population ecology of polar bears in Norway.pdf

Popul Ecol (2005) 47:267–275

DOI 10.1007/s10144-005-0231-2

ORIGINAL ARTICLE

Andrew E. Derocher

Population ecology of polar bears at Svalbard, Norway

Received: 21 September 2004 / Accepted: 11 August 2005 / Published online: 30 September 2005

The Society of Population Ecology and Springer-Verlag Tokyo 2005

Abstract The population ecology of polar bears at

Svalbard, Norway, was examined from 1988 to 2002

using live-captured animals. The mean age of both fe-

males and males increased over the study, litter pro-

duction rate and natality declined and body length of

adults decreased. Dynamics of body mass were sugges-

tive of cyclical changes over time and variation in body

mass of both adult females and adult males was related

to the Arctic Oscillation index. Similarly, litter produc-

tion rate and natality correlated with the Arctic Oscil-

lation index. The changes in age-structure, reproductive

rates and body length suggest that recovery from over-

harvest continued for almost 30 years after harvest

ended in 1973 and that density-dependent changes are

perhaps being expressed in the population. However, the

variation in reproduction and body mass in the popu-

lation show a relationship between large-scale climatic

variation and the upper trophic level in an Arctic marine

ecosystem. Similar change in other polar bear popula-

tions has been attributed to climate change, and further

research is needed to establish linkages between climate

and the population ecology of polar bears.

low density and high research costs for monitoring polar

bear populations result in an incomplete understanding

of their population dynamics. Researchers must piece

together a variety of elements to construct an insight

into the dynamics of a population: often with incomplete

time series or small sample sizes. Despite these limita-

tions, the population ecology of polar bears is reason-

ably well understood. Similar to other species of bears,

polar bears have delayed maturation, small litters and a

prolonged mother–oﬀspring bond that results in low

population growth rates (Bunnell and Tait 1981 ). Pop-

ulation growth, however, is most sensitive to changes in

adult female survival rate (Taylor et al. 1987 ), but nat-

ural variation in survival of adults is low and treating

adult survival as a constant has been suggested (Eber-

hardt 1977 ; Amstrup and Durner 1995 ). Reproductive

rates, body size and condition are the most dynamic

components of polar bear populations and play a major

role in determining population dynamics (Derocher and

Stirling 1995 , 1996 ; Stirling et al. 1999 ).

The Arctic sea ice habitat is a dynamic environment

and large temporal and spatial variation is common

(A ˚ dlandsvik and Loeng 1991 ; Shapiro et al. 2003 ; Bar-

ber and Iacozza 2004 ). Linkages between climate-driven

sea ice habitats and polar bears were ﬁrst established

from hunting returns in Greenland (Vibe 1967 ). More

recent research has linked the dynamics of polar bear

populations to climatic variability that reduce produc-

tivity of their primary prey, ringed seals (Phoca hispida)

(Stirling et al. 1982 ; Stirling 2002 ). Changes in the pro-

ductivity of ringed seals aﬀects polar bear natality and

oﬀspring survival through changes in body condition

(Ramsay and Stirling 1988 ; Derocher and Stirling 1994 ;

Stirling 2002 ). Recent studies of several northern and

Arctic birds and mammals have linked the dynamics of

populations to climatic indices such as the Arctic

Oscillation and the North Atlantic Oscillation (Forch-

hammer et al. 1998 ; Aanes et al. 2002 ; Post and

Forchhammer 2002 ; Forchhammer and Post 2004 ).

However, no such linkages have been made for polar

bears.

Introduction

Polar bears (Ursus maritimus) are a highly specialised

predator of seals that live in the ice-covered seas of the

Arctic (DeMaster and Stirling 1981 ). Remote locations,

A. E. Derocher

Department of Biological Sciences, University of Alberta,

Edmonton, T6G 2E9, Alberta, Canada

E-mail: derocher@ualberta.ca

Tel.: +1-780-4925570

Fax: +1-780-4929234

A. E. Derocher

Norwegian Polar Institute, Tromsø, 9296, Norway

268

The best studied polar bear populations are found

in the Beaufort Sea, north of Alaska and western

Canada, and in western Hudson Bay (Amstrup et al.

