Seabird Guano Is an Efficient Conveyer of Persistent Organic Pollutants (POPs) to Arctic Lake Ecosystems A . E V E N S E T , * ,†,‡ J . C A R R O L L , † G. N. CHRISTENSEN,† R. KALLENBORN,§ D. GREGOR,| AND G. W. GABRIELSEN⊥ Akvaplan-niva, Polar Environmental Centre, 9296 Tromsø, Norway, Norwegian College of Fishery Science, University of Tromsø, 9037 Tromsø, Norway, Norwegian Institute for Air Research, Polar Environmental Centre, 9296 Tromsø, Norway, Gartner Lee Limited, 3 Watson Road South, Suite 1, Guelph, ON, Canada, N1L 1E3, and Norwegian Polar Institute, Polar Environmental Centre, 9296 Tromsø, Norway
Migratory seabirds have been linked to localized “hotspots” of contamination in remote Arctic lakes. One of these lakes is Lake Ellasjøen on Bjørnøya in the Barents Sea. Here we provide quantitative evidence demonstrating that even relatively small populations of certain seabird species can lead to major impacts for ecosystems. In the present example, seabird guano accounts for approximately 14% of the contaminant inventory of the Lake Ellasjøen catchment area, approximately 80% of the contaminant inventory of the lake itself, and is approximately thirty times more efficient as a contaminant transport pathway compared to atmospheric long-range transport. We have further shown that this biological transport mechanism is an important contaminant exposure route for ecosystems, responsible for POPs levels in freshwater fish that are an order of magnitude higher than those in Arctic top predators. Given the worldwide presence of seabird colonies in coastal marine areas where resources are also harvested by humans, this biological transport pathway may be a greater source of dietary contamination than is currently recognized with consequent risks for human health.
Introduction Atmospheric long-range transport of pollutants (LRTP) from the world’s major chemical production and application regions is the main supplier of persistent organic pollutants (POPs) to the Arctic (1-3). The atmospherically transported contaminants presently identified in the Arctic are dominated by hexachlorocyclohexanes (HCHs), polychlorinated biphenyls (PCBs), polychlorinated bornanes/bornenes (“toxaphene”), and chlordane-related compounds (cyclodienecompounds). However, also present-use compounds, such as brominated flame retardants (e.g., polybrominated diphen* Corresponding author phone: +47 77 75 03 11; fax: +47 77 75 03 01; e-mail:
[email protected]. † Akvaplan-niva, Polar Environmental Centre. ‡ University of Tromsø. § Norwegian Institute for Air Research. | Gartner Lee Limited. ⊥ Norwegian Polar Institute. 10.1021/es0621142 CCC: $37.00 Published on Web 01/11/2007
xxxx American Chemical Society
yl ethers (PBDEs) and hexabromocyclododecane (HBCD)) have been identified in different environmental samples from remote areas (3-6). The pollutants enter Arctic lakes via precipitation and dry deposition to lake surfaces and through indirect deposition into catchment areas, followed by surface water runoff or groundwater seepage (7, 8). Recently, seabirds have been identified as potential transport vectors, supplying contaminants acquired and enriched through feeding in the marine environment to lakes. In the Canadian Arctic, eight lakes influenced by seabirds have been reported to contain sediment contaminant levels 10-60 times higher than in three nearby lakes with no significant seabird populations (9). Similarly, in the Norwegian Arctic, lake sediment contaminant levels comparable to the highest levels reported in the Canadian study have been reported (10). Guano from seabirds is suggested as the source of contamination due to higher δ15N-values in sediment (9) and biota (10) in these areas, presumably reflecting enriched δ15N-values in guano relative to other N sources in the environment. However, the role of seabirds in contaminant transport has not been confirmed through direct analyses of guano, nor have there been any elucidations of the connection between seabird guano, pollutant exchange, and ecosystem impacts. Here we report POPs levels in fresh guano from selected seabird species lending support to previous inferences of a linkage between seabirds and contaminated lakes. This new information is then used as part of a comparative analysis of contaminant supplies to two neighboring lakes, Ellasjøen and Øyangen, located on Bjørnøya (74°30′N, 19°00′E), a remote Arctic island in the Barents Sea. Lake Ellasjøen is located in the southern, mountainous part of Bjørnøya, while Øyangen is located 5 km further north, on the central plains of the island (details in 10). Both lakes receive contaminants from long-range atmospheric transport, and previous studies have shown that Bjørnøya belongs to the main deposition area for POPs in the Norwegian Arctic (11). However, the levels of POPs are several times higher in sediment and biota from Lake Ellasjøen than in comparable samples from Lake Øyangen, and there are also differences in contaminant patterns (10), indicating different sources to the two lakes. The southern margin of Bjørnøya contains some of the largest seabird colonies in the northern hemisphere, with more than a million seabirds present during the summer season (12). The birds spend most of their time in the breeding colony or out at sea on feeding expeditions. However, a small proportion of the seabirds use Lake Ellasjøen as a resting and bathing area throughout the ice-free season (JuneOctober). Seabirds are rarely present at Lake Øyangen. Thus, while direct precipitation of rain and snow into lakes and their catchment areas introduces LRTP to both lakes, only Lake Ellasjøen is influenced by seabirds. By describing and quantifying sources of contaminants to lake catchment areas (precipitation and guano) and to the individual lakes, we are able to relate specific contaminant sources to observed ecosystem impacts. We mainly focus on the seabird-impacted Lake Ellasjøen, but Lake Øyangen is an important reference lake where ecosystem impacts are linked solely to LRTP.
