Airborne organic contaminants in the Great Lakes ecosystem Atmospheric fluxes to the Great Lakes are a combination of dry and wet removal processes. Dry deposition is I . 5-5.0 times the wet deposition for the trace organics selectedchlorinated pesticides and polychlorinated biphenyls
Steven J. Eisenreich Brian B. Looney J. David Thornton University of Minnesota Minneapolis, Minn. 55455
The presence of trace organic pollutants in the Great Lakes, proven or thought to be toxic to aquatic organisms, birds, and humans, stirs our imagination and concern for the shortand long-term health of the lakes. Recent summaries of toxic substances in the Great Lakes have been prepared by Delfino (1) and Konasewich et al. ( 2 ) , and the saga of mirex in Lake Ontario has been updated by Kaiser (3). Rachel Carson’s “The Silent Spring,” published in 1962, heightened our awareness of the problem of pesticides in the environment, but it was 10-1 5 years before this awareness, along with improved analytical techniques, provided a clearer picture of the magnitude of the problem. Results of early fish testing showed that trace organics, particularly the chlorinated pesticides and polychlorinated biphenyls (PCBs), were concentrating in larger fish and bird species feeding on fish. Where were these compounds coming from and what threat did they pose to human health and the Great Lakes ecology? Unfortunately, although these questions are finally being addressed, the answers remain elusive, due mainly to the sheer magnitude of the problem. The trace organics which have caused this concern have similar physicochemical properties, including low water solubility, high octanol30
Environmental Science & Technology
water coefficients, high hydrophobicity, the ability to bioconcentrate, and chemical/microbiological stability. As an excellent example, both DDT and PCB were banned either voluntarily or by mandate in the early 1970s to reduce their concentrations in fish for human health reasons. DDT levels have dropped significantly in Lake Michigan fish, but PCB concentrations remain well above Food and Drug Administration guidelines of 2 PPm ( 1 ) . The International Joint Commission ( 4 ) noted that atmospheric emissions from anthropogenic sources were affecting the quality of the earth’s atmosphere and were likely being deposited in the Great Lakes. Data available on airborne inorganic pollutants (5-9) and organic pollutants (10-16) suggest that atmospheric deposition to the Great Lakes represents a sizable, if not the major source. The published and unpublished literature is reviewed relative to airborne organic pollutants, including atmospheric removal processes, trace organic concentrations, and deposition to the Great Lakes, with special emphasis on the PCBs. Additional details and sampling methodologies are presented in Eisenreich et al. (17 ) .
cant role in the total input. When this observation is coupled with the long hydraulic residence times of the lakes (Superior, -180 y; Erie, -3 y), the relative impact of trace organic inputs can be great, and the recovery long. The susceptibility of the Great Lakes to input of airborne organic pollutants (Table 1) is related specifically to the atmospheric burden of organics in the basin resulting from regional and local transport, the absence of effective in-lake removal processes, and the sensitivity of the biota. The Great Lakes are near and downwind of major pollution areas, and meteorological patterns accompanying edstward-moving low- or high-pressure systems distribute emitted pollutants to sensitive areas. In the absence of effective physical and chemical removal processes for many organics, persistence and bioconcentration result. Sorption of trace organics onto biotic or abiotic particles in the water column, followed by sedimentation, is the major in-lake removal process for many systems. However, the Great Lakes, except Lake Erie, have suspended solids concentrations less than 1 mg/L. Even for compounds having sediment/water partition coefficients of lo5 (Kp = ( a / g ) sed/(Pg/g) HzO), more than 90% of the water burden may exist in the dissolved fraction. Without significant biodegradation, the compounds may accumulate in the system.
