Circumpolar contaminant concentrations in polar bears (Ursus maritimus) and potential population-level effects

Environmental Research 151 (2016) 50–57

Contents lists available at ScienceDirect

Environmental Research journal homepage: www.elsevier.com/locate/envres

Circumpolar contaminant concentrations in polar bears (Ursus maritimus) and potential population-level effects R.J.M. Nuijten a,b,n, A.J. Hendriks a, B.M. Jenssen c,d, A.M. Schipper a,e a

Department of Environmental Science, Institute for Water and Wetland Research (IWWR), Radboud University (RU), NL-6500 GL Nijmegen, The Netherlands Department of Animal Ecology, Netherlands Institute of Ecology (NIOO-KNAW), 7608 PB Wageningen, The Netherlands c Department of Biology, Norwegian University of Science and Technology, NO-7491 Trondheim, Norway d Department of Arctic Technology, The University Centre in Svalbard, Longyearbyen, Norway e PBL Netherlands Environmental Assessment Agency, PO Box 303, 3720 AH Bilthoven, The Netherlands b

art ic l e i nf o

a b s t r a c t

Article history: Received 19 May 2016 Received in revised form 8 July 2016 Accepted 15 July 2016

Polar bears (Ursus maritimus) currently receive much attention in the context of global climate change. However, there are other stressors that might threaten the viability of polar bear populations as well, such as exposure to anthropogenic pollutants. Lipophilic organic compounds bio-accumulate and biomagnify in the food chain, leading to high concentrations at the level of top-predators. In Arctic wildlife, including the polar bear, various adverse health effects have been related to internal concentrations of commercially used anthropogenic chemicals like PCB and DDT. The extent to which these individual health effects are associated to population-level effects is, however, unknown. In this study we assembled data on adipose tissue concentrations of ∑PCB, ∑DDT, dieldrin and ∑PBDE in individual polar bears from peer-reviewed scientific literature. Data were available for 14 out of the 19 subpopulations. We found that internal concentrations of these contaminants exceed threshold values for adverse individual health effects in several subpopulations. In an exploratory regression analysis we identified a clear negative correlation between polar bear population density and sub-population specific contaminant concentrations in adipose tissue. The results suggest that adverse health effects of contaminants in individual polar bears may scale up to population-level consequences. Our study highlights the need to consider contaminant exposure along with other threats in polar bear population viability analyses. & 2016 Elsevier Inc. All rights reserved.

Keywords: Polar bear Ursus maritimus Population effects Pollution Bioaccumulation Arctic PCB DDT Dieldrin PBDE

1. Introduction Arctic ecosystems are threatened by multiple anthropogenic stressors, such as global warming, increased human activity and pollution (Huntington, 2009; Letcher et al., 2010; Bennett et al., 2015). Combined, these factors may constitute a threat to Arctic wildlife species, in particular endemic top predators with low population numbers, such as the polar bear (Ursus maritimus) (Jenssen, 2006; Stirling and Derocher 2012; Jenssen et al., 2015). Although climate change has received by far the most attention with respect to viability of polar bear populations, it is likely that also other stressors, such as environmental pollutants, may threaten the polar bear (Sonne, 2010; Jenssen et al., 2015; Dietz et al., 2015). The lipophilicity of many persistent organic pollutants (POPs) causes them to accumulate in tissue of animals (MacKay n Corresponding author at: Department of Animal Ecology, Netherlands Institute of Ecology (NIOO-KNAW), 7608 PB Wageningen, The Netherlands. E-mail address: [email protected] (R.J.M. Nuijten).

http://dx.doi.org/10.1016/j.envres.2016.07.021 0013-9351/& 2016 Elsevier Inc. All rights reserved.

and Fraser, 2000; Borgå et al. 2004). Due to trophic transfer the chemicals are transported along the food chain and the highest internal concentrations are generally reached in the highest trophic levels of a certain system (Skaare et al., 2000; Borgå et al. 2004; Sørmo et al., 2006). In the Arctic food chain high concentrations of POPs have been found in the apex predator, the polar bear (Lie et al. 2003; Verreault et al., 2005; McKinney et al., 2011). The dependency on a diet high in fat, their position in the food chain and the extreme seasonal dynamics, which force them to rely on their lipid stores for a large part of the year (Stirling and Derocher, 1993; Borgå et al. 2004), make polar bears particularly vulnerable to deleterious effects of anthropogenic pollution. Many studies have focused on the potential adverse health effects of contaminants in wildlife, fish and humans, including Arctic species (Bustnes et al., 2003; Van Oostdam et al., 2005; Jørgensen et al., 2006; Miljeteig et al., 2012). In polar bears, organochlorine contaminants have been associated with a variety of effects, including endocrine disruption of thyroid hormones, sex hormones and stress hormones, as well as effects on vitamin levels, liver, kidney and thyroid gland morphology, decreased bone

R.J.M. Nuijten et al. / Environmental Research 151 (2016) 50–57

51

CS

SB LP

NB

MC WH GB SH

VM LS

AB NW KB

FB

KS

BB

DS

BS

EG

Esri Fig. 1. Geographic location of the 19 currently recognized polar bear subpopulations (see Paetkau et al. (1999) and Peacock et al. (2015) for detailed information on data underlying this subdivision). WH: Western Hudson Bay, SH: Southern Hudson Bay, DS: Davis Strait, BB: Baffin Bay, FB: Foxe Basin, KB: Kane Basin, NW: Norwegian Bay, LS: Lancaster Sound, GB: Gulf of Boothia, MC: M’Clintock Channel, VM: Viscount Melville Sound, NB: Northern Beaufort Sea, SB: Southern Beaufort Sea, CS: Chukchi Sea, LP: Laptev Sea, KS: Kara Sea, BS: Barents Sea, EG: East Greenland (PBSG, 2015).