Materials and Methods We use three lines of evidence to establish the importance of guano in this system: mass-balance analysis of contaminant sources and sinks to lakes and catchment areas, analyses of contaminant patterns in lake samples and guano, and stable isotope patterns in biota. We have focused on a set of 12 polychlorinated biphenyls (ΣPCB) analyzed in all the relevant media (precipitation, lake VOL. xx, NO. xx, xxxx / ENVIRON. SCI. & TECHNOL.
PAGE EST: 6.9
9
A
water, guano, and biota) in the mass-balance assessment. These PCB congeners (CB 18, 28, 52, 101, 105, 118, 123, 138, 149, 153, 180 and 194) represent the most prevalent PCBs found in environmental materials. Twenty-four congeners were used in the contaminant pattern analysis. Sample Collection: Precipitation. Precipitation rates were determined for two areas: one was at Bjørnøya meteorological station, which is located far north on the island, and one was located close to Lake Ellasjøen. The meteorological station is located in a terrain that is very similar to the terrain around Lake Øyangen, and data from the station was therefore assumed to be representative for the situation around Lake Øyangen. Daily precipitation measurements at the Bjørnøya Meteorological Station, combined with data from precipitation samplers deployed around the lakes from 1999 to 2004 were used to derive annual flux estimates for snow and rainfall. To estimate the precipitation rate on the southern mountainous part of the island (to the Lake Ellasjøen catchment area) 15 passive precipitation gauges were installed in the catchment area of Lake Ellasjøen for measurement from late May to late September. Identical samplers (n ) 5) were installed at the meteorological station for comparison with those set at Ellasjøen. The gauges at Ellasjøen collected on average 3.15 times more precipitation (rain and condensed fog) than the gauges at the meteorological station. Average annual precipitation to the Ellasjøen area was calculated by multiplying the precipitation rate measured at Bjørnøya meteorological station by 3.15. Precipitation was assessed as snow during periods when air temperatures were below 0 °C, and about 60% of the precipitation in the area is snow (Norwegian Meteorological Institute). Rain samples (including also condensed fog) for contaminant analysis were collected using horizontal stainless steel plates (1 m2) mounted 1 m above the ground. The collected rainwater drained from the sampler into a 50-L steel barrel. One sampler was located close to Ellasjøen and one was close to the meteorological station. The samplers were operated during the summer seasons (May-October) 1999-2003. A total of 6 rain samples (at least 30 L in each sample) were analyzed during the sampling period. Snow collectors were operated at the meteorological station during the winter season (October-May). The snow collectors were wide, angled funnels made of aluminum plates surrounded with wind shadings (diameter of 2.5 m). The collected snow was drained into 50-L steel barrels. In addition to these snow collectors, old and new surface snow was collected at different locations in the catchment area of Ellasjøen and close to the meteorological station. Sixteen snow samples were collected and analyzed during the project period. Polyurethane foam (PUF) adsorbents were used to extract the contaminants from the rainwater/snow samples. Sample Collection: Seabirds. There are three seabird species that are important around Lake Ellasjøen; little auk (Alle alle), kittiwake (Rissa tridactyla), and glaucous gull (Larus hyperboreus). Monthly seabird population estimates for these species were derived from censuses carried out at the lake over five summer seasons (1999-2004). Guano samples were collected from seabirds (little auk, kittiwake, and glaucous gulls) killed with a shotgun (1999, 2001, and 2003). The birds were dissected on site and intestinal content (guano) was collected by squeezing gently on the posterior part of the intestine. Intestinal contents from several birds were pooled prior to analyses. In addition, two fresh guano samples from glaucous gull were collected from rocks in 2003. Samples were transferred to pre-burned glass jars and frozen to -20 °C. Literature values of species-specific food intake (13) and guano excretion rates (14-16) were used in the estimation of the amount of guano excreted by seabirds visiting the lake. B
9
ENVIRON. SCI. & TECHNOL. / VOL. xx, NO. xx, xxxx
Sample Collection: Lake Ecosystem. Zooplankton (90% Cyclops abyssorum), chironomid larvae (Chironomidae sp.), and Arctic char (Salvelinus alpinus) were collected from Lake Ellasjøen and Lake Øyangen over a 5-year period (19962001), according to methods described in Evenset et al. (10). Water samples from each lake were collected in July 1999 using a stainless steel filtration system. For suspended particles, silicon rubber hosing and a pump were used to pump (flow 3 L/min) water through 293 mm glass-fiber filters (Whatman GF/F). A prefilter (90 µm) was used to remove zooplankton and large particles from the water stream. Hence, lake suspended particulate matter consisted of phytoplankton, small zooplankton, and detritus. The dissolved fraction was collected on polyurethane foam (PUF) adsorbent. Sample volumes were from 386 to 400 L. Field blanks were collected directly after each sample had been collected, by repeating the sample procedure steps. No water was run through the blank samples. Filters and PUFs were wrapped in aluminum foil, packed in separate plastic bags, and stored at -20 °C directly after sampling. All filters, PUFs, and wrapping material were combusted at 450 °C for 8 h prior to use. Two samples of particulates (filters) and two of the dissolved fraction (PUFs) from each lake were analyzed for PCBs. Stable Isotope Analyses. Sub-samples of the samples analyzed for POPs, as well as some additional samples from each organism group, were analyzed for stable isotopes of carbon and nitrogen using the method described in Evenset et al. (10) (see Supporting Information). PCB Analyses. PCB analyses were performed by the Department of Applied Environmental Science, University of Stockholm, Sweden (water) and the Norwegian Institute for Air Research, Kjeller, Norway. The water samples were analyzed according to the method described by Axelman et al. (17) (see Supporting Information). The biota samples were analyzed by the Norwegian Institute for Air Research according to the method described by Derocher et al. (18) (see Supporting Information). Statistical Analyses. The multivariate relationship between the contaminants in fish and the explanatory variables (δ15N, % lipid, length, age) was analyzed by redundancy analysis (RDA). Before the analyses were performed, the wetweight concentrations were log-transformed to normalize the distribution and to reduce variance heterogeneity. In the RDA, the samples and the PCBs were assigned scores on ordination axes presented in a diagram, and arrows were used to represent the environmental variables pointing in the direction of maximum variance accounted for by the variable. Ordination techniques and rules for interpretation of redundancy diagrams are summarized and reviewed elsewhere (19, 20). The RDA was followed by a Monte Carlo permutation test to analyze whether the contaminant concentrations were significantly related to the explanatory variables. The Monte Carlo permutation test was run with 999 unrestricted permutations and an overall significance of p < 0.005. Correspondence analyses (CA) were used to analyze contaminant patterns in water, sediment, and invertebrates from the two lakes and the contaminant pattern in guano from the seabirds. The statistical analysis was performed using CANOCO, ver. 4.5.
