The mating system of polar bears: a genetic
E. Zeyl, J. Aars, D. Ehrich, L. Bachmann, and Ø. Wiig
Abstract: Parentage analysis data for 583 individuals genotyped at 27 microsatellite loci were used to study the mating
system of polar bears (Ursus maritimus Phipps, 1774) in the Barents Sea area. We discriminated statistically between full
and half-siblings identified through only one common parent. We document for the first time multiple paternity in polar
bears. We demonstrated for both sexes low fidelity to mating partners over time. We did not detect any significant differ-
ence between the age distribution of adult males at capture and the age distribution of males siring cubs. This might indi-
cate that the male’s age and size are less indicative of the reproductive success than previously thought. This is further
supported by a rather long mean litter interval of 3.9 years for males siring several litters. The mating system of polar
bears in the Barents Sea appears to be promiscuous, usually with a single successful father siring full siblings within a
year, but with consecutive litters of a mother being fathered by different males. We discuss how population density, land-
scape characteristics, and adult sex ratio might influence the mating system of polar bears. This is of particular importance
for management decisions such as, e.g., implementing sex ratios in hunting quotas.
´:Nous utilisons les re
´sultats d’analyses de filiation portant sur 583 individus ge
`27 loci microsatellites,
´tudier le syste
`me de reproduction des ours blancs (Ursus maritimus Phipps, 1774) de la re
´gion de la Mer de
Barents. Nous avons statistiquement discrimine
´entre les vrais-fre
`res et sœurs et les demi-fre
`res et sœurs identifie
´s par un
seul parent commun. Nous documentons, pour la premie
`re fois chez les ours blancs, un cas de parente
´multiple dans une
´e. Nous avons de
´une faible fide
´entre partenaires sexuels, lors de diffe
´rentes reproductions, quel que
soit le sexe. Nous n’avons pas de
´rence entre la distribution des a
ˆges des ma
ˆles adultes a
`la capture et la dis-
tribution des a
ˆges des ma
ˆles au moment ou
`ils sont pe
`res. Cela pourrait indiquer que l’a
ˆge des ma
ˆles et leur taille sont une
moindre indication du succe
`s de leur reproduction qu’on ne le croyait. Cela est supporte
´par un intervalle moyen assez
long (3,9 ans) entre les porte
´es pour les pe
`res ayant engendre
´es. Nos re
´sultats indiquent que le syste
reproduction des ours blancs de la mer de Barents semble lie
´; ainsi, de manie
´rale, un ma
ˆle est le
`re de tous les petits d’une porte
´e, mais les porte
´es successives d’une femelle ont des pe
´rents. Nous discutons
comment les variations de densite
´des populations et des caracte
´ristiques du paysage, ainsi que le rapport des sexes des
adultes peuvent influencer le syste
`me de reproduction des ours blancs. Ceci est important pour les de
´cisions de gestion des
populations, par ex. la fixation du rapport des sexes dans les quotas de chasse.
Information about the mating system and the reproductive
biology of both sexes is important for wildlife conservation
and management. Mating systems can be viewed as different
degrees of female monopolization by males (Le Boeuf
1991). In species where males do not contribute to parental
care the main factor influencing the ability of males to mo-
nopolize receptive females, and hence the male reproductive
success, is the temporal distribution of receptive females.
The operational sex ratio (i.e., number of adult males per re-
ceptive female; OSR) and female spatial distribution are im-
portant factors when there is little estrous synchronization
(Ims 1988). In general, the spatial distribution of females is
closely associated with resources necessary for producing
and rearing the young (Davies and Lundberg 1984; Gehrt
and Fritzell 1998). Among marine mammals such as pinni-
peds, the spatial and temporal distribution of females deter-
mine the mating system in that the degree of polygyny
(where males mate with the same restricted group of fe-
males in successive mating attempts; Clutton-Brock 1989)
varies with female clumping and estrous synchrony (Boness
1991). In land-breeding pinnipeds such as northern elephant
seals (Mirounga angustirostris (Gill, 1866)), the females are
spatially clumped. Accordingly, the mating system is highly
polygynous and the variance of lifetime reproductive suc-
cess is estimated to be four times greater among males than
females (Le Boeuf and Reiter 1988; Le Boeuf 1991). The
mating systems of ice-breeding seals with spatially scattered
distribution of females are less well known (Le Boeuf
1991). Promiscuity such as in harp seals (Pagophilus groen-
landicus (Erxleben, 1777); Kovacs 1995 in Kovacs et al.
1997) and slight polygyny or serial monogamy (where fe-
males (or males) usually mate with a single partner in suc-
cessive breeding attempts during a single breeding season
but mate with several different partners during their lifetime;
Clutton-Brock 1989) such as in hooded seals (Cystophora
cristata (Erxleben, 1777); Boness et al. 1988; McRae and
Kovacs 1994) has been reported.
Received 21 April 2009. Accepted 3 September 2009. Published
on the NRC Research Press Web site at cjz.nrc.ca on
4 December 2009.
E. Zeyl,1L. Bachmann, and Ø. Wiig. Natural History
Museum, National Centre for Biosystematics, University of
Oslo, P.O. Box 1172 Blindern, NO-0318 Oslo, Norway.
J. Aars. Norwegian Polar Institute, NO-9296 Tromsø, Norway.
D. Ehrich. University of Tromsø, Department of Biology,
NO-9037 Tromsø, Norway.
1Corresponding author (e-mail: email@example.com).
Can. J. Zool. 87: 1195–1209 (2009) doi:10.1139/Z09-107 Published by NRC Research Press
In many species, females may mate with several males
during a mating season (polyandry, where females mate
with the same restricted group of males in successive breed-
ing attempts; Clutton-Brock 1989). In mammals, various
benefits have been hypothesized for such behavior, e.g., im-
provement of fecundity by confusing paternity and thus re-
duction of the risk of infanticide from the mating males
(Hosken and Stockley 2003), or avoidance of inbreeding
(Stockley et al. 1993). In pinnipeds, the multiple copulations
by females have been interpreted as an avoidance strategy
toward male harassment in periods when females are spa-
tially clumped (Cassini 1999).
Mating with more than one male during an oestrus may
result in multiple paternity within a litter. Multiple paternity
is common in mammals (Baker et al. 1999; Burton 2002;
DeYoung et al. 2002; Crawford et al. 2008) and has also
been documented in several bear species, such as in Ameri-
can black bears (Ursus americanus Pallas, 1780; (Schenk
and Kovacs 1995; Sinclair et al. 2003). In Swedish brown
bears (Ursus arctos L., 1758), a minimum of 10 out of 69
litters (14.5%) were fathered by several males (Bellemain et
al. 2006a). However, the frequency of multiple paternity
may vary between (Baker et al. 1999; Burton 2002) and
within species, and may depend on population density (Say
et al. 1999). Multiple paternity is expected to be more com-
mon in species with synchronized oestrus and spatially clus-
tered receptive females, making it more difficult for males
to monopolize individual receptive females (Clutton-Brock
1989; Isvaran and Clutton-Brock 2007).
Polar bears (Ursus maritimus Phipps, 1774) are solitary
marine carnivores. Their breeding season extends from
March to May (Lønø 1970; Tumanov 2001; Rosing-Asvid
et al. 2002). Most ovulations are believed to occur in April
and May (Rosing-Asvid et al. 2002) and the oestrus is thus
not synchronous. Female polar bears enter oestrus every 2 or
3 years after weaning off their offspring (Ramsay and Stir-
ling 1986; Derocher and Stirling 1998), or after they have
lost their offspring before weaning (Ramsay and Stirling
1986). Polar bears live in a highly unstable environment;
sea-ice extent and characteristics are constantly changing
under the influence of temperature, wind, and sea currents
(Ferguson et al. 1998). Females in oestrus are not particu-
larly clustered, but their distribution is considered to be de-
termined by foraging opportunities. Males may be
distributed with the same relative densities as solitary fe-
males (Ramsay and Stirling 1986). The mating success of
female polar bears may depend both on the OSR and on the
density of available breeding males (Molna
´r et al. 2008). In
an area such as the Barents Sea with moderate or high den-
sities of polar bears and where hunting is prohibited (Aars et
al. 2009), it is unlikely that lack of opportunity to find a
male partner restricts female reproduction.