mineral density and impairment of both the reproductive and immune system (Haave et al., 2003; Oskam et al., 2003; Braathen et al., 2004; Oskam et al., 2004; Sonne, 2010; Gustavson et al., 2015). Recently, Dietz et al. (2015) indicated that although a temporal decrease in the levels of POPs, including PCBs, has been observed in polar bears during the past two decades, concentrations are still above toxic thresholds (i.e. risk quotient (RQ) 41) in all 11 subpopulations examined for reproductive traits, immunotoxic and carcinogenic effects (for subpopulations see Fig. 1). PCBs were the main contributor (Dietz et al., 2015). However, these studies have all focused on individual health effects only. There have been no efforts to examine if pollutants affect polar bears at the population level (Jenssen et al., 2015; Letcher et al., 2010). To assess the threat of contaminant exposure on the population level, effects need to be assessed on population-relevant endpoints such as survival, reproductive success or population density. Population effects of environmental pollution are of increasing scientific interest (Forbes et al., 2016), however for Arctic species only

a few published studies exist. In glaucous gull (Larus hyperboreus), a 10-fold increase in oxychlordane levels in the blood was found to reduce adult survival probability up to 29% (Bustnes et al., 2003). Later studies on organochlorine levels in the same species found evidence for sex-dependent threshold values for mortality in adult birds and an association between organochlorine level of the mother and hatching sex ratio (Erikstad et al., 2011, 2013). The aim of the present study was threefold: 1) to assemble data on internal contaminant concentrations in polar bears across the entire Arctic, 2) to compare these contaminant concentrations with toxicity thresholds for adverse individual health effects, and 3) to investigate potential associations between subpopulationspecific contaminant concentrations and population-level characteristics, based on the data that is currently available. In addition to subpopulation-specific contaminant concentrations, the role of other anthropogenic stressors, such as relevant global warming proxies (ice coverage), human population density and harvest rate were examined.

52

R.J.M. Nuijten et al. / Environmental Research 151 (2016) 50–57

2. Methods 2.1. Data collection A literature search in Google Scholar based on various combinations of the keywords “polar bear”, “Ursus maritimus”, “contaminant”, “pesticide” and “[internal/tissue] concentration”, provided peer-reviewed articles reporting measured concentration of POPs in polar bear tissue. Relevant cross-references were also included. When multiple studies published results on the same samples, the results were included only once. Based on data availability for the 19 different polar bear subpopulations, we focused on adipose tissue concentrations (ng
g  1 lipid weight) of the typical legacy contaminants dieldrin, polychlorinated biphenyls (∑PCBs) and dichlorodiphenyltrichloroethane (∑DDT). To cover more recent contamination, concentrations (ng
g  1 lipid weight) of polybrominated diphenyl ethers (∑PBDEs) were also included, despite information only being available for 11 subpopulations. Toxicity reference values (TRVs) for Arctic top-predators for several different effect categories were reviewed by Sonne (2010). From this review we extracted TRVs for ∑PCB, dieldrin, ∑DDT and ∑PBDE concentrations in adipose tissue. Where possible we used TRVs specific to polar bears, otherwise values for sledge dogs (Canis familiaris) or harbour seals (Phoca vitulina) were used (Sonne, 2010). Data on polar bear population variables were collected from peer-reviewed literature based on the keywords “survival”, “reproduction”, “litter size/frequency/interval”, “population viability”, “population size/density”. In addition the IUCN Polar Bear Specialist Group (PBSG) database was consulted (PBSG, 2015). From the search results obtained, polar bear subpopulation density (number of polar bears per 1000 km2), mean litter size and adult female survival rate were selected as the response variables. Data on other relevant endpoints were not available for a sufficient number of subpopulations. Population density of each subpopulation was obtained by dividing the most recent population count by the area of the subpopulation (PBSG, 2015). Other factors that may influence polar bear populations include global warming (Regehr et al., 2007; Stirling and Derocher, 2012), harvest (Taylor et al., 2005; Taylor et al., 2008) and human presence in the Arctic (Cardillo et al., 2004). Therefore we added data on sea ice, harvest and human population density as additional variables in our analyses. Harvest data per subpopulation (n
yr  1; actual 5-year mean) were also collected from the PBSG database (PBSG, 2015). Ice cover maps for every day between January 1st, 1999 until December 31st, 2013 were downloaded from the IMS Daily Northern Hemisphere database of the National Snow and Ice Data Centre (NSIDC), Boulder, USA (NSIDC, 2008). Human population density was extracted from the ‘Gridded Population of the World’ map (version 3) obtained from the Socioeconomic Data and Applications Centre (SEDAC) (CIESIN, 2005). 2.2. Data treatment Reported concentrations were assigned to polar bear subpopulations based on reported sampling locations and the polar bear subpopulation map (Fig. 1). To reduce a potential bias in tissue concentration based on age, we included samples from adult polar bears only. Because samples were approximately equally distributed between males and females, we pooled the samples irrespective of sex. A weighted average was calculated of all reported concentrations per subpopulation per pollutant. Concentrations were log10-transformed and weighing was based on the number of individual polar bear samples per reported concentration. It is known that concentrations of some contaminants

show a temporal decrease in the Arctic, and also that concentrations of lipophilic chemicals vary seasonally in polar bears due to fasting in summer (Dietz et al., 2007; Polischuk et al., 2002; Riget et al., 2010; Muir and de Wit, 2010). To identify potential biases in our data caused by sampling season or year, we related the concentrations per subpopulation to the proportion of the samples taken in winter and spring, as well as to the average sampling year (Figs. S1 and S2). For none of the contaminants, internal concentrations were related to sampling season or sampling year (p 40.05). Ice cover maps covered the whole Northern Hemisphere. To obtain subpopulation-specific maps of ice cover per day (1999– 2013), the ice cover data were extracted based on the polar bear subpopulation map (Fig. 1) using the geographic information system ArcMap 10.1. We calculated the variables ‘sea ice free period’ and ‘summer sea ice cover’ from the maps as an average over the years 1999–2013. Sea ice free period (days) has previously been associated with polar bear viability, as polar bears have been observed to abandon the ice to forage ashore below a certain threshold (Stirling et al., 1999; Regehr et al., 2010). Here we defined the sea ice free period as the number of days sea ice cover was o50% of the maximum for that year. The minimum summer sea ice cover (% of total potential ice cover in the respective area), which we defined as the average sea ice cover (%) on September 15th over the years 1999–2013, is also often used as a proxy for climate change and might constrain polar bear viability (Stroeve et al., 2012). The mean human population density (humans per 1000 km  2) in each of the polar bear subpopulation areas was calculated in ArcMAP 10.1 by overlaying the SEDAC data with the subpopulation map. 2.3. Data analyses Averaged internal contaminant concentrations per polar bear subpopulation were compared to TRVs for individual health effects reported in literature (Sonne, 2010). In addition, we explored whether the individual health effects could lead to effects on the population level by applying linear modelling between population variables and internal contaminant concentrations in the statistical program R version 3.2.2 (R Development Core Team). Polar bear population density, litter size and adult female survival were used as the response variables; ∑PCBs, ∑DDT and dieldrin concentrations in adipose tissue of adult polar bears were included as predictor variables (data available for 14 out of 19 subpopulations; n¼14). Additionally, harvest, summer sea ice cover, sea ice free period and human population density per subpopulation were included as predictor variables. Because of the smaller sample size for ∑PBDEs (n ¼11; see Table 1), we performed a separate analysis with ∑PBDEs included as a predictor variable. All predictor variables were standardized to zero mean and unit variance. Variance inflation factors (VIFs) were calculated among predictor variables to check for multi-collinearity. Based on a VIF threshold of 5, the variable sea ice free period (VIF ¼13.9) was left out of the final regression model (O’Brien, 2007). All other predictors were included with a VIF o5.0. We conducted multiple linear regression analyses based on all possible combinations of predictors (limited to a maximum of two predictors per model) and calculated Akaike information criterion corrected for small sample sizes (AICc) to compare between output models. In addition, we calculated model parameters by averaging across all best-supported models (i.e. all models where ΔAICc o2). In total, six separate regression analyses were conducted, two for each of the response variables (one with ΣPCB, ΣDDT and dieldrin and one with ΣPBDE also included). The samples sizes for the analyses with population density and mean litter size as the response variables were 14 (ΣPCB, ΣDDT and dieldrin) and 11 (ΣPBDE also included). The analyses with survival

R.J.M. Nuijten et al. / Environmental Research 151 (2016) 50–57

as the response variable had a sample size of 9 subpopulations (Table S1).

53

subpopulations exceeded the value for renal lesions (Fig. 2). 3.3. Potential population-level effects

3. Results 3.1. Tissue concentrations In total 13 peer-reviewed articles that reported data on polar bear adipose tissue concentrations of the contaminants ∑PCB, dieldrin, ∑DDT and ∑PBDE were included in this study (Norstrom et al., 1988, 1998; Bernhoft et al., 1997; Kucklick et al., 2002; Polischuk et al., 2002; Sandala et al., 2004; Muir et al., 2006; Sørmo et al., 2006; Verreault et al., 2006; Dietz et al., 2007; Gebbink et al., 2008; McKinney et al., 2011). Sampling dates ranged from 1982 to 2008. Concentrations per subpopulation are presented in Fig. 2 and Table S1 (mean 7sd). ∑PCB concentrations in polar bear adipose tissue ranged from 1.4
103 ng
g  1 lipid weight (lw) in Gulf of Boothia to 1.2
104 ng
g  1 lw in the Barents Sea subpopulation. ∑DDT concentrations ranged from 5.5
101 ng
g  1 lw, also in Gulf of Boothia, to 9.4
102 ng
g  1 lw in the Foxe Basin subpopulation. Dieldrin concentrations were lowest in polar bears from the Chukchi Sea and highest in those from M’Clintock Channel (5.3
101 and 2.4
102 ng
g  1 lw, respectively). The lowest ∑PBDE concentrations were also found in the Chukchi Sea subpopulation, while the highest were measured in Southern Hudson Bay polar bears (5.8 and 7.8
101 ng
g  1 lw, respectively). 3.2. Toxicity reference values Most TRVs were identified for ∑PCB (Sonne, 2010). In almost all subpopulations for which data were available, ∑PCB concentrations in adipose tissue exceeded the TRVs for liver lesions, decreased baculum bone mineral density (BMD) and deficiencies in the immune-response (13 of the 14 subpopulations). In addition, concentrations in several subpopulations were found to exceed the threshold for decreased testes size and decreased skull BMD (5 out of 14). The highest TRV for ∑PCB, which represented immunosuppression in harbour seals, might only be reached in polar bears from the Barents Sea subpopulation. Fewer TRVs were available for ∑DDT, dieldrin and ∑PBDE. However, for ∑DDT, concentrations in nine out of 14 subpopulations exceeded the TRV for liver lesions and for one subpopulation (Foxe Basin) the concentration exceeded the level for decreased skull BMD. Dieldrin TRVs for renal hyperplasia and increased risk of osteoporosis were exceeded in 10 of the 14 subpopulations for which data were available. For ∑PBDE, concentrations in three out of 11 subpopulations exceeded the TRV for decreased uterus size and two