Susceptibility to input The Great Lakes as a whole represent the largest body of freshwater in the world and the largest surface area (Figure 1). With lake surface area comprising from 27% (Ontario) to 64% (Superior) of the total basin area, Deposition processes and lake surface areas ranging from Vapor-particle distribution. Trace 19 000 km2 (Ontario) to 82 100 km2 organics exist in the atmosphere in the (Superior), it is not surprising that vapor phase and are adsorbed to paratmospheric input may play a signifi- ticulate matter. Theoretical consider0013-936X/81/0915-0030$01 .OO/O
@
1981 American Chemical Society
ations suggest that vapor-aerosol partitioning in the atmospheredepends on the vapor pressure, size, and surface area of the suspended particles, and the organic content (18). Vapor-aerosol distributions based on saturation vapor pressure and available surface area of atmospheric particulate matter were calculated for urban, rural, and clean environments by Junge (17): C. 0 (1) =Po c . 0 where @ = ratio of adsorbed organic vapor on aerosol to the total amount of vapor in air; 0 = aerosol surface area; PO = saturation vapor pressure; and c = constant that depends on heat of condensation and molecular weight (-0.13 for many organics). Only physical (nonspecific) adsorption is considered; therefore, Equation 1 should be used for trends only. Calculations show that C$io clean air environmentsis small as long as PO is greater than mm Hg. Therefore, in a clean atmosphere,many PCB isomers, DDT, and Hg should exist primarily in the vapor phase. Considering the range of atmospheric environments, organics having Po 1 mm Hg should exist almost entirely in the vapor phase, and those having PO 5 mm Hg should exist almost entirely in the particulate phase. In reality, most high-molecular-weight organics lie between these extremes, @
+
and their distribution and atmospheric lifetimes depend largely on the particle concentration and composition in the atmosphere. Sampling techniques for airborne organics do not adequately distinguish between vapor- and particle-phase species. The most commonly applied collection system employs a high- or low-volume sampler that collects particles on a glass-fiber filter (>95% collection efficiency for 0.1-pm particles), followed by an adsorbent trap. Polyurethane foam (19-21) and XAD-2 microreticular resins (22, 23) have been used to collect atmospheric pesticides and PCBs efficiently. High-volume collection of airborne organics employing the filter-adsorbent system may underestimate the proportion of particle-bound organics though, since they may be desorbed from particles collected on the filter, with the vapor retained by the adsorbent. Filter-retained organics contain the high-molecular-weight and lowvapor-pressure species, as has been demonstrated for a homologous series of alkanes ( 2 4 ) and PCBs ( I O ) in urban atmospheres. Cautreels and Van Cauwenberghe ( 2 4 ) found that the fraction occurring in the particle phase increased with increasing molecular weight and decreasing PO,and demonstrated that a large fraction of high-molecularweight species (e& PAH, phthalate
esters) occurred in the vapor phase, even in urban atmospheres. Therefore, the distribution of trace organics between the particle filter and the vapor adsorbent cannot be taken as unequivocal evidence of vapor-particle partitioning, hut rather the total atmospheric concentration. This observation points out the importance of using a backup adsorbent; otherwise volatile species will be lost. The distribution of airborne organics between vapor and particle phases strongly affects atmospheric removal processes. For PCBs, Junge (18) calculated that -90% of the atmospheric burden should occur in the vapor phase. PCB determinations in urban atmospheres show that -80-100% of the airborne PCBs are operationally in the vapor phase; Le., PCBs were collected in the backup adsorbent (17). Io the Great Lakes environment, 87100% of the atmospheric PCBs over the lakes and in urban areas existed in the “vapor” phase (IO, 13, 17, 23). Thus, for many trace organics, the vapor phase may contribute a sizable fraction of the atmospheric burden. The dominant particle size of atmospheric organics is largely unknown. Doskey and Andren (13) and Eisenreich et al. (16) suggest that PCBs associate with submicrometer particles. Van Vaeck et al. ( 2 5 ) show that aerosol PAHs in urban, rural, and seashore environments have mass Volume 15, Number 1. January I981
31
median diameters (mmd) of 0.7-1.4 p m with 30-70% of the mass occurring in size fractions less than 1 pm. Because of their higher surface area/ volume ratio and higher organic content, submicrometer particles may have higher concentrations than do large particles. However, high-mo-
Ethvlerm
a2
80.8
Environmental Science (L Technology
