
54. Raymond M , Rousset F ( 1995 ) An exact test for population differentiation . Evolution 49 : 1280 - 1283 .
55. Excoffier L , Laval G , Schneider S ( 2005 ) Arlequin ver . 3 .0: An integrated software package for population genetics data analysis . Evol Bioinform Online 1 : 47 - 50 .
56. Nei M ( 1987 ) Molecular evolutionary genetics . Columbia University Press, New York. 512 p.
57. Ewens WJ ( 1972 ) The sampling theory of selectively neutral alleles . Theor Popul Biol 3 : 87 - 112 .
58. Watterson G ( 1975 ) On the number of segregating sites in genetical models without recombination . Popul Biol 7 : 256 - 276 .
59. Weir BS , Cockerham CC ( 1984 ) Estimating F-statistics for the analysis of population structure . Evolution 38 : 1358 - 1370 .
60. Corander J , Marttinen P ( 2006 ) Bayesian identification of admixture events using multi-locus molecular markers . Mol Ecol 15 : 2833 - 2843 .
61. Meirmans PG ( 2006 ) Using the AMOVA framework to estimate a standardised genetic differentiation measure . Evolution 60 : 2399 - 2402 .
62. Posada D , Crandall KA ( 1998 ) MODELTEST: testing the model of DNA substitution . Bioinformatics 14 : 817 - 818 .
64. Zink RM , Barrowclough GF ( 2008 ) Mitochondrial DNA under siege in avian phylogeography . Mol Ecol 17 : 2107 - 2121 .
65. Wilson GA , Rannala B ( 2003 ) Bayesian inference of recent migration rates using multilocus genotypes . Genetics 163 : 1177 - 1191 .
66. Evanno G , Regnaut S , Goudet J ( 2005 ) Dectecting the number of clusters of individuals using the software STRUCTURE: a simulation study . Molec Ecol 14 : 2611 - 2620 .
67. Rice WR ( 1989 ) Analzying tables of statistical tests . Evolution 43 ( 1 ): 223 - 225 .
68. Beerli P , Felsenstein J ( 2001 ) Maximum likelihood estimation of a migration matrix and effective population sizes in n subpopulations using a coalescent approach . Proc Natl Acad Sci USA 98 : 4563 - 4568 .
69. Swofford DL ( 2000 ) PAUP*. Phylogenetic Analysis Using Parsimony (*and other methods) . Sinauer Associates , Sunderland, MA, 4 p.
70. Heled J , Drummond AJ ( 2008 ) Bayesian inference of population size history from multiple loci . BMC Evol Biol 8 : 289 .
71. Rogers AR , Harpending H ( 1992 ) Population growth makes waves in the distribution of pairwise genetic differences . Mol Biol Evol 9 : 552 - 569 .
72. Harpending HC ( 1994 ) Signature of ancient population growth in a low resolution mitochondrial-DNA mismatch distribution . Hum Biol 66 : 591 - 600 .
73. Tajima F ( 1989 ) Statistical method for testing the neutral mutation hypothesis by DNA polymorphism . Genetics 123 : 585 - 595 .
74. Fu YX ( 1997 ) Statistical tests of neutrality of mutations against population growth, hitchhiking and background selection . Genetics 147 : 915 - 925 .
75. Kuhner MK ( 2006 ) Lamarc 2.0: maximum likelihood and Bayesian estimation of population parameters . Bioinformatics 22 : 768 - 770 .
76. Librado P , Rozas J ( 2009 ) DnaSP v5: a software for comprehensive analysis of DNA polymorphism data . Bioinformatics 25 : 1451 - 1452 .
77. Rogers AR ( 1995 ) Genetic evidence for a Pleistocene population explosion . Evolution 49 : 608 - 615 .
78. Drummond AJ , Rambaut A ( 2007 ) BEAST: Bayesian evolutionary analysis by sampling trees . BMC Evolutionary Biology 7 : 214 .
79. Saarma U , Ho SYW , Pybus OG , Kaljuste M , Tumanov IL , et al. ( 2007 ) Mitogenetic structure of brown bears (Ursus arctos L.) in north-eastern Europe and a new time-frame for the formation of European brown bear lineages . Mol Ecol 16 : 401 - 413 .
80. Talbot SL , Shields GF ( 1996 ) A phylogeny of the bears (Ursidae) inferred from complete sequences of three mitochondrial genes . Mol Phylogenet Evol 5 : 567 - 575 .
81. Colony R , Thorndike AS ( 1984 ) An estimate of the mean field of Arctic sea ice motion . J Geophys Res 89 (C6): 10623 - 10629 .
82. Peacock E , Derocher AE , Obbard ME , Lunn NF ( 2010 ) Polar bear ecology and management in Hudsoy Bay in the face of climate change . In: Loseto LL , Ferguson SH , Gilchrist G, editors. Climate change and top predators in Hudson Bay . New York : Springer. pp. 93 - 116 .