The results suggested that a model containing only ∑PCB or only ∑DDT as a predictor was the most parsimonious in explaining the variance in polar bear density between subpopulations (Tables 1 and 2 respectively). These best-supported models accounted for 43% and 61% (R2) of the variation in density between subpopulations, respectively. Several other models were within ΔAICc of 2.0, including the predictor variables dieldrin, human population density and ∑PBDE (Tables 1 and 2). Model-averaged parameter estimates (Table 3) showed negative relationships for all predictors, indicating lower population density in subpopulations with higher concentrations of ∑PCB, ∑DDT, ∑PBDE and dieldrin and higher human population density. For litter size, the null-model (comprised of an intercept only) was the most parsimonious in both analyses (Table S2 and S3). In the analyses for adult female survival, the most parsimonious models also only contained an intercept. However, here, another model with the predictor ‘September ice’ included was within a ΔAICc of 2.0 (Table S4 and S5). Model-averaged parameter estimates indicated a negative relationship to polar bear population density (Table S6).

4. Discussion 4.1. Tissue concentrations Large differences in concentrations were present between subpopulations (Fig. 2). Possible explanations include differences in transport routes and proximity to sources, or differences in trophic interactions among regions (Kleivane et al., 2000; Kannan et al., 2005; McKinney et al., 2009; St. Louis et al., 2011). For example in the Barents sea, polar bears have been reported to frequently feed on harp seals (Pagophilus groenlandicus), which have higher ∑PCB concentrations than their preferred prey species in most other regions, the ringed seal (Phoca hispida) (Stirling and McEwan, 1975; Kleivane et al., 2000; McKinney et al., 2012). A similar influence of diet composition on internal concentrations of contaminants has been suggested for East Greenland polar bears (McKinney et al., 2013). For dieldrin and ∑DDT, the highest concentrations were found in the Canadian Arctic, with a particularly high concentration of ∑DDT in Foxe Basin. No explanation for this high concentration was found in literature, which is based on data from only one study (n¼ 10) (Norstrom et al., 1988, 1998). Excluding Foxe Basin from the regression analyses for population density did not change the results. 4.2. Toxicity reference values

Table 1 Overview of the 10 best supported models (þ intercept-only model) for the regression analysis with population density as the response variable and all predictor variables except ∑PBDE (n¼14). k refers to the number of parameters in the model. The models within ΔAICc of 2.0 from the most parsimonious model are indicated in bold. Model

k

AICc

ΔAICc

Akaike weight

R2

i þ PCB i þ Dieldrin þPCB i þ DDT þ PCB i þ DDT i þ HumanPopDens þ PCB i þHarvestþ PCB i þSept_ice þDDT i þDDT þ HumanPopDens i þSept_ice þPCB i þDDT þ Dieldrin i

2 3 3 2 3 3 3 3 3 3 1

70.96 71.36 71.92 72.04 72.29 73.83 74.51 74.69 74.99 75.09 75.55

0.00 0.40 0.96 1.08 1.33 2.87 3.55 3.73 4.04 4.13 4.59

0.206 0.169 0.127 0.120 0.106 0.049 0.035 0.032 0.027 0.026 0.021

0.43 0.56 0.54 0.39 0.53 0.47 0.45 0.44 0.43 0.43 0.00

Toxicity reference values for impaired individual health were frequently exceeded (Fig. 2). This is in accordance with results previously published by Sonne et al. (2009), who calculated Risk Quotients (RQ) based on extrapolated rat toxicity thresholds to assess the risk of contaminant concentrations in adipose tissue for effects on reproductive traits in East Greenland polar bears in the years 1990, 2000 and 2006. In 1990, the RQs of both dieldrin and ∑PCB concentrations were 41, indicating that concentrations were high enough to cause adverse health effects (Sonne et al., 2009). In 2000 the same was found for ∑PCBs and PFOS (perfluoroctanesulfonate) and in 2006 RQ's were 41 for dieldrin, ∑PCBs and PFOS (Sonne et al., 2009). Using the same approach, Dietz et al. (2015) recently showed that ∑PCBs, although banned from production, were by far the main contributor to the RQs for all three effect categories assessed in their study (immune,

54

R.J.M. Nuijten et al. / Environmental Research 151 (2016) 50–57

Fig. 2. Means and standard deviations (error bars) of concentrations in adipose tissue (ng·g-1 lipid weight) of ∑PCB (a), ∑DDT (b), dieldrin (c) and ∑PBDE (d) per subpopulation (n¼ 14 for ∑PCB, ∑DDT and dieldrin, n ¼ 11 for ∑PBDE). For sample sizes within subpopulations, see Table S1. Subpopulations are ordered from high (GB) to low (MC) polar bear population density. Horizontal lines represent toxicity threshold values (TRVs) above which a certain health effect was observed (see labels). Solid horizontal lines represent TRVs specific to polar bears, evenly dashed horizontal lines represent TRVs for harbour seals and unevenly dashed horizontal lines represent TRVs for sledge dogs. See Fig. 1 for subpopulation full names.