lecular-weight n-alkanes, carboxylic acids, and PAHs were found to have a significant mass in the >I-pm-diameter particles. Since these particles have higher deposition velocities and washout ratios ( 2 6 ) ,the flux of high-molecularweight organics in the particulate phase may be dominated by deposition of large particles. That is, even though trace organics are concentrated on submicrometer particles, most of their deposition may be associated with larger particles. Over-lake particles may have a particle-size distribution more favoring submicrometer particles since larger ones are removed closer to the soutce. For example, greater than 70% of the TOC (total organic carbon) mass in airborne particles over Lake Michigan occurred in the submicrometer range with organic carbon comprising 7.5-54% of the particle mass depending on source ( 1 4 ) . Vapor partitioning across air-water interface. Transfer of vapors across the air-water interface may be predicted from a two-film diffusion model ( 2 7 ) . In this model, the air and water reservoirs are assumed to be well mixed except for thin, stagnant films of air and water at the interface. The rate of :" governed by the molecular
1.1 x 10-3
:
diffusion across these interfacial layers and is driven by the concentration gradients between equilibrium concentrations at the interface and the concentrations at the interface in the bulk air and water reservoirs. For steady-state transfer, the flux (F) is given by ( 2 8 ) :
F
KoL(C - P / H )
1 _ - -1
(2)
+-
RT (3) KOL KL HKG where F = flux (mol/m2.h); KL,KC = liquid- and gas-phase mass transfer coefficients (m/h); KOL = overall mass transfer coefficient (m/h); H = Henry's law constant (atm.m3/mol); C = solute concentration in liquid phase (mol/m3); P = solute partial pressure (atm); T = absolute temperature (K); R = gas constant. Resistance to gas-phase transfer occurs in both the liquid and gas films. For solutes of increasing molecular weight, K L and KG decrease although their ratio may remain constant. Mackay et al. (28) have shown that for HZ 5X resistance lies almost totally in the liquid phase, and the flux is: F KL(C - P/H) (4, If H 5 5 X then resistance lies almost totally in the gas phase, and the flux is: KG(CH - P ) F= (5) RT For H values between these extremes, resistance to mass transfer occurs in both the gas and liquid film. Table 2 shows volatilization parameters for selected organic compounds. Knowledge of Henry's law constant is essential to predict air-water exchange because the organic concentration in equilibrium with the vapor phase must be calculated. If the water is undersaturated with respect to the atmosphere, then there would be a net transfer of vapor to the water. The rate and amount of mass transferred depends on whether the compound is gasor liquid-phase controlled. Mackay and Yuen ( 2 9 ) summarized the tendency of organic compounds to partition intoor out of water (Table 3). As indicated, the resistance to transfer depends on the phase from which the pollutant tends to partition; if the pollutant is unable to establish high concentrations, diffusion rates are fast. Further discussion of the transfer of gases across air-water interfaces can be found in References 13, 18, and 26-30. In theory, the equilibrium concentration of a trace organic in air and
water may be calculated if H is known. However, vapor-phase organics of low water solubility (e.g., PCBs) entering the water column may remain dissolved and unassociated, bind with dissolved or colloidal organics, adsorb onto a particle surface, or be absorbed into organic detritus. Only the dissolved, unassociated species equilibrates with the organic vapor in the atmosphere. To determine whether organics present as vapors in air are in equilibrium with the organics in water, the concentration of the "dissolved" species must be known. Analytically, this measurement is, at best, difficult at ambient concentrations in the environment of0.1-5 x w 9 g / L . Wet deposition of airborne organics. Wet removal of airborne organic pollutants occurs with the scavenging of particles by and partitioning of organic vapor into rain and snow. The relative importance of these processes depends on the distribution of the organic matter between vapor and aerosol, particle-size distribution, and Henry's law constant. Wet-removal rates for organic vapors are generally low and atmospheric residence times are long. As a result, airborne organics should be relatively uniformly distributed through the troposphere away from sources ( 3 1 ) . A falling raindrop should attain equilibriumwith a trace organic vapor in a -IO-m fall (26). Therefore, the washout of organic vapors may be viewed as an equilibrium partitioning, and a washout ratio defining the scavenging efficiency can be written:
and surface fluxes as:
-
F = 0 1 . J X, = W . J-A',
(7) where J = rainfall intensity; A', = concentration of organic vapor in air; and CY = solubility coefficient. Washout ratios may also be determined from field measurement using the relationship: W = (pg/m3) rain/(pg/m3)air (8) The field determination will correspond to the washout of gases and particles; when Wfield is much larger than a,then scavengingof particles is important in deposition, or H is incorrect. Table 4 lists some washout coefficients, determined in the field (31). which were calculated from H values. Note that the values for DDT and PCBs are much higher than those predicted from a.This suggests that wet removal of airborne DDT may be due to particle scavenging. For PCBs, H estimated from saturation Po and
MMlc...
