Research Article

Effects of Earlier Sea Ice Breakup on Survival and
Population Size of Polar Bears in Western Hudson Bay
ERIC V. REGEHR,1 United States Geological Survey, Alaska Science Center, 1011 E. Tudor Road, Anchorage, AK 99503, USA
NICHOLAS J. LUNN, Wildlife Research Division, Science and Technology Branch, Environment Canada, 5320 122 Street NW,
Edmonton, AB T6H 3S5, Canada
STEVEN C. AMSTRUP, United States Geological Survey, Alaska Science Center, 1011 E. Tudor Road, Anchorage, AK 99503, USA
IAN STIRLING, Canadian Wildlife Service, 5320 122 Street NW, Edmonton, AB T6H 3S5, Canada

ABSTRACT Some of the most pronounced ecological responses to climatic warming are expected to occur in polar marine regions, where
temperature increases have been the greatest and sea ice provides a sensitive mechanism by which climatic conditions affect sympagic (i.e., with
ice) species. Population-level effects of climatic change, however, remain difficult to quantify. We used a flexible extension of Cormack–Jolly–
Seber capture–recapture models to estimate population size and survival for polar bears (Ursus maritimus), one of the most ice-dependent of
Arctic marine mammals. We analyzed data for polar bears captured from 1984 to 2004 along the western coast of Hudson Bay and in the
community of Churchill, Manitoba, Canada. The Western Hudson Bay polar bear population declined from 1,194 (95% CI ¼ 1,020–1,368) in
1987 to 935 (95% CI ¼ 794–1,076) in 2004. Total apparent survival of prime-adult polar bears (5–19 yr) was stable for females (0.93; 95% CI
¼ 0.91–0.94) and males (0.90; 95% CI ¼ 0.88–0.91). Survival of juvenile, subadult, and senescent-adult polar bears was correlated with spring
sea ice breakup date, which was variable among years and occurred approximately 3 weeks earlier in 2004 than in 1984. We propose that this
correlation provides evidence for a causal association between earlier sea ice breakup (due to climatic warming) and decreased polar bear survival.
It may also explain why Churchill, like other communities along the western coast of Hudson Bay, has experienced an increase in human–polar
bear interactions in recent years. Earlier sea ice breakup may have resulted in a larger number of nutritionally stressed polar bears, which are
encroaching on human habitations in search of supplemental food. Because western Hudson Bay is near the southern limit of the species’ range,
our findings may foreshadow the demographic responses and management challenges that more northerly polar bear populations will experience
if climatic warming in the Arctic continues as projected. (JOURNAL OF WILDLIFE MANAGEMENT 71(8):2673–2683; 2007)

DOI: 10.2193/2006-180
KEY WORDS capture–recapture, climate change, Cormack–Jolly–Seber, Hudson Bay, polar bear, population size, sea ice,
survival, Ursus maritimus.

Recent changes in climatic conditions have been associated
with changes in physiology, phenology, and distribution for
a wide range of species (Stenseth et al. 2002, Walther et al.
2002, Parmesan and Yohe 2003). The effects of climatic
change on population size and trend, however, have been
difficult to quantify because the robust estimation of
demographic parameters requires longitudinal individualbased data (Williams et al. 2002, Amstrup et al. 2005a),
which are often expensive and difficult to collect. We used
long-term data for polar bears (Ursus maritimus) collected
under 2 sampling protocols to simultaneously evaluate
survival in relation to climatic conditions and derive
estimates of population size.
Polar bears depend upon sea ice for survival; it is their only
predictable substrate for foraging and is also critical to
movement and some aspects of reproduction (Amstrup
2003). Polar bears are most common on annual sea ice that
occurs over shallow coastal or interisland waters, where
biological productivity is high (Amstrup and DeMaster
1988, Stirling and Lunn 1997, Durner et al. 2004). In many
parts of the Arctic, warming temperatures and altered
atmospheric circulation have resulted in significant changes
to the extent, duration, and character of annual sea ice
(Vinnikov et al. 1999, Lindsay and Zhang 2005, Stroeve et
al. 2005). These changes have raised concerns regarding the
1

E-mail: eregehr@usgs.gov

Regehr et al.

 

Polar Bears in Western Hudson Bay

long-term conservation of polar bears (Stirling 2002,
Derocher et al. 2004, Aars et al. 2006, Regehr et al. 2006).
The Western Hudson Bay (WH) polar bear population
occurs near the southern limit of the species’ range and is
relatively discrete from adjacent populations (Derocher and
Stirling 1990, Stirling et al. 2004). In winter and spring, the
WH population disperses over the ice-covered bay to hunt
seals (Phoca hispida, Erignathus barbatus, and P. vitulina;
Iverson et al. 2006). In summer and autumn when Hudson
Bay is ice-free, the WH population is confined to a restricted
area of the western coast, where polar bears are cut off from
their seal prey and must rely on fat reserves for approximately
4 months until freeze-up. Pregnant females are fooddeprived for 8 months because they must remain on land
in maternal dens, give birth, and nurse until their cubs are
large enough to venture onto the sea ice in the early spring
(Ramsay and Stirling 1986). In the past 50 years, spring air
temperatures in western Hudson Bay have increased by 2–38
C (Skinner et al. 1998, Gagnon and Gough 2005).
Consequently, the sea ice now breaks up approximately 3
weeks earlier than it did 30 years ago (Stirling and Parkinson
2006). This forces the WH population off the sea ice earlier,
shortening the spring foraging period when seals are most
available and reducing polar bears’ ability to accumulate the
fat reserves needed to survive while stranded onshore.
Previous studies have shown a correlation between rising
temperatures, earlier sea ice breakup, and declining recruitment and body condition for polar bears in western Hudson
2673

 Figure 2. Number of individual polar bears captured in western Hudson
Bay, Canada, by the Canadian Wildlife Service (CWS) and the Manitoba
Department of Conservation (MDOC), 1984–2004. Bar heights are
stacked. For example, in 1994 185 individual polar bears were captured:
110 by the CWS, 75 by the MDOC.

Figure 1. Western Hudson Bay, Canada, showing the management area for
the Western Hudson Bay polar bear population and the study area where
the Canadian Wildlife Service captured polar bears during the ice-free
season, 1984–2004.

Bay (Derocher and Stirling 1996, Stirling et al. 1999,
Stirling and Parkinson 2006). Based on forecasts of
continued warming and progressively earlier breakup
(Zhang and Walsh 2006), Stirling and Parkinson (2006)
predicted that conditions will become increasingly difficult
for the WH population. Conversely, in recent years an
increasing number of polar bears have been sighted around
Churchill, Manitoba, Canada, and around Inuit communities along the Nunavut coast of western Hudson Bay. Some
have interpreted the increased sightings as evidence that the
WH population is growing, leading to the implementation
in 2005 of a 19% increase in the annual harvest quota
(Stirling and Parkinson 2006).
We hypothesized that sea ice breakup influences polar bear
survival because of its direct effect on the time available for
foraging, and that survival rates would reflect both interannual
variation in breakup date and the long-term trend toward
earlier breakup. We further hypothesized that declines in
survival contributed to a decline in population size; increased
sightings of polar bears do not reflect a larger population but
rather are the result of nutritionally stressed polar bears
encroaching upon human settlements in search of food.

