Environ. Sci. Technol. 2002, 36, 5057-5061
Latitudinal Fractionation of Polybrominated Diphenyl Ethers and Polychlorinated Biphenyls in Frogs (Rana temporaria) A R N O U T F . H . T E R S C H U R E , * ,† P E R L A R S S O N , † J U H A M E R I L A¨ , ‡ A N D K. INGEMAR JO ¨ NSSON§ Chemical Ecology and Ecotoxicology, Department of Ecology, Lund University, Ecology Building, S-223 62 Lund, Sweden, and Ecological Genetics Research Unit, Department of Ecology & Systematics, P.O. Box 65, FIN-00014, University of Helsinki, Helsinki, Finland
The flame retardant polybrominated diphenyl ethers (PBDEs) have become ubiquitous environmental pollutants. The environmental distribution of PBDEs is much less studied than that of the polychlorinated biphenyls (PCBs). To compare the environmental fate of the PCBs with their partial substitute, the PBDEs, common frogs (Rana temporaria) were collected along a ∼1500-km-long latitudinal gradient of the Scandinavian Peninsula and their livers analyzed for PCBs and PBDEs. Mean concentrations of total PCBs and BDE47 ranged from 9200 to 92 900 and 30 to 120 ng kg-1 fresh weight, respectively, whereas BDE99 was detected in less than 50% of the frogs. PCB concentrations were higher than that of the PBDEs, and the differences decreased in the northern latitudes. Moreover, the pollutant concentrations in frog livers were negative functions of latitude. The observed scatter and regression slopes imply several influencing factors (such as habitat, exposure route, uptake, metabolism, excretion, etc.) and indicate release events instead of the grasshopper effect. Biological variables such as gender, age, body size, and lipid content did not influence pollutant concentrations in the frog livers. The total PCB and BDE47 concentrations in frogs were highly correlated. Hence, their environmental fate is analogous and our results highlight the need to examine the potential role of xenobiotics on amphibian populations.
Introduction In conjunction with the global amphibian decline (1), frogs are recognized as important bioindicators of environmental quality (2). However, the presence of persistent organic pollutants (POPs), such as polychlorinated biphenyls (PCBs) and polybrominated diphenyl ethers (PBDEs), in amphibians has received remarkably little attention (3), despite the fact that frogs are cosmopolitan and act as both prey and predator. In addition, frog densities can be high depending on season, habitat, food supply, and geography (4). Being poikilothermic, these long-lived animals have a lower metabolic rate than * Corresponding author. Phone: + 46 (0)46 2224598. Fax: + 46 (0)36 2223790. E-mail:
[email protected]. † Chemical Ecology and Ecotoxicology, Department of Ecology, Lund University. ‡ University of Helsinki. § Theoretical Ecology, Department of Ecology, Lund University. 10.1021/es0258632 CCC: $22.00 Published on Web 10/26/2002
2002 American Chemical Society
birds or mammals. Furthermore, they use the liver for deposition of lipids (3). These are all qualities that favor the accumulation of the lipid-soluble pollutants PCBs and PBDEs in frogs. PCBs have been widely used as flame retardants, coolants, and lubricants in electrical equipment. The total world production is estimated at 1.5 million metric tonnes (5). However, by the mid-1970s, usage had ceased by most industrialized countries. PCBs are known for moderate toxic effects on development and metamorphosis of several frog species (ref 6 and ref 7 in ref 3) and environmental persistence and are subject to long-range transport (8). PBDEs, unlike PCBs, are still applied as additive flame retardants to electrical equipment used for consumer products. The increasing global consumption was estimated at 67 125 metric tonnes in 1999 (9), which is comparable to maximum PCB production during the early 1970s. PBDEs have similar physical-chemical properties as PCBs and thus are suspected of a similar behavior in the environment (10). The level of BDE47 in human breast milk is known to double every 5 years since 1972, whereas PCB levels are declining (11). PBDEs have recently also been reported in freshwater (12, 13), marine biota (10, 14), and sediment (14, 15), and adverse effects have been documented on biota (16, 17). Because of their toxic, bioaccumulative, and long-range transport potential, PBDEs are of growing environmental concern. Nevertheless, compared to PCBs, their spatial distribution in the environment is currently not well documented. The atmosphere is, however, recognized as the major transport route for POPs in the environment, and the behavior of PCBs in the atmosphere is well understood (18). Recently, PBDEs have also been reported in remote air (19) and atmospheric deposition (20). Accordingly, the long-range and worldwide atmospheric transport of POPs has been described by the “global fractionation theory” (8, 21); i.e., organic compounds become latitudinally fractionated according to their volatility and condense at different ambient temperatures. Volatile compounds therefore travel more easily to the cold polar regions than less volatile pollutants, both through either one single release event, followed by deposition, or through a series of repeated air-surface exchange events (“the grasshopper effect”). Different matrixes such as soil, water, and air have been analyzed for POPs to study spatial trends and the global fractionation theory (22-24). Since airborne pollutants, such as PCBs and PBDEs, may ultimately deposit on the earth’s surface where they become available to organisms via different exposure routes, a latitude-related dominance difference of persistent pollutants in biota is observed (8, 21). To investigate the long-range transport hypothesis for PBDEs, we compared PCB and PBDE concentrations in frogs along a 1500-km-long south-to-north gradient. Results are applied to the global fractionation theory, and the possible role of these xenobiotics in the global decline of amphibian populations is discussed.
Experimental Section Sampling Sites and Sampling Techniques. From 1998 to 2000, 12-34 mature common frogs (Rania temporaria) were collected at each of seven locations over a ∼1500-km-long south-to-north gradient on the Scandinavian Peninsula (Figure 1) during the early phase of the spawning season (April-June depending on locality), just after hibernation. All localitiessexcept for locality 5, situated in the city center of Umeåswere rural or near to pristine sites, located at least 20 km from the nearest town. The frogs had a similar sex VOL. 36, NO. 23, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5057
nological method (27). In short, from each individual, the number of resting lines on hematoxylin-stained cross sections of decalcified phalanges were counted, assuming that each resting line corresponded to one hibernation. Because of the resorption of innermost resting lines in periosteal bone in part of the individuals, the number of missing resting lines was estimated (28). Quality Assurance and Statistics. Procedural blanks (n ) 21) were processed simultaneously with the liver samples and used for limits of quantification (LOQ), defined as the signal-to-noise (S/N) ratio of >10 and in case of interference defined as three times average blank sample levels. Results from the mass spectrometry analysis were used as controls for the GC/ECD analysis, and concentrations were not found to be significantly different between both analytical methods. Pollutant concentrations below the LOQ were excluded from statistical analysis. All variables, including log transformed concentrations, were tested for normality with P-P plots using the Tukey’s proportional estimation. Regression analysis and regression analysis of variance between biological variables and pollutant concentrations were performed using a general linear model (GLM) multivariate procedure. Interactions between factorsslocation and gendersas well as their individual effects were investigated. For regression analysis, we specified the independent variablessage, length, weight, and lipid contentsas covariates. All tests had a 0.05 significance level and a 95% confidence interval.