Results and Discussion Mass Balance Calculations. During the summer season, there is a constant flux of birds between Lake Ellasjøen and bird cliffs of southern Bjørnøya. Based on our censuses we estimated that at any time during the day there are approximately 6500 little auks, 250-2000 kittiwakes, and 70 glaucous gulls present on/around Lake Ellasjøen. During time spent both above and on the lake, seabirds excrete waste
TABLE 1. Average Concentrations of Stable Isotope and PCB Concentrations in Seabird Intestinal Content (IC) and Fresh Guano Samples from Glaucous Gull Collected from Rocks (R) (12 Congeners)a species Little auk (IC) Kittiwake (IC) Glaucous gull (IC) Glaucous gull (R) a
% lipid
niso
4.1 (6.0 - 2.2) 3.7 ( 0.7 (2.9 - 4.8) 3.1 ( 1.6 (1.1 - 5.9) 1.8 ( 0.9 (2.4, 2.7)
6 6 5
δ15N (‰)
δ13C (‰)
sum PCB (ng/g ww)
-20.3 ( 0.7 (-19.8 to -21.3) -20.1 ( 0.4 (-19.4 to -20.7) -20.5 ( 1.0 (-19.3 to -21.9)
25 ( 9 (19.2 - 31.3) 124 ( 41 (76.4 - 157) 1,460 ( 1,846 (203 - 3533) 638 ( 439 (368, 948)
nPOPs
11.6 ( 0.7 (10.7 - 12.7) 11.9 ( 0.4 (11.4 - 12.7) 14.8 ( 0.8 (13.9 - 15.9)
2 5 5 2
Individual measurements represent pooled samples of Guano from (5-15) individuals (ranges are given in parentheses).
products from the digestive tract (guano) directly into the limnic system and into the lake catchment area. The PCB levels in the guano samples varied considerably both within and among these species in accordance with the trophic status of their preferred food items (Table 1). Little auk (ΣPCB in guano ) 25 ( 9 ng/g wet weight (ww)) is a small seabird that feeds mainly on different marine zooplankton (amphipods), while the larger kittiwake (ΣPCB in guano ) 124 ( 41 ng/g ww) usually feeds on zooplankton and small fish. The large glaucous gull is an opportunistic feeder, consuming prey species higher in the food chain, resulting in considerably higher PCB concentrations in guano (ΣPCB ) 1224 ( 1243 ng/g ww) compared to levels in either little auk or kittiwake. The fresh guano samples from glaucous gull had lipid contents and PCB concentrations similar to those of samples of intestinal content. A large between sample variability (Table 1) is not surprising because this species feeds on anything from zooplankton to marine mammal carcasses. Our measured PCB levels in guano, when combined with species-specific guano excretion rates and seabird numbers at the lake (Table 2), yield an estimate of approximately 4.4 grams/yr as the total amount of PCBs associated with guano excretion at Lake Ellasjøen (Table 3). According to our calculations little auk contributes only 20% to the total although this species represents approximately 80% of the total number of seabirds at the lake. This is largely a consequence of the low PCB content of guano connected to this species’ preference for lower trophic level prey species. Glaucous gull represents only 1% of the number of seabirds at the lake but is responsible for approximately 50% of the total annual PCB supply in guano. Despite uncertainties in these parameter values, it is clear that population size alone is a poor indicator of contaminant contributions from seabirds. To put this contaminant source term into context, we compare it to the supply of LRTP to lake catchment areas quantified using data from multi-year measurements of PCB concentrations and deposition rates of snow- and rainfall (Table 2). Some PCB may also enter the lake via dry deposition, and previous work has shown that dry deposition may be an important contaminant source, especially to eutrophic lakes in cold and low-precipitation areas, although it usually contributes far less than precipitation and runoff from the watershed (21). At Lake Ellasjøen the precipitation rate is relatively high and there is a high frequency of fog on Bjørnøya during the summer (up to 75% of the days) (22). Further, even though seabirds are supplying nutrients to Lake Ellasjøen, both lakes on Bjørnøya are oligotrophic. Therefore as a first approximation, dry deposition will be of minor importance compared to wet deposition. LRTP introduces 28 g/yr of PCBs to the Lake Ellasjøen catchment area compared to only 3 g/yr to the Lake Øyangen