STUDY AREA
Historically, Hudson Bay has been covered with annual ice
for approximately 8 months of the year and has been ice-free
2674

for 4 months (Jul–Nov; Gough et al. 2004). The Canadian
Wildlife Service (CWS) captured free-ranging polar bears
during the ice-free period, mainly in a 12,000-km2 study
area between the community of Churchill and the Nelson
River (Fig. 1). The coastline of the study area was
characterized by raised beaches surrounded by mud flats,
sedge and grass meadows, and willow (Salix spp.); the
interior was characterized by wet tundra with stands of scrub
spruce (Picea spp.) along lake edges, riparian corridors, and
areas of discontinuous permafrost (Dredge and Nixon
1992).
During spring breakup, counterclockwise marine currents
in Hudson Bay carry the last large ice floes along the
Manitoba and Ontario coastline (Saucier et al. 2004),
resulting in a high degree of fidelity of the WH population
to the CWS study area (Derocher and Stirling 1990, Lunn
et al. 2004, Stirling et al. 2004). Although the management
boundary for the WH population extended approximately
500 km to the north of Churchill, polar bears were seldom
seen along that section of the coast until freeze-up begins in
late autumn (Derocher and Stirling 1990; N. J. Lunn and I.
Stirling, CWS, unpublished data). While on land, polar
bears were greatly concentrated relative to their winter
distribution on the sea ice and were easily sighted against a
snow-free background (Lunn et al. 2002). Adult males
predominated on coastal beaches, whereas single females,
females with dependent young, and subadults generally used
inland areas (Derocher and Stirling 1990, Lunn et al. 2004,
Stirling et al. 2004).

METHODS
From 1984 to 2004, we (the CWS) used helicopters to
capture polar bears each year between late August and early
October (Stirling et al. 1989). Although sample size varied
with project funding (Fig. 2), we attempted to evenly
distribute capture effort over the entire study area in every
The Journal of Wildlife Management

 

71(8)

 year after 1986. We captured all observed polar bears
regardless of sex, age, or reproductive status, with the
exception of some pregnant females that took refuge in dens,
which they had excavated in anticipation of maternal
denning, and a few individuals that entered water or
otherwise could not safely be sedated. We ear-tagged all
captured polar bears with a unique identification number
and applied permanent tattoos to both sides of the inner
surface of the upper lip. We extracted a vestigial premolar
from polar bears .1 year old for age determination (Calvert
and Ramsay 1998). Cubs-of-the-year (COYs; approx. 8
months old in autumn) were always with their mothers and
we could visually age them without error (Ramsay and
Stirling 1988). In our analyses, we excluded data for polar
bears that we located by radiotelemetry and data for a few
previously marked polar bears that we could not individually
identify.
The Manitoba Department of Conservation (MDOC)
captured problem bears in and around Churchill as part of
the community’s Polar Bear Alert Program (Lunn et al.
2002). Beginning in 1984, this program followed explicit
rules regarding when to haze, capture, or kill polar bears that
were considered a threat to human life or property.
Therefore, all polar bears that entered the Churchill vicinity
were treated similarly. The MDOC immobilized polar bears
from the ground and either detained them in a holding
facility or transported them out of town. Marking and aging
procedures were similar to those used by the CWS. The
Animal Care Committee of the CWS (Prairie and Northern
Region) and the University of Alberta BioSciences Animal
Policy and Welfare Committee approved the immobilization and handling protocol for free-ranging polar bears. The
MDOC and Parks Canada issued annual permits under
which we conducted our research.
We estimated the date of spring sea ice breakup in the
WH management area using the methods of Stirling and
Parkinson (2006) and derived the covariate ice for use in
capture–recapture models by standardizing the Julian date of
breakup (Franklin 2001). We used ice rather than a largescale climatic index, such as the Arctic Oscillation, because
ice directly quantified our hypothesized link between
environmental conditions and polar bear survival (Hallett
et al. 2004). We used standard regression techniques to
describe trends in the sex and age composition of the CWS
and MDOC capture samples (e.g., Zar 1996, Hosmer and
Lemeshow 2000). Supplemental materials for the analyses
in this paper are provided in Regehr et al. (2007).
Goodness-of-Fit
We used program RELEASE (Burnham et al. 1987) to
investigate patterns in both the CWS-only and the
combined CWS and MDOC data, and to quantify how
well the standard Cormack–Jolly–Seber (CJS) model fit
various subsets of the data (Regehr et al. 2007). This
allowed us to identify homogenous strata (i.e., groups of
polar bears with similar survival and recapture probabilities)
and to establish a general model that allowed for all
detectable sex- and age-based variation. We estimated the
Regehr et al.

 

Polar Bears in Western Hudson Bay

variance inflation factor (^c ) by applying program RELEASE
to each independent stratum in the global model and
dividing the summed chi-square statistics by the total
degrees of freedom (e.g., Sendor and Simon 2003).
Valid estimation of population size and trend requires a
consistent, well-defined study population (Hines and
Nichols 2002). If the area exposed to capture effort changes
with time, demographic parameters will reflect individuals
added to and lost from the study population as the effective
boundaries change. We captured free-ranging polar bears in
a geographically restricted portion of the study area from
1984 to 1986. From 1987 to 2004, we attempted to search
the entire study area each year. We quantified the
spatiotemporal distribution of capture effort by overlaying
the study area with 25-km2 grid cells and calculating t-like
statistics to determine whether the relative distribution of
capture locations changed for sequential 3-year periods
(Amstrup et al. 2004, 2005b).
Capture–Recapture Analysis
The CJS model conditions on first capture and estimates the
probabilities of survival (/) and recapture (p) that are most
likely to explain the observed capture history data for an
open population (Lebreton et al. 1992). We fit CJS models
using program MARK (Cooch and White 2005) and Rlanguage software for the general regression approach to
capture–recapture (McDonald et al. 2005). The 2 approaches fit identical likelihoods and result in identical
parameter estimates, but the latter approach makes it easier
to construct complex models with time-dependent individual covariates.
Model selection.—We based model selection on
Akaike’s Information Criterion (AIC; Akaike 1981), biological realism, and model interpretability. We corrected
AIC for small sample size (AICc) and used ^c ¼ 1.0 from the
goodness-of-fit analysis (Anderson and Burnham 2002).
When appropriate, we based inference regarding important
hypotheses on the strength of evidence across multiple
models. For pairwise comparisons, we quantified relative
support using DAICc, where DAICc ,2 indicated similar
support for both models and DAICc .10 indicated strong
support for the lower-AICc model (Burnham and Anderson
2002). For each fitted model, we also considered the
magnitude and variance of the estimated parameters. This
was necessary because AIC attempts to optimize the overall
tradeoff between model fit and precision; it does not indicate
how many of a model’s parameters are estimable, nor does it
indicate which of a model’s parameters explain appreciable
variation in the data (Stephens et al. 2005).
Model fitting.—We estimated survival and population
size from the most supported model for the combined CWS
and MDOC data, which we developed in several steps. We
began by modeling the CWS-only data because they
represented a random sample from free-ranging polar bears,
whereas the MDOC data were nonrandom in that polar
bears effectively sampled themselves by coming into
Churchill. This simplified the analysis and allowed us to
perform initial investigations of biological hypotheses using
2675