reproductive and carcinogenic effects; Dietz et al., 2015). In addition, the level of endocrine-disrupting chemicals in polar bears was recently linked to penile bone mineral density in polar bears across several subpopulations (Sonne et al., 2015). Similar to the RQ-studies based on the rat toxicity thresholds and the correlative study on penile bone density (Sonne et al., 2015), our results indicate that despite declining concentrations of most legacy contaminants in the Arctic environment, the residue concentrations of the legacy compounds are still high enough to pose a threat to polar bear population viability. 4.3. Potential population level effects Although based on a limited amount of population data, our regression analyses suggested negative relationships between

polar bear population density and adipose tissue concentrations for ∑PCB, ∑DDT, ∑PBDE and dieldrin, and human population density (Table 3). Interestingly, such negative linear relationships were absent for mean litter size (Tables S2 and S3) where the intercept-only model appeared to be the most parsimonious model. For adult female survival, only the variable September sea ice cover appeared in the best supported models (ΔAICc o2; Tables S4 and S5). Toxicant concentrations in polar bears might be too low to induce adult mortality, as lethal concentrations are generally an order of magnitude higher than concentrations causing sub-lethal adverse effects (Korsman et al., 2012). The absence of a relationship between mean litter size (as a proxy for reproduction) and contaminant concentrations in adipose tissue might reflect that litter size is not a particularly sensitive endpoint, as it does not include reproductive failure (i.e., litter size of zero). Possibly,

R.J.M. Nuijten et al. / Environmental Research 151 (2016) 50–57

Table 2 Overview of the 10 best supported models ( þ intercept-only model) for the regression analysis with population density as the response variable and all predictor variables (including ∑PBDE; n ¼ 11). k refers to the number of parameters in the model. The models within ΔAICc of 2.0 from the most parsimonious model are indicated in bold. Model

k

AICc

ΔAICc

Akaike weight

R2

i þ DDT i þ PBDE i þDDT þ PBDE i þDDT þ PCB i þSept_ice þDDT i þPCB i þDDT þ HumanPopDens i þDDT þ Dieldrin i þHarvestþ PCB i þDDT þ Harvest i

2 2 3 3 3 2 3 3 3 3 1

55.91 57.51 58.96 59.26 59.30 59.81 59.10 60.32 60.84 61.12 62.27

0.00 1.60 3.05 3.36 3.40 3.90 4.09 4.41 4.93 5.21 6.36

0.327 0.147 0.071 0.061 0.060 0.047 0.042 0.036 0.028 0.024 0.014

0.61 0.55 0.68 0.67 0.67 0.44 0.65 0.64 0.62 0.61 0.00

toxicant effects on reproductive output (and ultimately on population density) occur via an increased breeding interval or reduced juvenile survival instead of a decreased litter size. Female polar bears with poor body condition as a consequence of high pollutant levels may skip reproduction for one or more consecutive years to be able to feed on the sea ice in winter. This coping mechanism in polar bears with poor body condition was also discussed in a recent study where the relationship between climate change and polar bear litter size was modelled (Molnár et al., 2011). Unfortunately, there was insufficient data to use reproduction frequency or birth interval as a response variable in our analysis. Surprisingly, some of the additional explanatory variables (e.g. sea ice cover and harvest) did not appear in the final regression models, although these factors have been shown to affect polar bear populations (Taylor et al., 2006; Regehr et al., 2007; Stirling and Derocher, 2012). For sea ice, other sea ice-related variables such as ice thickness, the rate of decrease in ice cover or break-up date might be more relevant for polar bear population density than the summer sea ice cover that was used here. These data should be investigated in future studies. Also, climate change effects on for example sea ice cover and thickness vary regionally, and have been shown to have differential effects on polar bear subpopulations (Rode et al., 2014). This might have obscured an effect in this pan-Arctic comparison between subpopulations. Harvest did not appear in the final regression model either, indicating that current harvest levels do not affect polar bear population density. However, studies on the risk of harvest for polar bear subpopulations are ambiguous (Taylor et al., 2005, 2006). The most difficult part of correlative studies like the study described here is the accessibility and the quality of the data that is available. This is especially the case when studying a species that lives in a harsh environment, occurs in very low densities and has large home ranges, such as the polar bear. The data used in this

55

study was collected from the most reliable sources (peer-reviewed literature and the IUCN polar bear specialist group). However, limitations apply. First, the data collection (both for the contaminant levels and for the polar bear population variables) was scattered over a long time period (1982–2008). Within this time period there might have been changes in these or other variables that might have affected relationships between variables in different directions. We tested for such potential biases by checking for time-trends in the data. No trend over time was found (not for season, Fig. S1, nor for year, Fig. S2). Second, because of the limited amount of data on subpopulation sizes, we included all available estimates, regardless of their method of collection (mark-recapture analysis, distance sampling or modelling efforts; PBSG, 2015). Possibly this may have introduced uncertainty in our results. Third, we did not have data on other potentially relevant predictors of polar bear population characteristics such as prey density and availability. Although there is some literature on seal densities for certain subpopulations (Smith, 1975; Smith and Stirling 1978; Kingsley et al., 1985; Stirling and Oritsland 1995; Bengtson et al., 2005), region-specific data for most subpopulations is currently not available.

