Environ. Sci. Technol. 1996, 30, 3464-3472
Spatial and Temporal Trends of Atmospheric Organochlorine Vapors in the Central and Upper Great Lakes C A R R I E L . M O N O S M I T H †,‡ A N D M A R K H . H E R M A N S O N * ,§ Michigan Department of Natural Resources, Air Quality Division, P.O. Box 30028, Lansing, Michigan 48909, and Department of Health, West Chester University, West Chester, Pennsylvania 19383
From November 1990 to October 1991, monthly air samples were collected from Saginaw Bay, Sault Ste. Marie (SSM), and Traverse City, MI, to identify spatial and temporal variation in organochlorine (OC) vapors in the central and upper Great Lakes region. Average annual and maximum concentrations of ∑PCB were highest at Traverse and lowest at SSM. PCB 8+5 dominated the congener mix, except at Traverse in the summer. Maxima of HCB, R- and γ-HCH, DDT, and DDE were observed at Traverse or Saginaw, except one unusual event at SSM in May. All compound concentrations varied seasonally at all sites except R-HCH. Correlations measuring regression strength between air temperature and OC concentrations show Saginaw affected most by local sources, and SSM affected the least. Among compounds with high vapor pressure, correlations were low at Traverse indicating nonlocal sources. The spatial and temporal differences observed may limit prediction of regional atmospheric trends of these compounds.
Introduction Organochlorine industrial compounds and pesticides are of concern to environmental scientists because of possible carcinogenic and other effects that these contaminants may have on aquatic and marine organisms and on humans. The identification of these compounds in remote areas of the world where they have had limited or no use (1-3), shows that atmospheric processes distribute them over broad areas. In the Great Lakes region and other areas, investigators have identified the atmosphere as source, sink, or distributor of organochlorine compounds (OCs) (4-13). However, differences in sampling frequencies, durations of studies, and sizes of networks make comparisons between * Corresponding author e-mail address:
[email protected]. † Michigan Department of Natural Resources. ‡ Present address: Environmental Assistance Division, Michigan Department of Environmental Quality, P.O. Box 30457, Lansing, MI 48909-7957. § West Chester University.
3464
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 12, 1996
studies difficult. Our study was initiated in 1990 to investigate variations in OC contaminant concentrations at three sites during 1 year in the central and upper Great Lakes region in order identify spatial and temporal variability. The sites are not known to be contaminated by local sources but are probably influenced by deposition to or out-gassing of compounds from the Great Lakes, their bays, and channels or are influenced by a variety of possible atmospheric processes that may deliver contaminants from distant sources.
Sampling Sites, Materials, and Methods Simultaneous monthly samples were collected over 48 h at sites near Grand Traverse Bay, Saginaw Bay, and Sault Ste. Marie (SSM), MI (see Figure 1) from November 1990 through October 1991. The same dates were sampled each month. Samples were collected using Andersen PS-1 PUF samplers. Particles >0.1 µm were collected using glass fiber filters (GFF). OC vapors were concentrated onto a matrix of 4.5 cm layers of polyurethane foam (PUF) surrounding a layer of XAD-2 resin. PUF was cleaned before sampling by extraction in dichloromethane (DCM) for 24 h and then in 1/1 (v/v) acetone/hexane for 24 h. The XAD-2 was cleaned by rinsing with water and then by sequential extraction for 24 h each in methanol, DCM, acetone, hexane, DCM, and 1/1 acetone/hexane, a procedure modified from Franz et al. (14). After cleaning, the PUF and XAD-2 were packed into glass cylinders precleaned by heating for 4 h at 450 °C. The packed cylinders were frozen during storage and shipment. A sampler flow rate of 0.20 m3/min was used to collect a target volume of 576 m3. Linear calibrations between orifice flow and pressure drop behind the vapor trap were determined at the beginning of the project and checked quarterly using an orifice transfer standard. Two internal audits of all sampler flows were conducted during the project. Compounds of Interest (COI). Our COI include vapor phase PCB congeners and selected pesticides and related isomers. PCBs are probable carcinogens and have the potential to adversely affect reproduction, behavior, and the immune and endocrine systems in humans and wildlife (15). All of the