 a relatively unbiased data set. Starting with the general
model, we evaluated models with restricted parameterizations for p while retaining the general parameterization
for /. To reflect interannual variation in sample size, we
modeled p as time dependent and as a function of the annual
number of helicopter hours flown in capture operations. We
fit models with sex- and age-specificity in p to reflect
potential differences in sighting probability due to the
spatial segregation of polar bears while on land. Throughout
the study, the sex ratio in the CWS capture sample
remained approximately 1:1 despite observations of field
crews that there were fewer large males in the study area
since the late 1990s. We hypothesized that as the number of
male polar bears declined, a larger proportion of them
occupied the limited coastal beaches where they were highly
visible. We modeled this change in the relative p of female
and male polar bears as an additive sex effect, which differed
for the periods 1985–1995 and 1996–2004.
After establishing a parsimonious parameterization for p,
we evaluated models with restricted parameterizations for /.
We hypothesized a sex effect in survival due to sex-selective
harvest (Derocher et al. 1997). Based on the life history of
polar bears, we considered 5 age classes: COYs, juveniles
(COYs and yearlings combined), subadults (2–4 yr), primeadults (5–19 yr), and senescent-adults (!20 yr; Derocher
and Stirling 1992, 1996; Amstrup 2003). We first modeled
/ for prime-adult polar bears while retaining the general
parameterization for other age classes, based on the
hypothesis that interannual variation in survival was lowest
for prime-adults (Bunnell and Tait 1981, Amstrup and
Durner 1995, Eberhardt 2002). We then modeled / for the
other age classes and evaluated support for combining age
classes.
We modeled temporal variation in / using time-constant,
time-dependent, and linear trend parameterizations. We
investigated the relationship between sea ice and survival by
modeling /j (i.e., survival from the autumn of calendar yr j
to the autumn of yr jþ1) as a linear function of ice breakup
date in the spring of year j (the covariate ice). We also
modeled /j using the centered, 3-year running mean of ice
breakup date to investigate whether survival was a
cumulative function of environmental conditions over
several years. Finally, we modeled /j for COYs using
breakup date in the year jÀ1 to investigate whether COY
survival was predominately a function of birth weight,
which is related to maternal body condition at the time of
den entry the preceding year (Derocher and Stirling 1998).
We constructed all CJS models using the logit link function
(Cooch and White 2005). After establishing a parsimonious parameterization for /, we revisited the parameterization for p to ensure that it was still valid (Lebreton et al.
1992).
Once we had identified the most supported model for the
CWS-only data, we extended the CJS analysis to the
combined CWS and MDOC data. Because the CWS often
avoided capturing polar bears in the immediate vicinity of
Churchill, it is possible that some polar bears were
2676

consistently missed by the CWS but were included in the
MDOC sample. Also, the nonrandom sex and age
composition of the MDOC sample suggested that some
strata within the WH population (e.g., subad M) were
concentrated around Churchill at the time of sampling. We
estimated survival and population size from the combined
data to ensure that our analysis represented all possible polar
bears. To minimize potential bias associated with differences
in sampling protocols, we explicitly allowed CJS model
parameters to differ for polar bears that came into Churchill
versus polar bears that remained to the south in the CWS
study area. For both / and p, we modeled capture-history
dependence following capture by the MDOC (e.g., the
individual covariate trap 0 ) using temporary, permanent, and
sex- and age-specific effects (Nichols et al. 1984, Pradel
1993). Age-specific effects in year j were conditional upon
age in year j. After identifying a final model that allowed for
major sources of variation in the combined data, we revisited
the relationship between sea ice and survival and other
important hypotheses.
Survival.—The CJS model produces estimates of total
apparent survival (/), which is the cumulative probability of
remaining alive and available for capture (Lebreton et al.
1992). Because the CJS model right-censors data for
animals that were not released following capture, our
estimates of / do not reflect the mortality contribution of
problem bears that were killed in Churchill. Polar bears
from the WH population were harvested as part of an
annual, regulated hunt by Inuit from the communities of
Arviat, Whale Cove, and Rankin Inlet, located north of
Churchill along the Nunavut coast of western Hudson Bay
(Fig. 1). We estimated natural apparent survival (/N) by
adjusting the CJS estimates of / using tag-return data from
the harvest (Regehr et al. 2007).
Population size.—We estimated population size (N) by
applying a Horvitz–Thompson estimator to recapture
probabilities using the formula
Nˆ j ¼

nj
X
Iij
i¼1

pˆ ij

;

where nj is the total number of polar bears captured at
occasion j, Iij is an indicator variable equal to 1 if the ith
polar bear was captured at occasion j and 0 otherwise, and pˆ ij
is the estimated recapture probability of the ith polar bear at
occasion j (McDonald and Amstrup 2001, Williams et al.
2002). This approach is a generalization of the multiple-age
Jolly–Seber model (Pollock 1981, Pollock et al. 1990) and
has several advantages: 1) it retains the flexibility of the CJS
modeling framework, 2) it is straightforward to estimate the
size of multiple strata within the population, and 3) it is
possible to estimate N from models in which pij is a function
of individual covariates. We estimated the variance of Nˆ j
using the method of Taylor et al. (2002).
For interpretive purposes, we needed to summarize Nˆ j in a
way that reduced the variability (and presumably the bias) of
the point estimates but preserved biologically meaningful
trends. We fit a curve through Nˆ j using a variable span
The Journal of Wildlife Management

 

71(8)

 bivariate smoother (S-PLUS 6.2 function supsmu with
default parameters; Insightful Corporation 2001) and used a
bootstrap procedure to estimate the variance of the
smoothed curve (Manly 1997). We resampled the individual
capture histories and covariates with replacement to
generate 500 new data sets, and fit the most supported
model to each. We then reprocessed the resulting Nˆ j with
the smoother and assumed that the distribution of smoothed
values represented the sampling distribution of true
population size at each occasion j. We also used Nˆ j from
the CJS analysis to derive estimates of mean annual
population growth rate (kj) by fitting a least-squares
regression of ln(Nˆ j) versus year, weighted by 1/(Nˆ j), and
exponentiating the slope coefficient (b) from the regression:
^
kj ¼ eb.

RESULTS
The CWS data consisted of 3,306 captures of 1,963 polar
bears (Fig. 2), including 19 that were not released. The
mean date of capture was 10 September. The MDOC data
consisted of 1,417 captures of 963 polar bears, including 95
that were not released (mostly killed as problem bears). The
mean date of capture was 31 October. From 1984 to 1990,
the number of polar bears captured by the MDOC was
variable (annual x¯ ¼ 51; SD ¼ 22) but did not exhibit a linear
^ ¼ À1.71 bears/yr, SE(
^ ¼ 4.41, P ¼ 0.71). From
c b)
trend (b
1991 to 2004, the number of polar bears captured by the
^ ¼ 4.7 bears/yr, SE(
^ ¼ 1.66, P ¼
c b)
MDOC increased (b
0.02).
From 1984 to 2003, 792 polar bears were harvested along
the Nunavut coast of western Hudson Bay. Most of the
harvest occurred toward the end of the ice-free season, as
polar bears left the CWS study area and moved northward
along the coastline in anticipation of freeze-up (median date
of harvest was 5 Nov). The number of polar bears in the
harvest that carried research marks deployed by the CWS or
MDOC was variable (annual x¯ ¼ 22; SD ¼ 7) but did not
^ ¼ 0.14 bears/yr, SE(
^ ¼ 0.29, P ¼
c b)
exhibit a linear trend (b
0.64). The proportions of marked polar bears in the CWS,
MDOC, and harvest samples varied with time (5-yr period
model vs. time-constant model; G ¼ 111.6, df ¼ 3, P ,
0.001) but were consistently similar for the 3 samples
(sample 3 5-yr period model vs. 5-yr period model; G ¼
14.2, df ¼ 8, P ¼ 0.08). Overall, the marked proportions in
the CWS, MDOC, and harvest samples were 0.59 (SE ¼
0.01), 0.59 (SE ¼ 0.01), and 0.55 (SE ¼ 0.02), respectively.
The sex and age composition of the samples is described in
Regehr et al. (2007).
The covariate ice was based on breakup data for 1984–
2003 only (Fig. 3d), because CJS models do not estimate
survival following the final sampling occasion. Using linear
regression, the mean sea ice breakup date in 2003 (21 Jun)
^ ¼À1.03 d/
occurred 19.5 days earlier than in 1984 (12 Jul; b
^ ¼ 0.47, P ¼ 0.04). The trend toward earlier
c b)
yr, SE(
breakup during the current study was consistent with the
entire available time series (linear regression from 1971 to
^ ¼À0.59 d/yr, SE(
^ ¼ 0.19, P , 0.001). From 1984
c b)
2004; b
Regehr et al.