1986 ; Stirling et al. 1999 ; Stirling 2002 ). The dynamics

of reproduction in these two populations are substan-

tially diﬀerent. Large declines in the condition and

reproductive rates of polar bears in western Hudson

Bay have been linked to possible density-dependent

responses following recovery from over-harvest (Der-

ocher and Stirling 1992 , 1995 ) and more recently to

climate change (Stirling and Derocher 1993 ; Stirling

et al. 1999 ). In contrast, the dynamics of condition and

reproduction in the Beaufort Sea are more closely tied

to recovery from over-harvest and climatic variation

(Amstrup et al. 1986 ; Stirling 2002 ). Svalbard, north of

Norway, is the only other polar bear population with

a long-term ecological research program. The dynam-

ics of this population are only partially understood

and are related to a history of intensive harvest that

resulted in a severely depleted population (Larsen

1986 ).

Global concerns over rapidly increasing harvest levels

of polar bears in the 1960s culminated in the signing of

the International Agreement on Polar Bears in 1973 that

resulted in management regimes being imposed on most

populations (Prestrud and Stirling 1994 ). Interpretation

of the Agreement by the ﬁve member states varied

widely. Polar bears in Russia had already received total

protection from harvest in 1956 (Prestrud and Stirling

1994 ). Norway introduced a total ban on harvest in

1973, while Canada and the United States implemented

inventory programs and harvest monitoring. While

extensive research on polar bears has been conducted in

North America, all of these populations still undergo

extensive harvest, which is a signiﬁcant source of mor-

tality (Amstrup et al. 1986 ; Taylor et al. 1987 ; Lee and

Taylor 1994 ; Amstrup and Durner 1995 ). In Norway,

the termination of hunting and the International

Agreement spurred research on the population (Larsen

1986 ; Wiig 1998 ; Mauritzen et al. 2002 ). Polar bears in

Svalbard are the only studied population without an

ongoing harvest.

No meaningful population estimates are available for

polar bears in the Svalbard–Barents Sea area but esti-

mates derived in the 1970s–1980s, when the population

was thought to be recovering from over-harvest (Larsen

1986 ), were of questionable reliability when produced

and with the passage of 20 years, have little relevance for

the current state of the population. Despite the lack of

harvest, conservation issues pertaining to polar bears in

the Barents Sea have centred upon the possible eﬀects of

anthropogenic pollutants. In particular, levels of per-

sistent organic pollutants in polar bears were thought to

negatively inﬂuence the immune system (Bernhoft et al.

2000 ; Lie et al. 2003 ) and hormone homeostasis (Skaare

et al. 1999 ) and may have aﬀected the population

(Derocher et al. 2003 ). Levels of pollutants declined in

the 1990s (Henriksen et al. 2001 ) but the eﬀects on the

population remain unclear.

In this paper, I examine the population ecology of

polar bears in Svalbard and examine the age structure,

reproductive rates and dynamics and body size dynam-

ics. I also examine relationships between the population

parameters and the Arctic Oscillation climatic index.

Study area and methods

Polar bears were live-captured during late March to

mid-May in 1988–2002 in the Svalbard area (78N,

20E) eastward to the central Barents Sea (70N, 44E;

Fig. 1 ). Polar bears in Svalbard are part of the Barents

Sea population and are linked genetically and through

movements to the bears in the western Russian Arctic

(Wiig 1995 ; Paetkau et al. 1999 ; Mauritzen et al. 2001 ).

In general, sampling was constrained to nearshore areas

due to helicopter range restrictions imposed by the dis-

tribution of fuel caches. Sampling intensity of the pop-

ulation varied between years due to do environmental

conditions and funding but the method of searching for

animals was similar over time. Eﬀorts were made to

capture each animal observed unless the conditions were

deemed unsafe. The sample is believed to be represen-

tative of the population.

