Environ. Sci. Technol. 2002, 36, 2146-2153
Influence of Environmental Variables on the Spatial Distribution of PCBs in Norwegian and U.K. Soils: Implications for Global Cycling S . N . M E I J E R , * ,† E . S T E I N N E S , ‡ W . A . O C K E N D E N , †,§ A N D K . C . J O N E S † Environmental Science Department, IENS, Lancaster University, Lancaster LA1 4YQ, U.K., and Department of Chemistry, Norwegian University of Science and Technology, NO-7491 Trondheim, Norway
This paper reports the influence of environmental variables on soil concentrations of polychlorinated biphenyls (PCBs) and their global fractionation. Soils were sampled from remote woodland (coniferous and deciduous) and grassland locations on a latitudinal transect through the United Kingdom and Norway. Different processes control PCB concentrations and burdens in coniferous, deciduous, and grassland soil systems; these are discussed, with emphasis on the influence of canopy scavenging and soil organic matter (OM) content. In general, concentration differences between sites were 1-2 orders of magnitude for lighter PCBs and 2-3 orders of magnitude for heavier PCBs, when expressed on a pg g-1 dry weight basis. These differences decreased by up to an order of magnitude when expressed as pg g-1 OM. The dataset suggests that the more volatile PCBs are moving toward equilibrium with the OM burden of the soil compartment on a European regional scale, while the distribution of the “stickier”, heavier homologues appears to still be primarily influenced by their preferential deposition closer to source areas. The relative concentration of the tri- and tetra-PCBs increases with latitude, while that of the hepta- and octa-PCBs decreases, consistent with the global fractionation theory. However, the regression slopes are quite shallow, with high scatter, implying that many environmental and soilrelated factors (such as precipitation, organic carbon content and type, other soil properties, local sources, etc.) are also influencing the observed congener patterns. Temperature-driven fractionation, while clearly operating and detectable, needs to be considered in this broader context.
Introduction Soils play an important role in the global fate and distribution of persistent organic pollutants (POPs). They are a major reservoir and sink for POPs due to their large capacity for these compounds (1, 2). Soil burdens of POPs are a function of inputs and loss mechanisms. In background soils, atmo* Corresponding author phone: +44 1524 593922; fax: +44 1524 593985; e-mail:
[email protected]. † Lancaster University. ‡ Norwegian University of Science and Technology. § Present address: The Scientific World, Cherwell Innovation Centre, 77 Heyford Park, Upper Heyford, Bicester, Oxfordshire OX25 5HD, U.K. 2146
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002
spheric deposition can be assumed to be the sole input mechanism for POPs. Inputs will vary with proximity to sources (such as cities, waste disposals, incinerators, contaminated sites) and deposition processes will vary with land use (vegetation type) and environmental variables such as temperature and precipitation as well as the properties of the chemical. Loss mechanisms include biodegradation, volatilization, and burial in deeper soil layers (e.g., through bioturbation). Permanent retention in soil organic matter will also lead to decreased availability of POPs for exchange with the atmosphere. Loss rates will be influenced by soil properties, environmental conditions, and the physical/ chemical properties of the compounds in question. It is not clear precisely how each of these input and loss processes contributes to the overall burdens of POPs in soil. Likewise, it is not known how much these processes influence fate and transport on a regional or global scale, although it is thought that air-soil exchange is important in the global cycling of POPs because soil can act as a potential source to the atmosphere (3). The ideas of “global distillation” and “cold condensation” were first introduced in the 1970s to describe the process whereby semivolatile chemicals such as POPs would volatilize from warm source/usage areas, undergo long-range atmospheric transport (LRAT), and