4.4. Implications The findings of our study suggest that many polar bear subpopulations are exposed to contaminant levels at or above the level where individual adverse health effects occur, and that these effects may scale up to the population level. The mechanism (s) behind an effect on population density as investigated here requires further data collection and analyses. Although many studies have shown associations between pollutant levels and adverse health effects in individuals, the challenge of linking such individual effects quantitatively to population-level effects remains a priority in conservation biology.

Acknowledgements This study was conducted as part of a 4-year project on the accumulation of pollutants in the Arctic food chain, funded by the Netherlands Polar Program (NPP) by the Netherlands Organization for Scientific Research (NWO) (Project number: 866.13.007).

Appendix A. Supplementary material Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.envres.2016.07.021.

Table 3 Mean ( þ SE) model-averaged parameter estimates for the best-supported models (i.e. within ΔAICc of 2.0) for the analyses with population density as the response variable. All predictor variables were standardized before the analyses (mean ¼ 0, SD ¼1). Analysis

Variable

Estimate

SE

z-value

p-value

Pop. Dens  PCB þ DDTþ Dieldrin þHuman Population Density þ Harvest þ September Ice

i PCB Dieldrin DDT Human pop. dens

3.04  1.56  0.27  0.53  0.14

0.63 0.95 0.57 0.87 0.42

4.30 1.56 0.45 0.59 0.32

o0.000 0.119 0.654 0.556 0.746

Pop. Dens  PBDE þ PCB þDDT þDieldrin þ Human Population Density þHarvestþ September Ice

i DDT PBDE

3.524  1.80  0.77

0.68 1.34 1.21

4.47 1.31 0.62

o0.000 0.192 0.537

56

R.J.M. Nuijten et al. / Environmental Research 151 (2016) 50–57

References Bengtson, J.L., et al., 2005. Ringed and bearded seal densities in the eastern Chukchi Sea, 1999–2000. Polar Biol. 28 (11), 833–845. Bennett, J.R., et al., 2015. Polar lessons learned : long-term management based on shared threats in Arctic and Antarctic environments. Front. Ecol. Environ. Bernhoft, A., et al., 1997. Organochlorines in polar bears (Ursus maritimus) at Svalbard. Environ. Pollut. 95 (2), 159–175. Borgå, K., et al., 2004. Biological and chemical factors of importance in the bioaccumulation and trophic transfer of persistent organochlorine contaminants in Arctic marine food webs. Environ. Toxicol. Chem./SETAC 23 (10), 2367–2385. Braathen, M., et al., 2004. Relationships between PCBs and thyroid hormones and retinol in female and male polar bears. Environ. Health Perspect. 112 (8), 826–833. Bustnes, J.O., et al., 2003. Ecological effects of organochlorine pollutants in the Arctic: a study of the Glaucous Gull. Ecol. Appl. 13 (2), 504–515. http://dx.doi. org/10.1890/1051-0761(2003)013[0504:EEOOPI]2.0.CO;2. Cardillo, M., et al., 2004. Human population density and extinction risk in the world's carnivores. PLoS Biol. 2 (7), 909–914. CIESIN, 2005. Gridded Population of the World, Version 3 (GPWv3): Population Density Grid. Dietz, R., et al., 2007. Age and seasonal variability of polybrominated diphenyl ethers in free-ranging East Greenland polar bears (Ursus maritimus). Environ. Pollut. 146 (1), 166–173. Dietz, R., et al., 2015. Physiologically-based pharmacokinetic modelling of immune, reproductive and carcinogenic effects from contaminant exposure in polar bears (Ursus maritimus) across the Arctic. Environ. Res. 140, 45–55. http://dx. doi.org/10.1016/j.envres.2015.03.011. Erikstad, K.E., Sandvik, H., Reiertsen, T.K., Bustnes, J.O., Strøm, H., 2013. Persistent organic pollution in a high-Arctic top predator: sex-dependent thresholds in adult survival. P Roy Soc Lond B Bio 280 (1769), 20131483. Erikstad, K.E., et al., 2011. High levels of organochlorines may affect hatching sex ratio and hatchling body mass in arctic glaucous gulls. Funct. Ecol. 25 (1), 289–296. Forbes, V.E., et al., 2016. Assessing the risks of pesticides to threatened and endangered species using population modeling: a critical review and recommendations for future work. Environ. Toxicol. Chem. Gebbink, W.A., et al., 2008. Tissue-specific congener composition of organohalogen and metabolite contaminants in East Greenland polar bears (Ursus maritimus). Environ. Pollut. 152 (3), 621–629. Gustavson, L., et al., 2015. Hydroxylated polychlorinated biphenyls decrease circulating steroids in female polar bears (Ursus maritimus). Environ. Res. 138, 191–201. http://dx.doi.org/10.1016/j.envres.2015.02.011. Haave, M., et al., 2003. Polychlorinated biphenyls and reproductive hormones in female polar bears at Svalbard. Environ. Health Perspect. 111 (4), 431–436. Huntington, A.H., 2009. A preliminary assessment of threats to arctic marine mammals and their conservation in the coming decades. Mar. Policy 33, 77–82. Jenssen, B.M., 2006. Endocrine-disrupting chemicals and climate change: a worstcase combination for arctic marine mammals and seabirds? Environ. Health Perspect. 114 (Suppl. 1), S76–S80. Jenssen, B.M., et al., 2015. Anthropogenic flank attack on polar bears: Interacting consequences of climate warming and pollutant exposure. Front. Ecol. Evol. 3, 1–7, Available at: 〈http://journal.frontiersin.org/Article/10.3389/fevo.2015. 00016/abstract〉. Jørgensen, E.H., et al., 2006. Toxicokinetics and effects of PCBs in Arctic fish: a review of studies on Arctic charr. J. Toxicol. Environ. Health Part A 69 (1–2), 37–52. Kannan, K., Yun, S.H., Evans, T.J., 2005. Chlorinated, brominated, and perfluorinated contaminants in livers of polar bears from Alaska. Environ. Sci. Technol. 39 (23), 9057–9063. Kingsley, M.C.S., Stirling, I., Calvert, W., 1985. The Distribution and Abundance of Seals in the Canadian High Arctic, 1980–82. Can. J. Fish. Aquat. Sci. 42 (6), 1189–1210. Kleivane, L., Severinsen, T., Skaare, J.U., 2000. Biological transport and mammal to mammal transfer of organochlorines in Arctic fauna. Mar. Environ. Res. 49 (4), 343–357. Korsman, J.C., et al., 2012. Modelling the impact of toxic and disturbance stress on white-tailed eagle (Haliaeetus albicilla) populations. Ecotoxicology 21 (1), 27–36. Kucklick, J.R., et al., 2002. Persistent organochlorine pollutants in ringed seals and polar bears collected from northern Alaska. Sci. Total Environ. 287 (1–2), 45–59. Letcher, R.J., et al., 2010. Exposure and effects assessment of persistent organohalogen contaminants in arctic wildlife and fish. Sci. Total Environ. 408 (15), 2995–3043. Lie, E., et al., 2003. Geographical distribution of organochlorine pesticides (OCPs) in polar bears (Ursus maritimus) in the Norwegian and Russian Arctic. Sci. Total Environ. 306 (1–3), 159–170, Available at: 〈http://www.scopus.com/inward/re cord.url?eid ¼2-s2.0-0037403566&partnerID ¼ tZOtx3y1〉. MacKay, D., Fraser, A., 2000. Bioaccumulation of persistent organic chemicals: mechanisms and models. Environ. Pollut. 110 (3), 375–391. Mckinney, M.A., et al., 2013. Global change effects on the long-term feeding ecology and contaminant exposures of East Greenland polar bears. Glob. Change Biol. 19 (8), 2360–2372. McKinney, M., Peacock, E., Letcher, R.J., 2009. Sea ice-associated diet change and brominated contaminants. Environ. Sci. Technol. 43 (12), 4334–4339.