Depending on social and ecological conditions, the timing
of first reproduction might be delayed well beyond sexual
maturity (Say et al. 1999). Male polar bears attain 97% of
their asymptotic body mass at approximately 13 years of
age (Derocher et al. 2005). Although age of maturity is
about 5 years, it is expected that middle-aged males will
have preferential access to females owing to increased com-
petitiveness (Bunnell and Tait 1981; Ramsay and Stirling
1986). In addition, most polar bear populations are harvested
by indigenous people, and harvest is usually male-biased
(Derocher et al. 1997). Removal of problem bears is also
typically male-biased (Dyck 2006). In a long-term perspec-
tive, this could result in a lack of sexually mature males in
the population, which could impair fecundity (McLoughlin
et al. 2005; Taylor et al. 2008). Knowing how reproductive
success of males varies with age would prove valuable to
population dynamics modelling and management.
In the present study, we use genetic and parentage data
from the Barents Sea to gain information about the mating
system of polar bears, which is currently known only from
behavioral observations (Ramsay and Stirling 1986; Wiig et
al. 1992; Rosing-Asvid et al. 2002). First, we assess evi-
dence for multiple paternity. Traditional methods to detect
multiple paternity require either that both parents are identi-
fied or that incompatible alleles are assessed within litters
consisting of at least three siblings (Burton 2002; Bellemain
et al. 2006a; Crawford et al. 2008). Here, two statistical
likelihood approaches were applied in addition to the tradi-
tional method (Bellemain et al. 2006a) to discriminate be-
tween half-siblings (HS) and full siblings (FS) within litters.
These methods were employed to discriminate between HS
and FS identified by at least one common parent (father or
mother) assigned through parentage analysis. Considering
the male-biased OSR in the Barents Sea population and ear-
lier observations of females consorting successively with
several males (Wiig et al. 1992), we expect multiple pater-
nity to occur. However, taking into account the long breed-
ing period and the unpredictability of the spatial distribution
of females during the mating season owing to large temporal
changes in distribution of sea ice, multiple paternity may
also be relatively rare. Second, we investigate to which de-
gree polar bears breed with the same mate in different years.
We also estimate the time span between litters sired by the
same male. Intense wounding and scaring (Ramsay and Stir-
ling 1986) of males during the breeding season certainly in-
dicates high male competition for mating opportunities.
Young males are hypothesized to have lesser competitive
abilities than older males (Ramsay and Stirling 1986) and to
be less attractive for females (Derocher et al. 2005). We
consequently expect low fidelity between mating partners
and a short time span between litters sired by reproductively
succesful males. Furthermore, we expect a propensity of
males siring litters to be older than the mean age of adult
males at capture. Finally, we compare the accuracy of the
different methods employed for discriminating between HS
and FS relationships.
Materials and methods
Study area, animal sampling, laboratory procedures, and
The Barents Sea population of polar bears has been de-
fined as the animals occupying the area between longitudes
of 108E and 608E, and latitudes of 728N and 838N (Wiig
and Derocher 1999); an area that belongs to Norway and
Russia. The population has been protected from harvesting
since 1973 (Prestrud and Strirling 1994). The mean density
of polar bears in the Barents Sea area is moderate to high
(1.1 bear per 100 km2in August 2004) compared with
mean densities reported in several other surveyed areas, but
1196 Can. J. Zool. Vol. 87, 2009
Published by NRC Research Press
local density varies considerably (Aars et al. 2009). The
present study is based on animals captured during spring in
the years 1995–2006. A detailed description of the study
area and the sampling procedure was provided elsewhere
(Wiig 1995; Zeyl et al. 2009). The animal-handling methods
used had previously been granted approval by the Norwe-
gian Animal Health Authority (Oslo, Norway).
The laboratory methods and basic population genetics
analysis tools used in this study have been previously de-
scribed (Zeyl et al. 2009). In brief, DNA was isolated from
tissue samples following a standard chloroform–phenol pro-
tocol for 151 samples (Sambrook and Russell 2001) or the
manufacturer’s instructions of the DNeasy tissue kit
(Qiagen, Hilden, Germany) for 432 samples. Twenty-seven
bear-specific polymorphic microsatellite loci (Crompton et
al. 2008) were amplified in six multiplex polymerase chain
reactions for 583 samples and 126 blind replicates, i.e.,
21% of the samples. The mean error rate per locus was
0.004 (range = 0–0.045) for allelic dropout (ADO) and
0.005 (range = 0–0.023) for false alleles. Two of 89
mother–offspring pairs identified by field observations
showed one incompatibility each, which is likely to result
from the mistyping of adjacent alleles corresponding to an
estimated error rate of 8.3 10–5 per locus. Evidence for
significant scoring errors was not found when comparing
the consensus of 126 samples with the consensus of the 126
blind replicates. The mean number of alleles per locus esti-
mated using GENALEX version 6 (Peakall and Smouse 2006)
was 8.04 (SD = 3.07, range = 2–15), the mean observed het-
erozygosity was 0.61 (SD = 0.24, range = 0.02–0.85), and
the mean expected heterozygosity was 0.62 (SD = 0.24,
range = 0.02–0.85). The level of resolution was high with
an estimated probability of identity of 6.74 10–23, follow-
ing Paetkau and Strobeck (1994). The probability of identity
among siblings, accounting for the presence of relatives in
the data set, remained low (1.92 10–9), following Waits
et al. (2001). No individuals shared identical genotypes. De-
viations from Hardy–Weinberg and linkage equilibriums
were tested for each locus using the GENEPOP version 3.4
software (Raymond and Rousset 1995). No deviation from
Hardy–Weinberg equilibrium was observed for the adult in-
dividuals (FIS = 0.003, p= 0.0798). There was no indication
that loci were physically linked (Zeyl et al. 2009).
We herein use the results of DNA-based parentage analy-
ses from 583 individuals (309 females, 271 males, and 3
mislabeled samples for which the reference number did not
correspond to any individual in the field data; Zeyl et al.
2009), performed using the software Parente (Cercueil et al.
2002). To avoid the incorrect rejection of parentages, a max-
imum of three incompatibilities between parent–offspring
pairs was accepted. Even in the case of a low mean error
rate of only 1% over all loci up to 40.7% of erroneous gen-
otypes can be expected, following Bonin et al. (2004). Pa-
rentage was assigned when the probability of being the true
parent was >0.5, a rather conservative cutoff value. In addi-
tion, two mother–offspring relationships were assigned be-
cause the mothers had been captured together with the cubs.
The respective probabilities of being the true mother in
these cases were low (0.2404 and 0.4802), but no genotype
incompatibility was observed. In all other cases of parentage
assigned with only low probability, we noted that several
relatives of the parent–offspring pairs were included in the
data set, which might explain the low probabilities for as-
signing the true parent (Jones and Ardren 2003; Morrissey
and Wilson 2005). The mean probability of being the true
parent for assigned parentage was high (0.97 for mother–off-
spring and 0.96 for father–offspring). Relatedness between
individuals was computed using ML-Relate (Kalinowski et
al. 2006) on all genotyped individuals (n= 583). Allele fre-
quencies were estimated with the software ML-Relate from
412 bears captured as adults, as these should represent the
reproductive part of the population.
Discriminating between full siblings and half-siblings
Directly assessing incompatible alleles within groups of
offspring (the traditional method) allows identifying HS
only in cases where three or more siblings were assigned to
a common parent (parent 1). If, in addition to the two alleles
from parent 1, three or more alleles were detected, this was
taken as an indication that the siblings were born from at
least two different parents of the sex opposite to that of the
parent 1 (i.e., that at least one of them was a HS). If no
more than two additional alleles were detected, this would,
however, not prove that the siblings were FS. Thus, the tra-
ditional method provides a minimum estimate of the number
of HS. We report the results of this method only for compar-
ison with two other statistical methods that can be applied to
the entire data set.