Results and Discussion FIGURE 1. Sampling locations in the Scandinavian Peninsula region: (1) Lund, (2) Blekinge, (3) Lindrågen, (4) Uppsala, (5) Umeå, (6) Ammarna1 s, and (7) Kiruna. Location details are presented in Table 1. ratio and covered different year classes. After capture, the animals were maintained at ∼+4 °C, and within 1-2 days transferred to laboratory where they were anaesthetized with an overdose of MS-222 and dissected. Livers were immediately frozen and kept at -25 °C until lipid extraction. Sample Extraction and Cleanup. After thawing, octachloronaphthlene (OCN, 100 µL at 8.5 ng mL-1) was added as an extraction standard and the frog livers were homogenized individually in a Bligh and Dyer solution (25), with an UltraTurrax mixer. Dichloromethane was used instead of chloroform. The extracts were evaporated until dry in a vacuum centrifuge (Savant), and lipid weight was measured gravimetrically. Afterward, 400 µL of n-hexane was added. Fractionation and purification was performed on an open column (26). Finally, the eluted fraction containing all compounds of interest was evaporated until dry in a vacuum centrifuge (Savant), redissolved in 75 µL of isooctane, and analyzed. Instrumentation. Quantitative analysis of PCBs and PBDEs was performed on a Varian 3400 gas chromatograph using an electron capture detector (ECD). The instrument was equipped with a 10-m methyl-deactivated precolumn connected to a 30-m DB5-HT capillary column (0.25-mm i.d., 0.1-µm film thickness, J&W Scientific), and injection was by a septum-equipped programmable injector (SPI), fitted with a high-performance insert. The injector was 80 °C, followed by a 150 °C min-1 increase to 300 °C (17 min). The column temperature was programmed as follows: 80 °C (1 min), 15 °C min-1 to 300 °C (15 min). Qualitative analysis of two randomly chosen samples from each location and all blanks (n ) 21) were performed on a Finnigan TSQ 700 mass spectrometer (20). Age Determination. The age of the individuals was determined from bone cross sections using a skeletochro5058
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 23, 2002
PBDE and PCB Concentrations. Several PCBs and BDE47 were found in greater than 90% of the individuals (Table 1). BDE99 was, however, detected in less than 50% of the frogs, and therefore, these results were not used in the statistical analysis. Both PCB52 and PCB153 are the dominant PCB congeners in the frog livers, and reported PCB levels in frogs are often higher or comparable to those found in other organisms (29, 30). In this case, the highest concentrations were observed in Lund; 13 700 and 9000 ng kg-1 fresh weight for PCB52 and PCB153, respectively. We could only compare the mean PCB52 concentrations (Table 1) with one other study in which PCB52 levels in frog livers from two rural sites in the northern parts of Siberia ranged between 650 and 7190 ng kg-1 fresh weight (31). Hence, our results are similar. PBDEs have to our knowledge not yet been reported in Anurans (tail-less frogs) from the wild, although it has recently been shown that consumption of insects containing PBDEs is a possible uptake route for frogs (32). Relationships of PBDE Concentrations with Biological Variables and PCB Levels. No significant relationships between the biological variables gender (MANOVA: P > 0.16 all cases), body size, age, lipid content (ANCOVA: P > 0.05 all cases), and pollutant concentrations in the frog livers were observed. This corresponds with previous results (33). It follows that the variance in the pollutant levels of the amphibians can be explained more by geographical and physical variables. Noteworthy, other factors such as latitudinal variations in habitat, prey, and length of hibernation period may also influence POP concentrations. As in the case of salmonids from the Great Lakes (34) and freshwater fishes in Virginia (12), the total PCB and BDE47 concentrations were positively correlated (Pearson correlation: r ) 0.692, P < 0.001), indicating similar contaminant sources (Figure 2). Since these PBDE congeners also dominate in biota from various other parts of the world (10, 12-15), the global environment appears to have become ubiquitously contaminated with PBDEs in a manner comparable to that of PCBs. Spatial Trend and Comparison to the Global Fractionation Theory. For comparison between the spatial PCB and
TABLE 1. Summary Information of Geographical, Physical, Biological, and Concentration Data location
(1) Lund
(2) Blekinge 56.20 280.35 18 (12, 6) 4 (4, 4) 65.1 ( 6.0 32.9 ( 9.2 0.4 ( 0.1 2.7 ( 4.8
(3) Lindrågen 59.90 278.22 12 (8, 4) 5 (5, 5) 76.3 ( 4.2 d 0.4 ( 0.3 5.0 ( 1.9
(4) Uppsala 60.00 278.88 29 (14, 15) 4 (4, 4) 74.5 ( 5.9 47.6 ( 14.6 0.8 ( 0.3 2.1 ( 0.6
(5) Umeå
(7) Kiruna
65.54 272.25 34 (22, 12) 7 (7, 7) 80.8 ( 4.1 60.0 ( 11.1 1.1 ( 0.2 1.5 ( 0.3
67.51 271.75 28 (13, 15) 7 (7, 7) 69.2 ( 2.9 37.1 ( 5.5 0.6 ( 0.1 2.8 ( 0.8
55.42 281.20 26 (14, 12) 3 (3, 4) 65.8 ( 5.8 33.8 ( 11.4 0.4 ( 0.1 3.1 ( 2.0
BDE47 BDE99
Polybrominated Diphenyl Ether (PBDE)e 123 ( 100(24)f 58 ( 2(16) 94 ( 5(4) 42 ( 22(29) 90 ( 85(29) 34 ( 91(28) 85 ( 73(9) 52 ( 3(2) 34 ( 29(5) 28 ( 11(13) 178 ( 90(22) 84 ( 10(17)
PCB52 PCB153 PCB183 PCB201 PCB206
2002 ( 480 2289 ( 1974 564 ( 872 173 ( 259 16 ( 20
Polychlorinated Biphenyl (PCB)e 1657 ( 335 198 ( 286 111 ( 205 1825 ( 1170 2082 ( 1883 982 ( 694 215 ( 114 421 ( 602 150 ( 100 123 ( 96 145 ( 140 77 ( 48 10 ( 8 14 ( 15 5(3
64.27 276.10 30 (16, 14) 5 (6, 5) 80.5 ( 6.5 62.7 ( 15.9 1.2 ( 0.4 2.3 ( 1.4
(6) Ammarna1 s
latitude (°N) temp (K, 1961-1990)a no. of samples (M, F) frog age (years; M, F)b frog length (mm)c frog weight (g)c fresh wt liver (g)c fat content liver (% of fw)c
109 ( 128 1895 ( 1093 544 ( 831 97 ( 120 10 ( 14
76 ( 57 334 ( 208 36 ( 25 26 ( 13 2(1
26 ( 10(27) (0)g 52 ( 37 350 ( 283 36 ( 27 32 ( 19 2(2
a Mean temperatures calculated from reference norms collected by weather stations positioned in the same regions from the Swedish Meteorological and Hydrological Institute (SMHI) during 1961-1990. b Mean age from both males and females. Age is expressed in whole years. c Mean ( SD. d Not measured. e Mean ( s.d. in ng kg-1 fresh weight liver. f Values in parentheses are the number of samples quantified’ i.e., levels are above the LOQ. g Penta-BDE99 was not detected in any of the samples.
FIGURE 2. Correlation between the logarithm of total PCB and BDE47 concentrations (ng kg-1 fresh weight) in all samples (r2 ) 0.480, n ) 177). PBDE trend, it might be more appropriate to distinguish between single congeners, as recently proposed (34). Indeed, in our study, the total PCB concentration is the sum of five congeners, whereas the PBDE concentration is given by just one congener, BDE47. As predicted by the global fractionation theory, the PCB and PBDE concentrations in frog liver declined with increasing latitude (Figure 3, Table 1). Sellstro¨m (39) observed a similar trend in freshwater fish from Sweden, indicating that southern Sweden is more contaminated with PBDEs than northern Sweden. Moreover, decreasing PBDE levels in cod liver in the North Sea from south to north (22360 to 2-68 ng g-1 lipid, respectively) have been found (40). These observations, together with our results indicate a gradient with PBDE levels declining from south to north, with high levels in the source area, i.e., the south of Sweden, and low levels in the Arctic. Because of the negative correlation between latitude and mean ambient temperature across the study populations (Pearson correlation: r ) -0.931, P < 0.001), the inference is the same if the latitude in these comparisons is substituted with mean temperatures calculated from reference norms of 1961-1990 (Table 1). Hence, the results support the global fractionation theory and we hypothesize that similar to PCBs there is an ongoing atmospheric transport of PBDEs to remote areas. The latitude-related dominance difference was, however, opposite to that predicted by the theory (8, 21). Namely,