Polar bears (‡1 year of age) were captured using a

helicopter and remote injection of the drug Zoletil

(Stirling et al. 1989 ). A vestigial premolar tooth was

extracted from all bears for age determination (Calvert

and Ramsay 1998 ). Ages were unavailable for 54 bears

and an estimated age was used for those analyses where I

could allocate the bear to a speciﬁc group (e.g., adult

females ‡4 years old with cubs-of-the-year) but these

animals were excluded from age-speciﬁc analyses. The

sex, reproductive status and a series of standardised

morphometric measure were collected from each bear.

Body length (cm) was measured as the dorsal straight-

line distance from the tip of the nose to the caudal end of

Arctic Ocean

80 °

Frans Josef

Land

Barents Sea

Svalbard

75 N

Novaya

Zemlya

0 200 400

20 E

40 °

60 °

Fig. 1 Map of the study area showing Svalbard, Norway, Franz

Josef Land and Novaya Zemlya, Russia, where the Barents Sea

polar bear population is located

269

the last tail vertebrae with bears lying sternally recum-

bent with the back legs straight behind and the front legs

ﬂexed at the elbows with the forelegs forward and par-

allel to the body. Axillary girth (cm) was measured as

the circumference around the chest at the axilla with a

rope tightened with a tension of about 0.5 kg. Body

mass was estimated from a regression model developed

for the study population using axillary girth and body

length (Derocher and Wiig 2002 ).

Age-speciﬁc reproductive parameters were calculated

based on the methods of Stirling et al. ( 1980 ) and

Ramsay and Stirling ( 1988 ) and include cubs-of-the-year

of both sexes. Age-speciﬁc mean litter size (LS x ) was

calculated from:

The Arctic Oscillation shares some features with the

North Atlantic Oscillation but the Arctic Oscillation

has a more northern centre of action that includes most

of the Arctic (Tremblay 2001 ). Higher than normal sea-

level pressure over the Arctic results in weaker westerly

winds in the upper atmosphere and colder conditions in

northern areas. In contrast, lower sea-level pressure

over the Arctic results in a warming pattern and an

inﬂux of warmer Atlantic water into the Arctic.

Monthly values for the Arctic Oscillation were

obtained from http://www.cpc.ncep.noaa.gov/products/

precip/CWlink/daily_ao_index/ao_index.html and were

ﬁtted to mass data for adult females (‡4 years old)

without cubs-of-the-year and adult males (‡10 years

old). These age groups were selected to minimise age-

related and reproductive status-related variation in the

mass data. Adult females with cubs-of-the-year were

excluded from the analyses because they den over-

winter while lone females and females with older oﬀ-

spring remain active. The mean of body mass for each

sex was used as a single value for each year and was

considered as an index of condition.

To assess the role of climate on the population, I

created a spring and winter index of Arctic Oscillation. I

used the mean of the Arctic Oscillation index for April–

June for a spring index and used a winter index (Octo-

ber–January) and monthly values of Arctic Oscillation

for the 12 months preceding capture in the multiple

stepwise regressions. Variables were retained in the

model if signiﬁcant at P £ 0.05.

P cubs-of-the-year with females age x

LS x ¼

P females age x with cub-of-the year litters :

In instances where no litters from a female of age x

were observed, LS x was assigned a value of 0.000 if fe-

males of that age class were captured. Age-speciﬁc rate

of litter production (LP x ) was calculated as

LP x ¼ P females age x with cub-of-the-year litters

P females age x

An estimate of natality rate (N x ), the product of LS x

and LP x , was derived from:

N x ¼ P cubs-of-the-year with females age x

P females age x

The above equations and notation were used to be

consistent with previous literature. However, because

the natality rate applies to a period roughly 4 months

after birth, it reﬂects recruitment to the spring.

I used SAS statistical software (SAS Institute 1989 ) for

all analyses. Statistical signiﬁcance was set to P £ 0.05.