The second method uses maximum-likelihood estimates
of relatedness and performs a statistical test based on simu-
lated genotypes to assess the significance of the likelihood
ratio (LR) between the most likely relationship and an alter-
native relationship (Kalinowski et al. 2006). The pairwise
relationship tests are implemented in the software ML-
Relate (Kalinowski et al. 2006). For each pair of siblings
we tested either FS against HS or HS against FS, de-
pending on which relationship had the highest likelihood.
The pvalue of each hypothesis was computed using
100 000 simulated genotypes. If the pvalue was £0.05,
we accepted the relationship with the highest likelihood;
otherwise we considered the relationship as uncertain.
The third method uses also LRs to compare the evidence
for two alternative relationship hypotheses (Mayor and
Balding 2006). It differs from the ML-Relate method by tak-
ing into account the genotype of the assigned parent when
computing likelihoods. Analysis were run in R version 2.8.0
(R Development Core Team 2008). Mayor and Balding
(2006) showed that if the maternal genotype is known, only
22–24 loci are required to obtain misclassification rates be-
low 2%, applying a decision rule based on LRs larger or
smaller than 1. We used a somewhat more stringent decision
rule, as our objective was to minimize the risk of errone-
ously accepting a HS relation when a FS relation is true.
Thus, individuals were considered FS when LRs were ‡2,
and HS when LRs were £0.5. Between those values, the re-
lationship was considered uncertain.
In case of incongruent outcomes between the methods, we
accepted the results from Mayor’s method (method 3), as it
was considered more powerful (see comparison between the
Rejecting the FS hypothesis for siblings from a litter im-
plies multiple paternity. Rejection of the HS hypothesis for
Zeyl et al. 1197
Published by NRC Research Press
siblings from different litters suggests that the mother had
mated successfully with the same male in different years.
Siblings were only attributed to the same litter when they
had been captured as dependent young with their mother.
When offspring were not captured as dependent young with
their mother, their estimated year of birth is uncertain (see
below). In such cases, assigning siblings as FS could indi-
cate either that the siblings originated from a single litter or
that repeated mating between the same partners occurred in
None of the used methods accommodates for genotyping
errors. Thus loci, at which offspring showed incompatibility
with their assigned parent owing to erroneous genotyping
(such as, e.g., allelic dropout), were removed from the anal-
ysis. Removed loci were UarMU50,G10C–UarMU61,
G10B,LIST11016–G10B,LIST11020, and G10C for individ-
uals O6, O30, O33, O65, O77, and O79, respectively, in Ta-
ble 1. Removed loci were G10D–G10P–G10B,LIST11020–
G10P, and LIST11020 for individuals O100, O105, and
O107, respectively, in Table 2. Missing loci owing to ampli-
fication failure were also excluded from the analysis with
Mayor’s method, as it does not allow for missing informa-
tion. Excluded loci were UarMU61 for individual M36 in
Table 1 and UarMU50,G10D,UarMU10, and MSUT3–
G10H for individuals O28, F16, O119, and O120, respec-
tively, in Table 2.
Both likelihood methods assume that neither of the two
individuals being compared are inbred. This appears to be a
reasonable assumption for the polar bear data set, as mating
can be considered random within the studied population (E.
Zeyl, J. Aars, D. Ehrich, L. Bachmann, and Ø. Wiig, unpub-
lished data). The applied likelihood methods further assume
that no migrants are entering the population, in other words
allele frequencies do not change from one generation to the
next. The no-migrant assumption is likely to be violated.
However, we consider this only a minor bias because mi-
grants are likely to originate from neighboring populations
with similar allele frequencies (Paetkau et al. 1999).
Age of males siring litters and time span between litters
Age was known with certainty only for bears captured as
dependent offsprings with their mother. For bears captured
later in life, age was estimated by the count of cementum
growth layers of a vestigial premolar tooth (Calvert and
Ramsay 1998) or in a few cases based on field observations
of body size and tooth wear. In the Barents Sea polar bears,
counts of cementum growth layers result in imprecise age
estimates; usually the age of younger bears (<7 years) is
overestimated, while the age of older bears (‡7 years) is
underestimated (Christensen-Dalsgaard 2006). Age estimates
based on count of cementum growth layers were available
for 544 samples. For 35 samples, age was estimated from
field observations (including 22 samples from which only a
minimum age estimate could be deduced according to the
categories adult, subadult, or cub, and year of capture). Age
estimates were lacking for four samples. In the present
study, the minimum age difference for parent–offspring
pairs was set to 4 years because this age has been reported
as the earliest age of female reproduction in the Barents Sea
population (Derocher 2005), and because males in the
Barents Sea population, according to Lønø (1970), can be
sexually mature at 3.5 years.
Sixty-one fathers were included to determine the age dis-
tribution of males siring cubs. The distribution of age at re-
production was compared with the distribution of age at
capture of 190 males that could potentially have been mat-
ing (‡4 years of age at capture; Fig. 1). When an adult
male had been captured several times, it was included sev-
eral times in the age distribution of captured bears. Simi-
larly, when a male had fathered several litters, it was
included several times in the distribution of age at reproduc-
tion. The males with several litters were used to estimate the
time span between successive litters.
Multiple paternity and mate fidelity
From the total data set of the 583 individuals, parentage
analysis revealed 132 mother–offspring relations and 75
father–offspring relations (Zeyl et al. 2009). Twenty-six lit-
ters with two siblings and one litter with three siblings from
26 mothers were identified in the field and confirmed genet-
ically (Table 1). For two additional litters observed with
their mothers in the field, genetic data were available only
for the cubs and not for the mothers. Consequently, only
the ML-Relate method could be applied for those two litters.
Both parents were assigned for eight litters. The 29 litters
were tested for multiple paternity (Table 1). Of those, 27 lit-
ters were confirmed to consist of FS. In one litter, the two
siblings were confirmed to be HS, showing a case of multi-
ple paternity (M27; Table 1). Neither of the two fathers was
identified. Multiple paternity was likely also in one more lit-
ter, also with unidentified fathers. However, only Mayor’s
method supported the two siblings being HS in this case
(M36; Table 1).
Twenty females had two to four litters (consisting of one
to three offspring) hypothesized to be from different years.
Among those 20 females, it was certain that 1 female (M7)
reproduced with the same male in two different years (F8;
Table 1), as two of the offspring were captured in 2006 as
cubs, whereas the third sibling was captured in 2006 at an
estimated age of 4 years (cementum growth layers). It is
likely that two additional females (M38 and M39; Table 1)
had offspring from an identical but unidentified male in dif-
ferent years because two of their respective offspring (O87–
O89 and O90–O93), born in different years, turned out to be
FS. The age difference between offspring O87–O89 (5 years)
and O90–O93 (9 years) makes it unlikely that they were
born in a single litter. However, because of the uncertainty
of age estimates (Christensen-Dalsgaard 2006), this cannot
be firmly ruled out.
Among 17 males siring two to four offspring, 8 males had
several offspring (FS) born in the same year. In these cases,
the mother was identified genetically (M1–M8; Table 1) and
had also been observed with the cubs in the field. Twelve
males had sired two or three litters in different years. One
male certainly had offspring twice with the same female in
different years (F8 and M7; Table 2). If the offsprings’ birth
years were estimated correctly, another male is likely to
have reproduced with the same female in different years
(F17, estimated age difference between O117 and O118 was
4 years based on cementum growth layers; Table 2).