subsequently condense onto surfaces such as soil, vegetation, or snow at low temperatures, effectively accumulating in the Polar Regions (4, 5). Wania and Mackay took this idea a step further and postulated that a “global fractionation” effect would occur, whereby a POPs mixture would become fractionated during LRAT, based on the ambient temperature and the physical/chemical properties of the individual compounds, causing more volatile compounds to be preferentially transported to and deposited in higher latitudes, whereas less volatile compounds would remain closer to source regions (5, 6). They envisaged that this would happen either in one single event of release, followed by deposition, or alternatively in a series of “hops” through repeated air-surface exchange (the “grasshopper effect”). It is important to note that, although both the “single hop” and the “grasshopping” scenario would lead to fractionation, repeated air-surface exchange (grasshopping) would cause the fractionation effect to be more pronounced and would also be more likely to lead to a net increase of POPs in polar regions (i.e., cold condensation). The aim of the present study was 2-fold: (1) to investigate the influence of environmental and soil-related variables on soil burdens along a latitudinal transect and (2) to find evidence for the global fractionation theory in soils and to gain insight into how some of these variables may affect this process. PCBs were selected as the test compounds because of their wide range of physical/chemical properties. Several studies have looked for evidence of the global fractionation theory by studying spatial trends of POPs in different matrixes such as soil, air, water, and vegetation (6-12). For example, a previous much smaller study has looked at spatial trends of POPs in soils on a latitudinal transect from the south of the U.K. to the north of Norway (9). Although some evidence for global fractionation was found, the results were confounded by the fact that the concentrations and congener patterns in U.K. soils (mainly low organic carbon grassland soils) were very different from the Norwegian soils (high organic carbon forest soils). It was suggested that more forest samples from the U.K. and grassland samples from Norway were needed in order to investigate whether these differences were due to differences in land use and soil type or due to spatial differences. It has 10.1021/es010322i CCC: $22.00
2002 American Chemical Society Published on Web 04/17/2002
FIGURE 1. Map showing all sampling sites. been shown that forest canopies are very efficient scavengers of POPs, giving rise to greater atmospheric deposition in forest soils as compared to grassland soils at the same site (13, 14). For example, it has been calculated that in Germany 70% of soil-borne POPs could be present in the soils of forests, which cover 30% of the country’s land area (15). In the present study, spatial trends of PCBs in U.K. and Norwegian soils were investigated further to gain insight into the environmental and soil-related variables that influence those trends. We analyzed background soils from 41 sites collected in 1998 on a latitudinal transect from the south of the U.K. to the north of Norway. Where possible (at 25 sites), both grassland (GL) and woodland (WL) soil was sampled to investigate the difference between these soil types/land uses. The influence of soil type, land use, temperature, and latitude on PCB concentrations and congener patterns are therefore discussed to set the global fractionation effect into a broader context.
Methods Sampling Sites. The soil samples were taken from remote sites (i.e., away from towns, roads, or other human activity). Whenever possible, both GL and WL soil samples were collected, the two samples being collected not more than 1 km apart. All sites are shown in Figure 1. The samples were collected using a stainless steel hand-held corer that was cleaned before and after each sample using moss or other vegetation from the site. The first two cores were always discarded. The sampling depth was 0-5 cm after removal of the litter layer. Three cores, taken over an area of several square meters, were bulked together to form one sample. The samples were wrapped in aluminum foil twice and sealed in two plastic bags to minimize the possibility for contamination. One extra core of known volume was collected at each site and wrapped separately to be used for bulk density determination. During the sampling trip (∼10 days), the samples were stored in a cool box. Upon arrival in Lancaster, the samples were immediately transferred to a freezer where they were stored until analysis. Sample Extraction and Analysis. The three cores were mixed together after which a subsample of between 10 and 60 g wet weight of soil (corresponding to between 2 and 50