McKinney, M.A., et al., 2011. Flame retardants and legacy contaminants in polar bears from Alaska, Canada, East Greenland and Svalbard, 2005–2008. Environ. Int. 37 (2), 365–374. McKinney, M.A., et al., 2012. Trophic transfer of contaminants in a changing arctic marine food web: Cumberland sound, Nunavut, Canada. Environ. Sci. Technol. 46 (18), 9914–9922. Miljeteig, C., et al., 2012. Eggshell thinning and decreased concentrations of vitamin E are associated with contaminants in eggs of ivory gulls. Sci. Total Environ. 431, 92–99. http://dx.doi.org/10.1016/j.scitotenv.2012.05.018. Molnár, P.K., et al., 2011. Predicting climate change impacts on polar bear litter size. Nat. Commun. 2, 186. Muir, D.C.G., et al., 2006. Brominated flame retardants in polar bears (Ursus maritimus) from Alaska, the Canadian Arctic, East Greenland, and Svalbard. Environ. Sci. Technol. 40 (2), 449–455, Available at: 〈http://www.ncbi.nlm.nih.gov/ pubmed/16468388〉. Muir, D.C.G., de Wit, C.A., 2010. Trends of legacy and new persistent organic pollutants in the circumpolar arctic: Overview, conclusions, and recommendations. Sci. Total Environ. 408 (15), 3044–3051. http://dx.doi.org/10.1016/j. scitotenv.2009.11.032. Norstrom, R.J., et al., 1988. Organochlorine contaminants in arctic marine food chains: Identification, geographical distribution, and temporal trends in polar bears. Environ. Sci. Technol. 22 (9), 1063–1071. Norstrom, R.J., et al., 1998. Chlorinated hydrocarbon contaminants in polar bears from Eastern Russia, North America, Greenland, and Svalbard: Biomonitoring of Arctic pollution. Arch. Environ. Contam. Toxicol. 35, 354–367. NSIDC, 2008. IMS Daily Northern Hemisphere Snow and Ice Analysis at 4 km and 24 km Resolutions. O’Brien, R.M., 2007. A caution regarding rules of thumb for variance inflation factors. Qual. Quant. 41 (5), 673–690. Oskam, I.C., et al., 2003. Organochlorines affect the major androgenic hormone, testosterone, in male polar bears (Ursus maritimus) at Svalbard. J. Toxicol. Environ. Health Part A 66 (22), 2119–2139, Available at: 〈http://www.ncbi.nlm.nih. gov/pubmed/14710596〉. Oskam, I.C., et al., 2004. Organochlorines affect the steriod hormone cortisol in freeranging polar bear (Ursus maritimus) at Svalbard, Norway. J. Toxicol. Environ. Health Part A: Curr. Issues 67 (12), 959–977. Paetkau, D., et al., 1999. Genetic structure of the world's polar bear populations. Mol. Ecol. 8 (10), 1571–1584. PBSG, 2015. IUCN Polar Bear Specialist Group. Peacock, E., et al., 2015. Implications of the circumpolar genetic structure of polar bears for their conservation in a rapidly warming Arctic. PLoS One 10 (1), 1–30. Polischuk, S.C., Norstrom, R.J., Ramsay, M.A., 2002. Body burdens and tissue concentrations of organochlorines in polar bears (Ursus maritimus) vary during seasonal fasts. Environ. Pollut. 118 (1), 29–39. Regehr, E.V., et al., 2007. Effects of earlier sea ice breakup on survival and population size of polar bears in Western Hudson Bay. J. Wildl. Manag. 71 (8), 2673–2683. Regehr, E.V., et al., 2010. Survival and breeding of polar bears in the southern Beaufort Sea in relation to sea ice. J. Anim. Ecol. 79 (1), 117–127. Riget, F., et al., 2010. Temporal trends of legacy POPs in Arctic biota, an update. Sci. Total Environ. 408 (15), 2874–2884. http://dx.doi.org/10.1016/j. scitotenv.2009.07.036. Rode, K.D., et al., 2014. Variation in the response of an Arctic top predator experiencing habitat loss: Feeding and reproductive ecology of two polar bear populations. Glob. Change Biol. 20 (1), 76–88. Sandala, G.M., et al., 2004. Hydroxylated and methyl sulfone PCB metabolites in adipose and whole blood of polar bear (Ursus maritimus) from East Greenland. Sci. Total Environ. 331 (1–3), 125–141. Skaare, J.U., et al., 2000. Organochlorines in top predators at Svalbard—occurrence, levels and effects. Toxicol. Lett. 112–113, 103–109. Smith, T.G., 1975. Ringed seals in James Bay and Hudson Bay: population estimates and catch statistics. Arctic, 170–182. Smith, T.G., Stirling, I., 1978. Variation in the density of ringed seal (Phoca hispida) birth lairs in Amundsen Gulf, Northwest Territories. Can. J. Zool. 56 (6), 1066–1070. Sonne, C., et al., 2009. Reproductive performance in East Greenland polar bears (Ursus maritimu)s may be affected by organohalogen contaminants as shown by physiologically-based pharmacokinetic (PBP)K modelling. Chemosphere 77 (11), 1558–1568. http://dx.doi.org/10.1016/j.chemosphere.2009.09.044. Sonne, C., 2010. Health effects from long-range transported contaminants in Arctic top predators: An integrated review based on studies of polar bears and relevant model species. Environ. Int. 36 (5), 461–491. http://dx.doi.org/10.1016/j. envint.2010.03.002. Sonne, C., et al., 2015. Penile density and globally used chemicals in Canadian and Greenland polar bears. Environ. Res. 137, 287–291. http://dx.doi.org/10.1016/j. envres.2014.12.026. Sørmo, E.G., et al., 2006. Biomagnification of polybrominated diphenyl ether and hexabromocyclododecane flame retardants in the polar bear food chain in Svalbard, Norway. Environ. Toxicol. Chem./SETAC 25 (9), 2502–2511, Available at: 〈http://www.ncbi.nlm.nih.gov/pubmed/16986806/nISI:000239912100029〉. St. Louis, V.L., et al., 2011. Differences in mercury bioaccumulation between polar bears (Ursus maritimus) from the Canadian high- and sub-arctic. Environ. Sci. Technol. 45 (14), 5922–5928. Stirling, I., McEwan, E.H., 1975. The caloric value of whole ringed seals (Phoca hispida) in relation to polar bear (Ursus maritimus) ecology and hunting behavior. Can. J. Zool. 53 (8), 1021–1027.