Values are presented as means ±1SE. Some information

was not available for all animals, resulting in varying

sample sizes between analyses. Ages were log 10 trans-

formed for statistical analyses to normalise data.

To test for temporal trends in body length, I used the

year of birth so that trends would relate to the cohort

rather than the year of capture. Year of birth was deter-

mined from the age at capture subtracted from the year of

capture. Females attain 97% of their asymptotic body

length at 4.4 years of age in the Svalbard population

(Derocher andWiig 2002 ) and to reduce age-related biases

I used females ‡5 years of age for body length analyses.

Males in the population attain 97% of their asymptotic

body length at 6.2 years of age (Derocher and Wiig 2002 )

but growth is slower than in females, so males ‡8 years of

age were included in body length analyses.

I used multiple regression to explore the relation-

ships between the mean body mass for adult females

and adult males and the Arctic Oscillation index. The

Arctic Oscillation is a mode of climate variability in the

Northern Hemisphere north of 20N related to sea-le-

vel pressure variations (Thompson and Wallace 1998 ).

Results

Age structure

The age structure of the captured population was con-

structed from 553 females and 509 males and revealed

Fig. 2 Age structure of female and male polar bears sampled near

Svalbard, 1988–2002

270

that bears 2–5 years of age were under-represented in

the sample (Fig. 2 ). The mean age of females (‡3 years

old, the age of independence) increased over 1988–2001

(linear regression, F 1,395 =24.51, P<0.0001, R 2 =0.06;

Fig. 3 ). A similar pattern was noted for males (‡3 years

old) over 1990–2001 (linear regression, F 1, 351 =5.91,

P<0.016, R 2 =0.02; Fig. 3 ). Comparing the early and

late parts of the study, the mean age of females (‡3 years

old) was 9.1±0.5 years (n=72 for 1988–1993) and in-

creased (t-test, P<0.0001) to 11.9±0.4 years (n=208

for 1998–2001). A similar pattern was observed in males

(‡3 years old) where the mean age from 1988–1993 was

10.8±1.0 years (n=40) and increased (t-test, P=0.008)

to 12.8±0.4 years (n=223) in 1998–2001. To compare

with studies in Canada, I included the mean ages of

females (‡1 year old), which was 8.0±0.6 years (n=84)

for 1988–1993 and rose to 10.8±0.4 years (n=230) for

1998–2001. For males (‡1 year old), in 1988–1993 the

mean age was 8.4±0.9 years (n=54) and rose to

11.7±0.4 years (n=246) in 1998–2001.

Table 1 Age-speciﬁc litter size (LS x ), litter production (LP x ) and

natality (N x ) rates for female polar bears captured in the Svalbard

area in 1993–2002. Sample size (n litters) refers to the number of

females of age x with cub-of-the-year litters and n females refers to

the number of females of age x of all reproductive classes

Age (years) LS x n litters LP x N x

n females

1.00

0.143 0.143

1.50

0.100 0.150

1.62

0.650 1.050

1.84

0.559 1.029

1.70

0.400 0.680

1.58

0.444 0.704

2.00

0.375 0.750

1.86

0.333 0.619

2.20

0.357 0.786

1.33

0.273 0.364

1.62

0.444 0.722

2.00

0.462 0.923

2.00

0.333 0.667

–

0.000 0.000

1.00

0.250 0.250

1.60

1.000 1.600

1.00

0.200 0.200

1.67

0.375 0.625

22+

1.00

0.000 0.067

Reproductive rates

Overall

1.72 106

0.375 0.643 283

The earliest age of ﬁrst reproduction in females was

4 years of age (Table 1 ). However, only three of 27

females (11%) produced cubs before 6 years of age,

after which litter production rate was notably higher

and similar to older age classes. Mean litter size of

cubs-of-the-year did not vary between years (ANOVA,

P=0.37) and was 1.72 cubs/litter (SE=0.05, n=106)

with annual means ranging over 1.71–2.00 cubs/litter

(Fig. 4 ). Singleton litters comprised 32.1%, twins

64.1% and triplets 3.8% of the sample. The mean litter

size for yearlings was 1.52 yearlings/litter (SE=0.07,

n=58) and varied signiﬁcantly (ANOVA, F 10,47 =3.43,

P=0.0019) between 1.00 yearlings/litter and 2.00 year-

lings/litter between years (Fig. 4 ). Singleton yearling

litters comprised 48.3% and twin yearling litters 51.7%

of the sample.