g dry weight (dw)) was weighed out and mixed with Na2SO4 to remove water. The sample was then transferred to a preextracted paper thimble and spiked with a recovery standard containing 13C12 PCBs 28, 52, 101, 138, 153, and 180. The samples were extracted for 16 h using dichloromethane (DCM). The extracts were cleaned on an activated alumina/ silica column containing 10 g of alumina topped with 9 g of silica and 1 cm of Na2SO4, eluted with 140 mL of hexane/ DCM (1:1). The sample was then exchanged into ∼3 mL hexane and shaken with approximately 2.5 mL of sulfuric acid to remove lipids and waxes. The extract was further cleaned on a 6 g Biobead (SX-3) column to remove any remaining lipids. The sample was eluted with hexane/DCM (1:1), the first 15 mL were discarded, and the 15-45 mL fraction was collected. Finally, the sample volume was reduced to 25 µL in dodecane containing PCB 30 and 13C12 PCB 141 as internal standards for volume correction. The samples were analyzed on a Fisons GC-MS with an EI+ source in selected ion mode (SIM), as described elsewhere (16, 17). The following PCB congeners were routinely detected in all samples: tri-PCB 18, 31, 28, 22; tetra-PCB 52, 49, 44, 74, 70; penta-PCB 95, 90/101, 99, 87, 110, 123, 118; hexa-PCB 151, 149, 153/132, 141, 138, 158; hepta-PCB 187, 183, 180, 170; and octa-PCB 199, 203, 194. Reference to the sum of homologue groups or the total sum of PCB considers the aforementioned congeners. Water content of the soils was determined by drying an aliquot of the soil at 105 °C until constant weight was achieved. Dry bulk density (BD) was determined by drying and weighing a soil core of known volume. Organic matter (OM) content was determined by loss on ignition at 450 °C. Quality Control. Recoveries were monitored in all samples using the 13C12 PCB mixture added to the sample prior to extraction. Average recoveries were 85% for 13C12 PCB 28, 90% for 13C12 PCB 52, 90% for 13C12 PCB 101, 101% for 13C12 PCB 138, 103% for 13C12 PCB 153, and 106% for 13C12 PCB 180. Blanks (extraction of a thimble filled with Na2SO4) were included at a rate of one for every five samples and were treated in exactly the same manner as the samples. All results are blank corrected. The limit of detection (LOD) was calculated as three times the standard deviation (SD) of the mean blank. In case the concentration of a compound was VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2147
TABLE 1. Sample Information and Soil Concentrations (pg g-1 dry weight) for U.K. and Norwegian Soil Sampling Sites soil concentrations (pg g-1 dry wt) a
temperature data site
soil type
1
GL WL GL WL GL WL GL WL WL GL WL GL WL WL GL WL WL GL WL GL WL GL WL WL GL WL GL WL GL WL GL WL WL GL WL GL WL GL WL WL WL WL GL WL WL GL WL GL WL WL GL WL WL WL GL WL GL WL WL GL WL WL WL WL GL
2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41a 41b 41c
forest typeb C M M C D D D C D D C C D C C C C D C
long
lat
mean annual
mean Jan
mean July
max 98
min 98
% OM
dry BD (g cm-3)
sum tri
sum tetra
sum penta
sum hexa
sum hepta
sum octa
sum PCB
-4.5
50.6
9.0
4.4
14.7
26.6
-6.4
50.8
9.9
5.0
16.0
26.2
-3.9
-3.5
51.0
8.5
3.4
14.1
-3.8
52.2
8.9
3.8
14.6
25.9
-6.7
-4.0 -3.8
52.5 52.6
9.75 9.72
4.8 4.8
15.4 15.8
27.2 26.0
-5.9 -2.3
53.1
9.63
4.0
15.9
28.0
-4.4
-2.8
54.0
9.10
3.7
15.2
23.9
-4.5
-3.0 -3.6
54.4 55.2
8.26 8.63
2.9 3.3
14.2 14.6
25.8 25.1
-5.9 -3.2
-5.5
55.8
8.58
4.0
13.6
-6.2
56.1
9.15
4.8
13.8
23.0
-2.3
-4.7 -5.1
56.5 56.7
7.69 8.71
2.8 3.7
13.6 13.6
24.0
-9.1
-6.0
57.3
8.35
4.0
13.2
-4.7
57.6
8.02
2.7
14.6
26.9
-11.1
-5.2
57.9
7.1
2.5
12.8
8.3 6.4
58.3 58.5
6.50 6.30
-2.0 -0.8
15.5 13.9
23.8 24.0
-13.9 -15.1
8.5
59.1
5.00
-4.0
15.1
23.8
-14.9
10.8
59.3
6.40
-3.5
16.0
24.4
-16.6
5.7 11.2 6.8 5.4
59.7 60.3 60.6 61.3
6.70 3.80 5.00 6.50
1.2 -7.2 -3.0 0.0
13.8 15.2 14.0 13.7
23.9 24.8 21.1
-19.4 -9.5 -7.1
11.8 10.2 6.5 10.3
61.3 61.6 61.8 61.8
2.00 -0.30 4.50 1.00
-10.7 -9.7 -1.5 -9.9
13.8 10.4 13.0 11.2
25.5 19.4 25.4 20.3
-27.4 -22.5 -12.9 -26.8
8.7 12.1
61.8 64.0
2.50 2.80
-9.7 -6.0
13.9 13.0
23.2 25.2
-22.6 -32.9
10.5 13.6 16.0
64.1 65.0 67.0
4.50 1.00 -1.00
-2.5 -9.5 -10.0
12.8 11.5 10.5
25.2 22.1
-24.6 -31.0
14.7
67.4
4.40
-3.2
12.9
26.1
-14.0
16.0 19.8
68.0 69.1
3.30 1.00
-3.5 -10.2
13.2 13.0
28.0 27.0
-14.2 -26.5
18.6 25.2 28.0 19.0
69.8 69.8 70.5 75.0
3.00 0.00 0.00 -2.50
-3.0 -12.0 -8.5 -15.0
11.8 12.7 12.5 10.0
25.9 24.5 22.6
-22.0 -33.0 -39.0
0.55 0.14 0.84 0.25 0.64 0.78 0.59 0.16 0.49 0.72 0.40 0.80 0.65 0.10 0.82 0.19 0.31 0.74 0.21 0.27 0.10 0.48 0.60 0.15 0.40 0.38 0.46 0.11 0.47 0.12 0.46 0.22 0.14 0.31 0.13 0.68 0.13 0.53 0.10 0.13 0.19 0.12 0.55 0.14 0.14 0.34 0.19 0.72 0.30 0.54 0.54 0.12 0.13 0.15 0.32 0.15 0.64 0.45 0.10 0.7 0.18 0.12 0.07 0.14 1.52 1.51 0.79
63 300 24 62 25 25 47 120 47 NA NA 37 36 130 74 81 110 60 150 300 180 76 53 160 63 88 79 300 65 1200 NA 130 650 77 260 33 NA 41 1000 140 110 1300 66 160 210 79 120 43 180 40 49 180 240 140 45 190 26 NA 400 66 160 210 160 140 13 13 16
49 270 34 410 20 150 39 320 73 NA NA 41 64 270 57 400 240 34 210 120 240 43 30 360 56 120 NA 340 110 1200 40 340 1100 55 960 NA NA NA 1700 560 270 1800 37 160 200 55 160 25 220 45 22 270 490 91 26 130 15 NA 440 36 150 300 180 270 7.6 8.4 8.7
260 1500 110 2200 34 690 130 1100 420 77 880 190 340 1700 620 3400 2400 94 2000 690 1400 210 270 1300 43 360 68 1500 310 2700 210 1300 4800 750 4400 180 3400 230 4400 2000 1600 3500 300 920 1100 250 650 38 1300 310 36 650 740 460 81 490 42 400 1400 39 470 1600 810 900 9.6 9.8 10
680 2400 370 5100 150 2400 600 3000 1000 190 2000 470 990 4900 1100 6700 4200 210 4200 1400 2800 590 680 3200 270 1000 890 2400 560 3600 590 2000 11000 1900 8000 1600 6400 830 7200 4800 3800 7100 620 1900 2500 510 1200 150 2700 470 52 1300 1500 1200 250 800 140 630 2400 130 930 2500 1600 1100 15 17 16
390 660 120 1800 57 1200 320 1500 570 71 830 150 310 1400 350 1900 1500 48 1500 470 920 190 180 1600 180 510 790 710 160 1200 140 650 4200 850 3100 1200 2300 360 2400 1800 1500 2600 220 680 870 140 280 37 790 120 17 430 480 320 74 190 29 210 710 35 290 660 510 270 4 4.6 5.3
380 110 21 310 18 460 110 520 290 13 210 38 93 270 94 450 390 15 340 140 200 38 33 500 22 130 180 130 19 180 8.1 150 930 170 680 340 480 68 450 380 290 470 130 160 180 27 52 7.1 130 20 5.7 96 73 70 15 35 5.5 53 110 5.3 33 150 72 33 2.1 1.5 1.8
1800 5300 680 9800 300 4900 1200 6600 2400
-3.8
19 97 14 49 13 18 32 65 25 11 28 16 17 82 15 38 31 20 56 60 96 31 19 83 26 25 27 96 42 97 28 86 95 49 96 9 96 19 95 85 37 95 23 75 76 58 77 15 64 15 25 94 97 60 15 94 6 20 81 13 83 90 95 96 4 4 7
-3.9
C C C M C C D M C D C C C M D D C D D M D
a NA: some or all of the congeners in this homologue group were not analyzed due to chromatographic interferences. deciduous; M ) mixed.
below LOD, the value of 1/2LOD was inserted. Concentrations for most congeners were above the detection limit, except for several tri- and tetrachlorinated PCBs in about half of the samples and for most congeners in the Bear Island samples (site 41). The reproducibility of the method was good. Replicate extraction (n ) 6) of selected samples showed an average RSD of 9% for site 8-GL (range 0.8-29.4%) and 14% for site 29-WL (range 3.0-27.4%). 2148
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002
b
920 1800 8700 2300 13000 8800 460 8400 3100 5700 1100 1200 7100 630 2200 5400 1200 10000 4600 22000 3800 17000
17000 9700 7600 17000 1400 4000 5100 1100 2400 300 5400 1000 180 2900 3600 2300 490 1800 250 5600 310 2000 5400 3300 2800 51 55 57
C ) coniferous; D )
Results and Discussion Table 1 summarizes sample information for all sites, including location, temperature, soil properties, and homologue concentrations. Variation in Environmental and Soil-Related Factors. There was very wide variation in soil properties, notably in OM content, in the sample set. OM varied between 3% and
FIGURE 2. Variation of temperature with latitude along the transect.
FIGURE 3. Concentration of total PCB versus latitude: (A) expressed as pg g-1 dry weight and (B) expressed as pg g-1 organic matter. 60% for the GL soils and between 15% and 97% for the WL soils, with this range being well-distributed with latitude (i.e., both low and high organic matter soils were found along the entire transect). Soil BD varied between 0.07 and 1.52 g cm-3; as expected, there was an inverse relationship between BD and OM content. The substantial variations in OM and BD need to be borne in mind when considering data presentation and processes. Figure 2 shows the temperature variations with latitude. Temperature data were obtained from the U.K. Meteorological Office and the Norwegian Meteorological Institute. Average temperatures for each site were compiled using daily air temperature data from the nearest meteorological station (generally not more than 20 km from the site). Mean annual, mean January, and mean July temperatures were averaged over several decades, depending on availability of temperature data (1961-1990 for Norway and 1960-1996 for the
FIGURE 4. Comparison of different soils types/land uses at one U.K. site: (A) results expressed as pg g-1 dry weight and (B) expressed as pg g-1 organic matter. U.K.). These temperatures were considered representative of general latitudinal trends. Mean annual temperature varied between -2.5 °C (site 41) and +9.9 °C (site 2) and decreased with increasing latitude. The mean July and maximum 1998 temperatures showed quite small differences along the transect, while the mean January and minimum 1998 values varied widely with latitude. A larger range of temperatures was observed in Norway than in the U.K., especially for the minimum 1998 temperature. Temperature will clearly affect several aspects of POP behavior, notably degradation within soil (18) and volatilization from it (19) as well as key environmental processes such as C turnover rates (20) and vegetation type. VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2149
FIGURE 5. Log concentration (pg g-1 dry weight) versus percent organic matter showing woodland (WL) and grassland (GL) samples separately. General Comments on Soil Concentrations. Because of the large amount of data collected in this study, only the sum of the tri- to octa-homologue groups and the total PCB concentrations are shown at each site in Table 1. Results for individual PCB congeners are available in the Supporting Information. In general, differences in concentration between sites were 1-2 orders of magnitude for the lighter PCBs and around 2-3 orders of magnitude for the heavier PCBs, when expressed on a picograms per gram dry weight basis. Generally, the highest total PCB concentrations are found in the south of Norway, mainly in WL samples (e.g., sites 18WL and 19-WL). The lowest concentrations were found on Bear Island (site 41). Most concentrations for this site were below detection limit; therefore, the results are not taken into account in subsequent figures or calculations. Expressing the results on a picogram per gram OM basis, the differences between the sites decreased as much as an order of magnitude. On an OM basis, the lowest concentrations for 2150
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002
total PCB were found at sites 31-GL and 41, with the highest concentrations in U.K. WL sites 3, 8, and 9. However, concentrations at WL sites 18 and 19 were still among the highest, when the results are expressed on a picogram per gram OM basis. This preliminary discussion highlights that location and soil/vegetation type are key factors influencing background soil PCB concentrations Influence of Soil and Land Use Variables on the Spatial Distribution of PCBs. Figure 3A shows the concentration of total PCB in picograms per gram dry weight versus latitude, distinguishing the WL and GL samples. Concentrations in the WL samples were significantly higher on this basis. The same data are expressed as picograms per gram OM in Figure 3B; this clearly reduces the differences between the land use types, although WL concentrations were still generally higher. This might be explained by a difference in scavenging/ deposition mechanisms between WL and GL soils. The burden of PCBs measured in these surface (0-5 cm) WL and
FIGURE 6. Relative concentration (% of total PCB) of the homologue groups versus latitude. GL soils will principally be a function of net cumulative deposition, degradation losses, and other processes (such as C turnover/burial and leaching), which take these compounds below the 0-5 cm surface sampling depth. Deposition fluxes of PCBs and similar POPs are greater in forests than adjacent grasslands (13), because of greater scavenging from the atmosphere (the “forest filter” effect) (14). Figure 3, parts A and B, shows similar spatial concentration trends, with higher concentrations south of 60 °N (U.K. and the south of Norway) and a clear drop off north of that. It is believed that this largely reflects two issues. The first is proximity to the source; global usage (and hence emission) of PCBs was greatest between 30 and 60 °N (21, 22), with ca. 95% of global usage occurring in this “temperate industrial band”. Major users were the U.K., Germany, and parts of Eastern Europe (21), which will impact the U.K. and southern Norway soils to a greater extent than those in more remote locations. The second might be called the “orographic effect”, enhanced wet deposition in southern Norway compared to central and northern Norway, a phenomenon which has previously been shown to enhance deposition of many heavy metals (23) and PAHs (24). PAH levels in southern Norwegian forest soils are ca. 10 times higher than in central Norway, with long-range atmospheric transport (LRAT) from elsewhere in Europe being the dominant source (24). It is likely that the same process is important for PCBs, but probably
to a lesser extent, because they occur more in the air gas phase than most PAHs (25, 26). Data for GL, and deciduous and coniferous WL soils all collected at one site (site 7) in the U.K. are shown in Figure 4. Differences between the soils are apparent when the data are expressed on a dry weight basis but disappear for the WL soils when expressed on an OM basis. However, the GL soil concentrations remain about a factor of 2 lower when normalized to % OM, indicating that factors other than OM content of the soil (e.g., scavenging, degradation, soil mixing) influence land use differences in concentration. Further Consideration of the Role of OM and Implications for Global Cycling. To further investigate the influence of land use and OM content, PCB concentrations (pg g-1 dw) are plotted against % OM for the tri- to octa-homologue groups in Figure 5 (note the y axis is on a log scale). There is a discernible relationship between concentration and % OM for the lighter homologues, becoming less clear with increasing degree of chlorination. The scatter in the data is similar for GL and WL samples, with both land use types following similar trends. For all of the homologue groups, except the tri-PCBs, the WL concentrations lie slightly above the GL concentrations, likely reflecting the processes discussed previously. These trends are interpreted as evidence that, over time, the more volatile PCBs are moving toward equilibrium with the OM burden of the soil compartment, VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2151
on a European regional and possibly global scale. Because lighter PCBs have lower air-surface media partition coefficients (for which KOA is the commonly used surrogate) and higher vapor pressures than heavier PCBs, we envisage lighter PCBs will be subject to greater LRAT (6), greater atmospheric mixing and greater mixing within soils (27, 28), than the heavier homologues. This will enable the lighter homologues to approach equilibrium with the global soil OM pool more rapidly than their heavier counterparts. The distribution of the “stickier”, heavier homologues is seemingly still primarily influenced by their preferential deposition closer to source and their greater susceptibility to the forest filter effect (14). These observations would explain the comparative amounts of scatter on the plots of different homologues (i.e., greatest for heavier homologues) as well as the WL/GL differences between homologues in Figure 5. Evidence for Global Fractionation. The relative concentration of individual homologue groups (expressed as a percentage of the total PCB concentration) was regressed against latitude and against the various measures of temperature discussed previously (i.e., mean annual, mean January, mean July, minimum 1998, and maximum 1998). Figure 6 provides an example of the output, with the relative homologue concentration plotted against latitude. The relative concentration of the more volatile PCBs (e.g., triand tetra-PCBs) increases with latitude, while that of the less volatile PCBs (e.g., hepta- and octa-PCBs) decreases. This is consistent with the global fractionation theory, which predicts a preferential transport of more volatile compounds to higher latitudes (5, 6). As far as the authors are aware, this is the first reported evidence for global POP fractionation in soils. However, the regression slopes are quite shallow and there is a fair amount of scatter, implying that many of the environmental and soil-related processes discussed earlier (and other factors, such as precipitation, organic carbon content and type, other soil properties, local sources, etc.) are also influencing the observed congener patterns. Temperature-driven fractionation, while clearly operating and detectable, needs to be considered in this broader context. To investigate the significance of the observed trends, the linear regressions were evaluated statistically and results are shown in Table 2 for each homologue group. The tri- and tetra-PCBs show a statistically significant increasing trend with latitude and a significant decreasing trend with temperature (p < 0.05), whereas the hepta- and octa-PCBs show the reverse. The penta- and hexa-PCBs show no statistically significant (p < 0.05) trend (i.e., the relative concentrations of these homologue groups are equally distributed along the transect). Interestingly, Table 2 suggests that the annual or minimum temperatures are more important than maximum temperatures in influencing fractionation in soil. However, as noted previously, there is little variation in maximum temperature between the sites (see Table 1 and Figure 2). Also, low air temperatures are probably important in terms of deposition processes, whereas volatilization from soil would be governed by higher temperatures. The relationship between the slope of the latitudinal trend and compound physical/chemical properties was investigated by carrying out regressions between the relative amounts of all individual congeners detected and latitude. The value of the slope of each regression is plotted against the log of vapor pressure (PL, values from Falconer and Bidleman (29)) and the log of the octanol/air partition coefficient (KOA, values from Harner and Bidleman (30)) in Figure 7. The upper and lower 95% confidence limits are also shown. When the confidence limits include the zero line, the slope is not significantly different from zero at the p < 0.05 level. Figure 7 shows a trend where slopes are negative for compounds with low vapor pressure and high 2152
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 36, NO. 10, 2002
TABLE 2. Statistical Information for Regressions of Relative Concentrations of the Homologue Groups against Latitude and Temperature slopea
std error
tri 0.323 tetra 0.223 penta 0.215 hexa -0.136 hepta -0.343 octa -0.212
0.094 0.057 0.125 0.107 0.105 0.062
tri -0.530 tetra -0.308 penta -0.256 hexa 0.093 hepta 0.502 octa 0.308 -0.247 -0.140 -0.121 -0.015 0.241 0.164
r2
lower 95% upper 95%
Latitude 0.147 0.135 0.191 0.109 0.037 -0.035 0.021 -0.351 0.121 -0.553 0.133 -0.335
sign.
0.512 0.336 0.464 0.078 -0.133 -0.089
0.001 0.000 0.091 0.208 0.002 0.001
Mean Annual Temperature 0.153 0.150 -0.835 0.095 0.138 -0.499 0.200 0.021 -0.654 0.171 0.004 -0.248 0.168 0.104 0.167 0.099 0.112 0.111
-0.225 -0.118 0.142 0.434 0.837 0.504
0.001 0.002 0.203 0.588 0.004 0.003
Mean January Temperature 0.096 0.091 -0.438 0.060 0.079 -0.260 0.120 0.013 -0.360 0.103 0.000 -0.220 0.103 0.067 0.036 0.060 0.090 0.045
-0.056 -0.020 0.119 0.189 0.446 0.283
0.012 0.023 0.319 0.884 0.022 0.008
tri -1.252 tetra -0.559 penta -0.505 hexa 0.111 hepta 1.117 octa 0.577
Mean July Temperature 0.365 0.150 -1.980 0.245 0.075 -1.047 0.459 0.016 -1.420 0.393 0.001 -0.671 0.387 0.099 0.347 0.231 0.076 0.116
-0.524 -0.070 0.411 0.893 1.887 1.037
0.001 0.026 0.276 0.778 0.005 0.015
tri 0.083 tetra 0.259 penta -0.382 hexa -0.479 hepta 0.154 octa 0.158
Maximum 1998 Temperature 0.304 0.001 -0.525 0.189 0.034 -0.120 0.293 0.026 -0.967 0.290 0.042 -1.058 0.257 0.006 -0.359 0.181 0.012 -0.204
0.691 0.638 0.203 0.101 0.668 0.521
0.786 0.176 0.196 0.104 0.551 0.386
tri -0.175 tetra -0.138 penta -0.044 hexa 0.081 hepta 0.113 octa 0.098
Minimum 1998 Temperature 0.058 0.141 -0.291 0.035 0.229 -0.207 0.064 0.008 -0.171 0.063 0.026 -0.045 0.054 0.065 0.005 0.037 0.099 0.024
-0.059 -0.068 0.083 0.207 0.220 0.173
0.004 0.000 0.492 0.204 0.040 0.011
tri tetra penta hexa hepta octa
a
Bolded numbers are significant at the p < 0.05 level.
KOA and positive for compounds with high vapor pressure and low KOA. These data can be interpreted as an indication that, although the lighter compounds manage to make the journey to higher latitudes more readily than the heavier PCBs, much of the total PCB burden in the environment is retained by the soil close to source (see Figure 3). It seems likely that the OM-rich soils of southern Norway are a particularly very effective trap for these compounds, significantly hindering revolatilization following deposition. Although global fractionation is clearly happening in soils, it is unclear whether compounds that have reached the far north have done so following their initial (primary) release and LRAT (a “single hop”), or following several “hops” of repeated air-surface exchange (the so-called grasshopper effect). The shallow slopes of the regression lines and the greater influence of low temperatures on fractionation in soil would suggest a single hop mechanism, as repeated air-surface exchange would lead to a more pronounced fractionation and therefore to steeper slopes. These issues are discussed further by Bignert et al. (12), based on their investigation of spatial and temporal trends of POPs in Swedish biota.
Literature Cited
FIGURE 7. Slopes of regression of relative concentration with latitude versus physical/chemical properties: (A) log vapor pressure (PL) and (B) log octanol/air partition coefficient (KOA).
Acknowledgments We thank Irena Twardowska and Patrick Poupart for their help with the sampling and Andy Sweetman and Gareth Thomas (Lancaster University) and Todd Gouin and Don Mackay (Trent University, Canada) for helpful discussions. We also thank the EU for financial support (Global-SOC project, Contract No. 4-Ct97-0638) and our partners on the Global-SOC project for many interesting exchanges.
Supporting Information Available Two tables showing soil concentrations of individual PCB congeners for the U.K. and Norway, respectively. This material is available free of charge via the Internet at http:// pubs.acs.org.
(1) Harrad, S. J.; Sewart, A. P.; Alcock, R.; Boumphrey, R.; Burnett, V.; Duarte-Davidson, R.; Halsall, C.; Sanders, G.; Waterhouse, K.; Wild, S. R.; Jones, K. C. Environ. Pollut. 1994, 85, 131-146. (2) Wild, S. R.; Jones, K. C. Environ. Pollut. 1995, 88, 91-108. (3) Harner, T.; Bidleman, T. F.; Jantunen, L. M. M.; Mackay, D. Environ. Toxicol. Chem. 2001, 20, 1612-1621. (4) Rahn, K. A.; Heidam, N. Z. Atmos. Environ. 1981, 15, 13451348. (5) Wania, F.; Mackay, D. Ambio 1993, 22, 10-18. (6) Wania, F.; Mackay, D. Environ. Sci. Technol. 1996, 30, 390-396. (7) Agrell, C.; Okla, L.; Larsson, P.; Backe, C.; Wania, F. Environ. Sci. Technol. 1999, 33, 1149-1156. (8) Lead, W. A.; Steinnes, E.; Jones, K. C. Environ. Sci. Technol. 1996, 30, 524-530. (9) Lead, W. A.; Steinnes, E.; Bacon, J. R.; Jones, K. C. Sci. Total Environ. 1997, 193, 229-236. (10) Ockenden, W. A.; Sweetman, A. J.; Prest, H. F.; Steinnes, E.; Jones, K. C. Environ. Sci. Technol. 1998, 32, 2795-2803. (11) Calamari, D.; Bacci, E.; Focardi, S.; Gaggi, C.; Morosini, M.; Vighi, M. Environ. Sci. Technol. 1991, 25, 1489-1495. (12) Bignert, A.; Olsson, M.; Persson, W.; Jensen, S.; Zakrisson, S.; Litzen, K.; Eriksson, U.; Haggberg, L.; Alsberg, T. Environ. Pollut. 1998, 99, 177-198. (13) Horstmann, M.; McLachlan, M. S. Atmos. Environ. 1998, 32, 1799-1809. (14) Wania, F.; McLachlan, M. S. Environ. Sci. Technol. 2001, 35, 582-590. (15) McLachlan, M. S.; Horstmann, M. Environ. Sci. Technol. 1998, 32, 413-420. (16) Thomas, G. O.; Sweetman, A. J.; Parker, C. A.; Kreibich, H.; Jones, K. C. Chemosphere 1998, 36, 2447-2459. (17) Ockenden, W. A.; Prest, H. F.; Thomas, G. O.; Sweetman, A.; Jones, K. C. Environ. Sci. Technol. 1998, 32, 1538-1543. (18) Sinkkonen, S.; Paasivirta, J. Chemosphere 2000, 40, 943-949. (19) Hippelein, M.; McLachlan, M. S. Environ. Sci. Technol. 2000, 34, 3521-3526. (20) Edwards, C. A.; Bohlen, P. J. Biology and ecology of earthworms, 3rd ed.; Chapman and Hall: New York,1996; p 426. (21) Breivik, K.; Sweetman, A.; Pacyna, J. M.; Jones, K. C. Sci. Total Environ., in press. (22) Breivik, K.; Sweetman, A.; Pacyna, J. M.; Jones, K. C. Sci. Total Environ., in press. (23) Steinnes, E.; Allen, R. O.; Petersen, H. M.; Rambaek, J. P.; Varskog, P. Sci. Total Environ. 1997, 205, 255-266. (24) Aamot, E.; Steinnes, E.; Schmid, R. Environ. Pollut. 1996, 92, 275-280. (25) Bidleman, T. F. Environ. Sci. Technol. 1988, 22, 361-367. (26) Cotham, W. E.; Bidleman, T. F. Environ. Sci. Technol. 1995, 29, 2782-2789. (27) Cousins, I. T.; Gevao, B.; Jones, K. C. Chemosphere 1999, 39, 2507-2518. (28) Cousins, I. T.; Mackay, D.; Jones, K. C. Chemosphere 1999, 39, 2519-2534. (29) Falconer, R. L.; Bidleman, T. F. Atmos. Environ. 1994, 28, 547554. (30) Harner, T.; Bidleman, T. F. J. Chem. Eng. Data 1996, 41, 895899.
Received for review December 12, 2001. Revised manuscript received March 11, 2002. Accepted March 11, 2002. ES010322I
VOL. 36, NO. 10, 2002 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2153