1198 Can. J. Zool. Vol. 87, 2009
Published by NRC Research Press
Comparison between the methods
Siblings for which both parents were assigned through pa-
rentage analysis allowed us to assess the accuracy of the two
statistical methods: ML-Relate and Mayor’s method. In one
case, ML-Relate erroneously classified a pair of HS as being
FS (O100 and O36, F9; Table 2). Mayor’s method did not
reach any incorrect conclusions with the applied settings for
acceptance. ML-Relate was inconclusive in three cases
where both parents were assigned (M4 and F3, M6 and F8,
M8 and F5, F6; Tables 1 and 2) and in one case where only
one parent was assigned (F8; Table 2). Mayor’s method was
only inconclusive in the M4–F3 case, when using the geno-
type of the mother (M4; Table 1) but provided a correct re-
sult when using the genotype of the father (F3; Table 2).
Overall, ML-Relate was inconclusive in 17 of 114 tests
(15%) and Mayor’s method was inconclusive in 3 of 126
tests (2%). Thus, we decided to rely on the results provided
by Mayor’s method in case of inconsistencies between
methods. In total, four inconsistencies between the two
methods were observed (Tables 1 and 2). The traditional
method did not reveal any contradiction with the results
shown by the two other methods. However, very little infor-
mation was gained using this method, as it was only applica-
ble in 11 siblings groups out of 41 defined through the
mother and in 7 siblings groups out of 17 from a common
Mayor’s method was inconclusive for the siblings O91
and O92 (M39; Table 1); however, the LR was in favour of
FS (LR = 1.26). Mayor’s method was also inconclusive for
the siblings O52 and O53 (M21; Table 1). The LR was
slightly in favour of HS (LR = 0.9). In both cases, ML-Relate
concluded FS. For the siblings O7 and O8, the LR test was
inconclusive (LR = 0.59) when using the mother’s genotype
(M4; Table 1) but was strongly in favour of FS (LR = 956)
when using the father’s genotype (F3; Table 2). Thus, we ac-
cepted FS. In that case, the relatedness between the identified
parents was high (R= 0.19), possibly leading to the inconclu-
sive outcome of one of the LR tests.
Both methods identified the case of multiple paternity
(M27; Table 1). The relatedness between the siblings was
R= 0.2317, which corresponds to what is expected for HS.
ML-Relate was not able to distinguish between HS and FS
for siblings O82 and O83 (PFS = 0.08; see M36, Table 1).
However, Mayor’s method indicated this was another case
of multiple paternity (LR = 0.3014). The relatedness be-
tween the siblings was high for HS with R= 0.4261. Thus,
we considered this case a likely multiple paternity.
The mean relatedness among the 583 genotyped bears
was R= 0.0428 (SD = 0.06, n= 169 653 pairwise compari-
sons). The mean relatedness between litter mates was close
to the mean expected relatedness for FS: R= 0.47 (SD =
0.013, n= 31 pairwise comparisons from 29 litters consist-
ing of 28 duplets and 1 triplet). Offspring sharing the same
mother who were hypothesized not to originate from the
same litter had a mean relatedness of R= 0.28 (SD =
0.012, n= 56 pairwise comparisons), a value close to the
expected relatedness for HS. Confirmed FS had a mean re-
latedness of R= 0.486 (SD = 0.010, n= 33 pairwise com-
parisons). Identified reproductive pairs had a mean
relatedness of R= 0.039 (SD = 0.049, n= 21 pairwise com-
We revealed one case of incestuous mating. F18 success-
fully mated with his daughter O119, resulting in the inbred
offspring O121. The relatedness between F18 and O121
was R= 0.7878, and the relatedness between O119 and
O121 was R= 0.6251 (Table 2). F18 was heterozygous at
14 of 27 loci; O119 (which is the mother of O121) was het-
erozygous at 13 of 26 loci (amplification failure of 1 locus);
and O121 was heterozygous at 14 of 27 loci. Manual exami-
nation of the raw data indicates that individual O121 can
have received one allele from F18 and one allele from
O119 at each locus. Such high values of relatedness are in
line with the assumption of inbreeding. However, when indi-
viduals are inbred, a correction for co-ancestry is required to
differentiate between HS and FS. This was not done in our
analyses. Nevertheless, Mayor’s method appeared more ro-
bust than ML-Relate in detecting HS in case of co-ancestry
Age distribution of males siring cubs and mean time
span between litters
Comparison of the distribution of the males’ age at cap-
ture and the age at reproduction revealed that males from
10 to 14 years of age were slightly under-represented among
successful fathers, while young males (4–9 years) and bears
aged 15–19 years were slightly over-represented (Fig. 1).
However, the differences were not significant (ANOVA,
F[1,335] = 0.0472, p= 0.8281), indicating that the mean age
at capture was not different than the mean age of reproduc-
tion. The mean time span between litters sired by the same
male was 3.9 years (n= 18, range 1–11 years). Five males
sired litters with an interval of 6 years or more (6, 6, 9, 9,
and 11 years, respectively). They were between 13 and
23 years old when they sired their last observed litter (13,
17, 15, 23, and 17 years, respectively).
Copulations with multiple males within one breeding pe-
riod are known to occur in female polar bears (Ramsay and
Stirling 1986; Wiig et al. 1992). We document for the first
time that this behavior can result in multiple paternities.
One multiple paternity was unambiguously identified among
29 litters and in another case multiple paternity was likely.
The frequency of multiple paternities in polar bears was ex-
pected to be low because of the rather long breeding period
in the population, implying asynchronous oestrus among fe-
males. Furthermore, the predictability of female locations in
the mating season is low because of the large changes in
sea-ice distribution. However, a larger data set is needed to
statistically compare the frequency of multiple paternity in
polar bears with that reported in brown bears (14.5%; Belle-
main et al. 2006a). Multiple paternities in bears occur in
general at relatively low rates in comparison with other
mammals such as rodents (Baker et al. 1999; Burton 2002;
Crawford et al. 2008), insectivores (Stockley et al. 1993), or
social carnivores (Randall et al. 2007). In the gray red-
backed vole (Clethrionomys rufocanus (Sundevall, 1846)),
Zeyl et al. 1199
Published by NRC Research Press
1200 Can. J. Zool. Vol. 87, 2009
Published by NRC Research Press
Table 1. Full-sibling (FS) and half-sibling (HS) assignments (based on traditional, ML-Relate, and Mayor’s methods) of Barents
Mother Offspring Birth Father Traditional 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
M1aO1j2003 F7 na #
O2j2003 F7 #
M2aO3j2004 F1 na #
O4j2004 F1 #
M3aO5j2006 F2 na #
O6j2006 F2 #
M4a,bO7j2006 F3 na *
O8j2006 F3 *
M5aO9j2006 F4 na #
O10j2006 F4 #
O12j2002 F6 ##
O13j2002 F6 *#
O15j2006 F8 ##
O16j2006 F8 ##
O18j2006 F5 &#
O19j2006 F5 *#
M9 O20 1994 na &
M10 O22 1995 na &
O23j2000 F19 &
M11 O24j1998tF11 na *
O25 2002 *
O28j2003 F12 &&
M13 O29 1995t
O30j2002ft F20 & & &
O31 2003 &&#
O32 2003 &&#
M14 O33 1993
O34 1993 # &&&&&
O35 1997t& & &&&*
O37 2002 *&&
O38 2002 &&&
O39j2005 F21 & & *
M15 O40 1996 na #
O41 1996 #
M16 O42 1997 na #
O43 1997 #
M17 O44 1997 na #
O45 1997 #
M18 O46 1997 na #
O47 1997 #
M19 O48 1997 na #
O49 1997 #
M20 O50 1997 na *
O51 1997 *
M21eO52 2003 na #
O53 2003 #
M22 O54 2004 na #
O55 2004 #
M23 O56 2005 na *
O57 2005 *
M24 O56 2006 na #
O57 2006 #
M25 O58 2006 na #
O59 2006 #
Sea polar bears (Ursus maritimus) as determined through the mother.
& & &&&&
&&& & & & &&&
&#& & & &&#&
&# * & & & &# &
&&* & & & &&&
Zeyl et al. 1201
Published by NRC Research Press
1202 Can. J. Zool. Vol. 87, 2009
Published by NRC Research Press
Table 1 (concluded).