TABLE 2. Parameters and Values Used in Mass Balance Calculations parameter catchment area (km2) lake surface area (km2) lake depth (max.) (m) lake volume (V) (m3) water retention time (yr) (τ) rain flux (Fr) (L km-2 yr-1) snow flux (Fs) (L km-2 yr-1) PCB-concentration lake water (g/L) PCB-concentration rain (Cr) (g/L) PCB-concentration snow (Cs) (g/L) Little Auk (SB) (no. seabird days/year)a Glaucous Gull (SB) (no. seabird days/year)b Kittiwake (SB) (no. seabird days/year)c guano excretion rate (gex)LA (g/d) guano excretion rate (gex)GG (g/d) guano excretion rate (gex)KW (g/d) PCB concentration guano (Cg)LA(g/g ww) PCB concentration guano (Cg)GG(g/g ww) PCB concentration guano (Cg)KW(g/g ww)
Øyangen
Ellasjøen
2.2 6.1 0.35 0.72 5 34 0.72 × 106 11.3 × 106 0.9 1.6 1.3 × 108 4.2 × 108 2.4 × 108 7.5 × 108 23 × 10-12 129 × 10-12 3.0 × 10-9 4.6 × 10-9
3.0 × 10-9 4.6 × 10-9 598 × 103 6.4 × 103 146 × 103 50 262 80 25 × 10-9 1224 × 10-9 124 × 10-9
a The following average numbers were used in calculations of seabird days: 70 bird/day for 3 months. b The following average numbers were used in calculations of seabird days: 6500 birds/day for 3 months.c The following average numbers were used in calculations of seabird days: June 500 birds/day, July 2000 birds/day, August 2000 birds/day, September 250 birds/day.
catchment area (Table 3). The large difference in LRTP supplies between the two lakes is partly due to the larger size of the Lake Ellasjøen catchment area but also because this lake is located in the much wetter, mountainous coastal area of Bjørnøya compared to Lake Øyangen’s location in the much drier central plains region. Thus, while guano contributes only 14% of the total contaminant load (4.4 g/(4.4 g + 28 g)) compared to 86% from LRTP to the Lake Ellasjøen catchment area, in absolute terms the amount from guano is still significant. Considering the lake itself, the direct deposition from guano may in fact be a more important source of PCBs than the indirect supply of LRTP. This is a reasonable hypothesis given that LRTP exchanges from catchment areas into Arctic lakes is considered to be a relatively inefficient transfer mechanism due to factors that include ice cover during a large fraction of the melting season, and limited retention time (due to stratification) of snowmelt in lakes (7, 8). Evaluating the composition and quantity of PCBs present in both lakes and their linkage to sources, a consistent picture VOL. xx, NO. xx, xxxx / ENVIRON. SCI. & TECHNOL.
9
C
TABLE 3. PCB Budgets (12 Congeners) for Both Lakesa equations
Ellasjøen
guano supply AG ) ΣSB × gex × Cg ΣSB ) no. seabird days/year gex ) guano excretion rates (g/d) Cg ) PCB level (g/g ww) LRTP supply to lake catchment areas (g/yr) AY ) [(Cr × FrY) + (Cs × FsY)]Ca Cr ) PCB concentration in rain (g/L) Cs ) PCB concentration in snow (g/L) FrY ) rain flux (L km-2 yr-1) FsY ) snow flux (L km-2 yr-1) Ca ) catchment basin area (km2)
Øyangen
AG(LA) ) 0.8 AG(GG) ) 2.1 AG(KW) ) 1.5 Y)E 28 3.0 × 10-9 4.6 × 10-9 4.2 × 108 7.5 × 108 2.2
Lake Øyangen (surf. area ) 0.35 km2; depth (max) ) 5 m) supplies of LRTP (g/yr) LØ(LRTP) ) [(CØ × VØ)/tØ] CØ ) PCB concentration (diss. + part. fractions) (g/L) tØ ) water retention time (yr) VØ ) lake volume (m3) LRTP catchment to lake exchange (no units) Ex(LRTP) ) LØ(LRTP)/AØ
Y)Ø 3 3.0 × 10-9 4.6 × 10-9 1.3 × 108 2.4 × 108 6.1
2.0 × 10-2 23 × 10-12 0.9 0.72 × 106 6.7 × 10-3
Lake Ellasjøen (surf. area ) 0.72 km2; depth (max) ) 34 m) TOTAL supply (g/yr) LE(TOTAL) ) [(CE × VE)/tE] CE ) PCB concentration (diss. + part. fractions) (g/L) tE ) water retention time VE ) lake volume supplies from guano(g/yr) LE(G) ) LE(TOTAL) - (Ex(LRTP) × AE) supplies of LRTP(g/yr) LE(LRTP) ) LE(TOTAL) - LE(G) guano to lake exchange (no units): Ex(G) ) LE(G)/AG AG ) total PCB supplies from seabird guano
9.2 × 10-1 129 × 10-12 1.6 11.3 × 106 7.3 × 10-1 1.9 × 10-1 166 × 10-3 4.4
a All values except exchange coefficients (Ex) are given in g/year. LA ) little auk (Alle alle); GG ) glaucous gull (Larus hyperboreus); KW ) kittiwake (Rissa tridactyla).