 

Polar Bears in Western Hudson Bay

to 2003, there was not a statistically significant change in the
date of freeze-up in western Hudson Bay (Gagnon and
Gough 2005).
Goodness-of-Fit
Program RELEASE indicated that future recapture rates
were consistently lower for first-time captures than for
recaptures, which suggests lower survival of young polar
bears (Cooch and White 2005). We also found evidence for
positive capture-history dependence (i.e., increased p
following capture) among young males in the combined
data (Pradel 1993). The standard CJS model adequately fit
the CWS-only and the combined data after partitioning by
sex and 2 age classes: prime-adults versus all other ages
(Regehr et al. 2007). We began the CJS analysis with a
model based on these 4 independent strata, which we
further generalized to allow for an additive effect in /
between juveniles and an aggregate age class of subadults
and senescent-adults. We estimated ^c ¼ 0.72 (v2 ¼ 197.8, df
¼ 274) for the CWS-only data and 0.81 (v2 ¼ 251.5, df ¼
311) for the combined data, and we used ^c ¼ 1.0 for all
modeling because the goodness-of-fit analysis did not detect
any overdispersion or un-modeled heterogeneity in the WH
data after partitioning by sex and age (Burnham and
Anderson 2002).
The spatiotemporal analysis of capture locations suggested
that we evenly distributed capture effort throughout the
CWS study area from 1987 to 2004. Prior to 1987, we were
more likely to capture polar bears in the northern part of the
study area and along the coast. This was corroborated by the
proportion of adult males in the CWS capture sample,
which decreased from 0.37 (1984–1986) to 0.25 (1987–
1989) as capture operations expanded inland to include
more of the habitat typically used by females and family
groups. In the last years of the study, more polar bears were
again captured along the coast. However, Global Positioning System flight logs (available 2001–2004) indicated that
we searched the entire study area in these years and that the
shift in capture locations reflected a change in polar bear
distribution.
Survival
We estimated survival and population size from the most
supported model for the combined CWS and MDOC data.
That model included 7 beta coefficients for / and 24 for p,
all of which were individually identifiable (Regehr et al.
2007).
Total apparent survival probability (^
uj) was time-constant
for prime-adult polar bears, at 0.93 (95% CI ¼ 0.91–0.94)
for females and 0.90 (95% CI ¼ 0.88–0.91) for males.
Survival of juvenile, subadult, and senescent-adult polar
bears varied from year to year as a function of sea ice
^ice ¼ 0.23, 95% CI ¼ 0.07–0.38; Fig. 3). For
breakup date (b
these age classes, survival decreased by 2–5% for each week
earlier than average that the sea ice broke up. We derived
^ j using the average breakup date for
estimates of average u
1984–2003 (Table 1), where the associated confidence
intervals represent sampling variance only, not the inter2677

 Figure 3. Total apparent survival and 95% confidence intervals for (a) juvenile, (b) subadult, and (c) senescent-adult polar bears in western Hudson Bay,
Canada, estimated from the most supported model fit to capture–recapture data collected by the Canadian Wildlife Service and the Manitoba Department of
Conservation (MDOC), 1984–2004. Reduced survival rates for subadults and senescent-adults captured around Churchill by the MDOC are not shown. (d)
Timing of sea ice breakup in the Western Hudson Bay polar bear management area, which was the best predictor of survival for juvenile, subadult, and
senescent-adult polar bears.

Table 1. Average total apparent survival and 95% confidence intervals for
polar bears in western Hudson Bay, Canada, estimated from the most
supported model fit to capture–recapture data collected by the Canadian
Wildlife Service and the Manitoba Department of Conservation (MDOC),
1984–2004.
F

M

Age class

Survival

95% CI

Survival

95% CI

Juv (0–1 yr)
Subad (2–4 yr)
Subad (2–4 yr)a
Prime-ad (5–19 yr)
Senescent-ad (!20 yr)
Senescent-ad (!20 yr)a

0.70
0.86
0.78
0.93
0.81
0.72

0.65–0.74
0.82–0.89
0.74–0.82
0.91–0.94
0.77–0.84
0.66–0.78

0.62
0.81
0.72
0.90
0.75
0.65

0.57–0.66
0.76–0.84
0.68–0.76
0.88–0.91
0.70–0.79
0.58–0.71

a

Reduced survival rates for subad and senescent-ad captured around
Churchill by the MDOC.
2678

annual variation in survival due to changing environmental
conditions. For subadult and senescent-adult polar bears,
capture around Churchill was associated with a permanent
^trap 0 ¼ À0.49,
decrease of 8–14% in both total survival (b
95% CI ¼ À0.21 to À0.77) and natural survival (Table 2).
Population Size
Recapture probabilities (p) were time dependent and
differed for females and males by an additive sex effect.
The magnitude of the sex effect differed for the periods
1985–1995 and 1996–2004. Polar bears that were captured
around Churchill were more likely to be recaptured on
subsequent occasions, conditional upon survival. This effect
was stronger for juveniles and subadults (odds ratio ¼ 2.68,
95% CI ¼ 2.19–3.28) than for prime-adults and senescentadults (odds ratio ¼ 2.01, 95% CI ¼ 1.73–2.33).
The Journal of Wildlife Management

 

71(8)

 Table 2. Average natural apparent survival for polar bears in western
Hudson Bay, Canada, estimated by adjusting total apparent survival rates
from the most supported model fit to capture–recapture data collected by
the Canadian Wildlife Service and the Manitoba Department of
Conservation (MDOC), 1984–2004, using tag-return data from the
harvest.
Age class

F

M

Juv (0–1 yr)
Subad (2–4 yr)
Subad (2–4 yr)a
Prime-ad (5–19 yr)
Senescent-ad (!20 yr)
Senescent-ad (!20 yr)a

0.73
0.92
0.82
0.93
0.82
0.72

0.71
0.94
0.78
0.94
0.82
0.68

a

Reduced survival rates for subad and senescent-ad captured around
Churchill by the MDOC.

We derived final estimates of population size by fitting a
smoothed curve to the 1987–2004 point estimates of
population size (Nˆ j ; Fig. 4). We excluded Nˆ j for 1985 and
1986 because incomplete sampling of the CWS study area
resulted in a large negative bias for those years. The WH
polar bear population declined by approximately 22% from
1,194 (95% CI ¼ 1,020–1,368) in 1987 to 935 (95% CI ¼
794–1,076) in 2004. For this period, the mean annual
population growth rate (kj) was 0.986 (95% CI ¼ 0.978–
0.995), which indicates that the probability of a stable or
increasing population (kj ! 1) was negligible (P ¼ 0.002).
Ancillary estimates of kj derived from CJS models other
than the most supported model indicated that the trend in
population size was insensitive to model selection.