Mother Offspring Birth Father Traditional 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
M26 O60 2006 na #
O61 2006 #
M27fO62 2006 na &
O63 2006 &
M28 O64 1986t
M29 O66 1987t
O67 1987ft &
M30 O68 1987t
M31 O70 1988t
M32 O72 1989t
O73 1996 &
M33 O74 1993t
M34 O76 1995 na &
M35 O78 1991t
O79 2004 &#
O80 2004 &#
O82 2006 &*
O83 2006 &*
M37 O84 1989t
O86 1996 &&
O88 1994 *&
O91 1994 &#*
O92 1994 &# &
(M40)iO94 2001 F22 na #
O95 2001 ?F22 #
(M41) O96 1992 na #
O97 1992 #
Note: The mothers in parenthesis were determined from field data only. Siblings for which both parents were assigned through parentage analysis
when age was determined from tooth cementum growth layers and ‘‘ft’’ indicating when age was determined from field measures of body size
denotes no indication that the siblings were sired by different fathers, and ‘‘different’’ denotes that there was indication that the siblings were sired
‘‘&’’s; inconclusive assignments are indicated by ‘‘*’’s.
aThe ML-Relate test for siblings with both parents assigned by parentage analysis is identical in Tables 1 and 2.
bBoth methods were inconclusive. FS was accepted following Mayor’s method using the father’s genotype.
cFemale that mated twice with the same male. O15 and O16 were captured as dependent cubs in 2006; O14 was captured as subadult in 2006.
dInconsistency between the two methods. Mayor’s results were accepted (see text).
eMayor’s method was inconclusive.
fMultiple paternity accepted.
gLikely case of multiple paternity.
hSame litter or the female mated twice with the same male.
iThe probability of being the true father (?F22) was low in this case (p= 0.3), but there were no incompatibilities. See parentage analysis (Zeyl et al. 2009).
are indicated by ‘‘j’’s. The offsprings’ year of birth appears in boldface type when first captured as juveniles together with their mother, with ‘‘t’’ indicating
and tooth-wear examination). The fathers in boldface type also appear in Table 2. For the traditional method, ‘‘na’’ denotes not applicable, ‘‘negative’’
by at least two different fathers. For ML-Relate and Mayor’s methods, FS pairwise relationships are indicated by ‘‘#’’s and HS pairwise relationshipsby
Zeyl et al. 1203
Published by NRC Research Press
Table 2. Full-sibling (FS) and half-sibling (HS) assignments (based on traditional, ML-Relate, and Mayor’s methods) of Barents
Sea polar bears (Ursus maritimus) as determined through the father.
Father Offspring Birth Mother Traditional 12345612 345 6
F1aO3j2004 M2 na ##
O4j2004 M2 ##
F2aO5j2006 M3 na ##
O6j2006 M3 ##
F3aO7j2006 M4 na *#
O8j2006 M4 *#
F4aO9j2006 M5 na ##
O10j2006 M5 ##
F5aO18j2006 M8 na ##
O19j2006 M8 ##
&&& && &
O11j2002 M6 &#*&##
O12j2002 M6 &## & # #
O13j2002 M6 &*# & ##
&&& && &
O1j2003 M1 &#&& #&
O2j2003 M1 &#&&# &
O98j2005 M42 & & & & & &
&# # &# #
O99 2004t&*&& &&
O15j2006 M7 #*## &#
O16j2006 M7 #&# # &#
F9c,dO100 1995 (M43) na #&
F10 O101 2002 (M44) na *&
F11 O103 1989t
O24j1998tM11 *& &&
O104j1999 M45 * & & &
F12 O105 1992t
O106 2001 (M46) * & & &
O28j2003tM12 && & &
F14 O109 1989t
F15 O111 1990t
O114 1995t&& & &
O115 1999t#& &&
O117 1991t&# & #
O118 1995t*# &#
O120j1998ft M47 # # & #
O121j2003tO119 # #
Note: Siblings for which both parents were assigned through parentage analysis are indicated by ‘‘j’’s. The offsprings’ year of birth appears in
boldface type when first captured as juveniles together with their mother, with ‘‘t’’ indicating when age was determined from tooth cementum
growth layers and ‘‘ft’’ indicating when age was determined from field measures of body size and tooth-wear examination). The mothers in bold-
face type also appear in Table 1; mothers in parenthesis were determined from field data only. For the traditional method, ‘‘na’’ denotes not ap-
plicable, ‘‘negative’’ denotes no indication that the siblings were born by different mothers, and ‘‘different’’ denotes that there was indication that
the siblings had at least two different mothers. For ML-Relate and Mayor’s methods, FS pairwise relationships are indicated by ‘‘#’’s and HS
pairwise relationships by ‘‘&’’s; inconclusive assignments are indicated by ‘‘*’’s. Data in italic type indicate a father who mated incestuously with
aThe ML-Relate test for siblings with both parents assigned by parentage analysis is identical in Tables 1 and 2.
bMale mated twice with the same female. O15 and O16 were captured as dependent cubs in 2006. O14 was captured as subadult in 2006.
cInconsistency between the two methods. We accepted Mayor’s results (see text).
dInconsistency with the parentage analysis (with assigned mothers).
eSame litter or male mated twice with the same female.
fInbreeding case. The female O119 is believed to be the mother of O121.
1204 Can. J. Zool. Vol. 87, 2009
Published by NRC Research Press
the proportion of litters showing multiple paternities was
positively correlated with local density of males around fe-
males in oestrus (Ishibashi and Saitoh 2008). Further studies
are required to examine whether the frequencies of multiple
paternity in polar bears vary among populations with differ-
Uncertainty in age estimation limited our ability to distin-
guish between litters of siblings identified through a com-
mon parent. However, our results indicated that both
females and males rarely mated with the same partner in dif-
ferent years. Most consecutive litters were sired by different
males. Both female and male polar bears show area fidelity
during the breeding season; however, gene flow appears to
be slightly male-biased within the Barents Sea population
(Zeyl et al. 2009). One could therefore expect that specific
females frequently mated with the same neighboring male
in different years. However, a relatively high number of
available males in most areas may be sufficient to counter-
act this. Fierce male competition because of a male-biased
operational sex ratio should work in the same direction. Ac-
cordingly, we detected only one case of incestuous mating.
Mating between father and daughter has not been docu-
mented previously in polar bears and has been rarely re-
ported in other bear species. A putative case has been
mentioned in black bears (Costello et al. 2008). Father–
daughter matings were detected in 2 out of 95 litters in
Swedish brown bears (Bellemain et al. 2006b). The spatial
structure of bears, associated with low recruitment, male-
biased dispersal, and male turnover, appears to prevent high
rates of incestuous mating. Thus, there is no need for active
inbreeding avoidance mechanisms as suggested by Belle-
main et al. (2006b) and Costello et al. (2008). The mean re-
latedness observed between polar bear mating pairs is
consistent with the indication that female bears generally do
not mate with close relatives, as reported in brown bears
(Bellemain et al. 2006b).
It has previously been suggested that male polar bears are
polygynous (DeMaster and Stirling 1981) and that females
are polyandrous (Ramsay and Stirling 1986; Wiig et al.
´r et al. 2008). Our results indicate that at least
in the Barents Sea, the mating system of polar bears is pro-
miscuous (where males mate with any accessible receptive
female and there is no continuing bond between individual
males and females after mating has occurred. Females usu-
ally mate with different males in successive breeding at-
tempts. In some species, females mate with several different
males during each period of receptivity, whereas in others,
they typically mate with a single male; Clutton-Brock
1989). Since this conclusion is only based on a limited num-
ber of litters, it should be taken with caution. Nevertheless,
the results are in accordance with Clutton-Brock (1989),
who predicted a promiscuous mating system when females
are widely and unpredictably distributed. The mating system
of a species is plastic: there is a correlation between the
mating system and population densities (Kokko and Rankin
2006). In swift foxes (Vulpes velox (Say, 1823)), the mating
system is polygynous in high-density areas, whereas it is
monogamous in low-densities areas (Kamler et al. 2004). In
European water voles (Arvicola terrestris (L., 1758)), con-
secutive litters were found fathered by a single male in low-
density populations and usually by different males in a high-
density population (Aars et al. 2006). The Barents Sea has a
moderate to high polar bear density compared with several
other areas, but local densities are variable (Aars et al.