Pollutant partitions into atmosphere; experiences liquid-phase resistance Pollutant tends to partition more equally; both resistances are important Pollution partitions into water; experiences gasphase resistanc
solubilities may be lower by 102-103 (13). Washout ratios for aerosols are usually -IO6. The physical hasis for this is that the raindrop can be considered a concentrating agent by taking IO6 cloud droplets with a mean spacing of -I mm between them and placing them into a volume of -I mm3. Thus, a I-mm raindrop is the product of -106 collisions with IO-pm cloud droplets. Aerosols that exhibit W 5 IOs are probably insoluble, relatively young, or have dimensions between 0.1-1.0 pm. Many high-molecular-weight organics bound to particles have sizable mass fractions in the submicrometer range (25). Scott ( 3 2 ) summarized the wet removal process as follows: Aerosol removal rates depend on solubility of aerosol, size of aerosol, air concentration, and W: I O 5 < W < I O 6 for rain or warm clouds, W 105 for snow or cold clouds; gas removal rates depend on solubility (Henry's law), air concentration, and W: I O o < W < IO4. Additional details of the wet deposition process are available (26, 31, 3 2 ) . Dry deposition of aerosol-bound organics. The dry deposition of aerosol-bound organics in airborne particles onto a receptor surface depends on the type of surface, resistance to mass transfer in the deposition layer (a few mm), particle size, air concentration, and micro- and macrometeorology. A detailed discussion of the dry deposition process to water is beyond the scope of this paper; readers should consult Sievering et al. (7), Slinn et al. (26),Slinn and Slinn ( 3 3 ) , and Galloway et al. ( 3 1 ) . Particles are deposited on surfaces by Brownian diffusion (mmd < 0.3 pm), inertial impaction-interception (mmd = 0.5-5 pm), and gravitational
Vashout ratios for selected race organics
,roclor
> '
5.9 x 102 2-13 x 104 5 x 104 1-5x 104 4 X 10' i X 10'
6.3 8.3
1.