of the role of seabird mediation in Lake Ellasjøen emerges. Normalizing the PCB content in water from both lakes for water residence time, the annual PCB loads in water from Lake Ellasjøen and Lake Øyangen are markedly different: 924 mg/yr in Lake Ellasjøen and 19 mg/yr in Lake Øyangen (Table 3). Accounting for the contaminant load in Lake Øyangen with our previously quantified LRTP source term (3 g/yr), we derive an estimate of the exchange coefficient for LRTP of only 6.7 × 10-3 (0.02 g/ 3 g), indicating that only a very small fraction of PCBs deposited in the catchment area (0.7%) is transferred into the lake (Table 3). Assuming the same LRTP exchange coefficient for Lake Ellasjøen, then from the total PCB lake load of 924 mg/yr, only 190 mg/yr is accounted for by LRTP while the remaining 734 mg/yr is attributable to seabirds (Table 3). Compared to the total PCB supply of 4.4 g in guano, the exchange coefficient of guanoderived PCBs is 166 × 10-3 (0.73 g/ 4.4 g), or more than 30 times the exchange coefficient of LRTP. It is certainly plausible that the exchange coefficient for LRTP is higher for Lake Ellasjøen than Lake Øyangen due to differences in catchment area, hydrology, lake morphology, and productivity, but the large difference between exchange coefficients for guano and LRTP indicates that guano deposition is by far the more efficient contaminant transfer mechanism to lakes inhabited by seabirds. Contaminant Patterns. Previous investigations of biological transport vectors have relied primarily on the relative composition of PCB compounds (congener pattern) to identify, but not quantify, associated impacts on the environment (23, 24). Also in our study, we find that PCB congener patterns provide support for our conclusion that seabird guano is responsible for most of the PCB load in Lake D
9
ENVIRON. SCI. & TECHNOL. / VOL. xx, NO. xx, xxxx
Ellasjøen. The contaminants that are in guano have already passed through one or more biomagnification cycles in the marine ecosystem (i.e., from phytoplankton to zooplankton to fish). This has caused increased contaminant levels in the guano, but also leads to changes in the contaminant pattern compared to the pattern in the abiotic environment of the Barents Sea. The metabolic processes in animals tend to increase concentrations of the more persistent in relation to less persistent compounds. The PCB-pattern in guano is clearly dominated by the most persistent congeners, PCB 153 followed by PCB 138, 118, 180, and 99 (Figure 1a), while precipitation samples are dominated by PCB 18, 105, 101, and 138 (Figure 2b). In both Bjørnøya lakes, the patterns in water (dissolved + particulate fractions) (Figure 2a) and biota (Figure 1b and c) are dominated by the same congeners that are dominant in guano samples (CB 153, 138, 118, and 180), but the relative share of these congeners is higher in Ellasjøen than in Øyangen. In addition, the samples from Lake Ellasjøen had a lower share of congeners that are more easily metabolized (28, 31, 52, 90, 101, 105, 110, 118, 149, and 156) than the samples from Lake Øyangen (Figure 1b and c, Figure 2a). To further establish guano as a source for contaminants to Lake Ellasjøen we compared the pattern in the guano samples to the pattern in samples of water, sediment, and invertebrates from the two lakes, but not fish because they are able to metabolize some of the compounds. A Correspondence Analysis (CA) on relative PCB concentrations indicates that the PCB pattern in abiotic samples and invertebrates from Lake Ellasjøen is more similar to the pattern in guano than corresponding samples from Lake Øyangen. The pattern resemblance is strongest between
FIGURE 2. Relative congener pattern in lake water from Lake Ellasjøen and Lake Øyangen (a), and rainwater (b).
FIGURE 1. Relative congener pattern in guano from little auk (Alle alle), kittiwake (Rissa tridactyla), and glaucous gull (Larus hyperboreus) (a), in zooplankton (b), and Arctic char (c) from Lake Ellasjøen and Lake Øyangen.
FIGURE 3. Relation between sum PCB (24 congeners) and δ15Nvalues in organisms from Lake Ellasjøen and Lake Øyangen. Range of PCB concentrations (4637-9127 ng/g lipid weight; Σ42 congeners) measured in the blubber of female polar bears from Svalbard (25) is shown independent of δ15N-value.
guano from kittiwake, which is the dominating species around Lake Ellasjøen, and the samples from the lake (see Figure 1S in the Supporting Information). The relative ratios of different PCB congeners provide an additional indication of different contaminant sources for the two lakes. For instance, the average ratio between the persistent congener CB 99 and the relatively easily degradable CB 101 (CB 99/CB 101) is 12.6 in Arctic char from Lake Ellasjøen and 2.9 in Arctic char from Lake Øyangen. The same ratio is 106.4 in guano from kittiwake, but only 0.5 in rainwater from Bjørnøya. However, we cannot exclude the possibility that increased cytochrome P450 activity in char as a result of the elevated contaminant levels in Lake Ellasjøen is responsible for the high PCB99/PCB101 ratio in the fish. Biomagnification. Having shown that contaminant loads in Lake Ellasjøen and Lake Øyangen are primarily attributable to different sources, we are able to contrast the impact of
these different sources on the lake ecosystems. The food chains in the two lakes are short and consist of relatively few species (algae, zooplankton, Chironomids (Chironomidae sp.), tadpole shrimps (Lepidurus arcticus, only in Øyangen) with resident Arctic char (Salvelinus alpinus)) as the top predator. Using δ15N as a proxy for trophic level, an exponential least-square best-fit regression analysis of δ15N values versus contaminant concentrations in lake organisms shows that the slopes of the curves (biomagnification) for Lake Ellasjøen and Lake Øyangen are identical (0.7 in both lakes). However, the δ15N values in Lake Ellasjøen are approximately 8 units higher than in comparable samples from Lake Øyangen, indicating that the baseline value in Lake Ellasjøen is higher than that in Lake Øyangen as s a result of the guano input (Figure 3). The guano that enters the lake has already gone through several biomagnification steps in the marine food chain and thus has a stable isotope VOL. xx, NO. xx, xxxx / ENVIRON. SCI. & TECHNOL.
9
E
FIGURE 4. Triplot based on a Redundancy Analysis (RDA) of PCBs (log-transformed) in fish samples with respect to δ15N and lipid content (arrows). The individual scores of the samples from Lake Ellasjøen (n ) 20) and Lake Øyangen (n ) 16) are mapped with circles and squares, respectively. The first RDA axis accounted for 92.9% of the variation in the data set (99.6% of this variation could be explained by the environmental variables), whereas the second RDA axis accounted for 0.4%. Numbers given in the RDA plot correspond with the PCB congener nomenclature (IUPAC no.). signal that reflects the seabird diet (Table 1). Primary producers utilize the nutrients brought into the lake with guano (especially nitrogen and phosphorus) and develop an isotopic signal that reflects this nutrient source. The guano is therefore affecting the baseline δ15N-signal in Lake Ellasjøen at the same time as it introduces previously biomagnified contaminants. A Redundancy Analysis (RDA) followed by a Monte Carlo permutation test was performed to investigate the contaminant pattern in Arctic char from the two lakes, and to explore how differences between the lakes may be accounted for by the environmental variables length, age, lipid, and trophic level (δ15N). Only δ15N and lipid contributed significantly to the model (δ15N explained 92.0% of the variance in the data set (p ) 0.001), lipid explained 7.7% of the variance (p ) 0.001)). Length and age were not significant explanatory variables (p > 0.05), and were therefore excluded from the analyses. The main correlation pattern between dependent variables (indicated by thin arrows) and independent variables (indicated by bold arrows) is presented, together with the scores from the individual samples in Figure 4. The Arctic char samples from the two lakes separate quite clearly along the horizontal axis. The first RDA-axis accounts for 92.9% of the variation in the data set, whereas the second axis accounts for 0.4%. The triplot also shows that δ15N and lipid form two distinct, uncorrelated gradients. Concentrations of the most persistent compounds (PCB 180, 187, 170, 153, 138, 99, 118) are significantly correlated with δ15N-values, whereas the less persistent compounds (PCB 18, 31, 52) are more closely correlated with lipid content (Figure 4). In a previous study we compared levels of PCBs in the fish populations from Lake Ellasjøen and Lake Øyangen by means of an analysis of covariance (ANCOVA). The results from that analysis showed that if PCB concentrations in Arctic char were adjusted for trophic level (δ15N) and % lipid, the concentration differences between the lakes disappear (10). These results clearly indicate that without inputs from seabirds (i.e., no difference in stable isotope concentrations), contaminant concentrations would have been comparable in the two lakes. This indicates that an additional PCB load, which we attribute to seabird guano, is responsible for a striking increase in PCB levels in Arctic Char from Lake Ellasjøen (Figure 3). Average PCB concentrations in Arctic Char from Lake Ellasjøen (1.6 ( 1.5 mg/kg ww) exceed by a factor of 4 even the least restrictive United States Environmental F
9
ENVIRON. SCI. & TECHNOL. / VOL. xx, NO. xx, xxxx
Protection Agency guidelines for the consumption of fish by adults (U.S. EPA noncancer health endpoint for consumption of fish by adults ) 0.39 mg/kg ww). On a lipid-normalized basis the PCB levels in fish from Lake Ellasjøen are substantially higher than levels in top predator marine mammals from the Barents Arctic (Figure 3) (25). Pilot studies of Arctic char from Lake Ellasjøen have indicated that these levels are high enough to cause potential harmful effects on endocrine function and cellular homeostasis (26). Error Propagation. We performed an error propagation analysis to evaluate the uncertainties in parameter values used in the mass balance assessment. The analysis showed that significant uncertainty is linked to the determination of POPs fluxes into the catchment areas via precipitation. Despite the number and widespread distribution of sample collectors, the mean concentration value for ΣPCB in precipitation exhibited a high standard deviation that when propagated through the calculations resulted in standard deviations of (80% for LRTP supplies to lake catchment areas (AY). This alone has little impact on comparisons between the two lakes because the same ΣPCB concentrations for rain (Cr) and snow (Cs) were used to calculate LRTP fluxes into both lake catchment areas. Uncertainty in these terms does, however, affect the determination of the catchment area-lake exchange coefficient (Ex(LRTP)), a term that is used to quantify the amount of ΣPCB entering Lake Ellasjøen from seabird guano (LE(G)). However, even if the transfer efficiency is increased by a factor of 2, guano remains the dominant contaminant source to Lake Ellasjøen (∼ 60%). The resulting degree of uncertainty in the Lake Ellasjøen ΣPCB load estimate attributed to guano is (40%. The contaminant load in guano was not directly used in the mass-balance calculations (only to calculate transfer efficiency from guano), thus changing the guano calculations will not affect the massbalance results. Further the mass balance results are strongly supported by the patterns of stable isotopes and PCB congeners as well as ΣPCB concentrations in all measured environmental compartments. Taken together, the evidence supports our contention that guano is the dominant source of contaminants to Lake Ellasjøen in comparison to LRTP supplies from the catchment area. Biological Transport. Enhanced nutrient levels and the concomitant exchange of contaminants to remote Arctic lakes is a phenomenon which has previously not been well characterized because current monitoring procedures have traditionally focused on the need to detect regional patterns and trends in atmospheric LRTP. We provide quantitative evidence demonstrating that even relatively small populations of certain seabird species lead to major impacts for ecosystems. In the present example, seabird guano accounts for ∼14% of the contaminant inventory of the Lake Ellasjøen catchment area, but ∼80% of the contaminant inventory in the lake itself. We have further shown that this biological transport mechanism is an important contaminant exposure route for ecosystems, leading to extremely high levels of contaminants in lake biota. The evidence presented herein together with the identification of similar areas in the Canadian Arctic (9) suggests that the redistribution and focusing of contaminants in localized areas by seabirds may be far more widespread than currently known. Furthermore, this transport route which includes extra biomagnification steps is not limited to Arctic lake ecosystems, as seabird colonies are also present in coastal areas throughout the world. Although it is likely that contaminant effects on biota are not as pronounced in marine areas due to more efficient contaminant dilution processes, elevated stable isotope levels have also been observed in some coastal areas affected by seabirds (27). While data on contaminant levels in these areas are not yet available, our findings indicate that biological transport by seabirds may
be an important factor determining contaminant concentrations also in coastal organisms. Furthermore, it should not be assumed that the impact of this process on human populations whose diets include resources harvested from seabird inhabited areas is inconsequential.
Acknowledgments The project was financed by the Norwegian Research Council (projects 135064/720, 153411/720, 135632/720). We acknowledge the contributions of the Norwegian Coast Guard and the Rescue helicopters (330-squadron) for safe transport to and from Bjørnøya, as well as employees at Bjørnøya meteorological station for practical assistance and accommodation. Department of Applied Environmental Science, Stockholm University, contributed with PCB analyses in water. We thank Dr. Miriam Diamond, University of Toronto, for useful discussions on Arctic lake transport processes. Three anonymous reviewers provided valuable advice and suggestions that have significantly improved the presentation of this work. We also thank Dr. Derek Muir, Prof. Dag Hessen, Prof. Will Ambrose, and Prof. Ian Lerche for valuable comments on earlier versions of the manuscript.
Supporting Information Available Description of chemical analyses of water and biota. A figure showing the output of a correspondence analysis. This material is available free of charge via the Internet at http:// pubs.acs.org.
Literature Cited (1) de March, B. G. E.; de Wit, C.; Muir, D. C. G.; Braune, B. B.; Gregor, D. J.; Norstrom, R. J.; Olsson, M.; Skaare, J. U.; Stange, K. Persistent Organic Pollutants. In AMAP Assessment Report: Arctic Pollution Issues; Arctic Monitoring and Assessment Programme (AMAP): Oslo, Norway, 1998; Ch. 6, pp 183-372. (2) Macdonald, R. W.; Barrie, L. A.; Bidleman, T. F.; Diamond, M. L.; Gregor, D. J.; Semkin, R. G.; Strachan, W. M. J.; Li, Y. F.; Wania, F.; Alaee, M.; Alexeeva, L. B.; Bascus, S. M.; Bailey, R.; Bewers, J. M.; Gobeil, C.; Halsall, C. J.; Harner, T.; Hoff, J. T.; Jantunen, L. M. M.; Lockhart, W. L.; Mackay, D.; Muir, D. C. G.; Pudykiewicz, J.; Reimer, K. J.; Smith, J. N.; Stern, G. A.; Schroeder, W. H.; Wagemann, R.; Yunker, M. B. Contaminants in the Canadian Arctic: 5 years of progress in understanding sources, occurrence and pathways. Sci. Tot. Environ. 2000, 254, 93-234. (3) AMAP. AMAP Assessment 2002: Persistent Organic Pollutants in the Arctic; Arctic Monitoring and Assessment Programme: Oslo, Norway, 2004; 309 pp. (4) Wang, X. M.; Ding, X.; Mai, B. X.; Xie, Z. Q.; Xiang, C. H.; Sun, L. G.; Sheng, G. Y.; Fu, J. M.; Zeng, E. Y. Polybrominated diphenyl ethers in airborne particulates collected during a research expedition from the Bohai Sea to the Arctic. Environ. Sci. Technol. 2005, 39, 7803-7809. (5) Evenset, A.; Christensen, G. N.; Kallenborn, R. Selected chlorobornanes, polychlorinated naphthalenes and brominated flame retardants in Bjørnøya (Bear Island) freshwater biota. Environ. Pollut. 2005, 136, 419-430. (6) Muir, D. C. G.; Backus, S.; Derocher, A. E.; Dietz, R.; Evans, T. J.; Gabrielsen, G. W.; Nagy, J.; Norstrom, R. J.; Sonne, C.; Stirling, I.; Taylor, M. K.; Letcher, R. J. Brominated Flame Retardants in Polar Bears (Ursus maritimus) from Alaska, the Canadian Arctic, East Greenland, and Svalbard. Environ. Sci. Technol. 2006, 40, 449-455. (7) Helm, P. A.; Diamond, M. L.; Semkin, R.; Strachan, W. M. J.; Teixeira, C.; Gregor, D. A mass balance model describing multiyear fate of organochlorine compounds in a high Arctic lake. Environ. Sci. Technol. 2002, 36, 996-1003. (8) Diamond, M.; Bhavsar, S. P.; Helm, P. A.; Stern, G. A.; Alaee, M. Fate of organochlorine contaminants in arctic and subarctic lakes estimated by mass balance modelling. Sci. Tot. Environ. 2005, 342, 245-259.