2009). A lower density of available breeders within a popu-
lation that can, e.g., be caused by excessive hunting may in-
duce a change in the mating system of the animals within
the area. If human hunting of polar bears is not male-biased,
or if no hunting occurs such as in the Barents Sea popula-
tion, the adult sex ratio become close to 1:1. As many fe-
males in the mating season are with cubs and unavailable
for mating, more males than females will be available for
Fig. 1. Histogram representing (a) the relative age distribution densities of adult male polar bears (Ursus maritimus) at capture and (b) the
relative age distribution densities of males when they sired cubs. For illustration purpose, we used five age categories (4–9, 10–14, 15–19,
20–24, and 25+ years). The number of observations (n) is shown above each bar.
Zeyl et al. 1205
Published by NRC Research Press
mating (Bunnell and Tait 1981; Ramsay and Stirling 1986).
Male-biased OSR is expected to favor numerous mating en-
counters for females (DeYoung et al. 2002). Indeed, field
observations indicate that female polar bears in oestrus may
copulate with several males (Ramsay and Stirling 1986;
Wiig et al. 1992). In contrast, very low densities and a
closer to even OSR, e.g., in areas with male-biased hunting
could lead to a more monogamous mating system.
In black bears, paternity analyses revealed that male re-
productive success was dominated by intermediate-aged
bears (Costello et al. 2008), which indicates that most males
would have a relatively short reproductive tenure. This is
congruent with the observation of Ramsay and Stirling
(1986) who reported that male polar bears associated with
adult females were, on average, older (median = 10.5 years
of age) than solitary males (median = 8 years of age). Tak-
ing into account the growth pattern in male polar bears, we
expected that males belonging to the age class 15–19 would
sire most litters. It also corresponds to the period when the
foreleg guard hairs, which are hypothetized to be a sexual
ornament, reach their maximum length (Derocher et al.
2005) and when most wounds and scars are observed on
males (A.E. Derocher et al., submitted).2Sexual dimorphism
can be maintained through sexual selection, with larger body
size in males being correlated to higher reproductive success
through a better access to females (Andersson 1994). In gen-
eral, the size difference in polar bears is large (Derocher et
al. 2005), indicating that male competition for females is
fierce. Our data set was not large enough to test with high
statistical power for smaller differences in age for male re-
productive success, but showed that such differences seemed
small, contrary to our expectation. Some care should be
taken in any conclusion based on a comparison between the
age distribution of males at capture and the age distribution
of males when siring cubs. The youngest males are under-
represented in the capture data owing to lower trapability
(Derocher 2005). Older males that sired cubs may have
lower survival rates and less chance of being captured. The
main conclusion that reproductive success of adult males of
different ages were unlikely do differ profoundly still seems
to hold. Our results suggest that the male’s size and age
might be less indicative of the reproductive success than
The distribution of female polar bears during the mating
season is influenced by the distribution and accessibility of
their main prey (Stirling et al. 1993), which in the spring is
ringed seal (Pusa hispida (Schreber, 1775); Derocher et al.
2002). The prime habitat of ringed seals in Svalbard is near
glacier fronts in sheltered fjords and bays (Smith and Ly-
dersen 1991). The distribution of males in spring appears to
be correlated with the distribution of females available for
reproduction (Ramsay and Stirling 1986; Stirling et al.
1993). Consequently, some areas have higher local bear den-
sities than others (Wiig et al. 1992). This might influence
the pattern of age-related reproduction in males because the
level of competition between males may vary with density.
Indeed, in some species, fluctuations in population densities
might alter the outcome in the breeding success of males of
different age classes. In the Saint Kilda population of pro-
miscuous Soay sheep (Ovis aries L., 1758), the level of
male–male competition for mates varies with population
size and sex ratio. Old males sire larger sibships (i.e., a
group of offsprings produced by a male) than young males
at low population size when polygyny of old males is also
maximal. However, the size of sibships sired by young
males is small at any population size (Pemberton et al.
1999). In polar bears, both scramble competition (males dis-
perse to find receptive females) and contest competition
(males fight to establish dominance over other males to get
access to females) are believed to occur (A.E. Derocher et
al., submitted).2Sequestration and herding of females to-
ward low-density areas could be advantageous for younger
males because it reduces the probability of encountering
older males with superior fighting abilities. Some field ob-
servations indicate that males might herd females in the
Barents Sea population (Wiig et al. 1992), and this behavior
has also been reported in Canadian polar bears (Ramsay and
Stirling 1986). The mountains and glaciers from the Sval-
bard archipelago might offer multiple herding opportunities
for young males. Thus, the observed absence of age-related
breeding success in male polar bears could be related to the
mechanism of intermale competition to gain breeding oppor-
tunities. The landscape in the Barents Sea might favor a
high proportion of young males being able to gain breeding
opportunities. The relation between age and breeding suc-
cess of males may vary between populations according to
landscape characteristics, to the distribution of seals, and lo-
cal bear densities.
Limitation of the statistical analysis
Costello et al. (2008) found that ML-Relate was likely to
infer genetic relationships between individuals to be closer
than they were. This seemed indeed to explain some incon-
sistencies between ML-Relate and Mayor’s method in this
study. However, ML-Relate also failed to distinguish be-
tween alternative relationships (FS and HS) in cases where
siblings were FS. This might be due to a violation of the as-
sumption of co-ancestry between parents, to the presence of
groups of relatives in the data set, or this could also arise
because some parents might present genotypes with rare al-
leles. Indeed, this would influence the discriminating power
of the methods. Further studies evaluating the robustness of
those methods to hypothesis violation would be welcome.
The analysis performed with Mayor’s method appeared to
be more reliable than the results based on ML-Relate. This
is not surprising, as Mayor’s method uses the additional in-
formation provided by the genotype of one of the parents.
Moreover, both ML-Relate and Mayor’s method are able to
provide more information than the traditional method of pa-
rental allele counts, which is applicable only within groups
of n‡3 siblings when searching for multiple paternities
and studying the mating system of a species. These statisti-
cal methods allow the detection of HS within litters with
only two siblings, and thus are likely to provide a better es-
timate of the minimal rate of multiple paternities than the
2A.E. Derocher, J. Aars, M. Andersen, and Ø. Wiig. Mating ecology of polar bears at Svalbard. Submitted for publication.
1206 Can. J. Zool. Vol. 87, 2009
Published by NRC Research Press
We are grateful to Lianne Dance (previously Mayor) for
kindly providing the scripts necessary to perform the analy-
sis. We thank the anonymous reviewers for their construc-
tive comments. This work was supported by the Natural
History Museum of the University of Oslo, Norway, and the
Norwegian Research Council through the National Centre
for Biosystematics (project no. 146515/420). Samples were
provided by the Norwegian Polar Institute (Tromsø, Nor-
Aars, J., Dallas, J.F., Piertney, S.B., Marshall, F., Gow, J.L., Telfer,
S., and Lambin, X. 2006. Widespread gene flow and high ge-
netic variability in populations of water voles Arvicola terrestris
in patchy habitats. Mol. Ecol. 15(6): 1455–1466. doi:10.1111/j.
Aars, J., Marques, T.A., Buckland, S.T., Andersen, M., Belikov, S.,
Boltunov, A.N., and Wiig, Ø. 2009. Estimating the Barents Sea
polar bear subpopulation size. Mar. Mamm. Sci. 25(1): 35–52.
Andersson, M. 1994. Sexual selection. Princeton University Press,
Baker, R.J., Makova, K.D., and Chesser, R.K. 1999. Microsatellites
indicate a high frequency of multiple paternity in Apodemus
(Rodentia). Mol. Ecol. 8(1): 107–111. doi:10.1046/j.1365-294X.
Bellemain, E., Swenson, J.E., and Taberlet, P. 2006a. Mating stra-
tegies in relation to sexually selected infanticide in a non-social
carnivore: the brown bear. Ethology, 112(3): 238–246. doi:10.
Bellemain, E., Zedrosser, A., Manel, S., Waits, L., Taberlet, P., and
Swenson, J.E. 2006b. The dilemma of female mate selection in
the brown bear, a species with sexually selected infanticide.
Proc. R. Soc. Lond. B Biol. Sci. 273(1584): 283–291. doi:10.
Boness, D.J. 1991. Chap. 1. Determinants of mating systems in the
Otariidae (Pinnipedia). In Behaviour of pinnipeds. Edited by
Deane Renouf. Chapman and Hall, Bristol, UK. pp. 1–44.
Boness, D.J., Bowen, W.D., and Oftedal, O.T. 1988. Evidence of
polygyny from spatial patterns of hooded seals (Cystophora cris-
tata). Can. J. Zool. 66(3): 703–706. doi:10.1139/z88-104.
Bonin, A., Bellemain, E., Bronken Eidesen, P., Pompanon, F., Broch-
mann, C., and Taberlet, P. 2004. How to track and assess genotyp-
ing errors in population genetics studies. Mol. Ecol. 13(11): 3261–
3273. doi:10.1111/j.1365-294X.2004.02346.x. PMID:15487987.
Bunnell, F.L., and Tait, D.E.N. 1981. Population dynamics of bears
— implications. In Dynamics of large mammal populations.
Edited by C.W. Fowler and B.R. Smith. John Wiley and Sons,
New York. pp. 75–98.
Burton, C. 2002. Microsatellite analysis of multiple paternity and
male reproductive success in the promiscuous snowshoe hare.
Can. J. Zool. 80(11): 1948–1956. doi:10.1139/z02-187.
Calvert, W., and Ramsay, M. 1998. Evaluation of age determina-
tion of polar bears by counts of cementum growth layer groups.
Ursus, 10: 449–453.
Cassini, M.H. 1999. The evolution of reproductive systems in pin-
nipeds. Behav. Ecol. 10(5): 612–616. doi:10.1093/beheco/10.5.
Cercueil, A., Bellemain, E., and Manel, S. 2002. PARENTE: com-
puter program for parentage analysis. J. Hered. 93(6): 458–459.
Christensen-Dalsgaard, S.N. 2006. Temporal patterns in age struc-
ture of polar bears (Ursus maritimus) in Svalbard, with special
emphasis on validation of age determination. M.Sc. thesis, Fa-
culty of Science, University of Tromsø and Norwegian Polar In-
stitute, Tromsø, Norway.
Clutton-Brock, T.H. 1989. Review lecture: Mammalian mating sys-
tems. Proc. R. Soc. Lond. B Biol. Sci. 236(1285): 339–372.
Costello, C.M., Creel, S.R., Kalinowski, S.T., Vu, N.V., and Quig-
ley, H.B. 2008. Sex-biased natal dispersal and inbreeding avoid-
ance in American black bears as revealed by spatial genetic
analyses. Mol. Ecol. 17(21): 4713–4723. doi:10.1111/j.1365-
Crawford, J.C., Liu, Z., Nelson, T.A., Nielsen, C.K., and Bloom-
quist, C.K. 2008. Microsatellite analysis of mating and kinship
in beavers (Castor canadensis). J. Mammal. 89(3): 575–581.
Crompton, A.E., Obbard, M.E., Petersen, S.D., and Wilson, P.J.
2008. Population genetic structure in polar bears (Ursus mariti-
mus) from Hudson Bay, Canada: implications of future climate
change. Biol. Conserv. 141(10): 2528–2539. doi:10.1016/j.
Davies, N.B., and Lundberg, A. 1984. Food distribution and a vari-
able mating system in the dunnock, Prunella modularis.J.
Anim. Ecol. 53(3): 895–912. doi:10.2307/4666.
DeMaster, D.P., and Stirling, I. 1981. Ursus maritimus. Mamm.
Species, 145(145): 1–7. doi:10.2307/3503828.
Derocher, A.E. 2005. Population ecology of polar bears at Sval-
bard, Norway. Popul. Ecol. 47(3): 267–275. doi:10.1007/
Derocher, A.E., and Stirling, I. 1998. Geographic variation in
growth of polar bears (Ursus maritimus). J. Zool. (Lond.),
245(1): 65–72. doi:10.1111/j.1469-7998.1998.tb00072.x.
Derocher, A.E., Stirling, I., and Calvert, W. 1997. Male-biased har-
vesting of polar bears in western Hudson Bay. J. Wildl. Manage.
61(4): 1075–1082. doi:10.2307/3802104.
Derocher, A.E., Wiig, Ø., and Andersen, M. 2002. Diet composi-
tion of polar bears in Svalbard and the western Barents Sea.
Polar Biol. 25: 448–452. doi:10.1007/s00300-002-0364-0.
Derocher, A.E., Andersen, M., and Wiig, Ø. 2005. Sexual dimorph-
ism of polar bears. J. Mammal. 86(5): 895–901. doi:10.1644/
DeYoung, R.W., Demarais, S., Gonzales, R.A., Honeycutt, R.L., and
Gee, K.L. 2002. Multiple paternity in white-tailed deer (Odocoileus
virginianus) revealed by DNA microsatellites. J. Mammal. 83(3):
Dyck, M.G. 2006. Characteristics of polar bears killed in defense
of life and property in Nunavut, Canada, 1970–2000. Ursus,
17(1): 52–62. doi:10.2192/1537-6176(2006)17[52:COPBKI]2.0.
Ferguson, S.H., Taylor, M.K., Born, E.W., and Messier, F. 1998.
Fractals, sea-ice landscape and spatial patterns of polar bears. J.
Biogeogr. 25: 1081–1092.
Gehrt, S.D., and Fritzell, E.K. 1998. Resource distribution, female
home range dispersion and male spatial interactions: group
structure in a solitary carnivore. Anim. Behav. 55(5): 1211–
1227. doi:10.1006/anbe.1997.0657. PMID:9632506.
Hosken, D.J., and Stockley, P. 2003. Benefits of polyandry: a life
history perspective. Evol. Biol. 33: 173–194.
Ims, R.A. 1988. The potential for sexual selection in males: effect
of sex ratio and spatiotemporal distribution of receptive females.
Evol. Ecol. 2(4): 338–352. doi:10.1007/BF02207565.
Ishibashi, Y., and Saitoh, T. 2008. Effect of local density of males
Zeyl et al. 1207
Published by NRC Research Press
on the occurrence of multimale mating in gray-sided voles
(Myodes rufocanus). J. Mammal. 89(2): 388–397. doi:10.1644/
Isvaran, K., and Clutton-Brock, T. 2007. Ecological correlates of
extra-group paternity in mammals. Proc. R. Soc. Lond. B Biol.
Sci. 274(1607): 219–224. doi:10.1098/rspb.2006.3723.
Jones, A.G., and Ardren, W.R. 2003. Methods of parentage analy-
sis in natural populations. Mol. Ecol. 12(10): 2511–2523. doi:10.
Kalinowski, S.T., Wagner, A.P., and Taper, M.L. 2006. ML-Relate:
a computer program for maximum likelihood estimation of relat-
edness and relationship. Mol. Ecol. Notes, 6(2): 576–579.
Kamler, J.F., Ballard, W.B., Lemons, P.R., and Mote, K. 2004.
Variation in mating system and group structure in two popula-
tions of swift foxes, Vulpes velox. Anim. Behav. 68(1): 83–88.
Kokko, H., and Rankin, D.J. 2006. Lonely hearts or sex in the city?
Density-dependent effects in mating systems. Philos. Trans. R.
Soc. Lond. B Biol. Sci. 361(1466): 319–334. doi:10.1098/rstb.