DISCUSSION
Western Hudson Bay provides a unique opportunity to
quantify the demographic response of a polar bear
population to changing environmental conditions, both
because of the availability of capture data for both freeranging polar bears and problem bears that came into
Churchill, and because of the long duration and intensity of
the study (approx. 80% of ad bore research marks). Several
lines of evidence indicate that the relationship between early
breakup and decreased survival was a biological phenomenon, common to all WH polar bears. First, although more
polar bears generally came into Churchill in years of early
^ ¼À0.87 bears/d, SE(
^ ¼ 0.49, P ¼ 0.09), direct
c b)
breakup (b
human-caused mortality among these bears had no impact
on survival estimates. Second, the most supported model
allowed for a decrease in survival for subadults and
senescent-adults captured around Churchill, in addition to
fluctuations in survival associated with breakup date. This
decrease was likely due to both an increased probability of 1)
natural mortality, because polar bears that came into town in
search of food were more likely to be in poor nutritional
condition, and 2) anthropogenic mortality, because some
polar bears were relocated to the north where they were
more likely to encounter Inuit hunters. Third, there was not
a significant correlation between breakup date and the
number of marked juvenile, subadult, and senescent-adult
^ ¼À0.10 bears/d, SE(
^ ¼
c b)
polar bears killed in the harvest (b
Regehr et al.

 

Polar Bears in Western Hudson Bay

Figure 4. Population size and 95% confidence intervals for polar bears in
western Hudson Bay, Canada, estimated from the most supported model fit
to capture–recapture data collected by the Canadian Wildlife Service and
the Manitoba Department of Conservation, 1984–2004. Smoothed values
and bootstrapped 95% confidence intervals are shown for the 1987–2004
point estimates of population size only, because the 1985–1986 point
estimates were biased by incomplete sampling of the study area and are not
valid.

0.08, P ¼ 0.21). Finally, CJS models fit to a reduced data set
that excluded individual capture histories for polar bears that
were either captured or killed by the MDOC, or killed in
the Nunavut harvest, corroborated support for breakup date
as the best predictor of juvenile, subadult, and senescentadult survival. A model analogous to the most supported
model for the combined CWS and MDOC data estimated
^ice ¼ 0.39 (95% CI ¼ 0.14–0.65; likelihood ratio test vs. a
b
null model excluding the covariate ice: v2 ¼ 9.37, df ¼ 1, P ¼
0.002).
The breakup–survival relationship likely reflects increased
starvation associated with entering the winter in poor
nutritional condition. Previous studies have suggested that
cub survival in the first year of life is related to maternal
body condition at the time of birth (Ramsay and Stirling
1988; Derocher and Stirling 1996, 1998). We found that
COY survival beyond the age of 8 months (when we first
captured them) was predominately a function of environmental conditions in the current year. As the study
progressed, the proportion of females in the CWS capture
sample remained stable (approx. 0.52) despite an increasing
proportion of females in the WH population (to 0.65 in
2004). It is likely that declines in juvenile survival shortened
the average inter-birth interval; females produced cubs, lost
them, and bred again instead of keeping cubs with them for
.2 years until weaned. This pattern, also observed in the
southern Beaufort Sea (Regehr et al. 2006), meant that adult
females may have been progressively more likely to be
pregnant and in dens where they were largely unavailable for
capture. Such reproductively based temporary emigration
should not have introduced a consistent bias into estimates
of female population size because recapture probabilities
were time dependent and the most supported model allowed
2679

 for a change in female p relative to male p toward the end of
the study (Kendall et al. 1997).
Survival of prime-adult polar bears was stable, most likely
because prime-adults were in better body condition than
other polar bears (E. Richardson, CWS, unpublished data),
able to divert resources from reproduction to survival in
times of nutritional stress (Bunnell and Tait 1981,
Eberhardt 2002), better at catching seals, and more able to
take seal kills away from subordinate polar bears. Natural
survival of senescent-adult polar bears (!20 yr), however,
was 12–25% lower than for prime-adults and was sensitive
to interannual variation in environmental conditions.
Although variation in survival has been shown for some
animals with long life spans and delayed maturation (e.g.,
Eberhardt 1985, Jorgenson et al. 1997, Cameron and Siniff
2004), our study is one of the few to have investigated
senescence in survival in a statistically rigorous way (Gaillard
et al. 1994, Pistorius and Bester 2002).
Evidence from capture–recapture, radiotelemetry, and tagreturn studies suggests that exchange between the WH
polar bear population and adjacent populations was low
(Derocher and Stirling 1990; Stirling et al. 1999, 2004).
Thus, it is likely that total apparent survival (/) in this study
closely approximated true survival, although the rightcensoring of capture histories for polar bears killed in
Churchill may have introduced a slight positive bias into our
estimates. For prime-adult polar bears, our estimates of /
were approximately 5% higher than the estimates reported
by Lunn et al. (1997) for all WH polar bears age !1 year.
Our estimates of natural apparent survival (/N) were similar
or slightly lower than those reported for adult polar bears in
other populations (Larsen 1986; Taylor et al. 2005, 2006)
and were at the lower limit assumed necessary to maintain
populations of long-lived mammals (Taylor et al. 1987,
Eberhardt 2002). The similarity between female and male
/N for most age classes supports the hypothesis that
observed sexual differences in polar bear survival are due
to sex-selective harvest (Ramsay and Stirling 1986, Derocher et al. 1997).
Our estimates of population size were robust to minor, unmodeled heterogeneity in recapture probabilities because
mean p was relatively high (approx. 0.25; e.g., Pollock et al.
1990, Pledger and Efford 1998). The mean percent relative
difference between the point estimates of N from the most
supported model and the bootstrapped estimates of N was
,1%, suggesting that small-sample bias was minimal.
Furthermore, our modeling approach explicitly allowed for
heterogeneity associated with differences between the CWS
and MDOC sampling protocols, by incorporating timedependent individual covariates for capture-history dependence. The increase in p following capture around Churchill
was likely due to recidivist behavior with respect to the
potential food reward at the Churchill dump (Lunn and
Stirling 1985) and to the fact that polar bears with fidelity to
the northern part of the CWS study area were potentially
exposed to sampling by both the CWS and MDOC.
Our point estimate of the size of the WH population in
2680