83. Iverson SJ , Stirling I , Lang SLC ( 2006 ) Spatial and temporal variation in the diets of polar bears across the Canadian Arctic: indicators of changes in prey popualtions and environment . In: Body IL, Wanless S , Camphyuysen CJ, editors. Top predators in marine ecosystems . Cambridge : Cambridge University Press . pp. 98 - 117 .
84. Stirling I , Calvert W , Andriashek D ( 1980 ) Population ecology studies of the polar bear in the area of southeastern Baffin Island . Canadian Wildlife Service Occasional Paper 44 : 31 p.
85. Stirling I , Kiliaan HPL ( 1980 ) Population ecology studies of the polar bear in northern Labrador Canadian Wildlife Service Occasional Paper 42 : 19 p.
86. Sahanatien V , Derocher AE ( 2012 ) Monitoring sea ice habitat fragmentation for polar bear conservation . Anim Conserv 15 : 397 - 406 .
87. Stirling I , Parkinson CL ( 2006 ) Possible effects of climate warming on selected populations of polar bears (Ursus maritimus) in the Canadian Arctic . Arctic 59 : 261 - 275 .
88. Obbard ME , Middel KR ( 2012 ) Bounding the Southern Hudson Bay polar bear subpopulation . Ursus 23 : 134 - 144 .
89. Dawson NG , Hope AG , Talbot SL , Cook JA ( 2013 ) A multilocus evaluation of ermine (Mustela erminea) across the Holarctic, testing hypotheses of Quaternary diversification in response to climate change . J Biogeog 40 : 415 - 429 .
90. Liu S , Lorenzen ED , Fumagalli M , Li B , Harris K , et al. ( 2014 ) Population genomics reveal recent speciation and rapid evolutionary adaptation in polar bears . Cell 157 : 1 - 10 .
91. Hope AG , Ho SYW , Malaney JL , Cook JA , Talbot SL ( 2014 ) Accounting for rate variation among lineages in comparative demographic analyses . Evolution 68 ( 9 ): 2689 - 2700 .
92. Hailer F , Kutschera VE , Hallstr om BM , Fain SR , Leonard JA , et al. ( 2013 ) Response to comment on "Nuclear genomic seqeunces reveal the polar bears are an old and distinct bear lineage ." Science 339 : 1522 .
93. Nakagome S , Mano S , Hasegawa M ( 2013 ) Comments on ''Nuclear genomic sequences reveal that polar bears are an old and distinct bear lineage .'' Science 339 : 1522 .
94. Cronin MA , Rincon G , Meredith RW , MacNeil MD , Islasns-Trejo A , et al. ( 2014 ) Molecular phylogeny and SNP variation of polar bears (U . maritimus), brown bears (U. arctos), and black bears (U. americanus) derived from genome sequences . J Hered 105 ( 3 ): 312 - 323 .
95. Hailer F , Kutschera VE , Hallstr om BM , Klassert D , Fain SR , et al. ( 2012 ) Nuclear genomic sequences reveal that polar bears are an old and distinct bear lineage . Science 336 : 344 - 347
96. Bidon T , Janke A , Fain SR , Eiken HG , Hagen SB , et al. ( 2014 ) Brown and polar bear Y chromosomes reveal extensive male-biased gene flow within brother lineages . Mol Biol and Evol 31 ( 6 ): 1353 - 1363 .
97. Lorenzen ED , Nogu es -Bravo D , Orlando L , Weinstock J , Binladen J , et al. ( 2011 ) Species-specific responses of Late Quaternary megafauna to climate and humans . Nature 479 : 359 - 364 .
98. Prost S , Guralnick RP , Waltari E , Fedorov VB , Kuzmina E , et al. ( 2013 ) Losing ground: past history and future fate of Arctic small mammals in a changing climate . Global Change Biol 19 : 1854 - 1864 .
99. Mourier T , Ho SY , Gilbert MTP , Willerslev E , Orlando L ( 2012 ) Statistical guidelines for detecting past population shifts using ancient DNA . Mol Biol Evol 29 : 2241 - 2251 .
100. Amstrup SC , Durner GM , McDonald TL , Mulcahy DM , Garner GW ( 2001 ) Comparing movement patterns of satellite-tagged male and female polar bears . Can J Zool 79 : 2147 - 2158 .
101. Laidre KL , Born EW , Gurarie E , Wiig , Dietz R , et al. ( 2012 ) Females roam while males patrol: divergence in breeding season movements of pack-ice polar bears (Ursus maritimus) . Proc Biol Sci 280 : 20122371 - 20122371 .
102. Stirling I , Andriashek D , Calvert W ( 1993 ) Habitat preferences of polar bears in the western Canadian Arctic in late winter and spring . Polar Rec 29 ( 168 ): 13 - 24 .
103. Lone K , Aars J , Ims RA ( 2012 ) Site fidelity of Svalbard polar bears revealed by mark-recapture positions . Polar Biol 36 ( 1 ): 27 - 39 .
104. Stirling I , McDonald TL , Richardson ES , Regehr EV , Amstrup SC ( 2011 ) Polar bear population status in the northern Beaufort Sea , Canada, 1971 - 2006 . Ecol Appl 21 : 859 - 876 .