6-35
x
10'b 2.9
-
7 x 1033 x 105 1-9 x 104
sedimentation (mmd > 5 pm). Because Brownian diffusion increases below 0.3 pm and inertial impactioninterception increases as particle size increases above OS pm, the minimum deposition velocity (vd) is in the 0.3-0.5-pm range ( 3 1 ) . In a simplified case, the flux of particles to a receptor surface is:
F = Vd'Ca (9) where C, = concentration of the organic in air. Tbe Vd depends strongly Volume 15, Number
1, January 1981
33
on the deposition surface, particle-size distribution, and wind speed. Slinn and Slinn (33) suggested that atmospheric particles with mass median diameters of micron size have v d to natural wa-
ters which are rate-limited by turbulent transfer through the constant flux layer and which can be approximated by:
vd(z = 10 m) = CD.10. PI0
= 1.3 x 10-3fi10 (io)
where 2 = height above water; CD = dimensionless drag coefficient; and p10 = mean velocity of the horizontal wind at IO m. They base the deposition model on the premise that deposition of aerosol to a water surface is enhanced by rapid particle growth in the denosition laver due to water condensation. Field values of average v d for airborne oraanics (Table 5 ) were measured bydetermining fluxes to solid surfaces coated with hydrophilic components or fluxes to water surfaces while simultaneously measuring atmospheric concentrations. For most surfaces, it is difficult to determine whether the organic flux is due only to particle deposition or a combination of gas and particle. The deposition velocities determined in this way ( v d = F/C,) are operational in nature, but may represent estimates of actual total dry flux to a water surface if polar,
hygroscopic coatings are used. Caution should be observed in choosing a v d . For estimation of total dry deposition, Vd values of 0.1-0.5 cm/s seem reasonable. Readers should consult References 6 , 7 , 1 3 , 1 7 , 2 6 , 3 1and 33-35 for more details. Concentrations in air 81precipitation A paucity of data exists on the occurrence of trace organiccontaminants in the atmosphere. Only recently has an appreciation developed of the relative impact of atmospheric deposition on water quality (e&, acid rain). Air masses circulating the globe become “polluted” by accumulating chemical components emitted from point (smoke stack) and nonpoint sources (sanitary landfills, urban and agricultural areas). Air masses moving in a general easterly direction transport airborne pollutants from urban/industrial centers to regions where deposition occurs. Few areas of the earth do not experience detectable concentrations of some atmospheric contaminants ( 3 6 ) . We reviewed the published and unpublished literature for the past 10-15 years to determine the identity and concentrations of airborne pollutants. Since atmospheric fluxes to the Great Lakes are a combination of dry and wet removal processes, the data cited are separated into air (vapor particle) and precipitation concentrations (rain snow). Most sampling methodologies do not adequately differentiate between vapor- and particlephase organics. The data base for airborne organics in the Great Lakes Basin was limited, so the summary includes most of that reported in the literature. Concentrationsare reported for chlorinated pesticides, PCBs, PAHs (polynuclear aromatic hydrocarbons), and phthalate esters in air and in precipitation typical of rural areas around the world or in the Great Lakes (Table 6). The number of data entries for an individual compound contributing to the range and mean value used for deposition calculations varied from one to 30. In many instances, especially for PAH, airborne contaminant data were available only for aerosol concentrations in urban areas. PAH concentrations in rain/ snow were estimated from washout ratios using literature citations, and concentrations from Andren and Strand (14) that were obtained over Lake Michigan. Identification of the present-day concentration of atmospheric components was a challenge. The case of DDT is a good example. Use of DDT in the US. and Ontario was banned in
+
+
Total WT
Dieldrln
HCB P.P’-
methoxychlw ol-Endosulfan &Endosulfan Total PAH Anthracene Phenanthrene Pyrene Benzo[a]
.01-0.1 0.1-0.3
0 0.1-1 0.1-4 0.1-1
anUwacene Perylene Benzo[a]
0.1-2 0.1-2
pyrene
34
Environmental Science a Technology
2
