1
Supporting information
Appendix S1.
Materials and methods
POP analysis. As detailed in Dietz et al., 2013a; Dietz et al., 2013b, brominated and chlorinated
POPs were extracted from homogenized polar bear adipose tissues by accelerated solvent
extraction, followed by gel permeation chromatography to remove lipids and other bioorganics,
and then subject to solid-phase clean up. Final extracts were analyzed by GC-MS in electron
impact ionization mode or by GC with election capture detector for polychlorinated biphenyls
(PCBs) and other organochlorines. Brominated POPs were analyzed by GC-MS in electron
capture negative ionization mode. Concentrations of POPs were calculated on a lipid weight
basis.
Quality control. For fatty acids, the National Institute of Standards and Technology (Charleston,
SC, USA) standard reference material (SRM) 1945 (pilot whale homogenate) and inter-day
duplicate polar bear samples were alternately extracted and analyzed with each batch of 24
samples. Our 1945 values were first renormalized to the fraction of sum of the 27 FAs reported
in a previous inter-laboratory QC exercise (Kucklick et al., 2010). These renormalized values
were on average within 14% of the median inter-laboratory comparison values. Duplicate polar
bear samples showed a precision of, on average, 4.7% (relative SD) for all dietary FAs. Although
no δ13C-FA values have been assigned to 1945, we also ran this SRM by GC-C-IRMS as a
precision check and to generate values to which researchers can compare in future δ13C-FA
research. For 14:0, 16:0, 16:1n-7, 18:1, 20:1, 22:1, 20:5n-3, and 22:6n-3, the δ13C-FA ratios (±
SD of 8 replicates) were -32.04 ± 0.41‰, -30.78 ± 0.49‰, -31.91 ± 0.78‰, -29.67 ± 0.57‰, 29.09 ± 0.60‰, -28.13 ± 0.61‰, -36.99 ±1.63‰, and -33.98 ± 1.09‰, respectively. Duplicate
2
polar bear sample runs (at 0.5 and 2.0 mg/ml total FAME) showed average precision (±SD) for
14:0, 16:0, 22:1, 20:5n-3, and 22:6n-3 of ±0.39‰ , ±0.32‰, ±0.28‰, ±1.25‰, and ±0.73‰,
respectively. The average mass of these individual FAMEs in a 1µl injection volume ranged
from 0.01ng to 0.15ng. Complete quality control details for contaminant analysis of 1945,
duplicates and blanks with this dataset were previously reported (Dietz et al., 2013a; Dietz et al.,
2013b).
Sample oxidation. Polar bear % 22:6n-3 FA values were grouped and examined by subjective
oxidation class. As 22:6n-3 is the longest chain and most highly unsaturated FA of those
monitored, it is the FA most subject to oxidative degradation. We thus examined the range of
22:6n-3 proportions by subjective oxidation class to determine whether any polar bear FA
profiles showed gross signs of oxidation (i.e., obviously low 22:6n-3) that could bias diet
estimates. Additionally, in an initial run, the permutation MANOVA also included subjective
oxidation class as a factor to test whether this factor significantly influenced diet estimates.
Simulation exercises. Simulation studies were run to determine the robustness of the model.
First, prey-on-prey modeling assessed how distinct the individual prey species FA signatures
were from one another. This simulation randomly divided an individual prey species into a
‘prey’ and a ‘predator’ dataset. The ‘prey’ dataset in combination with other prey species was
used to model via QFASA this ‘predator’ dataset. The random division was done 1000 times,
and QFASA was re-run on each new ‘prey’ and ‘predator’ combination. Prey-on-prey modeling
was done for each prey species to identify whether all prey were well distinguished dietary items
based on their FA signatures. A second simulation assessed how well QFASA modeling
estimated diets relative to a simulated diet. The simulation randomly divided the prey into a
‘simulation’ and a ‘prey’ dataset. The ‘simulation’ dataset was used to construct a ‘pseudo-bear’
3
signature using arbitrarily assigned prey proportions with additional prey randomly added for
noise (10%). The ‘prey’ dataset was then used to model via QFASA the ‘pseudo-bear’ diet. The
random division was done 1000 times, and QFASA was re-run on each new ‘prey’ dataset.
‘Pseudo-bear’ modeling results indicated how accurately and precisely the QFASA model
estimated a given EG polar bear diet.
Results
Potential confounding influence of oxidation on diet estimates. Proportions of 22:6n-3 in EG
polar bear tissues were grouped according to subjective oxidation class (Fig. S1). In the freshest
group (oxidation class 1), 22:6n-3 ranged from 3.53% to 10.41%. With the exception of one, two
and six samples in oxidation classes 3, 4, and 5, respectively, 22:6n-3 ranges were within the
range of oxidation class 1 samples, i.e., above 3%. This finding suggests that the EG polar bear
FA signatures were not generally substantially altered by oxidative degradation. We nonetheless
excluded the nine samples with 22:6n-3 values below 3% from diet analyses reducing the sample
size to 301. In this set of 301 polar bears, oxidation class was also not a significant factor in the
initial permutation MANOVA (p = 0.61), further confirming that our diet estimates were not
influenced by potential sample oxidation.
Discussion
Hunting records- Limited records for Ittoqqortoormiit exist between 1983-1987 and
complete total annual catch data are available only for harp, ringed, bearded and hooded seals
from 1993-2008 (A. Rosing-Asvid, unpublished data). Between 1993-2008, ringed and hooded
seal catches declined. Harp and bearded seal catches did not significantly change. Thus, although
declines in ringed seals and lack of change in harp and bearded seals paralleled our estimated EG
4
diet trends, hooded seal catch numbers did not. It is possible that catch data does not closely
reflect seal abundance and may instead reflect, for instance, changes in hunting effort.
Fatty acid carbon isotopes- evidence of dietary influence- First, 13C-fractionation should
result in depleted products relative to shorter chain and/or more saturated precursors due to the
kinetic isotope effect. However, we found that δ13C ratios were enriched with increasing chain
length in FA groups with the same degree of saturation, suggesting de novo synthesis is not the
major factor. The same enrichment pattern was also reported in birds (Budge et al., 2011), but
was reversed in lower trophic feeding species (Murphy & Abrajano, 1994; Bec et al., 2011;
Gladyshev et al., 2012). Second, in a controlled feeding trial with eider ducks, the essential FAs
20:5n-3 and 22:6n-3 did not differ in δ13C between diet and consumer adipose (Budge et al.,
2011). Although caution should be exercised in extrapolating these observations to polar bears,
we suggest that the deviation in δ13C ratios of 20:5n-3 and 22:6n-3 observed between EG polar
bears and ringed seals is likely due to bears consuming substantial amounts of other prey species,
in agreement with QFASA-generated diet estimates. Third, changing δ13C-FA profiles (18:1 and
20:1) in the spring-summer time series, but not in fall-winter, concurs with QFASA estimates
indicating change in spring-summer, but not fall-winter, diets. It is conceivable that declines in
δ13C-FA could also be driven by declining δ13C of atmospheric CO2 resulting from fossil fuel
emissions (the Suess effect; Francey et al., 1999). However, since only certain δ13C-FA became
more depleted over time, the changes likely instead represent ecological changes.
5
Table S1 Average (± SE) FA profiles (mass % of total FAME) for adipose tissues of East
Greenland polar bears collected from 1984-2011 (n = 301a).
FA (‘dietary’)b
mean ± SE
FA (‘dietary’)b
mean ± SE
Saturated FA (SFA)
Polyunsaturated FA (PUFA)
12:0
0.05 ± 0.0
16:2n-6
0.03 ± 0.00
0
13:0
0.02 ± 0.0
16:2n-4
0.10 ± 0.00
0
14:0
3.69 ± 0.0
16:3n-6
0.34 ± 0.00
3
iso15:0
0.24 ± 0.0
16:3n-4
0.10 ± 0.00
1
anti15:0
0.07 ± 0.0
16:4n-3
0.12 ± 0.00
0
15:0
0.24 ± 0.0
16:4n-1
0.07 ± 0.00
0
iso16:0
0.10 ± 0.0
18:2Δ5,11
0.07 ± 0.00
0
16:0
7.59 ± 0.0
18:2n-7
0.07 ± 0.00
8
7Me16:0
0.22 ± 0.0
18:2n-6
1.59 ± 0.01
0
iso17:0
0.11 ± 0.0
18:2n-4
0.06 ± 0.00
0
17:0
0.14 ± 0.0
18:3n-6
0.11 ± 0.00
0
18:0
1.94 ± 0.0
18:3n-4
0.06 ± 0.00
5
20:0
0.08 ± 0.0
18:3n-3
0.54 ± 0.01
18:3n-1
0.03 ± 0.00
Monounsaturated FA (MUFA) 0
14:1n-9
0.09 ± 0.0
18:4n-3
0.78 ± 0.02
0
14:1n-7
0.04 ± 0.0
18:4n-1
0.09 ± 0.00
0
14:1n-5
0.82 ± 0.0
20:2n-9
0.07 ± 0.00
1
15:1n-8
0.01 ± 0.0
20:2n-6
0.27 ± 0.00
0
15:1n-6
0.05 ± 0.0
20:3n-6
0.10 ± 0.00
0
16:1n-11
0.30 ± 0.0
20:4n-6
0.21 ± 0.00
0
16:1n-9
0.51 ± 0.0
20:3n-3
0.07 ± 0.00
1
11.3 ± 0.1
16:1n-7
20:4n-3
0.41 ± 0.01
7
16:1n-5
0.054 ± 0.0
20:5n-3
2.33 ± 0.07
0
17:1
0.21 ± 0.0
21:5n-3
0.33 ± 0.00
0
18:1n-11
4.45 ± 0.0
22:4n-6
0.07 ± 0.00
6
24.0 ± 0.1
18:1n-9
22:5n-6
0.08 ± 0.00
0
5
18:1n-7
4.51 ± 0.0
22:4n-3
0.07 ± 0.00
4
18:1n-5
0.36 ± 0.0
22:5n-3
4.59 ± 0.07
0
20:1n-11
3.51 ± 0.0
22:6n-3
6.29 ± 0.07
6
11.4 ± 0.1
20:1n-9
8
14.5 ± 0.12
∑SFA
20:1n-7
0.607 ± 0.0
1
66.11 ± 0.21
∑MUFA
22:1n-11
3.11 ± 0.1
0
19.07 ± 0.19
∑PUFA
22:1n-9
0.68 ± 0.0
2
15.56 ± 0.18
∑n-3 (omega-3)
22:1n-7
0.07 ± 0.0
0
∑n-6 (omega-6)
2.863 ± 0.01
a
Initially 310 polar bear samples were included. However, 9 were removed to ensure no
confounding influence of oxidation on the FA dataset (Appendix S1, Fig. S1).
b
The bolded values indicate ‘dietary’ FAs, those that were used in the QFASA modeling.
6
Table S2 Sample sizes by year and sex/age class for adipose tissues of East Greenland polar
bears collected from 1984-2011.
Yeara
All bears
Adult females
Adult males
Subadults
1984
9
4
0
5
1986
7
1
0
6
1987
4
0
0
4
1988
5
0
0
5
1989
7
0
2
5
1990
14
4
4
6
1991
5
2
0
3
1992
18
2
7
9
1993
20
4
3
13
1994
12
1
4
7
1995
14
2
4
8
1996
8
0
2
6
1998
1
0
0
1
1999
27
7
7
13
2000
35
8
9
18
2001
13
4
1
8
2002
1
0
0
1
b
2003
9
2
1
5
2004 b
10
1
4
4
2005
4
1
1
2
2006
17
3
9
5
b
2007
11
3
4
3
b
2008
12
5
5
1
2009
20
3
9
8
2010
8
2
4
2
2011 b
10
0
2
7
a
Year was considered as February-December plus January of the following year.
b
The sample size for “all bears” is one more than the sum of the individual sex/age classes as one
bear from each of these years was not identified by sex/age class.
7
Mass % DHA by Freshness Class
22:6n-3 (mass % of total FAME)
12.00
12
10.00
10
8.008
6.006
4.004
2.002
0.000
0
11
22
33
44
55
6
Subjective oxidation class
Fig. S1 Percentage of 22:6n-3 in individual East Greenland polar bear adipose FA signatures,
assigned according to visual extent of oxidation (subjective oxidation class) on a 1 to 5 scale: 1
= white/fresh (n = 135); 2 = mainly white/slight yellow tinge (n = 49); 3 = white-yellow (n = 52);
4 = mainly yellow/slight white tinge (n = 43); 5 = yellow/sometimes a bit dry (n = 23). The
dashed line indicates a cut-off point; samples with 22:6n-3 values below this line did not fall
within the range of 22:6n-3 values found for the freshest samples (subjective oxidation class 1),
and were thus considered to have potentially biased FA signatures due to FA oxidation. These 9
samples were not included in subsequent diet analysis reducing the total sample size to 301.
Abbreviations used: FAME; fatty acid methyl ester.
8
Fatty acid composition (mass % of total FA)
14.00
14
12.00
12
10.00
10
8.008
6.006
4.004
2.002
0.000
22:6n-3
22:6n-3
22:4n-6
22:4n-6
21:5n-3
21:5n-3
20:5n-3
20:5n-3
20:4n-3
20:4n-3
18:4n-3
18:4n-3
18:2n-6
18:2n-6
16:4n-1
16:4n-1
16:3n-4
16:3n-4
22:1n-9
22:1n-9
22:1n-11
22:1n-11
20:1n-7
20:1n-7
20:1n-9
20:1n-9
20:1n-11
20:1n-11
9
Fig. S2 Selected dietary fatty acids (average mass % ± SE) in the blubber of polar bear prey from Greenland and neighboring regions
grouped by species and collection location/year: bearded seal (black bars; left to right: Davis Strait (1994-2003), South Greenland
(2011)), harp seal (striped bars; left to right: Davis Strait (1994-2002), Greenland Sea (1999), Denmark Strait (2001)), hooded seal
(white bars; left to right: Davis Strait (1993-2001), Greenland Sea (1999), Jan Mayen (2001), Denmark Strait (2001)), narwhal (crosshatched bars; left to right: Baffin Bay/Foxe Basin/Lancaster Sound (1999-2001), East Greenland (2010)), ringed seal (grey bars; left to
right: Davis Strait (1994-2000), East Greenland (2002-2006)) and walrus (checkered bars; left to right: Foxe Basin (1993-1996), East
Greenland (2011)). Greenland Sea, Denmark Strait and Jan Mayen, Davis Strait, Baffin Bay, Foxe Basin and Lancaster Sound data are
from previous studies (Thiemann et al., 2008a; Thiemann et al., 2008b; Falk-Petersen et al., 2009). Only the proportions of the dietary
fatty acids which were published in these earlier studies are presented.
10
All bears-all years
1.000
100
(a)
0.800
80
0.600
60
47.5
30.6
0.400
40
16.7
0.200
20
4.5
0
0.000
0.7
0.0
All bears-all years
100
1.000000
(b)
80
0.800000
60
0.600000
40
Percent prey contribution to bear FA signature
0.400000
38.4 37.1
14.7
20
0.200000
9.2
0.6
0
100
0.0
All bears-all years
0.000000
1.000000
(c)
80
0.800000
60
0.600000
0.400000
40
39.0 35.2
14.9
20
0.200000
10.2
0.6
0
1.000000
100
0.000000
0.0
All bears-all years
(d)
80
0.800000
0.600000
60
40.1
40
0.400000
33.2
14.9 11.2
20
0.200000
0.6
0
100
0.0
All bears-all years
0.000000
1.000000
(e)
80
0.800000
54.0
60
0.600000
6.4
4.1
0.6
0.0
Narwhal
Walrus
20
0.200000
Bearded seal
35.0
Hooded seal
40
0.400000
0
Harp seal
Ringed seal
0.000000
11
Fig. S3 Influence on East Greenland polar bears diet estimates, generated by QFASA from
adipose tissues collected from 1983 to 2011, of spatiotemporal variation in prey library using (a)
all available prey from Greenland and neighbouring regions, (b) only Greenland prey, as
available, (c) only Greenland prey, as available, and only ringed seal from 2002, (d) only
Greenland prey, as available, and only ringed seal from 2004, (e) only Greenland prey, as
available, and only ringed seal from 2006. Italicized numbers above each bar indicate the mean
dietary contribution. Error bars represent ± SE of individual polar bear diet estimates.
12
1.00
100
1.00
(a)
Ringed seal
Percent prey contribution to bear FA signature
0.80
80
0.80
0.60
60
0.60
0.40
40
0.40
0.20
20
0.20
0
0.00
1.00
1001980
1985
1990
1995
2000
0.80
80
2005
0.00
1980
2010 1.002015
(c)
Hooded seal
0.60
0.40
40
0.40
0.20
20
0.20
1980
1985
1990
1995
2000
1985
1985
1990
1990
1995
1995
2000
2000
2005
2005
2010
2010
0.00
1980
2015
1980
2005
2010
2015
(d)
Bearded seal
0.80
0.60
60
0.000
1980
(b)
Harp seal
1985
1985
1990
1990
1995
1995
2000
2000
2005
2005
2010
2010
2015
Fig. S4 Percent contribution to East Greenland polar bear yearly mean (●) and individual (○)
diets from fall-winters of 1986-2009 of the following prey species: (a) ringed seal (overall mean
47.6%), (b) harp seal (30.4%), (c) hooded seal (16.4%) and (d) bearded seal (5.2%). Narwhal
and walrus data not shown as they were almost never consumed and showed no temporal trends.
No significant linear time trends were found.
13
References
Bec A, Perga ME, Koussoroplis A, Bardoux G, Desvilettes C, Bourdier G, Mariotti A (2011)
Assessing the reliability of fatty acid-specific stable isotope analysis for trophic studies. Methods
in Ecology and Evolution 2 651-659.
Budge SM, Wang SW, Hollmen TE, Wooller MJ (2011) Carbon isotopic fractionation in eider
adipose tissue varies with fatty acid structure: Implications for trophic studies. Journal of
Experimental Biology 214 3790-3800.
Dietz R, Rigét FF, Sonne C, Born EW, Bechshøft TO, McKinney MA, Letcher RJ (2013a) Three
decades (1983-2010) of contaminant trends in East Greenland polar bears (Ursus maritimus).
Part I: Legacy organochlorine contaminants. Environment International In press .
Dietz R, Rigét FF, Sonne C, Born EW, Bechshøft TO, McKinney MA, Muir DCG, Letcher RJ
(2013b) Three decades (1983-2010) of contaminant trends in East Greenland polar bears (Ursus
maritimus). Part II: Brominated flame retardants. Environment International In press .
Falk-Petersen S, Haug T, Hop H, Nilssen KT, Wold A (2009) Transfer of lipids from plankton to
blubber of harp and hooded seals off East Greenland. Deep-Sea Research Part II 56 2080-2086.
Francey RJ, Allison CE, Etheridge DM, Trudinger CM, Enting IG, Leuenberger M, Langenfelds
RL, Michel E, Steele LP (1999) A 1000-year high precision record of 13C in atmospheric CO2.
Tellus Series B 51 170-193.
Gladyshev MI, Sushchik NN, Kalachova GS, Makhutova ON (2012) Stable isotope composition
of fatty acids in organisms of different trophic levels in the Yenisei River. Plos One 7 e34059.
14
Kucklick JR, Schantz MM, Pugh RS, Porter BJ, Poster DL, Becker PR, Rowles TK, Leigh S,
Wise SA (2010) Marine mammal blubber reference and control materials for use in the
determination of halogenated organic compounds and fatty acids. Analytical and Bioanalytical
Chemistry 397 423-432.
Murphy DE, Abrajano TA (1994) Carbon isotope compositions of fatty acids in mussels from
Newfoundland estuaries. Estuarine and Coastal Shelf Science 39 261-272.
Thiemann GW, Iverson SJ, Stirling I (2008a) Polar bear diets and arctic marine food webs:
Insights from fatty acid analysis. Ecological Monographs 78 591-613.
Thiemann GW, Iverson SJ, Stirling I (2008b) Variation in blubber fatty acid composition among
marine mammals in the Canadian Arctic. Marine Mammal Science 24 91-111.