approximately 140 PCB congeners found in the environment are hydrophobic and cover a wide range of saturation liquid vapor pressure (p°L). We selected 25 PCB congeners represented by 14 GC peaks for detailed consideration (Table 1). These include congeners most concentrated in air samples in this work and elsewhere (10, 16-19) and in aquatic biota and sediments (20) in both remote regions and in areas of known contamination. The p°L of these congeners is high enough that the vapor phase is predominant in the atmosphere (21). In most of our air samples, these congeners represent about 60% of ∑PCB. We also consider ∑PCB, which is the sum of all 121 congeners analyzed here. Other COI include hexachlorobenzene (HCB), p,p′isomers of DDE and DDT, and R- and γ-isomers of hexachlorocyclohexane (HCH). HCB is a synthetic byproduct of various chemical manufacturing processes and may be used as a fungicide. HCB is included in environmental investigations because it is bioaccumulative and considered
S0013-936X(96)00093-4 CCC: $12.00
1996 American Chemical Society
30) and 2,2′,3,4,4′,5,6,6′-octachlorobiphenyl (PCB 204). Since all of the PCB 65 added as a surrogate appeared in the hexane fraction, it was used as an internal standard in the 40%/60% DCM/hexane fraction. The hexane fraction was analyzed by GC-ECD using a 30 m long, 0.25 mm diameter 5% phenyl, methyl polysiloxane coated column (film thickness was 0.25 µm). The initial oven temperature was 50 °C, increased to 125 °C at 5 °C/min, then to 215 °C at 0.3 °C/min, and then to 280 °C at 10 °C/min. The 40%/60% DCM/hexane fraction was analyzed similarly. The values of all compounds reported here are not blank or recovery corrected to be consistent with a later Michigan data set, so they differ from those reported earlier (29).
FIGURE 1. Sampling sites and principal cities in the study area.
to be a carcinogen (22). The other compounds are all pesticides, isomers, or degradation products. The occurrence of p,p′-isomers of DDT or DDE in atmospheric samples may result from local remnants of the compounds in soils or from active use. The latter could include the use of pesticides that are not banned but that are contaminated with DDT, such as dicofol (23). DDE is often the most concentrated of the DDT-related compounds (4). Only γ-HCH is used as a pesticide at present in a variety of applications in North America (24). It is among the most widely used pesticides worldwide, available either in a technical mixture dominated by the R-HCH isomer or as a product known as lindane dominated by the γ-HCH isomer. The latter has wider popularity in North America and throughout Europe (24, 25) and in certain regions of Africa (19) because the γ-HCH isomer is the only effective insecticide. The non-γ-HCH isomers are synthetic byproducts that result in background contamination when used. Oehme (26) suggests that global atmospheric R-HCH concentrations range from 500 to 1000 pg/m3. Bidleman et al. (27), however, show that atmospheric concentrations declined between 1980 and 1992 in the Canadian and Norwegian Arctic and in the Chukchi Sea, suggesting that sources of R-HCH to the atmosphere are declining as less of the technical mixture is used. At our Michigan sites, there would be little or no expected evaporation of R-HCH to the atmosphere since there would be no expected use. Analytical Methods. All samples were extracted in 1/1 (v/v) acetone/hexane for 20 h using Soxhlet extractors. Surrogate standards to measure analytical recovery of PCB (composed of 3,5-dichlorobiphenyl (PCB 14), 2,3,5,6tetrachlorobiphenyl (PCB 65), and 2,3,4,4′,5,6-hexachlorobiphenyl (PCB 166)) were added before extraction (32). Extracts were reduced to 1 mL and then eluted through a column of silica deactivated 3% with water. The first, nonpolar hexane eluent contained HCB, p,p′-DDE and all PCB, including the surrogate standards. The more polar second eluent, comprised of 40%/60% DCM/hexane included R-HCH, γ-HCH, and the p,p′-isomers of DDT. After eluent volumes were reduced, the hexane fraction was spiked with internal standards 2,4,6-trichlorobiphenyl (PCB
Congener-specific quantification of PCB was based on the internal standard method established by Mullin (30), modified to include p,p′-DDE, HCB, and certain PCB congeners. The COI in the 40%/60% DCM/hexane fraction were quantitated using the responses of individual standards against the internal standard (PCB 65) in a mixed pesticide standard. Air temperature values we report are averages of hourly observations from monitoring stations operated by various state agencies near our sampling sites.
Results Total PCB. ∑PCB concentrations measured at Saginaw, SSM, and Traverse appear in Figure 2a. During most months, the concentrations of ∑PCB were highest at Traverse with an annual arithmetic average concentration of 616 pg/m3, ranging from 75 to 2013 pg/m3. The average at Saginaw was 469 pg/m3 (ranging from 206 to 1080 pg/ m3) and was 198 pg/m3 at SSM (ranging from 28 to 883 pg/m3). Our results are similar to other remote locations in the Great Lakes region, including northern Wisconsin (ranging from 135 to 1820 pg/m3 with an average of 483 pg/m3 (13)) and Egbert, Ontario (ranging from 55 to 823 pg/m3 (4)), and the Green Bay, WI, region (June-October 1989) with a ∑PCB range of 65 to 2300 pg/m3 (5). The ∑PCB annual averages at the Traverse and Saginaw sites (616 and 469 pg/m3) were higher than the average of 270 pg/m3 observed at the University of WisconsinsGreen Bay site from May to August 1989 (10). They are also greater than the averages at all Green Bay sites (5) except for the over-water samples collected in the southern, most contaminated area. However, when the average ∑PCB at Traverse is calculated for the same months of the year as the Green Bay work was conducted (June-October), the value is 1105 pg/m3, comparable to the average concentrations of 1430 and 1160 pg/m3 measured by Hornbuckle et al. (5) at the over-water sites. The comparison is not entirely direct however, because, unlike Green Bay data, we included PCB 8+5 in ∑PCB. In our observations, PCB 8+5 contributed up to 30% of ∑PCB, similar to observations by others who have measured PCB 8 or 8+5 (17, 19). Seasonal variation in ∑PCB concentrations was observed at each of our sites, with higher concentrations in warmer summer months (4, 12, 13) (Figure 2a). At Saginaw, winter temperatures (Figure 2b) were variable while ∑PCB concentrations were consistent, indicating that while the ground was frozen or snow covered, contributing no PCB, other sources produced consistent concentrations. The high concentration of ∑PCB at SSM in May, with a secondary high value in August, both suggest influence by distant
VOL. 30, NO. 12, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3465
TABLE 1
Compounds and Saturation Liquid Vapor Pressures IUPAC No.
biphenyl structures
saturation liquid-vapor pressure (p°L) (Pa)a
ref
8+5 18+15 31+28 33 52 47+48 70+76 95 56+60 101 77+110 153+132+105 163+138 187+182
PCB 2,4′-dichloro + 2,3-dichloro 2,2′,5-trichloro + 4,4′-dichloro 2,4′,5-trichloro + 2,4,4′-trichloro 2′,3,4-trichloro 2,2′,5,5′-tetrachloro 2,2′4,4′-tetrachloro + 2,2′,4,5-tetrachloro 2,3′,4′,5-tetrachloro + 2′,3,4,5-tetrachloro 2,2′,3,5′,6-pentachloro 2,3,3′,4′-tetrachloro + 2,3,4,4′-tetrachloro 2,2′,4,5,5′-pentachloro 3,3′,4,4′-tetrachloro + 2,3,3′,4′,6-pentachloro 2,2′,4,4′,5,5′-hexachloro + 2,2′,3,3′,4,6′-hexachloro + 2,3,3′,4,4′ pentachloro 2,3,3′,4′,5,6-hexachloro + 2,2′,3,4,4′,5′-hexachloro 2,2′,3,4′5,5′,6 heptachloro + 2,2′,3,4,4′,5,6′-heptachloro
0.1505 0.0742 0.0338 0.026 0.0158 0.0151 0.00567 0.00523 0.00471 0.0033 0.0018 0.00758 0.00546 0.00295
21 21 21 21 21 21 21 21 21 21 21 21 21 21
HCB p,p′-DDT p,p′-DDE R-HCH γ-HCH
Other Compounds hexachlorobenzene p,p′-dichlorodiphenyltrichloroethane p,p′-dichlorodiphenyldichloroethylene R-hexachlorocyclohexane γ-hexachlorocyclohexane
0.127 0.000512 0.00341 0.227 0.0649
35 35 35 35 35
a
All at 25 °C.
FIGURE 2. (a) Total PCB concentrations from the study sites; (b) total PCB is the sum of up to 121 congeners.
sources, similar to an observation at another Lake Superior monitoring site over 300 km distant (6).
3466
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 12, 1996
Individual PCB Congeners. The average annual and average summer concentrations of the individual PCB congeners of interest are shown in Figure 3. The highest annual average and summer average occurred at Traverse, except for PCB congeners 33, 47+48, 56+60, and 187+182, where the highest averages occurred at Saginaw. SSM had the lowest averages for all PCB. At all sites, the highest annual average concentration was for PCB 8+5, also highest during summer months at Saginaw and SSM. Traverse showed higher levels of PCB 31+28 during the summer, indicating a seasonal shift in source. Average concentrations of PCBs 31+28, 33, 52, 47+48, and 101 at Saginaw vary less than 2 times from observations at Egbert (4). At Traverse, annual average concentrations of PCB 31+28, 47+48, 56+60, and 70+76 are comparable to over-water samples collected in northern and central Green Bay and the Green Bay land-based sites (5). For PCB congeners 101, 77+110, 153+132+105, 163+138, and 187+182, the average concentrations at Traverse equaled or exceeded those collected over the water of southern Green Bay, which were the highest observed there (5). The congener profile at Traverse, a land-based site, shows a lack of enrichment of the most volatile congeners, but shows an average ∑PCB concentration comparable to the overwater concentration of southern Green Bay. The similarities between the results from Traverse and the findings in Green Bay (5) suggest that evaporation from southern Green Bay or a source of equivalent magnitude that increases local concentrations of PCB by up to 7 times (5) may be influencing the high PCB concentrations seen at Traverse. Another source to Traverse could be the more distant PCB-contaminated urban and industrialized areas farther to the south on Lake Michigan. Other investigations have noted northward movement of ozone along the Michigan and Wisconsin shorelines of the lake (31-33). During July sampling at Traverse, when we observed several maximum COI concentrations, the ozone level at a nearby Lake Michigan Ozone Study monitoring site at Sleeping Bear Dunes National Lakeshore, 45 km to the WSW, exceeded
FIGURE 3. (a) Average annual and (b) average summer concentrations of selected PCB congeners.
the National Ambient Air Quality Standard by >25%. This high concentration was caused by the transport of ozone from the south (32). These meteorological conditions over Lake Michigan may also have transported our COI to Traverse from the south. Another source to Traverse may be Lake Michigan, shown to be a source of atmospheric PCB in the northern part of the lake (7).
Congener Weight Percents. The weight percent that each congener contributes to ∑PCB was calculated in order to make a direct comparison of PCB congener distribution between sites. The weight percent by season is shown in Figure 4. Seasonal averages were calculated based on similar monthly temperatures: Fall includes SeptemberNovember, winter includes December-February, spring
VOL. 30, NO. 12, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3467
FIGURE 4. Average seasonal weight percent of total PCB represented by selected PCB congeners. Total PCB is the sum of up to 121 congeners. (a) Winter is December-February; (b) Spring is March and April; (c) Summer is May-August; (d) Fall is September-November.
includes March and April, and summer includes MayAugust. At all sites, the more volatile PCB congeners are dominant. The annual average weight percent of ∑PCB by PCB 8+5 and PCB 31+28 was in the range of 11-13% and 5-10%, respectively. PCB 8+5 dominates at all sites during
3468
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 30, NO. 12, 1996
all seasons except for summer at Traverse when PCB 31+28 was 11% of ∑PCB in comparison to 7% for PCB 8+5. During the spring, PCB 8+5 reached a maxima of 30% of ∑PCB at Traverse and 21% of ∑PCB at Saginaw. In winter, the percent of PCB 8+5 was also higher at Traverse and Saginaw than it was in the summer and fall, resulting from
compounds with higher p°L remaining in the vapor phase and contributing a higher proportion of ∑PCB. We believe that this is also why the proportions of PCB 8+5 are higher at Traverse and Saginaw in spring, when the temperature averaged 7-8 °C lower than in the fall. At SSM, the weight percent of ∑PCB for congeners 8+5, 18+15, and 31+28 was more stable throughout the year than at Traverse or Saginaw. Closer examination of PCB 8+5 and ∑PCB in the spring data revealed different trends at all sites. At Saginaw, the weight percent of PCB 8+5 was similar during March and April (22% and 20%), and the ∑PCB concentrations were also similar (258 and 280 pg/m3), characteristic of a stable, consistent source. At both Traverse and SSM, the March and April weight percents of PCB 8+5 were also similar, but the ∑PCB concentrations varied by a factor of 3, decreasing at Traverse and increasing at SSM. The differences seen between these sites indicate that individual PCBs can vary with respect to air temperature, particularly during transitional seasons such as spring. Calculated congener weight percent of ∑PCB concentration from a combined data set in Bloomington, IN (18), which did not include PCB 8+5, show PCB 31+28 contributing nearly 20% of the 65 congener ∑PCB. A similar data set from Bermuda (16) again shows dominance of PCB 31+28 with approximately 18% of the total. In Brazzaville, Congo, PCB 8 was nearly 30% and PCB 31+28 was 10% of a 42-congener ∑PCB (19). In “typical” samples collected from the Chicago area in 1989 (17), PCB 8 comprised an average of about 9.5% of a 48-congener ∑PCB, while 31+28 comprised about 8.5%. These samples were considered to be characteristic of urban and rural air and are consistent with many of our annual results. HCB. The low water solubility and high p°L of HCB (34, 35) result in a long atmospheric residence time, leading to an average global concentration of 50-150 pg/m3 (36), much less variable than PCBs. Our values, ranging from 40 to 175 pg/m3 with one high outlier, are consistent with this range and show some seasonal variation (Figure 5a). Traverse shows highest concentrations occurring from June through September, similar to Saginaw where highest concentrations occur from April through August. At SSM, concentrations are low (40-70 pg/m3) through the cold months of December-April and are 2-3 times greater from June through September (100-137 pg/m3). The variation with season indicates that global backgrounds of HCB can be enhanced by other sources that may have a seasonal influence. In a recent, longer term sampling of HCB at other sites in Michigan (37), there was a larger range of values in summer than winter, also showing a seasonal influence on sources. The amount of HCB observed in May at SSM is twice any other measured in this study. Since this amount is not seen again at SSM or elsewhere, it suggests that this air mass was from an unusual location that seldom influences these Michigan sites. This value is characteristic of some concentrations observed at sites in urbanized areas on Ontario (11), suggesting that this air mass may have had urban origins. DDT and DDE. Concentrations of DDT were below detection during most winter months at Saginaw and Traverse and were above detection at SSM only in March, May, July, and August (Figure 5b). The November values at all sites are consistent with the unusually warm air
FIGURE 5. Concentrations from sampling sites: (a) HCB; (b) p,p′DDT; (c) p,p′-DDE; (d) r-HCH; (e) γ-HCH.
temperatures, except at SSM, suggesting air mass movement from the south, perhaps from an area where DDT is still used (38). The May value at SSM, which was the second highest observed, again suggests that the air mass was from
VOL. 30, NO. 12, 1996 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
3469
FIGURE 6. Correlation coefficients [r] of linear regressions between COI concentration and air temperature (°C), arranged in order of decreasing p°L. Total PCB appear at the end of the series.
a distant location where significant remnants of DDT still contribute to atmospheric vapors. DDE is most concentrated of the p,p′-isomers of DDT and related compounds that we measured. Our observations show no indication of new DDT use because the DDT/ DDE ratio is