the concentration of PCB52 decreased over latitude with a factor of 3 more than that of the almost 1000 times less volatile and 2 times heavier flame retardant BDE47 and the other four PCBs of interest (Table 2, Figure 3). Even though PCB52 and BDE47 are structurally similar, i.e., both congeners have either four chlorine or four bromine substitutes, the latter has a long-range transport pattern that is more comparable with that of the other PCBs and these congeners have also similar slopes (-0.04, -0.05 to -0.07, respectively). The vapor pressures and molecular masses of the hexa- to nona-PCB congeners are also more similar to that of BDE47 than that of PCB52 (Table 2). Hence, the hexa- to nonachlorobiphenyls and BDE47 are pollutants with a low mobility, conforming to a uniform latitudinal gradient as proposed by Wania and Mackay (21). The regression slopes are, however, shallow, which has recently also been found for PCBs in soils by Meijer et al. (22). It was postulated that several soil-related factors such as organic matter, soil type, land use, etc., influenced the observed patterns. However, the authors reported a fair amount of scatter in their data (r2 varied between 0.035 and 0.387), whereas although in our case the regression slopes are even shallower, the scatter is considerably less (Table 2). Hence, our results support the suggested single-release event instead of the grasshopper effect (22, 23) and imply additional or different factors affecting the observed spatial concentration trend in frog livers than those discussed by Meijer et al. (22). Although we found no significant relationships between several biological variables and pollutant concentrations (see previous section), other factors beside those discussed earlier, such as different exposure routes, uptake, metabolism, and excretion, have to be considered. Furthermore, the major part of the sampling area is within the “temperate industrial band”, i.e., between 30 and 60°N, where 95% of the global PCB usage occurred (41), causing proximity to potential sources such as urbanization. PBDEs, in contrast to PCBs, are used in open systems such as electronics throughout mainly the industrialized countries in the Northern Hemisphere, such as Europe and North America (9). Our reasoning is strengthened by the fact that for all the pollutants analyzed only PCB52 had no elevated concentrations at location Umeå. Moreover, while at the other six rural and remote locations the PCB52 concentrations were significantly higher than that of BDE47 (independent sample T-tests: t < -2.31, df > 10, P < 0.03 all cases), no statistical VOL. 36, NO. 23, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5059
FIGURE 3. Relationship between latitude (°N) and the natural logarithm of pollutant concentrations in frog livers (ng kg-1 fresh weight). Std dev are presented in Table 1. The slope of regression (B), the level of significance (p), and the regression coefficient (r2) obtained by linear regression are given in Table 2. The colored background represents the temperature gradient over latitude (not to scale); i.e., warmer (southern) regions are shown with shades of red and colder (polar) regions are shown with shades of blue.
TABLE 2. Physical-Chemical Properties of the PBDEs and PCBs (Molecular Weight, Vapor Pressure, Lipophilicity) and Summary of Regression Slopes, p-Values, and r2 for the Compounds Analyzed
compd
substitution pattern
MW
log Vp (Pa) at 25 °C (35, 36)
BDE47 BDE99 PCB52 PCB153 PCB183 PCB201 PCB206
2,2′,4,4′-tetrabromo 2,2′,4,4′,5-pentabromo 2,2′,5,5′-tetrachloro 2,2′,4,4′,5,5′-hexachloro 2,2′,3,4,4′,5′,6-heptachloro 2,2′,3,3′,4,5,5′,6′-octachloro 2,2′,3,3′,4,4′,5,5′,6,-nonachloro
485.8 564.7 292.0 360.9 397.3 431.8 466.2
-4.66 -4.90 -1.79 -3.17 -3.45 -3.76 -4.96
a
log KOW (37, 38)
slope ( B)
significance (p-value)
r2
5.87-6.16 6.46-6.97 5.84 6.92 7.20 7.62 8.09
-0.043 a -0.118 -0.064 -0.070 -0.051 -0.049