R.J.M. Nuijten et al. / Environmental Research 151 (2016) 50–57

Stirling, I., Derocher, A.E., 1993. Possible Impacts of Climatic Warming on Polar Bears. Arctic 46 (3), 240–245, Available at: 〈http://arctic.synergiesprairies.ca/ arctic/index.php/arctic/article/view/1348〉. Stirling, I., Oritsland, N.A., 1995. Relationships between estimates of ringed seal (Phoca hispida) and polar bear (Ursus maritimus) populations in the Canadian Arctic. Canadian J. Fish. Aquat. Sci. 52, 2594–2612. Stirling, I., Derocher, A.E., 2012. Effects of climate warming on polar bears: A review of the evidence. Glob. Change Biol. 18 (9), 2694–2706. Stirling, I., Lunn, M.J., Iacozza, J., 1999. Long-term trends in the population ecology of polar bears in Western Hudson Bay in relation to climate change. Arctic 52 (3), 294–306, Available at: 〈http://www.jstor.org/stable/40511782〉. Stroeve, J.C., et al., 2012. The Arctic's rapidly shrinking sea ice cover: a research synthesis. Clim. Change 110 (3–4), 1005–1027. Taylor, M.K., et al., 2005. Demography and viability of a hunted population of polar

57

bears. Arctic 58 (2), 203–214. Taylor, M.K., et al., 2006. Demographic parameters and harvest-explicit population viability. Analysis 70 (6), 1667–1673. Taylor, M.K., et al., 2008. Population parameters and harvest risks for polar bears (Ursus maritimus) of Kane Basin, Canada and Greenland. Polar Biol. 31 (4), 491–499. Van Oostdam, J., et al., 2005. Human health implications of environmental contaminants in Arctic Canada: a review. Sci. Total Environ. 351–352, 165–246. Verreault, J., et al., 2005. Chlorinated hydrocarbon contaminants and metabolites in polar bears (Ursus maritimus) from Alaska, Canada, East Greenland, and Svalbard: 1996-2002. Sci. Total Environ. 351–352, 369–390. Verreault, J., et al., 2006. Maternal transfer of organohalogen contaminants and metabolites to eggs of Arctic-breeding glaucous gulls. Environ. Pollut. 144 (3), 1053–1060.