(9) Blais, J. M.; Kimpe, L. E.; McMahon, D.; Keatley, B. E.; Mattory, M. L.; Douglas, M. S. V.; Smol, J. P. Arctic seabirds transport marine-derived contaminants. Science 2005, 309, 445. (10) Evenset, A.; Christensen, G. N.; Skotvold, T.; Fjeld, E.; Schlabach, M.; Wartena, E.; Gregor, D. A comparison of organic contaminants in two high Arctic lake ecosystems, Bjørnøya (Bear Island), Norway. Sci. Tot. Environ. 2004, 318, 125-141. (11) Pacyna, J.; Oehme, M. Long range transport of some organic compounds to the Norwegian Arctic. Atmos. Environ. 1988, 22, 243-257. (12) Theisen, F. Dokumentasjon og vurdering av verneverdier på Bjørnøya; Meddelelser no. 143; Norwegian Polar Institute: Tromsø, Norway, 1997; 96 pp + appendix. (13) Mehlum, F.; Gabrielsen, G. W. Energy expenditure and food consumption by seabird populations in the Barents Sea region. In Ecology of Fjords and Coastal Waters; Skjoldal, H. R., Hopkins, C., Erikstad, K. E., Leinaas, H. P., Eds.; Elsevier: Amsterdam, 1995; pp 457-470. (14) Gabrielsen, G. W.; Taylor, J. R. E.; Konarezewski, M.; Mehlum, F. Field and laboratory metabolism and thermoregulation in dovekies (Alle-alle). Auk 1991, 108, 71-78. (15) Golovkin, A. G. Colonial sea birds in marine biocenosis. Doctor of Biological Science Thesis, Moscow State University, Moscow, 1991. (16) Brekke, B.; Gabrielsen, G. W. Assimilation efficiency of adult kittiwakes and Bru ¨ nnich’s guillemots fed capelin and Arctic cod. Polar Biol. 1994, 14, 279-284. (17) Jeremiassen, J. D.; Eisenreich, S. J.; Paterson, M. J.; Beaty, K. G.; Hecky, R.; Elser, J. J. Biogeochemical cycling of PCBs in lakes of variable trophic status: A paired-lake experiment. Limnol. Oceanogr. 1999, 44, 889-902. (18) Wolfarth, B.; Lemdahl, G.; Olsson, S.; Persson, T.; Snowball, I.; Ising, J.; Jones, V. Early Holocene environment on Bjørnøya (Svalbard) inferred from multidisciplinary lake sediment studies. Pollut. Res. 1995, 14, 253-275. (19) Axelman, J.; Broman, D.; Na¨f, C. Vertical flux and particulate/ water dynamics of Polychlorinated Biphenyls (PCBs) in the open Baltic sea. Ambio 2000, 29, 210-216. (20) Derocher, A. E.; Wolkers, H.; Colborn, T.; Schlabach, M.; Larsen, T. S.; Wiig, Ø. Contaminants in Svalbard polar bear samples archived since 1967 and possible population level effects. Sci. Tot. Environ. 2003, 301, 163-174. (21) Gower, J. C. Introduction to ordination techniques. In Developments in Numerical Ecology; Legendre, P., Legendre, L., Eds.; NATO ASI, series G, Vol. 14; Springer-Verlag: Berlin, 1987; pp 3-64. (22) Ter Braak, C. J. F. Ordination. In Data Analysis in Community and Landscape Ecology; Jongman, R. G. H., Ter Braak, C. J. F., Van Tongeren, O. F. R., Eds.; Cambridge University Press: Cambridge, UK, 1995; pp 91-173. (23) Ewald, G.; Larsson, P.; Linge, H.; Okla, L.; Szarzi, N. Biotransport of organic pollutants to an inland Alaska lake by migrating sockeye salmon (Oncorhynchus nerka). Arctic 1998, 51, 40 - 47. (24) Kru ¨ mmel, E. M.; Macdonald, R. W.; Kimpe, L. E.; Gregory-Eaves, I.; Demers, M. J.; Smol, J. P.; Finney, B.; Blais, J. M. Delivery of pollutants by spawning salmon - Fish dump toxic industrial compounds in Alaskan lakes on their return from the ocean. Nature 2003, 425, 255. (25) Verreault, J; Muir, D. C. G.; Norstrom, R. J.; Stirling, I.; Fisk, A. T.; Gabrielsen, G. W.; Derocher, A. E.; Evans, T. J.; Dietz, R.; Sonne, C.; Sandala, G. M.; Gebbink, W.; Riget, F. F.; Born, E. W.; Taylor, M. K.; Nagy, J.; Letcher, R. J. Chlorinated hydrocarbon contaminants and metabolites in polar bears (Ursus maritimus) from Alaska, Canada, East Greenland, and Svalbard: 19992002. Sci. Tot. Environ. 2005, 351, 369-390. (26) Jørgensen, E. H.; Vijayan, M. M.; Killie, J.-E. A.; Aluru, N.; AasHansen, Ø.; Maule, A. Toxicokinetics and effects of PCBs in arctic fish: A review of studies on Arctic char. J. Toxicol. Environ. Health 2006, 69, 37-52. (27) Wainright, S. C.; Haney, J. C.; Kerr, C.; Golovkin, A. N.; Flint, M. V. Utilization of nitrogen derived from seabird guano by terrestrial and marine plants at St. Paul, Pribilof Islands, Bering Sea, Alaska. Mar. Biol. 1998, 131, 63-71.
Received for review September 5, 2006. Revised manuscript received December 5, 2006. Accepted December 10, 2006. ES0621142
PAGE EST: 6.9
VOL. xx, NO. xx, xxxx / ENVIRON. SCI. & TECHNOL.
9
G