Age-speciﬁc litter size, litter production rates and

natality rates were determined from pooled data

(Table 1 ). Natality varied between years from

0.256 cubs/female to 1.250 cubs/female per year. Litter

production rates declined over the study (linear regres-

sion, F 1,8 =7.66, P=0.024, R 2 =0.49) and, similarly,

natality rate declined (linear regression, F 1,8 =7.24,

P=0.028, R 2 =0.48; Fig. 5 ).

Fig. 3 Mean age (±SE) of female and male polar bears ‡3 years

old sampled in the Svalbard area by year of capture. Linear

regressions are shown as sloping lines

Fig. 4 Mean litter size of cubs-of-the-year and yearlings (±SE) for

polar bears in the Svalbard area, 1992–2002. The number of litters

sampled is indicated above each error bar for cubs and to the right

of the symbol for yearlings

271

Fig. 5 Dynamics of natality

rate (cubs/female per year) and

litter production rate

(proportion of reproductive

aged females with cubs) for

female polar bears at Svalbard,

Norway, from 1993 to 2002.

Solid and dashed lines indicate

the linear regressions

Fig. 6 Relationship between

natality and litter production

and the Arctic Oscillation index

for polar bears in Svalbard,

Norway, 1990–2002. The mean

of April–June Arctic Oscillation

values were used for litter

production and May for

natality. Solid and dashed lines

indicate linear regressions

Using multiple regression, litter production rate was

negatively correlated to the spring Arctic Oscillation

of the preceding year (F 1,8 =6.14, P=0.038, R 2 =0.43)

and natality rate was negatively correlated to the

Arctic Oscillation in May of the preceding year for

the same period (F 1,8 =5.34, P=0.049, R 2 =0.40;

Fig. 6 ).

males (P=0.76). The mean body mass of adult females

without cubs-of-the-year and adult males appeared to

demonstrate a cyclical pattern.

Using multiple regression, the body mass of females

without cubs was positively related to the Arctic

Oscillation index in July (partial R 2 =0.59) and nega-

tively in December (partial R 2 =0.14; multiple regres-

sion, F 2,10 =13.42, P=0.0015, R 2 =0.73). The body mass

of adult males was positively related to the Arctic

Oscillation index in April (partial R 2 =0.60) and

negatively in September (partial R 2 =0.14; multiple

regression, F 2,10 =13.93, P=0.0013, R 2 =0.74).

There was a decrease in the body length of adult fe-

males (‡5 years of age) by year of birth over the study

(F 1,322 =11.80, P<0.001, R 2 =0.04, slope= 0.23 cm/

year; Fig. 8 ). Similarly, there was a signiﬁcant negative

trend in body length of adult males ‡8 years old by year

of birth (linear regression, F 1,239 =7.17, P=0.008,

R 2 =0.03, slope= 0.35 cm/year; Fig. 8 ).

Dynamics of body size

Body mass of adult females (‡4 years old) without

cubs-of-the-year varied between years (1990–2002;

ANOVA, F 12,229 =2.93, P<0.001; Fig. 7 ). Similarly,

the body mass of adult males (‡10 years old) varied

between years over the same period (ANOVA,

F 12,190 =2.03, P=0.024; Fig. 7 ). There was no signiﬁ-

cantly linear trend in the body mass of either adult

females without cubs-of-the-year (P=0.57) or adult

Plik z chomika:

Inne pliki z tego folderu:

Inne foldery tego chomika: