Global changes are thought to affect most Arctic species, yet some populations are more at risk. Today, the Barents Sea ecoregion is suffering the strongest sea ice retreat ever measured; and these changes are suspected to modify food access and thus diet of several species. Biochemical diet tracers enable investigation of diet in species such as polar bears (Ursus maritimus). We examined individual diet variation of female polar bears in Svalbard, Norway, and related it to year, season (spring and autumn), sampling area and breeding status (solitary, with cubs of the year or yearlings). Sampling areas were split according to their ice cover: North-West (less sea ice cover), South-East (larger amplitude in sea ice extent) and North-East/South-West (NESW) as bears from that zone are more mobile among all regions of Svalbard. We measured fatty acid (FA) composition in adipose tissue and carbon (δ13C) and nitrogen (δ15N) stable isotopes in plasma and red blood cells. Females feeding in the North-West area had lower δ15N values than those from the NESW. In South-East females, δ13C values were lower in autumn compared to spring and females seemed less selective in their diet as depicted by large variances in stable isotope values. Considering the differences in FA composition and stable isotope values, we suggest that females from the North-West and South-East could ingest a higher proportion of avian prey. With regard to breeding status, solitary females had higher δ15N values and smaller variance in their stable isotopic values than females with cubs, suggesting that solitary females were more selective and prey on higher trophic level species (i.e. seals). Overall, our results indicate that prey availability for Svalbard polar bears varies according to geographical area and prey selectivity differs according to breeding status. Our findings suggest that complex changes in sea ice and prey availability will interact to affect Svalbard polar bear feeding patterns and associated nutrition.
Citation: Tartu S, Bourgeon S, Aars J, Andersen M, Ehrich D, Thiemann GW, et al. (2016) Geographical Area and Life History Traits Influence Diet in an Arctic Marine Predator. PLoS ONE 11(5): e0155980. https://doi.org/10.1371/journal.pone.0155980
Editor: Philippe Archambault, Université du Québec à Rimouski, CANADA
Received: December 14, 2015; Accepted: May 6, 2016; Published: May 19, 2016
Copyright: © 2016 Tartu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: Data are available from the Norwegian Polar Institute Dataset catalogue: https://data.npolar.no/dataset/.
Funding: This study was supported by the Norwegian polar institute and the Research Council of Norway (http://www.forskningsradet.no/en/Home_page/1177315753906). The grant 216568/E10 was received by HR. This collaboration was supported in part by the Fulbright Distinguished US Arctic Chairship-Norway awarded to JM Welker. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
The Arctic marine environment is characterized by a dynamic sea ice platform that varies in its seasonal and spatial distribution, as well as its form and type . Over the last decades, the Arctic has experienced the greatest warming ever measured and sea ice extent has undergone dramatic reductions [2–5]. The Arctic marine biota relies on sea ice as a physical platform to breed and to feed and is thus vulnerable to changes in ice characteristics . With ongoing climate warming and sea ice reduction, diet composition of pagophilic marine mammals is expected to change [7–9]. Yet, studying the diet of sea ice dependent species such as polar bears (Ursus maritimus) or seal species can be a challenge given their low density populations [10,11].
Diet habits of polar bears have been well documented, they are the most carnivorous ursid species and they mainly prey upon ringed and bearded seals (Pusa hispida and Erignathus barbatus, respectively) [12–14]. Nevertheless, they are also opportunistic and scavenger feeders . This includes seabirds, waterfowl, eggs, reindeer (Rangifer tarandus), fish and whale carcasses [15–24]. With ongoing reductions in sea ice availability, terrestrial feeding may become more frequent [20,22,24–27] although the energetic and ultimate ecological value of these foods for polar bears remains a matter of debate [24,27].
Among subpopulations that received a high monitoring intensity such as southern Beaufort Sea, western Hudson Bay and the Barents Sea’s subpopulations , the diet of polar bears from Svalbard, has been seldom investigated. Only a handful of studies have described diet through opportunistic observation of stomach contents, seal kills, scat analysis or fatty acid (henceforth ‘FA’) composition [12,19,29,30]. The Barents Sea area is experiencing the fastest loss of sea ice recorded throughout the Arctic, with a reduction of 41.8 ± 7.1 ice-free days per decade . There is however, a lack of information on population trends of marine mammals inhabiting this region, including polar bears . Polar bears from the Barents Sea subpopulation are suspected to be particularly vulnerable to the consequences of climate change, which will be primarily mediated by the access to sea ice habitat, and thus reduced foraging opportunities . Thus, it is crucial to understand better the dietary ecology of Svalbard polar bears.
Life-history traits may explain a significant amount of individual variability in diet choices and habitat use in polar bears . For instance, females undergo a fasting period associated with parturition from about four to eight months, varying with area . Only a few studies have reported differences in habitat use according to breeding status in female polar bears [32,34] but there is to our knowledge no evidence that females with different breeding status feed on different prey. However, we assume that females will have different energetic needs depending on the presence, age and number of offspring, which may influence prey selection. Their dependence on seasonal and variable habitat–sea ice–together with their complex biology makes polar bears an ideal model organism to study how diet may vary depending on environmental conditions and life history status.
Biochemical tracers, for instance FA profiles (i.e. the relative proportion of various FA) or nitrogen and carbon stable isotopes, have been used as proxies to describe the foraging niche and diet composition of free-ranging species [14,35–37]. This occurs because the chemistry of animal tissues reflects to some degree the biochemical properties of their food. Indeed, FA can provide insight into an organism’s diet because dietary FA are predictably incorporated into a consumer’s tissues. The FA composition in adipose tissue can provide an integrated record of dietary intake over a period of weeks to months, and perhaps longer in some species . Moreover, nitrogen stable isotope ratios (15N/14N expressed as δ15N) fractionate or change in a predictable fashion between trophic levels and so reflect trophic position . Carbon isotope ratios (13C/12C expressed as δ13C) remain stable and thus indicate isotopic signatures of sources of primary productivity , for example marine vs. terrestrial, pelagic vs. benthic, inshore vs. offshore. Stable isotopes in red blood cells integrate the diet of the last 2–3 months while the turnover of stable isotopes in plasma is in the order of 1–2 weeks [39,40]. Therefore, concurrent analyses of stable isotopes and FA can provide a robust understanding of marine mammal foraging behavior . Additionally, these diet tracers have several advantages compared to other methods to determine diet. For example, they can integrate diet over a longer time window than polar bear kill sites monitoring or scat analyses. Moreover, they require the capture of the individual, therefore diet can be related to other individual information (biometry, status, physiology, etc.) and finally diet tracers are not destructive, on the contrary to studying stomach content.
We investigated individual variation in the diet of female polar bears and related it to environmental parameters and life history. Due to spatio-temporal variation in sea ice extent we expected the diet of bears to differ between spring and fall, possibly reflecting a larger proportion of seals available during the spring. Additionally, we expected diet to differ between areas in Svalbard according to the timing, amplitude and duration of ice cover; which may influence prey availability. Then, we hypothesized that solitary females will have different feeding strategies than females with dependent offspring because of differences in mobility of the latter (e.g. reduced mobility, risk of infanticide) but also because of increased energetic needs for further reproductive events.
Materials and Methods
The National Animal Research Authority (NARA), Norway, approved research and capture protocols (permit number FOTS ID 7278).
Adult female polar bears (age 4–28 years) from the Barents Sea subpopulation were captured in April and September 2012 and 2013. The 112 samples collected (N = 33 in April 2012, N = 24 in September 2012, N = 29 in April 2013 and N = 26 in September 2013) represented 78 females with 26 animals being captured more than once. Females were sampled opportunistically throughout Svalbard and immobilized by remote injection of tiletamine hydrochloride and zolazepam hydrochloride (Zoletil ® 100; Virbac, France), delivered by a dart fired from a helicopter (Eurocopter AS350 Ecureuil). Blood was collected from the femoral vein using heparinised collecting tubes. We kept samples cool and out of sunlight until centrifuged within 10 h (3500 rpm, 10 minutes). Red blood cells and plasma were frozen and stored at -20°C and subsequently used to assess stable isotope values. Adipose tissue samples were collected using a 8 mm biopsy punch and consisted of a full-layer core from skin to muscle, taken approximately 15 cm lateral to the base of the tail . Tissue samples were stored in liquid nitrogen in the field, and kept at -80°C until analysis. Immobilization and handling procedures followed standard protocols [42,43]. Females were classified in three groups according to their breeding status: solitary (i.e. alone or together with a male in spring), with 1 or 2 cubs of the year (COYs; cubs younger than 1 year old) or with 1 or 2 yearlings (YRLs; cubs aged between 1 and 2 years). Additionally, we reported opportunistic observations of kills and carcasses for the captured females and identified the prey species when possible.
Sampling areas determination
Owing to physical and chemical processes, sea ice concentration, the timing of ice availability and the duration of the ice season, are irregular on the different coasts of Svalbard [44–46]. Specifically, the sea ice is less extended and its density lower along the West coast of Svalbard compared to the Eastern coast [44–46]. The South-East area of Svalbard, including the islands Barentsøya and Edgeøya, experiences the largest amplitude of sea ice retreat during summer [44–46]. Consequently, we separated females caught in the North-West from those caught in the South-East area. Bears caught in the remainder of the archipelago, i.e. Nordaustlandet and along the North-East and Southern coasts of Spitsbergen (henceforth the North-East/South-West diagonal: NESW) frequently move among all regions (J. Aars, unpublished data), we therefore pooled them into a third group (Fig 1). Only two females were captured in more than one of these areas; both were caught in the NESW area in spring and recaptured in the South-East area in autumn.
Female polar bears were captured in 2012 and 2013. Each circle represents a polar bear sampled for blood and adipose tissue. The three regions are represented by different colors: North-West (orange dots), North-East/South-West diagonal (blue dots) and South-East (grey dots).
Stable isotopes and FA determination
The determination methods for FA from adipose tissue  and stable isotopes in plasma and red blood cells  are detailed in S1 Appendix. We collected blood and adipose tissue from all the captured females. We measured stable isotopes in plasma and red blood cells for 112 samples (78 individuals) and FA data were derived from 83 samples (64 individuals) as in the remainder, fat biopsies were too small to determine FA composition (S1 Table provides an overview of the samples). From 75 FA determined in fat samples, we used 32 FA representing more than 0.2% of total mass % FA in subsequent analyses (S2 Table) . Overall, 33 FA were ≥ 0.2% of total FA. Percentage values for FA were transformed by calculating the log of the ratio of each FA to c18:0 prior to principal component analysis (PCA) . Since the log of 0 cannot be taken, 0 values were replaced with a small constant (0.005%) prior to transformation. We used PCA ordination plots to illustrate the FA patterns according to sampling area, season and breeding status. Ordination plots will order individuals that are characterized by PCA scores, so that individuals with similar PCA scores are near each other and individuals with dissimilar PCA scores are farther from each other. The 32 FAs (without c18:0) used in the present study included i-14:0, c14:0, 14:1n-5, c15:0, c16:0, 16:1n-11, 16:1n-9, 16:1n-7, 16:1n-5, i-17:0, 16:2n-4, c17:0, 18:1n-11, 18:1n-9, 18:1n-7, 18:1n-5, 18:2n-6, 18:3n-4, 18:3n-3, 18:4n-3, 20:1n-11, 20:1n-9, 20:1n-7, 20:2n-6, 20:4n-6, 20:4n-3, 20:5n-3, 22:1n-11, 22:1n-9, 21:5n-3, 22:5n-3 and 22:6n-3.
We conducted statistical analyses using R version 3.2.1 . We ran a PCA on the 32 FA (R package ade4, ). We generated FA principal components (PCs) for further analysis from the first, second and third axis of the PCA (projected inertia: PC1: 31.6, PC2: 16.7, PC3:12.7%, respectively). The three first axes extracted 61.0% of the total variance of the data cloud. To test for the effects of sampling area, year, season, breeding status and two-way interactions (fixed factors) on FA PCs 1, 2 and 3, δ13C, and δ15N values, we used generalized linear mixed models (GLMMs; R-package nlme version 3.1–121; . We used female identity (FemID) as a random factor to account for the repeated measurements (among seasons and/or years). To select the best GLMMs, we used an information-theoretic approach  based on Akaike’s information criterion corrected for small sample size (AICc, R package AICcmodavg version 2.0–3, ). We tested 15 biologically relevant candidate models based on the variables included (S3 Table). The best model was taken to be the one with the smallest AICc. Yet among models with ΔAICc < 2, we selected the most parsimonious one . We obtained parameter estimates and 95% confidence intervals (CIs) for the selected models. The level of significance was set to α ≤ 0.05. When the interaction term was significant, we disregarded the effects of the main factors on the response variable and we used least squares means method (lsmeans function in Lsmeans package, ) to identify significantly different terms in a biologically relevant frame. For example, if the interaction of year and season is significant, we will not compare bears captured in spring 2012 to those captured in autumn 2013. We will rather check if in 2012 there was an effect of season we do not observe in 2013. With regard to FA, we tested whether some FA, more abundant in certain prey or characteristic of some arctic food chains, would differ according to environmental factors and breeding status. To do so, we selected FA 22:1n-11 as a marker of pelagic prey  and 18:1n-9 and c16:0 as markers of bird prey [55,56]. We checked the proportions of FA 22:6n-3, c18 polyunsaturated FA (PUFA: 18:2n-6, 18:4n-3), 20:5n-3, 22:5n-3 and 16:1n:7 to determine whether the diet was based on a diatom food-chain characteristic of harp seals, (Pagophilus groenlandicus), or a dinoflagellate and prymnesiophyte (Phaeocystis pouchetii) food-chain characteristic of hooded seals, (Cystophora cristata) [57,58]. We also focused on FA known to be readily or poorly mobilized during periods of nutritional stress . We selected five FA, which were highly represented in polar bear adipose tissues in this study: FA 22:1n:11, 20:1n-9, 22:1n-9 for the least mobilized FA and FA 16:1n-7 and 20:5n-3 for the readily mobilized FA. We tested differences in FA mass % according to variables selected for PC1, PC2 and PC3 by AICc (Table 1) and least square means method . Finally, to test whether prey selectivity differed according to sampling area, year, season and breeding status we performed Levene variance tests on δ13C and δ15N values in red blood cells and plasma (levene.test function in lawstat package ). We assumed a smaller variance reflects a more specific diet.
Predictors of FA in adipose tissue
The variables which best predicted FA PC1 value were season and breeding status (Table 2, Fig 2A and 2B). In autumn, PC1 values in female polar bears were lower compared to females caught in spring. PC1 values were less in solitary females compared to that of females with COYs, but they did not differ between solitary females and those with YRLs (Table 2). The best predictors of FA PC2 values were season and sampling area, yet the effect of sampling area was not considered significant as the 95% CI included 0 (Table 2). FA PC2 values were also lower in females caught in autumn compared to those caught in spring (Table 2). Finally, the best predictor of FA PC3 was sampling area (Table 2, Fig 3). Females caught in the NESW diagonal and those caught in the South-East had lower FA PC3 values than females caught in the North-West (Table 2).
Ordination plots from fatty acids (FA) principal component analysis (PCA) scores grouped by (A) season and (B) breeding status. PCA was based on mass percentage of total dietary fatty acids (FAs) relative to FA 18:0, in female polar bears sampled in Svalbard in 2012–2013. The first and second axes explained 31.6% and 16.7% of the total variation, respectively. FA contribution to each axis appears on the left panel. On panels A and B, each dot represents an individual. Individuals with similar FA composition (PCA scores) are near each other and individuals with dissimilar FA composition are farther from each other, the center of the star linking all individuals of a same group represents the centroid of PCA scores for that group.
PCA was based on mass percentage of total dietary FAs relative to FA 18:0, in female polar bears sampled in Svalbard in 2012–2013. The second and third axes explained 16.7% and 12.7% of the total variation, respectively. See Fig 2 for the interpretation of an ordination plot.
Among the selected FA, we only reported results that were significantly different. In females caught in autumn, the proportion of FA c16:0 and 16:1n-7 was higher of 1.41% [0.66; 2.17] (values are parameter estimates and 95% CI) and 2.52% [1.32; 3.72] respectively, compared to females caught in spring (Fig 4A). Additionally, in females caught in autumn we observed a lower proportion of FA 20:1n-9 and 22:1n-9–2.95% [-4.11; -1.78] and -0.13% [-0.20; -0.07] respectively, compared to females caught in spring (Fig 4A). In solitary females, FA 18:1n-9 was lower of -2.92% [-5.25; -0.60] and FA 16:1n-7, 18:4n-3 and 20:5n-3 were higher of 2.73% [0.67; 4.80], 0.31% [0.07; 0.56] and 1.02% [0.18; 1.87] respectively compared to females with COYs (Fig 4B). With regard to the relationships between individual FA variation and sampling area (Table 3), we observed the following pattern: North-West females < NESW, South-East females, for FA 16:1n-7 and 22:5n-3. North-West females > South-East females for FA 18:2n-6, 20:1n-9, 22:1n-9 and 22:1n-11 (Fig 4C). South-East females had higher proportions of c16:0, 16:1n-7 and lower proportions of 20:1n-9, 22:1n-11, 22:1n-9, 22:6n-3 compared to NESW females (Fig 4C).
Median mass % of 33 fatty acids (FA) according to A) season, B) breeding status and C) sampling area. FA were obtained from adipose tissue. Sampling areas are North-West (NW), North-East/South-West diagonal (NESW) and South-East (SE). Breeding status indicates whereas the female was accompanied or not with cubs and their age: solitary females (Solitary), females with cubs of the year (with COYs) and females with yearlings (with YRLs). Female polar bears were sampled in Svalbard in 2012 and 2013. Bars are median values and standard errors.
Predictors of stable isotope values in plasma and red blood cells
Among the variables we included in our models to explain δ13C values in red blood cells, the null model (ΔAICc = 1.75) received strong support indicating that little variance in δ13C values in red blood cells was explained by the variables we selected (Table 1, Fig 5). In plasma, the best predictors of δ13C values were season, sampling area and their interaction (Table 4). The only term that had a significant effect on plasma δ13C values was the interaction of season and sampling area (Fig 5). The comparison of autumn δ13C values in NESW and South-East females to spring δ13C values in North-West females was not biologically relevant. Therefore, we considered the interaction term by testing whether plasma δ13C values would seasonally vary in each area. Least squares means method revealed that in the South-East area, females caught in autumn had plasma δ13C values lower of 1.22‰ [0.72; 1.71] than in spring. In females caught in the North-West and in the NESW diagonal, we did not observe seasonal differences in plasma δ13C values (-0.07‰ [-0.73; 0.58] and 0.21‰ [-0.54; 0.96], respectively).
Relationship between blood stable isotopes and (A) sampling area, (B) season and (C) breeding status. Nitrogen (δ15N) and carbon (δ13C) stable isotope values were measured in plasma (circles) and red blood cells (triangles). Seasons are spring (samples from April) and autumn (samples from September). Breeding status indicates whereas the female was accompanied or not with cubs and their age: solitary females (Solitary), females with cubs of the year (with COYs) and females with yearlings (with YRLs). We sampled female polar bears in Svalbard in 2012 and 2013. Circles/triangles represent medians and bars standard error.
The best predictors of δ15N in red blood cells were sampling area and breeding status (Table 4). Females captured in the North-West had δ15N values that were 1.01‰ [0.33; 1.69] lower compared to females from the NESW diagonal, yet δ15N values did not differ from that of South-East females (Table 4, Fig 5A). Additionally, females with YRLs had δ15N values that were lower by 0.66‰ compared to solitary females (Table 4). δ15N values were not different between solitary females and those with COYs (Table 4, Fig 5C). In plasma, season and breeding status were significantly related to δ15N values (Table 4). In females caught in autumn, plasma δ15N values were lower of -0.99‰ [-1.39; -0.59] compared to those caught in spring (Fig 5B). Similarly as in δ15N red blood cells values, plasma δ15N values in females with YRLs was lower of -0.94‰ [-1.56; -0.31] compared to that of solitary females and we observed no differences between females with COYs plasma δ15N values and those of solitary ones (Table 4, Fig 5C).
Selectivity in diet
For δ13C values’ variances measured in plasma we observed the following pattern: South-East > NESW and North-West (South-East, NESW: Levene statistic tests (LST) = 5.1, P = 0.027; South-East, North-West: LST = 7.6, P = 0.008 and North-West, NESW: LST = 0.4, P = 0.517, Fig 5A). For δ15N variance the pattern was South-East and North-West > NESW diagonal (South-East, NESW: LST = 12.9, P<0.001; South-East, North-West: LST = 0.6, P = 0.434 and North-West, NESW: LST = 9.6, P = 0.003, Fig 5A). With regard to δ15N and δ13C values in red blood cells, we did not find any variance difference according to sampling area (LST<3.8, P>0.057 for all tests).
In autumn, the variances of plasma δ15N and δ13C in female polar bears were larger than that of females caught in spring (plasma δ15N: LST = 13.4, P<0.001, plasma δ13C: LST = 16.5, P<0.001, for stable isotopes in red blood cells P>0.760 for all tests, Fig 5B). Sampling year was not related to the variances of plasma and red blood cells’ stable isotope values (LST<0.5, P>0.465, for all tests). Finally, the variances of red blood cells’ δ15N values in solitary females were smaller than that of females with COYs (red blood cells δ15N: LST = 8.0, P = 0.005, Fig 5C); and solitary females had a smaller plasma δ13C value variance than females with YRLs (LST = 6.1, P = 0.016, Fig 5C). Specifically, solitary females were more selective (less isotopic variance) than females with COYs and with YRLs, in terms of δ15N and δ13C values, respectively (Fig 5C). Variances of δ15N and δ13C values in plasma or red blood cells were not significantly different for any other comparison (P>0.058 for all tests).
Female polar bear habitat area use, life history stage and season are all important drivers of diet based on our isotope and lipid analysis over two years on Svalbard, in the Barents Sea. Previous studies have observed that seal species differ in availability according to locations in Svalbard and distance from the coast [12,61–63] and that in spring female polar bears of different life history stages tend to select different habitats [32,34]. This is the first evidence of such diet variations among female polar bears from the Barents Sea subpopulation. Diet shifts between spring and autumn were less pronounced than predicted. Among females from the South-East we observed a plasma δ13C depletion from spring to autumn which could result from the larger amplitude in sea ice extent in that region. Additionally, FA PCA scores and δ15N values in red blood cells were lower in fall than spring, which could mirror seasonal differences in trophic level.
Carbon, nitrogen and lipid sources in relation to sampling area
As expected, we observed diet variations among sampling areas (Fig 6). Since prey distribution may not be uniform in Svalbard [12,61–63], it seems likely that female polar bears in different areas (i.e. North-West, NESW diagonal and South-East) feed on prey species relative to their occurrence. Furthermore, predation on seals will depend on sea ice cover.
Diet tracers include δ15N and δ13C values in plasma (e.g δ13CP) and red blood cells (e.g δ15NRBC), and fatty acids principal component first axis values (FA PC3). For each area, seasonal differences in plasma δ13C values appear as “δ13CP_Sp = δ13CP_Aut”, Sp denotes spring and Aut denotes autumn. Each sign (+ or -) denotes direction in significant differences. For example, δ15NRBC values are lower in North-West females compared to NESW females but not different from South-East females. Whereas FA PC3 values are higher in North-West females compared to NESW and South-East females. The red stars indicates the positions of whale carcasses, the orange star indicates the position of a bear feeding on a walrus (Odobenus rosmarus) kill, the green stars a bird cliff where bears have been observed feeding on birds and the blue star represents the position of a bear with reindeer hair between the teeth.
The west coast of Spitsbergen typically has less sea ice year round than the eastern part due largely to the relatively warm western Spitsbergen current [26,44,45]. Less sea ice may drive polar bears to feed on prey other than seals to meet their energetic needs. North-West females exhibited lower red blood cell δ15N values than females from the NESW diagonal. The FA 22:1n-11 was higher in North-West females and can be used as a marker of pelagic prey . FA 22:1n-11 is higher in fish and seabirds than in seal species or whales [64,65]. Ingesting seabirds and their eggs would likely result in depletion in the δ15N values of plasma and red blood cells of the predator, given the lower δ15N values of seabirds compared to seals . In Svalbard, δ15N value in seabird muscles and eggs (including Rissa tridactyla, Somateria molissima, Uria lomvia, Alle alle) averages 11.4 ± 0.4 ‰, whereas in ringed seal muscles, δ15N value averages 14.5 ± 0.5 ‰ (E. Fuglei and D. Ehrich unpublished data).
In females from the South-East, we observed lower plasma δ13C values in autumn compared to spring. In comparison to the two other locations, the South-East area experiences the largest amplitude of sea ice retreat during summer . During the ice free season, some females from the South-East area could have been preying on reindeer, seabirds or molting waterfowl and their chicks, as already observed in Svalbard [16,67] which would lead to depleted δ13C values in the females. Waterfowl, such as pink-footed goose, barnacle goose or Brent goose (Anser brachyrhynchus, Branta leucopsis, Branta bernicla, respectively) are common in Svalbard  and feed on vegetation . δ13C values in muscles and eggs of pink-footed and barnacle geese, average -27.3 ± 1.5 ‰, in Svalbard reindeer -26.3 ± 0.6 ‰, while in ringed seals δ13C values average -19.8 ± 0.5 ‰ (E. Fuglei and D. Ehrich unpublished data). Additionally, South-East females exhibited high % of 16:1n-7and 16:0. The FA 16:0 is found in high proportions in waterfowl eggs [55,56]. Consequently, δ13C values and FA profile in South-East females could integrate a larger proportion of terrestrial food during the ice free season. Indeed, among the females that were captured in 2012 and 2013, some individuals were directly observed preying on seabirds (R. tridactyla), and one female had reindeer hair between her teeth. The females preying on seabirds and reindeer exhibited depleted δ13C values in plasma (Fig 7). Moreover, baleen whales are also expected to have depleted stable isotope values compared to seals . δ13C values in plasma obtained from female polar bears feeding on whale carcasses were slightly lower than values from a female that just fed on a seal (Fig 7). Therefore, the fact South-East females could prey on different seal species, whale remains and birds would therefore explain the larger variance in both δ13C and δ15N stable isotope values in plasma compared to females from the NESW and North-West (δ13C values only).
Three females with COYs were observed feeding on seabirds (green dot), one female had reindeer hair between the teeth (blue dot), one was observed feeding on a walrus (orange dot), four on three different whale carcasses (red dot) and two on unidentified seals (grey dot).
In the NESW diagonal, females seemed to be more selective. This suggests that the access and availability of their preferred prey was higher than in the two other regions. In polar bears, the main feeding period starts early April during the ringed seal pupping peak and lasts until the moulting period of adults in mid- or late June [71–73]. Female ringed seals rely on stable ice conditions to dig out birth lairs and prime breeding areas are found near glacier fronts that terminate at the sea [72,74]. These tidewater glaciers are more numerous in the NESW diagonal than in the North-West or South-East areas . Additionally, in the 1980s, a survey investigated the density of breeding female ringed seals in Svalbard . It appeared that higher densities of birth lairs were found in the western part of Spitsbergen and between Nordaustlandet and Spitsbergen . Consequently, females from the NESW area should have better access to ringed seals compared to the two other locations.
Additionally, the only FA measured in NESW females that was different from both North-West and South-East females, was 16:1n-7 (South-East>NESW>North-West). This FA can be characteristic of body-condition or of differently based food-chain, depicting the access to different type of seal species [57–59]. For example, other seal species could be more accessible in certain locations than others. Hooded seals could be predated by North-West female polar bears, as from October to January they forage in the North of Svalbard . This could in part explain the higher 18:2n-6 and lower 16:1n-7 observed in North-West females as this FA profile is characteristic from the dinoflagellate and the prymnesiophyte based food-chain of hooded seals . In the South-East, polar bears could also prey on harp seals . In June and July, harp seals have been observed in the South-East of Svalbard  and their FA signature is derived from a diatom based food chain characterized by higher % of 20:5n-3, 22:5n-3, 16:1n-7 and lower c18 PUFA, and South-East females bore higher % of 22:5n-3, 16:1n-7 and lower 18:2n-6 [57,58].
Season and diet
For polar bears, predation success in winter is thought to be low  then they undergo a hyperphagic period from early April to late June . As previously mentioned, the turnover time of stable isotopes in red blood cells and FA in adipose is likely in the order of weeks to several months while the turnover of stable isotopes in plasma is in the order of 1–2 weeks [36,38,80,81]. Therefore, several blood samples taken in April represent diet at the onset of the hyperphagic period. This may explain the higher δ15N values in plasma in comparison to females sampled in autumn and the more specific diet (as depicted by the smaller stable isotopic variances in spring than in autumn). In spring, polar bears will mainly prey on seals, whereas in autumn prey items may be more diversified (e.g. waterfowls and seabirds). Surprisingly, the FA pattern seems less specific in spring than in autumn (Fig 2A). Females were sampled at the beginning of the hyperphagic period, while some of them were still fasting and had not started building energy stores. Indeed, body condition index (BCI) in the same polar bears as used in the present study was lower in spring than in autumn (Bourgeon et al. submitted). When used as an energy source, the release of FA is selective according to carbon chain length and degree of unsaturation (reviewed in ). In spring, females had higher proportions of FA 20:1n-9 and 22:1n-9 which are amongst the least mobilized . Conversely, 16:1n-7 is a readily mobilized FA  and was in lower proportion in females caught in spring. The higher δ15N values measured in females caught in spring could also be the result of a higher proportion of females under nutritional stress. These individuals will catabolize their lean tissues, which will enrich their δ15N pool independently of their trophic level [82–84]. Consequently, it is likely that the observed stable isotopic and FA patterns measured in spring resulted from both nutritional stress of some individuals and diet switching.
Carbon, nitrogen and lipid sources in relation to breeding status
In spring, females of different reproductive status do not use the same habitat [32,34]. Solitary females extensively use pack ice and have a large home ranges, whereas females with COYs mainly remain on fast ice close to glacier fronts [32,34]. In spring, the restriction of females with young cubs to a certain type of habitat could result from habitat selection to reduce the risk of infanticide, reduced mobility and swimming capacities of young cubs, added to a need for access to a predictable food source such as ringed seal pups born in the fast ice habitat [32,34]. Telemetry data indicates that females with YRLs are the reproductive group with the most freedom to move around as they are less limited in their mobility compared to females with COYs and they are not involved in mating .
FA profiles in females with YRLs, suggest lipid sources close to that of solitary females. Yet, solitary females have higher δ15N values in plasma and red blood cells than females with YRLs. Lactation has been shown to deplete the mother’s δ15N pool  and female polar bears typically lactate until their cubs are ~1.5 years old [86,87]. Consequently, females with YRLs that have been lactating for more than a year are likely to have lower δ15N values in both plasma and red blood cells than solitary females. The FA profile of females with COYs could have resulted from a broader prey base, as evidenced by the larger variance in plasma δ15C stable isotope values, or could mirror fasting physiology. Females with COYs sampled in spring have just emerged from their dens and FA in adipose tissues integrate diet information over the previous weeks to the past few months . If some females with COYs have been sampled while fasting, we could expect the observed FA pattern with lower proportions of 16:1n-7 and 20:5n-3 FA . Several fasting females with COYs could explain the observed δ15N values (Solitary > with COYs > with YRLs), although lactation is supposed to deplete δ15N values in the mother pool, nutritional stress could lead to an increase of δ15N values. Moreover, we observed three females with COYs feeding on kittiwakes in autumn 2013 (Fig 5). This could increase the overall higher proportion of 18:1n-9 measured in females with COYs, as this FA is observed to be in high proportion in birds [55,56].
The diet tracer patterns observed in lone females as compared to those with cubs suggest that lone females are the most selective foragers. For example, they have smaller variances in stable isotopes, higher δ15N values compared to females with YRLs and different FA composition compared to females with COYs. Lone females are less limited in terms of mobility and nutritional needs (e.g. no lactation) and could thus focus their hunts on higher energy adult ringed seals, thus preparing for future reproduction and growth .
In prey-rich environments, large carnivorous mammals are energy maximizers that can afford to select their prey in terms of those species, age, and sex classes that are most abundant or easiest to hunt and kill, whereas in a prey-poor environment predators will maximize the number of prey without selection of species [89,90]. Females with dependent young have to develop additional strategies to increase their own survival and that of their offspring . This is what we observe in female polar bears and similar strategies have been observed in other carnivorous mammals. For example, in leopards (Panthera pardus), cheetahs (Acinonyx jubatus) and Eurasian otters (Lutra lutra), females with young tend to forage in different areas than solitary females and males, take fewer risks of encountering potential predators and be less selective on prey [90–93].
Conclusion and Perspectives
This study provides new understanding of the ecological factors contributing to the diet of female polar bears in Svalbard and our data show diet specialization over a small geographic scale. Polar bears’ movements follow a circannual pattern with season-specific area fidelity [94–96]. Considering this, even within a small portion of the Barents Sea polar bear subpopulation, some individuals may be more threatened by sea ice retreat according to their season-specific foraging area. In the North-West females have lower BCI than in other areas (Bourgeon et al. submitted), whereas in the South-East females experience the largest amplitude in seasonal sea ice retreat. We can postulate that females from those two areas may be more sensitive to climate changes than NESW females, which have a more consistent habitat and thus a more consistent prey availability. Our findings suggest that complex changes in sea ice, prey availability and life history status could all interact to determine female polar bear feeding patterns in Svalbard.
S1 Appendix. Fatty acid and stable isotope determination.
S1 Table. Number of samples available for stable isotopes in plasma and red blood cells (n = 112) and fatty acids in adipose tissue (n = 83) of female polar bears from Svalbard (2012–2013).
Samples are then sorted by year, sampling location, breeding status and season.
S2 Table. Average (± standard deviation, SD) mass % of fatty acids in polar bear adipose tissue.
Samples were collected from 83 females in Svalbard archipelago in 2012 and 2013. FA in bold are those selected for statistical analyses.
S3 Table. List of the 15 candidate models used for model selection.
We used model selection via AICc to determine the best predictors of nitrogen and carbon stable isotope values in plasma and red blood cells, and fatty acid composition in adipose tissue (using principal components values).
We carried out fieldwork in Svalbard under permit from the Governor of Svalbard. We thank the helicopter and Lance crews and Rupert Krapp for logistic help. We want to thank Karen Lone and Jade Vacquié-Garcia for their help in QGIS, Anette Wold and Matthew Rogers for their help in stable isotopes preparation and determination, and Nigel G. Yoccoz for his help in statistical analyses. S.D. Petersen and one anonymous reviewer provided useful comments that improved earlier versions of this manuscript.
Conceived and designed the experiments: HR JA. Performed the experiments: HR JA MA SB GWT JMW. Analyzed the data: ST. Contributed reagents/materials/analysis tools: HR JA. Wrote the paper: ST HR SB DE JA MA GWT JMW.
- 1. Simmonds I. Comparing and contrasting the behaviour of Arctic and Antarctic sea ice over the 35 year period 1979–2013. Ann Glaciol. 2015;56: 18–28.
- 2. Hansen J, Sato M, Ruedy R, Lo K, Lea DW, Medina-Elizade M. Global temperature change. Proc Natl Acad Sci. 2006;103: 14288–14293. pmid:17001018
- 3. Cavalieri D, Parkinson C. Arctic sea ice variability and trends, 1979–2010. Cryosphere. 2012;6: 881–889.
- 4. Wang M, Overland JE. A sea ice free summer Arctic within 30 years: An update from CMIP5 models. Geophys Res Lett. 2012;39: L18501.
- 5. Laidre KL, Stern H, Kovacs KM, Lowry L, Moore SE, Regehr EV, et al. Arctic marine mammal population status, sea ice habitat loss, and conservation recommendations for the 21st century. Conserv Biol. 2015;29: 724–737. pmid:25783745
- 6. Moore SE, Huntington HP. Arctic marine mammals and climate change: impacts and resilience. Ecol Appl. 2008;18: S157–S165. pmid:18494369
- 7. Víkingsson GA, Elvarsson BÞ, Ólafsdóttir D, Sigurjónsson J, Chosson V, Galan A. Recent changes in the diet composition of common minke whales (Balaenoptera acutorostrata) in Icelandic waters. A consequence of climate change? Mar Biol Res. 2014;10: 138–152.
- 8. Braune BM, Gaston AJ, Hobson KA, Grant Gilchrist H, Mallory ML. Changes in trophic position affect rates of contaminant decline at two seabird colonies in the Canadian Arctic. Ecotoxicol Environ Saf. 2015;115: 7–13. pmid:25666731
- 9. McKinney Ma, Pedro S, Dietz R, Sonne C, Fisk At, Roy D, et al. A review of ecological impacts of global climate change on persistent organic pollutant and mercury pathways and exposures in arctic marine ecosystems. 2016;
- 10. Wiig Ø, Aars J, Born EW. Effects of climate change on polar bears. Sci Prog. 2008;91: 151–173. pmid:18717367
- 11. Kovacs KM, Lydersen C, Overland JE, Moore SE. Impacts of changing sea-ice conditions on Arctic marine mammals. Mar Biodivers. 2010;41: 181–194.
- 12. Derocher AE, Wiig Ø, Andersen M. Diet composition of polar bears in Svalbard and the western Barents Sea. Polar Biol. 2002;25: 448–452.
- 13. Thiemann GW, Iverson SJ, Stirling I. Polar bear diets and arctic marine food webs: insights from fatty acid analysis. Ecol Monogr. 2008;78: 591–613.
- 14. Rogers MC, Peacock E, Simac K, O’Dell MB, Welker JM. Diet of female polar bears in the southern Beaufort Sea of Alaska: evidence for an emerging alternative foraging strategy in response to environmental change. Polar Biol. 2015;38: 1035–1047.
- 15. Stempniewicz L. The polar bear Ursus maritimus feeding in a seabird colony in Frans Josef Land. Polar Res. 1993;12: 33–36.
- 16. Stempniewicz L. Polar bear predatory behaviour toward molting barnacle geese and nesting glaucous gulls on spitsbergen. Arctic. 2006;59: 247–251.
- 17. Dyck MG, Romberg S. Observations of a wild polar bear (Ursus maritimus) successfully fishing Arctic charr (Salvelinus alpinus) and Fourhorn sculpin (Myoxocephalus quadricornis). Polar Biol. 2007;30: 1625–1628.
- 18. Smith PA, Elliott KH, Gaston AJ, Gilchrist HG. Has early ice clearance increased predation on breeding birds by polar bears? Polar Biol. 2010;33: 1149–1153.
- 19. Iversen M, Aars J, Haug T, Alsos IG, Lydersen C, Bachmann L, et al. The diet of polar bears (Ursus maritimus) from Svalbard, Norway, inferred from scat analysis. Polar Biol. 2013;36: 561–571.
- 20. Gormezano LJ, Rockwell RF. What to eat now? Shifts in polar bear diet during the ice-free season in western Hudson Bay. Ecol Evol. 2013;3: 3509–3523. pmid:24223286
- 21. Stempniewicz L, Kidawa D, Barcikowski M, Iliszko L. Unusual hunting and feeding behaviour of polar bears on Spitsbergen. Polar Rec. 2014;50: 216–219.
- 22. Iverson SA, Gilchrist HG, Smith PA, Gaston AJ, Forbes MR. Longer ice-free seasons increase the risk of nest depredation by polar bears for colonial breeding birds in the Canadian Arctic. Proc R Soc Lond B Biol Sci. 2014;281: 20133128.
- 23. Aars J, Andersen M, Brenière A, Blanc S. White-beaked dolphins trapped in the ice and eaten by polar bears. Polar Res. 2015;34.
- 24. Gormezano LJ, Rockwell RF. The energetic value of land-based foods in western hudson bay and their potential to alleviate energy deficits of starving adult male polar bears. PLoS ONE. 2015;10: e0128520. pmid:26061693
- 25. Rode KD, Regehr EV, Douglas DC, Durner G, Derocher AE, Thiemann GW, et al. Variation in the response of an Arctic top predator experiencing habitat loss: feeding and reproductive ecology of two polar bear populations. Glob Change Biol. 2014;20: 76–88.
- 26. Prop J, Aars J, Bårdsen B-J, Hanssen SA, Bech C, Bourgeon S, et al. Climate change and the increasing impact of polar bears on bird populations. Interdiscip Clim Stud. 2015;3: 33.
- 27. Rode KD, Robbins CT, Nelson L, Amstrup SC. Can polar bears use terrestrial foods to offset lost ice-based hunting opportunities? Front Ecol Environ. 2015;13: 138–145.
- 28. Vongraven D, Aars J, Amstrup S, Atkinson SN, Belikov S, Born EW, et al. A circumpolar monitoring framework for polar bears. Ursus. 2012;23: 1–66.
- 29. Lønø O. The Polar bear (Ursus maritimus Phipps) in the Svalbard area. 118. 1970; Available: http://brage.bibsys.no/xmlui/handle/11250/173883
- 30. Grahl-Nielsen O, Andersen M, Derocher AE, Lydersen C, Wiig O, Kovacs KM. Fatty acid composition of the adipose tissue of polar bears and of their prey: ringed seals, bearded seals and harp seals. Mar Ecol Prog Ser. 2003;265: 275–282.
- 31. Durner GM, Douglas DC, Nielson RM, Amstrup SC, McDonald TL, Stirling I, et al. Predicting 21st-century polar bear habitat distribution from global climate models. Ecol Monogr. 2009;79: 25–58.
- 32. Freitas C, Kovacs KM, Andersen M, Aars J, Sandven S, Skern-Mauritzen M, et al. Importance of fast ice and glacier fronts for female polar bears and their cubs during spring in Svalbard, Norway. Mar Ecol Prog Ser. 2012;447: 289–304.
- 33. Amstrup SC. Polar bear, Ursus maritimus. Wild Mamm N Am Biol Manag Conserv. 2003; Available: http://pubs.er.usgs.gov/publication/87203
- 34. Pilfold NW, Derocher AE, Richardson E. Influence of intraspecific competition on the distribution of a wide-ranging, non-territorial carnivore. Glob Ecol Biogeogr. 2014;23: 425–435.
- 35. Hobson KA. Tracing origins and migration of wildlife using stable isotopes: a review. Oecologia. 1999;120: 314–326.
- 36. Iverson SJ, Field C, Don Bowen W, Blanchard W. Quantitative fatty acid signature analysis: a new method of estimating predator diets. Ecol Monogr. 2004;74: 211–235.
- 37. Wang SW, Springer AM, Budge SM, Horstmann-Dehn L, Quakenbush LT, Wooller MJ. Carbon sources and trophic relationships of ice seals during recent environmental shifts in the Bering Sea. Ecol Appl. 2015;
- 38. Budge SM, Iverson SJ, Koopman HN. Studying trophic ecology in marine ecosystems using fatty acids: a primer on analysis and interpretation. Mar Mammal Sci. 2006;22: 759–801.
- 39. Hobson KA, Schell DM, Renouf D, Noseworthy E. Stable carbon and nitrogen isotopic fractionation between diet and tissues of captive seals: implications for dietary reconstructions involving marine mammals. Can J Fish Aquat Sci. 1996;53: 528–533.
- 40. Watt CA, Ferguson SH. Fatty acids and stable isotopes (δ13C and δ15N) reveal temporal changes in narwhal (Monodon monoceros) diet linked to migration patterns. Mar Mammal Sci. 2015;31: 21–44.
- 41. Thiemann G, Iverson SJ, Stirling I. Seasonal, sexual and anatomical variability in the adipose tissue of polar bears (Ursus maritimus). J Zool. 2006;269: 65–76.
- 42. Stirling I, Spencer C, Andriashek D. Immobilization of polar bears (Ursus maritimus) with Telazol® in the Canadian Arctic. J Wildl Dis. 1989;25: 159–168. pmid:2716095
- 43. Derocher AE, Wiig Ø. Postnatal growth in body length and mass of polar bears (Ursus maritimus) at Svalbard. J Zool. 2002;256: 343–349.
- 44. Vinje T, Kvambekk ÅS. Barents Sea drift ice characteristics. Polar Res. 1991;10: 59–68.
- 45. Pavlova O, Pavlov V, Gerland S. The impact of winds and sea surface temperatures on the Barents Sea ice extent, a statistical approach. J Mar Syst. 2014;130: 248–255.
- 46. Hop H, Poltermann M, Lønne OJ, Falk-Petersen S, Korsnes R, Budgell WP. Ice amphipod distribution relative to ice density and under-ice topography in the northern Barents Sea. Polar Biol. 2000;23: 357–367.
- 47. Budge SM, Penney SN, Lall SP. Estimating diets of Atlantic salmon (Salmo salar) using fatty acid signature analyses; validation with controlled feeding studies. Can J Fish Aquat Sci. 2012;69: 1033–1046.
- 48. R Core Team. R: A Language and Environment for Statistical Computing. Vienna, Austria; 2015. Available: http://www.R-project.org.
- 49. Dray AB, Dufour S. The ade4 package: implementing the duality diagram for ecologists. J Stat Softw. 2007;22: 1–20.
- 50. Pinheiro J, Bates D, Debroy S, Sarkar D, R core team. nlme: Linear and Nonlinear Mixed Effects Models [Internet]. 2015. Available: http://CRAN.R-project.org/package=nlme.
- 51. Burnham KP, Anderson DR, editors. Model selection and multimodel inference. New York, NY: Springer New York; 2004. Available: http://link.springer.com/10.1007/b97636.
- 52. Mazerolle MJ. AICcmodavg: Model selection and multimodel inference based on (Q)AIC(c) [Internet]. 2015. Available: http://CRAN.R-project.org/package=AICcmodavg.
- 53. Lenth RV, Hervé M. lsmeans: Least-Squares Means. 2015. Available: http://CRAN.R-project.org/package=lsmeans.
- 54. Loseto LL, Stern GA, Connelly TL, Deibel D, Gemmill B, Prokopowicz A, et al. Summer diet of beluga whales inferred by fatty acid analysis of the eastern Beaufort Sea food web. J Exp Mar Biol Ecol. 2009;374: 12–18.
- 55. Speake BK, Surai PF, Bortolotti GR. Fatty acid profiles of yolk lipids of five species of wild ducks (Anatidae) differing in dietary preference. J Zool. 2002;257: 533–538.
- 56. Akearok J. Stable isotope (δ15N and δ13C) and fatty acid profiles in eggs of three species of marine birds nesting sympatrically in the high arctic. M.Sc. Thesis, Carleton University. 2011. Available: https://curve.carleton.ca/.
- 57. Falk-Petersen S, Haug T, Nilssen KT, Wold A, Dahl TM. Lipids and trophic linkages in harp seal (Phoca groenlandica) from the eastern Barents Sea. Polar Res. 2004;23: 43–50.
- 58. Falk-Petersen S, Haug T, Hop H, Nilssen KT, Wold A. Transfer of lipids from plankton to blubber of harp and hooded seals off East Greenland. Deep Sea Res Part II Top Stud Oceanogr. 2009;56: 2080–2086.
- 59. Raclot T. Selective mobilization of fatty acids from adipose tissue triacylglycerols. Prog Lipid Res. 2003;42: 257–288. pmid:12689620
- 60. Gastwirth JL, Gel YR, Hui W. W, Lyubchich V, Miao W, Noguchi K. lawstat: Tools for Biostatistics, Public Policy, and Law. R package version 3.0. 2015. Available: http://CRAN.R-project.org/package=lawstat.
- 61. Krafft BA, Kovacs KM, Andersen M, Aars J, Lydersen C, Ergon T, et al. Abundance of Ringed Seals (Pusa hispida) in the Fjords of Spitsbergen, Svalbard, During the Peak Molting Period. Mar Mammal Sci. 2006;22: 394–412.
- 62. Merkel B, Lydersen C, Yoccoz NG, Kovacs KM. The world’s northernmost harbour seal population–how many are there? PLoS ONE. 2013;8: e67576. pmid:23844035
- 63. Kovacs KM, Aars J, Lydersen C. Walruses recovering after 60+ years of protection in Svalbard, Norway. Polar Res. 2014;33.
- 64. Iverson SJ, Springer AM, Kitaysky AS. Seabirds as indicators of food web structure and ecosystem variability: qualitative and quantitative diet analyses using fatty acids. Mar Ecol-Prog Ser-. 2007;352: 235.
- 65. Budge SM, Wooller MJ, Springer AM, Iverson SJ, McRoy CP, Divoky GJ. Tracing carbon flow in an arctic marine food web using fatty acid-stable isotope analysis. Oecologia. 2008;157: 117–129. pmid:18481094
- 66. Hobson KA, Fisk A, Karnovsky N, Holst M, Gagnon J- M, Fortier M. A stable isotope (δ13C, δ15N) model for the North Water food web: implications for evaluating trophodynamics and the flow of energy and contaminants. Deep Sea Res Part II Top Stud Oceanogr. 2002;49: 5131–5150.
- 67. Derocher AE, Wiig Ø, Bangjord G. Predation of Svalbard reindeer by polar bears. Polar Biol. 2000;23: 675–678.
- 68. Fox AD, Ebbinge BS, Mitchell C, Heinicke T, Aarvak T, Colhoun K, et al. Current estimates of goose population sizes in western Europe, a gap analysis and an assessment of trends. Ornis Svec. 2010;20: 115–127.
- 69. Fox A, Francis IS, Bergersen E. Diet and habitat use of Svalbard Pink-footed Geese Anser brachyrhynchus during arrival and pre-breeding periods in Adventdalen. ARDEA-Wagening-. 2006;94: 691.
- 70. Dehn L-A, Follmann EH, Thomas DL, Sheffield GG, Rosa C, Duffy LK, et al. Trophic relationships in an Arctic food web and implications for trace metal transfer. Sci Total Environ. 2006;362: 103–123. pmid:16387350
- 71. Kovacs KM, Lydersen C, Gjertz I. Birth-site characteristics and prenatal molting in bearded seals (Erignathus barbatus). J Mammal. 1996;77: 1085–1091.
- 72. Lydersen C. Status and biology of ringed seals (Phoca hispida) in Svalbard. NAMMCO Sci Publ. 1998;1: 46–62.
- 73. Carlens H, Lydersen C, Krafft BA, Kovacs KM. Spring haul-out behavior of ringed seals (Pusa hispida) in Kongsfjorden, Svalbard. Mar Mammal Sci. 2006;22: 379–393.
- 74. Lydersen C, Assmy P, Falk-Petersen S, Kohler J, Kovacs KM, Reigstad M, et al. The importance of tidewater glaciers for marine mammals and seabirds in Svalbard, Norway. J Mar Syst. 2014;129: 452–471.
- 75. Smith TG, Lydersen C. Availability of suitable land-fast ice and predation as factors limiting ringed seal populations, Phoca hispida, in Svalbard. Polar Res. 1991;10: 585–594.
- 76. Folkow LP, Mårtensson P-E, Blix AS. Annual distribution of hooded seals (Cystophora cristata) in the Greenland and Norwegian seas. Polar Biol. 1996;16: 179–189.
- 77. Nordøy ES, Folkow LP, Potelov V, Prischemikhin V, Blix AS. Seasonal distribution and dive behaviour of harp seals (Pagophilus groenlandicus) of the White Sea–Barents Sea stock. Polar Biol. 2008;31: 1119–1135.
- 78. Messier F, Taylor MK, Ramsay MA. Seasonal activity patterns of female polar bears (Ursus maritimus) in the Canadian Arctic as revealed by satellite telemetry. J Zool. 1992;226: 219–229.
- 79. Atkinson SN, Nelson RA, Ramsay MA. Changes in the body composition of fasting polar bears (Ursus maritimus): the effect of relative fatness on protein conservation. Physiol Zool. 1996; 304–316.
- 80. Hilderbrand GV, Farley SD, Robbins CT, Hanley TA, Titus K, Servheen C. Use of stable isotopes to determine diets of living and extinct bears. Can J Zool. 1996;74: 2080–2088.
- 81. Ramos R, González-Solís J. Trace me if you can: the use of intrinsic biogeochemical markers in marine top predators. Front Ecol Environ. 2012;10: 258–266.
- 82. Fuller BT, Fuller JL, Sage NE, Harris DA, O’Connell TC, Hedges REM. Nitrogen balance and δ15N: why you’re not what you eat during nutritional stress. Rapid Commun Mass Spectrom. 2005;19: 2497–2506. pmid:16106342
- 83. Cherel Y, Hobson KA, Bailleul F, Groscolas R. Nutrition, physiology, and stable isotopes: new information from fasting and molting penguins. Ecology. 2005;86: 2881–2888.
- 84. Hatch KA. The use and application of stable isotope analysis to the study of starvation, fasting, and nutritional stress in animals. In: McCue MD, editor. Comparative physiology of fasting, starvation, and food limitation. Springer Berlin Heidelberg; 2012. pp. 337–364. Available: http://link.springer.com/chapter/10.1007/978-3-642-29056-5_20.
- 85. Fuller BT, Fuller JL, Sage NE, Harris DA, O’Connell TC, Hedges REM. Nitrogen balance and δ15N: why you’re not what you eat during pregnancy. Rapid Commun Mass Spectrom. 2004;18: 2889–2896. pmid:15517531
- 86. Derocher AE, Andriashek D, Arnould JPY. Aspects of milk composition and lactation in polar bears. Can J Zool. 1993;71: 561–567.
- 87. Arnould JPY, Ramsay MA. Milk production and milk consumption in polar bears during the ice-free period in western Hudson Bay. Can J Zool. 1994;72: 1365–1370.
- 88. Pilfold NW, Derocher AE, Stirling I, Richardson E, Andriashek D. Age and sex composition of seals killed by polar bears in the eastern beaufort sea. PLoS ONE. 2012;7: e41429. pmid:22829949
- 89. Karanth KU, Sunquist ME. Prey selection by tiger, leopard and dhole in tropical forests. J Anim Ecol. 1995; 439–450.
- 90. Bothma J du P, Coertze RJ. Motherhood increases hunting success in southern Kalahari leopards. J Mammal. 2004;85: 756–760.
- 91. Ruiz-Olmo J, Margalida A, Batet A. Use of small rich patches by Eurasian otter (Lutra lutra L.) females and cubs during the pre-dispersal period. J Zool. 2005;265: 339–346.
- 92. Ruiz-Olmo J, Batet A, Jiménez J, Martínez D. Habitat selection by female otters with small cubs in freshwater habitats in northeast Spain. Lutra. 2005;48: 45.
- 93. Hunter JS, Durant SM, Caro TM. To flee or not to flee: predator avoidance by cheetahs at kills. Behav Ecol Sociobiol. 2006;61: 1033–1042.
- 94. Mauritzen M, Derocher AE, Wiig Ø. Space-use strategies of female polar bears in a dynamic sea ice habitat. Can J Zool. 2001;79: 1704–1713.
- 95. Mauritzen M, Derocher AE, Wiig Ø, Belikov SE, Boltunov AN, Hansen E, et al. Using satellite telemetry to define spatial population structure in polar bears in the Norwegian and western Russian Arctic. J Appl Ecol. 2002;39: 79–90.
- 96. Lone K, Aars J, Ims RA. Site fidelity of Svalbard polar bears revealed by mark-recapture positions. Polar Biol. 2012;36: 27–39.