Kovacs, K.M. 1995. Harp and hooded seal reproductive behavior
and energetics — A case study in the determinants of mating
systems in pinnipeds. In Whales, seals, fish and man. Edited by
A. Schytte-Blix, L. Walloe, and O. Ulltang. Elsevier Science B,
Amsterdam, the Netherlands. pp. 329–335.
Kovacs, K.M., Lydersen, C., Hammill, M.O., White, B.N., Wilson,
P.J., and Malik, S. 1997. A harp seal hooded seal hybrid.
Mar. Mamm. Sci. 13(3): 460–468. doi:10.1111/j.1748-7692.
Le Boeuf, B.J. 1991. Pinniped mating systems on land, ice and in
the water: emphasis on the Phocidae. In Behaviour of pinnipeds.
Edited by Deane Renouf. Chapman and Hall, Bristol, UK.
Le Boeuf, B.J., and Reiter, J. 1988. Lifetime reproductive success
in northern elephant seals. In Reproductive success. Edited by
T.H. Clutton-Brock. University of Chicago Press, Chicago.
Lønø, O. 1970. The polar bear (Ursus maritimus Phipps) in the
Svalbard area. Norsk Polarinstitutt Skrifter. No. 149.
Mayor, L.R., and Balding, D.J. 2006. Discrimination of half-siblings
when maternal genotypes are known. Forensic Sci. Int. 159(2-3):
141–147. doi:10.1016/j.forsciint.2005.07.007. PMID:16153794.
McLoughlin, P.D., Taylor, M.K., and Messier, F. 2005. Conser-
vation risks of male-selective harvest for mammals with low
reproductive potential. J. Wildl. Manag. 69: 1592–1600.
McRae, S.B., and Kovacs, K.M. 1994. Paternity exclusion by DNA
fingerprinting, and mate guarding in the hooded seal Cystophora
cristata. Mol. Ecol. 3(2): 101–107. doi:10.1111/j.1365-294X.
´r, P.K., Derocher, A.E., Lewis, M.A., and Taylor, M.K.
2008. Modelling the mating system of polar bears: a mechanistic
approach to the Allee effect. Proc. R. Soc. Lond. B Biol. Sci.
275(1631): 217–226. doi:10.1098/rspb.2007.1307.
Morrissey, M.B., and Wilson, A.J. 2005. The potential costs of ac-
counting for genotypic errors in molecular parentage analyses.
Mol. Ecol. 14(13): 4111–4121. doi:10.1111/j.1365-294X.2005.
Paetkau, D., and Strobeck, C. 1994. Microsatellite analysis of ge-
netic variation in black bear populations. Mol. Ecol. 3(5): 489–
495. doi:10.1111/j.1365-294X.1994.tb00127.x. PMID:7952329.
Paetkau, D., Amstrup, S.C., Born, E.W., Calvert, W., Derocher,
A.E., Garner, G.W., Messier, F., Stirling, I., Taylor, M.K.,
Wiig, Ø., and Strobeck, C. 1999. Genetic structure of the
world’s polar bear populations. Mol. Ecol. 8(10): 1571–1584.
Peakall, R., and Smouse, P. 2006. GENALEX 6: genetic analysis in
Excel. Population genetic software for teaching and research.
Mol. Ecol. Notes, 6(1): 288–295. doi:10.1111/j.1471-8286.2005.
Pemberton, J.M., Coltman, D.W., Smith, J.A., and Pilkington, J.G.
1999. Molecular analysis of a promiscuous, fluctuating mating
system. Biol. J. Linn. Soc. 68(1-2): 289–301. doi:10.1111/j.
Prestrud, P., and Strirling, I. 1994. The international polar bear
agreement and the current status of polar bear conservation.
Aquat. Mamm. 20: 113–124.
Ramsay, M.A., and Stirling, I. 1986. On the mating system of polar
bears. Can. J. Zool. 64: 2142–2151. doi:10.1139/z86-329.
Randall, D.A., Pollinger, J.P., Wayne, R.K., Tallents, L.A., John-
son, P.J., and Macdonald, D.W. 2007. Inbreeding is reduced by
female-biased dispersal and mating behavior in Ethiopian
wolves. Behav. Ecol. 18(3): 579–589. doi:10.1093/beheco/
Raymond, M., and Rousset, F. 1995. GENEPOP (version 1.2): po-
pulation genetics software for exact tests and ecumenicism. J.
Hered. 86: 248–249.
R Development Core Team. 2008. R: A language and environment
for statistical computing. R Foundation for Statistical Comput-
ing, Vienna, Austria. Available from http://www.r-project.org/
[accessed 30 October 2008].
Rosing-Asvid, A., Born, E.W., and Kingsley, M.C.S. 2002. Age at
sexual maturity of males and timing of the mating season of po-
lar bears (Ursus maritimus) in Greenland. Polar Biol. 25: 878–
Sambrook, J., and Russell, D.W. 2001. Molecular cloning: a la-
boratory manual. Cold Spring Harbor Laboratory Press, New
Say, L., Pontier, D., and Natoli, E. 1999. High variation in multiple
paternity of domestic cats (Felis catus L.) in relation to environ-
mental conditions. Proc. R. Soc. Lond. B Biol. Sci. 266: 2071–
Schenk, A., and Kovacs, K.M. 1995. Multiple mating between
black bears revealed by DNA fingerprinting. Anim. Behav.
50(6): 1483–1490. doi:10.1016/0003-3472(95)80005-0.
Sinclair, E.A., Black, H.L., and Crandall, K.A. 2003. Population
structure and paternity in an american black bear (Ursus ameri-
canus) population using microsatellite DNA. West. N. Am. Nat.
Smith, T.G., and Lydersen, C. 1991. Availability of suitable land-
fast ice and predation as factors limiting ringed seal populations,
Phoca hispida, in Svalbard. Polar Res. 10(2): 585–594. doi:10.
Stirling, I., Andriashek, D., and Calvert, W. 1993. Habitat prefer-
ence of polar bears in the western Canadian Arctic in late winter
and spring. Polar Rec. (Gr. Brit.), 29(168): 13–24. doi:10.1017/
Stockley, P., Searle, J.B., Macdonald, D.W., and Jones, C.S. 1993.
Female multiple mating behaviour in the common shrew as a
strategy to reduce inbreeding. Proc. R. Soc. Lond. B Biol. Sci.
254(1341): 173–179. doi:10.1098/rspb.1993.0143.
Taylor, M., McLoughlin, P.D., and Messier, F. 2008. Sex-selective
harvesting of polar bears Ursus maritumus. Wildl. Biol. 14: 52–
Tumanov, I.L. 2001. Reproductive biology of captive polar bears.
Ursus, 12: 107–108.
1208 Can. J. Zool. Vol. 87, 2009
Published by NRC Research Press
Waits, L.P., Luikart, G., and Taberlet, P. 2001. Estimating the
probability of identity among genotypes in natural populations:
cautions and guidelines. Mol. Ecol. 10(1): 249–256. doi:10.
Wiig, Ø. 1995. Distribution of polar bears (Ursus maritimus) in the
Svalbard area. J. Zool. (Lond.), 237(4): 515–529. doi:10.1111/j.
Wiig, Ø., and Derocher, A.E. 1999. Application of aerial survey
methods to polar bears in the Barents Sea. In Marine mammal
survey and assessment methods. Edited by G.W. Garner, S.C.
Amstrup, J.L. Laake, B.F.J. Manly, L.L. McDonald, and D.G.
Robertson. A.A. Bolkema Publishers, Rotterdam, Holland.
Wiig, Ø., Gjertz, I., Hansson, R., and Thomassen, J. 1992. Breed-
ing behaviour of polar bears in Hornsund, Svalbard. Polar Rec.
(Gr. Brit.), 28(165): 157–159. doi:10.1017/S0032247400013474.
Zeyl, E., Aars, J., Ehrich, D., and Wiig, Ø. 2009. Families in space:
relatedness in the Barents Sea population of polar bears (Ursus
maritimus). Mol. Ecol. 18(4): 735–749. doi:10.1111/j.1365-
Zeyl et al. 1209
Published by NRC Research Press