1995 was 1,236 (SE ¼ 190), similar to the previous estimate
of 1,199 (SE ¼ 251; Lunn et al. 1997). Because polar bears
are K-selected and populations cannot fluctuate radically
among years, we fit a curve through the point estimates
using a relatively insensitive smoothing algorithm, which
resulted in a lower mean coefficient of variation (9%)
compared to the point estimates (12%). We assessed the
contribution of the MDOC data to estimates of population
size by deriving secondary estimates from the CWS-only
data. Whereas estimates of adult population size were
similar for the 2 data sets (x¯ % relative difference approx.
2%), estimates of total population size were larger for the
combined data. This difference increased from 73 polar
bears in 1987 to 173 polar bears in 2004 (i.e., in 2004 the
smoothed estimate of total population size was 762 for the
CWS-only data vs. 935 for the combined data). Apparently,
the CWS missed an increasing number of young polar bears
as more subadults came into Churchill each year.
We attribute the decline of the WH polar bear population
to increased natural mortality associated with earlier sea ice
breakup and to the continued harvest of approximately 40
polar bears per year (Lunn et al. 2002), which at some point
ceased to be sustainable. We found no support for
alternative explanations. Long-term observations suggest
that the WH population continues to exhibit a high degree
of fidelity to the CWS study area during the early part of the
ice-free season (Stirling et al. 1977, 1999, 2004; Taylor and
Lee 1995), which precludes permanent emigration as a cause
for the decline. Consistently similar proportions of researchmarked polar bears in the CWS, MDOC, and harvest
samples suggest that the same polar bears were first exposed
to capture around the CWS study area and were
subsequently exposed to harvest. This refutes the hypothesis
that large-scale changes in the onshore distribution of polar
bears in western Hudson Bay have created a separate
subpopulation that spends the entire ice-free season along
the Nunavut coast. Finally, capture–recapture data collected
from 2003 to 2005 for the adjacent southern Hudson Bay
population included very few polar bears with a previous
capture in the WH management area, which suggests that a
major distributional shift to the southeast did not occur (M.
E. Obbard, Ontario Ministry of Natural Resources,
unpublished data).
Polar bears in western Hudson Bay are likely to be among
the first to show population-level effects of climatic
warming because they occur near the southern limit of the
species’ range (Lunn et al. 2002) and because recent
temperature increases in the region have been among the
largest in the Arctic (Ferguson et al. 2005). We propose the
following explanation for the concurrent observations of an
increasing number of polar bears coming into Churchill and
a declining WH population. Sea ice breakup in western
Hudson Bay has been occurring progressively earlier due to
rising air temperatures (Gagnon and Gough 2005). This
shortens the time that polar bears can hunt seals on the sea
ice, thereby forcing them ashore in poorer nutritional
condition (Stirling et al. 1999) and increasing their risk of
The Journal of Wildlife Management

 

71(8)

 starvation. As polar bears exhaust their fat reserves toward
the end of the ice-free period, they are more likely to
encroach upon human settlements in search of alternative
food sources to sustain themselves until freeze-up (Stirling
and Parkinson 2006).

MANAGEMENT IMPLICATIONS
Although climatic warming may initially improve conditions
for polar bears in high-latitude regions of heavy ice
(Derocher et al. 2004), forecasted declines in the sea ice
for most parts of the Arctic are long term and severe
(Parkinson 2000, Comiso 2003, Arctic Climate Impact
Assessment 2004, Holland et al. 2006). Ultimately, we
predict that more northerly polar bear populations will
experience declines in demographic parameters similar to
those observed in western Hudson Bay, along with changes
in distribution and other currently unknown ecological
responses (Derocher et al. 2004, Aars et al. 2006). If sea ice
loss becomes the limiting factor determining population
growth rate, reduced harvest (e.g., Taylor et al. 2006) may
be insufficient to slow population declines or impractical due
to increasing human–polar bear conflicts. Innovative and
flexible management will be necessary to mitigate the
negative effects of climatic warming on polar bears in ways
that are socially and economically acceptable.

ACKNOWLEDGMENTS
We are grateful to the CWS, the MDOC, and the Natural
Sciences and Engineering Research Council for their longterm support of polar bear studies in western Hudson Bay,
and to the United States Geological Survey and University
of Wyoming for supporting analyses of data resulting from
these studies. The Churchill Northern Studies Centre, the
Discovery Channel, the National Fish and Wildlife
Foundation (Washington, D.C.), Nunavut Wildlife Research Trust, Ontario Ministry of Natural Resources, Parks
Canada, Sea World, the World Society for the Protection of
Animals, World Wildlife Fund (Canada), and the World
Wildlife Fund Arctic Programme provided financial and
logistical support. We thank the MDOC, the Nunavut
Department of Environment, and the Ontario Ministry of
Natural Resources for access to unpublished data. We thank
D. Andriashek, W. Calvert, A. E. Derocher, M. Kay, S.
Miller, M. A. Ramsay, E. Richardson, and C. Spencer for
assistance in the field and laboratory. T. McDonald at
WEST, Inc. (Cheyenne, WY) provided advice with the data
analysis.

LITERATURE CITED
Aars, J., N. J. Lunn, and A. E. Derocher. 2006. Polar bears. Proceedings of
the 14th Working Meeting of the World Conservation Union Species
Survival Commission (IUCN/SSC) Polar Bear Specialists Group, 20–24
June 2005, Seattle, Washington, USA. International Union for
Conservation of Nature and Natural Resources Species Survival
Commission Occasional Paper 32, Gland, Switzerland.
Akaike, H. 1981. Likelihood of a model and information criteria. Journal of
Econometrics 16:3–14.
Amstrup, S. C. 2003. Polar bear. Pages 587–610 in G. A. Feldhammer, B.
Regehr et al.

 

Polar Bears in Western Hudson Bay

C. Thompson, and J. A. Chapman, editors. Wild mammals of North
America: biology, management, and conservation. Second edition. Johns
Hopkins University Press, Baltimore, Maryland, USA.
Amstrup, S. C., and D. P. DeMaster. 1988. Polar bear—Ursus maritimus.
Pages 39–56 in J. W. Lentfer, editor. Selected marine mammals of
Alaska: species accounts with research and management recommendations. Marine Mammal Commission, Washington, D.C., USA.
Amstrup, S. C., and G. M. Durner. 1995. Survival rates of radio-collared
female polar bears and their dependent young. Canadian Journal of
Zoology 73:1312–1322.
Amstrup, S. C., G. M. Durner, I. Stirling, and T. L. McDonald. 2005a.
Allocating harvests among polar bear stocks in the Beaufort Sea. Arctic
58:247–259.
Amstrup, S. C., T. L. McDonald, and G. M. Durner. 2004. Using satellite
radiotelemetry data to delineate and manage wildlife populations.
Wildlife Society Bulletin 32:661–679.
Amstrup, S. C., T. L. McDonald, and B. F. J. Manly, editors. 2005b.
Handbook of capture–recapture analysis. Princeton University Press,
Princeton, New Jersey, USA.
Anderson, D. R., and K. P. Burnham. 2002. Avoiding pitfalls when using
information-theoretic methods. Journal of Wildlife Management 66:
912–918.
Arctic Climate Impact Assessment. 2004. Impacts of a warming Arctic.
Cambridge University Press, Cambridge, United Kingdom.
Bunnell, F. L., and D. E. N. Tait. 1981. Population dynamics of bears:
implications. Pages 75–98 in T. D. Smith and C. Fowler, editors.
Dynamics of large mammal populations. John Wiley and Sons, New
York, New York, USA.
Burnham, K. P., and D. R. Anderson. 2002. Model selection and
multimodel inference. Second edition. Springer-Verlag, New York,
New York, USA.
Burnham, K. P., D. R. Anderson, G. C. White, C. Brownie, and K. H.
Pollock. 1987. Design and analysis methods for fish survival experiments
based on release–recapture. American Fisheries Society Monograph 5,
Bethesda, Maryland, USA.
Calvert, W., and M. A. Ramsay. 1998. Evaluation of age determination of
polar bears by counts of cementum growth layer groups. Ursus 10:449–
453.
Cameron, M. F., and D. B. Siniff. 2004. Age-specific survival, abundance,
and immigration rates of a Weddell seal (Leptonychotes weddellii)
population in McMurdo Sound, Antarctica. Canadian Journal of Zoology
80:601–615.
Comiso, J. C. 2003. Warming trends in the Arctic from clear sky satellite
observation. Journal of Climate 16:3498–3510.
Cooch, E., and G. C. White. 2005. Program MARK: a gentle introduction.
Fifth edition. ,http://www.phidot.org/software/mark/docs/book/.. Accessed 22 Aug 2006.
Derocher, A. E., N. J. Lunn, and I. Stirling. 2004. Polar bears in a warming
climate. Integrative and Comparative Biology 44:163–176.
Derocher, A. E., and I. Stirling. 1990. Distribution of polar bears (Ursus
maritimus) during the ice-free period in western Hudson Bay. Canadian
Journal of Zoology 68:1395–1403.
Derocher, A. E., and I. Stirling. 1992. The population dynamics of polar
bears in western Hudson Bay. Pages 1150–1159 in D. R. McCullough
and R. H. Barrett, editors. Wildlife 2001: populations. Elsevier Applied
Science, London, United Kingdom.
Derocher, A. E., and I. Stirling. 1996. Aspects of survival in juvenile polar
bears. Canadian Journal of Zoology 74:1246–1252.
Derocher, A. E., and I. Stirling. 1998. Maternal investment and factors
affecting offspring size in polar bears (Ursus maritimus). Journal of
Zoology (London) 245:253–260.
Derocher, A. E., I. Stirling, and W. Calvert. 1997. Male-biased harvesting
of polar bears in western Hudson Bay. Journal of Wildlife Management
61:1075–1082.
Dredge, L. A., and F. M. Nixon. 1992. Glacial and environmental geology
of northeastern Manitoba. Geological Survey of Canada Memoir 432,
Ottawa, Ontario, Canada.
Durner, G. M., S. C. Amstrup, R. Neilson, and T. McDonald. 2004. The
use of sea ice habitat by female polar bears in the Beaufort Sea. Minerals
Management Service Outer Continental Shelf Study MMS 2004–014,
Anchorage, Alaska, USA.
2681

 Eberhardt, L. L. 1985. Assessing the dynamics of wild populations. Journal
of Wildlife Management 49:997–1012.
Eberhardt, L. L. 2002. A paradigm for population analysis of long-lived
vertebrates. Ecology 83:2841–2854.
Ferguson, S. H., I. Stirling, and P. McLoughlin. 2005. Climate change and
ringed seal (Phoca hispida) recruitment in Western Hudson Bay. Marine
Mammal Science 21:121–135.
Franklin, A. B. 2001. Exploring ecological relationships in survival and
estimating rates of population change using program MARK. Pages 290–
296 in R. Field, R. J. Warren, H. Okarma, and P. R. Sievert, editors.
Wildlife, land, and people: priorities for the 21st century. Proceedings of
the Second International Wildlife Management Congress, The Wildlife
Society, 28 June–2 July 1999, Bethesda, Maryland, USA.
Gagnon, A. S., and W. A. Gough. 2005. Trends in the dates of ice freezeup and breakup over Hudson Bay, Canada. Arctic 58:370–382.
Gaillard, J.-M., D. Allaine, D. Pontier, N. G. Yoccoz, and D. E. L.
Promislow. 1994. Senescence in natural populations of mammals: a reanalysis. Evolution 48:509–516.
Gough, W. A., A. R. Cornwell, and L. J. S. Tsuji. 2004. Trends in seasonal
sea ice duration in Southwestern Hudson Bay. Arctic 57:299–305.
Hallett, T. B., T. Coulson, J. G. Pilkington, T. H. Clutton-Brock, J. M.
Pemberton, and B. T. Grenfell. 2004. Why large-scale indices seem to
predict ecological processes better than local weather. Nature 430:71–75.
Hines, J. E., and J. D. Nichols. 2002. Investigations of potential bias in the
estimation of lambda using Pradel’s (1996) model for capture–recapture
data. Journal of Applied Statistics 29:573–587.
Holland, M. M., C. M. Bitz, and B. Tremblay. 2006. Future abrupt
reductions in the summer Arctic sea ice. Geophysical Research Letters
33:L23503.
Hosmer, D. W., and S. Lemeshow. 2000. Applied logistic regression. John
Wiley and Sons, New York, New York, USA.
Insightful Corporation. 2001. S-PLUS 6 for Windows guide to statistics.
Volume 1. Insightful Corporation, Seattle, Washington, USA.
Iverson, S. J., I. Stirling, and S. L. C. Lang. 2006. Spatial and temporal
variation in the diets of polar bears across the Canadian Arctic: indicators
of changes in prey populations and environment. Pages 98–117 in I. L.
Boyd, S. Wanless, and C. J. Camphuysen, editors. Top predators of
marine ecosystems. Cambridge University Press, Cambridge, United
Kingdom.
Jorgenson, J. T., M. Festa-Bianchet, J.-M. Gaillard, and W. D. Wishart.
1997. Effects of age, sex, disease, and density on survival of bighorn
sheep. Ecology 78:1019–1032.
Kendall, W. L., J. D. Nichols, and J. E. Hines. 1997. Estimating temporary
emigration using capture–recapture data with Pollock’s robust design.
Ecology 78:563–578.
Larsen, T. 1986. Population biology of the polar bear (Ursus maritimus) in
the Svalbard area. Norsk Polarinstitutt Skrifter 184, Oslo, Norway.
Lebreton, J.-D., K. P. Burnham, J. Clobert, and D. R. Anderson. 1992.
Modeling survival and testing biological hypotheses using marked
animals: a unified approach with case studies. Ecological Monographs
62:67–118.
Lindsay, R. W., and J. Zhang. 2005. The thinning of Arctic sea ice, 1988–
2003: have we passed a tipping point? Journal of Climate 18:4879–4894.
Lunn, N. J., S. Schliebe, and E. W. Born, editors. 2002. Polar bears.
Proceedings of the Thirteenth Working Meeting of the IUCN/SSC
Polar Bear Specialist Group, 23–28 June 2001, Nuuk, Greenland.
International Union for Conservation of Nature and Natural Resources
Species Survival Commission Occasional Paper 26, Gland, Switzerland.
Lunn, N. J., and I. Stirling. 1985. The significance of supplemental food to
polar bears during the ice-free period of Hudson Bay. Canadian Journal
of Zoology 63:2291–2297.
Lunn, N. J., I. Stirling, D. Andriashek, and G. B. Kolenosky. 1997. Reestimating the size of the polar bear population in Western Hudson Bay.
Arctic 50:234–240.
Lunn, N. J., I. Stirling, D. Andriashek, and E. Richardson. 2004. Selection
of maternity dens by female polar bears in western Hudson Bay, Canada
and the effects of human disturbance. Polar Biology 27:350–356.
Manly, B. F. 1997. Randomization, bootstrap, and Monte Carlo methods
in biology. Second edition. Chapman and Hall, London, United
Kingdom.
McDonald, T. L., and S. C. Amstrup. 2001. Estimation of population size
2682

using open capture–recapture models. Journal of Agricultural, Biological,
and Environmental Statistics 6:206–220.
McDonald, T. L., S. C. Amstrup, E. V. Regehr, and B. F. J. Manly. 2005.
Examples. Pages 196–265 in S. C. Amstrup, T. L. McDonald, and B. F.
J. Manly, editors. Handbook of capture–recapture analysis. Princeton
University Press, Princeton, New Jersey, USA.
Nichols, J. D., J. E. Hines, and K. H. Pollock. 1984. Effects of permanent
trap response in capture probability on Jolly–Seber capture–recapture
model estimates. Journal of Wildlife Management 48:289–294.
Parkinson, C. L. 2000. Variability of Arctic sea ice: the view from space, an
18-year record. Arctic 53:341–358.
Parmesan, C., and G. Yohe. 2003. A globally coherent fingerprint of
climate change impacts across natural systems. Nature 421:37–42.
Pistorius, P. A., and M. N. Bester. 2002. A longitudinal study of senescence
in a pinniped. Canadian Journal of Zoology 80:395–401.
Pledger, S., and M. Efford. 1998. Correction of bias due to heterogeneous
capture probability in capture–recapture studies of open populations.
Biometrics 54:888–898.
Pollock, K. H. 1981. Capture–recapture models allowing for age-dependent
survival and capture rates. Biometrics 37:521–529.
Pollock, K. H., J. D. Nichols, J. E. Hines, and C. Brownie. 1990. Statistical
inference for capture–recapture experiments. Wildlife Monographs 107.
Pradel, R. 1993. Flexibility in survival analysis from recapture data:
handling trap-dependence. Pages 29–37 in J.-D. Lebreton and J. D.
Nichols, editors. Marked individuals in the study of bird populations.
Birkhauser Verlag, Basel, Switzerland.
Ramsay, M. A., and I. Stirling. 1986. On the mating system of polar bears.
Canadian Journal of Zoology 64:2142–2151.
Ramsay, M. A., and I. Stirling. 1988. Reproductive biology and ecology of
female polar bears (Ursus maritimus). Journal of Zoology (London) 214:
601–634.
Regehr, E. V., S. C. Amstrup, and I. Stirling. 2006. Polar bear population
status in the southern Beaufort Sea. U.S. Geological Survey Open-File
Report 2006–1337, Reston, Virginia, USA.
Regehr, E. V., N J. Lunn, S. C. Amstrup, and I. Stirling. 2007.
Supplemental materials for the analysis of capture-recapture data for polar
bears in western Hudson Bay, Canada, 1984–2004. U.S. Geological
Survey Data Series 304, Reston, Virginia, USA.
Saucier, F. J., S. Senneville, S. Prinsenberg, F. Roy, G. Smith, P. Gachon,
D. Caya, and R. Laprise. 2004. Modelling the sea ice–ocean seasonal
cycle in Hudson Bay, Foxe Basin and Hudson Strait, Canada. Climate
Dynamics 23:303–326.
Sendor, T., and M. Simon. 2003. Population dynamics of the pipistrelle
bat: effects of sex, age and winter weather on seasonal survival. Journal of
Animal Ecology 72:308–320.
Skinner, W. R., R. L. Jefferies, T. J. Carleton, R. F. Rockwell, and K. F.
Abraham. 1998. Prediction of reproductive success and failure in lesser
snow geese based on early season climatic variables. Global Change
Biology 4:3–16.
Stenseth, N. C., A. Mysterud, G. Ottersen, J. W. Hurrell, K.-S. Chan, and
M. Lima. 2002. Ecological effects of climate fluctuations. Science 297:
1292–1296.
Stephens, P. A., S. S. Buskirk, G. D. Hayward, and C. Martinez del Rio.
2005. Information theory and hypothesis testing: a call for pluralism.
Journal of Applied Ecology 42:4–12.
Stirling, I. 2002. Polar bears and seals in the eastern Beaufort Sea and
Amundsen Gulf: a synthesis of population trends and ecological
relationships over three decades. Arctic 55:59–76.
Stirling, I., C. Jonkel, P. Smith, R. Robertson, and D. Cross. 1977. The
ecology of the polar bear (Ursus maritimus) along the western coast of
Hudson Bay. Canadian Wildlife Service Occasional Paper 33, Edmonton, Alberta, Canada.
Stirling, I., and N. J. Lunn. 1997. Environmental fluctuations in arctic
marine ecosystems as reflected by variability in reproduction of polar bears
and ringed seals. Pages 167–181 in S. J. Woodin and M. Marquiss,
editors. Ecology of arctic environments. Blackwell Scientific, Oxford,
United Kingdom.
Stirling, I., N. J. Lunn, and J. Iacozza. 1999. Long-term trends in the
population ecology of polar bears in western Hudson Bay in relation to
climatic change. Arctic 52:294–306.
Stirling, I., N. J. Lunn, J. Iacozza, C. Elliott, and M. Obbard. 2004. Polar
bear distribution and abundance on the southwestern Hudson Bay coast
The Journal of Wildlife Management

 

71(8)

 during open water season, in relation to population trends and annual ice
patterns. Arctic 57:15–26.
Stirling, I., and C. L. Parkinson. 2006. Possible effects of climate warming
on selected populations of polar bears (Ursus maritimus) in the Canadian
Arctic. Arctic 59:261–275.
Stirling, I., C. Spencer, and D. Andriashek. 1989. Immobilization of polar
bears (Ursus maritimus) with Telazolt in the Canadian Arctic. Journal of
Wildlife Diseases 25:159–168.
Stroeve, J. C., M. C. Serreze, F. Fetterer, T. Arbetter, W. Meier, J.
Maslanik, and K. Knowles. 2005. Tracking the Arctic’s shrinking ice
cover: another extreme September minimum in 2004. Geophysical
Research Letters 32:L04501.
Taylor, M., D. P. DeMaster, F. L. Bunnell, and R. E. Schweinsburg. 1987.
Modeling the sustainable harvest of female polar bears. Journal of
Wildlife Management 51:811–820.
Taylor, M. K., J. Laake, H. D. Cluff, M. Ramsay, and F. Messier. 2002.
Managing the risk from hunting for the Viscount Melville Sound polar
bear population. Ursus 13:185–202.
Taylor, M. K., J. Laake, P. D. McLoughlin, E. W. Born, H. D. Cluff, S. H.
Ferguson, A. Rosing-Asvid, R. Schweinsburg, and F. Messier. 2005.
Demography and viability of a hunted population of polar bears. Arctic
58:203–214.
Taylor, M. K., J. Laake, P. D. McLoughlin, and H. D. Cluff. 2006.

Regehr et al.

 

Polar Bears in Western Hudson Bay

Demographic parameters and harvest-explicit population viability analysis
for polar bears in M’Clintock Channel, Nunavut, Canada. Journal of
Wildlife Management 70:1667–1673.
Taylor, M. K., and J. Lee. 1995. Distribution and abundance of Canadian
polar bear populations: a management perspective. Arctic 48:147–154.
Vinnikov, K. Y., A. Robock, R. J. Stouffer, J. E. Walsh, C. L. Parkinson,
D. J. Cavalieri, J. F. B. Mitchell, D. Garrett, and V. F. Zakharov. 1999.
Global warming and northern hemisphere sea ice extent. Science 286:
1934–1937.
Walther, G.-R., E. Post, P. Convey, A. Menzel, C. Parmesan, T. J. C.
Beebee, J.-M. Fromentin, O. Hoegh-Guldberg, and F. Bairlein. 2002.
Ecological responses to recent climate change. Nature 416:389–395.
Williams, B. K., J. D. Nichols, and M. J. Conroy. 2002. Analysis and
management of animal populations. Academic Press, San Diego,
California, USA.
Zar, J. H. 1996. Biostatistical analysis. Third edition. Prentice Hall, Upper
Saddle River, New Jersey, USA.
Zhang, X., and J. E. Walsh. 2006. Toward a seasonally ice-covered Arctic
Ocean: scenarios from the IPCC AR4 model simulations. Journal of
Climate 19:1730–1747.
Associate Editor: Alpizar-Jara.

2683