the early 1970s following peak usage in the late 1960s. As a result, decreases in DDT concentrations or deposition have been reported for northern Atlantic air and water (37), Lake Michigan fish ( I ) ,and Lake Superior sediments (38). The airborne concentration range quoted for DDT in rural areas in the early 1970s was perhaps 0.1-10 ng/m3; the present value is much less, 0.01-0.1 ng/m3. DDT is still being detected in rainlsnow in the Great Lakes Basin even though usage was to have ended (39, 40). The concentration values of chlorinated hydrocarbons (including PCBs) in precipitation in the Great Lakes Basin have been provided mostly by Canadian effort (12, 39). Compounds which had surprisingly high concentrations at one or more sites and which deserved more attention are a-BHC, y-BHC (Lindane), methoxyclor, DDT, a and @-endosulfan,and PCBs. Although DDT levels remain low at 1-5 ng/L, its presence is surprising and signifies a newly exposed or new atmospheric source. No spatial or temporal trends in atmospheric concentrations are discernable. The atmospheric concentrations of di-n-butyl phthalate (DBP) and di-2-(ethylhexyl) phthalate (DEHP) were taken from studies of Giam and co-workers (40, 41) with rain concentrations estimated from washout ratios. Atmospheric deposition Total deposition (wet dry) of airborne organics to urban and rural environments has been reported for remarkably few compounds. Data available on PCBs in the Great Lakes Basin (10-13. 15, 16) suggest that atmospheric input may represent 60-90% of the total PCB input to Lakes Superior and Michigan, considering all sources. Recently there has been much international interest regarding the importanceof atmospheric deposition of nutrients, metals, acid components, and trace organics. The objective of this section is to estimate the deposition of selected trace organics, including PCBs, PAHs, and phthalates, to each of the Great Lakes. The estimation of atmospheric deposition of trace organics is hampered by several limitations: 1) an inadequate data base on atmospheric concentrations of trace organics; 2) inadequate knowledge of the distribution between vapor and particulate forms in the atmosphere; 3) lack of understanding of the dry deposition process to a water surface; 4) inadequate micro- and macrometeorological information about the lakes during dry
and wet deposition; 5) lack of appreciation for the episodic nature of atmosphericdeposition of trace organics to water; 6 ) inadequate understanding of the temporal and spatial variations in atmospheric concentrations and deposition. However, it is possible to apply reasonable parameterizations to wet and especially dry deposition to estimate total deposition. Table 6 lists the trace organic concentrations in air and precipitation selected to estimate deposition to the lakes. For each organic, one value was chosen as the “best guess” concentration to be used. This approach involves a certain degree of subjectivity and precludes estimating deposition on a
lake-by-lake basis, except for PCBs, which will be discussed separately in the next section. However, other data points can be substituted as they become available. The air and precipitation concentrations of trace organics were then used to estimate wet and dry deposition based on these simplified equations: Wetflux
FW=CRIS.J.SA (11)
Dry flux FD C v p .
Vd. S A
(12)
where Fw and FD = wet and dry deposition, respectively; CRp = concentration in rain and snow; CVIp = concentration in vapor and particulate phase; J = annual precipitation; Vd =
OlIndosulfan
+
Phenanthrene
Volume 15. Number 1, January I981 35
deposition velocity; S A = lake surface area. The wet flux of airborne trace organics, annual precipitation (70-85 cm/y), and surface area for each of the Great Lakes may be calculated from the concentrations given in Table 7. An example calculation is given below for PCB deposition to Lake Superior: Fw = 30 X g/m3. 0.8 m/y IO4 m2/ha. 8.21 X 106 ha = 2.0 X IO3 kg/y
.
Estimating dry deposition to natural water remains the primary enigma in evaluating the importance of the atmospheric pathway for the input of particles and gases to the Great Lakes. Particularly enlightening discussions of the dry deposition process appear in References 7, 13, 26, 33, 34, and 42. A simple parameterizationdescribing the dry deposition process was chosen, with the critical parameter being deposition velocity. The range of V, values reported for transfer of airborne gases and submicron particles to a water surface, whether calculated or measured, is -0.1-0.6 cm/s (see dry deposition section). For the estimation of dry fluxes to the Great Lakes, an intermediate, V, = 0.3 cm/s, was selected. Choice of a particular V,, is open to much discussion, and different values may be selected where appropriate. Given these caveats, the dry deposition of airborne trace organics to each of the Great Lakes may be calculated from the air concentration, deposition velocity, and surface area of the lake. An example calculation for PCB deposition to Lake Superior is given below:
Fo = I X g/m3. (0.3 cm/s) .3.15X 107s/y.10-2m/cm .104m2/ha-8.21 X 106ha = 7.8 X IO4 kg/y The dry and wet deposition of each organic to the Great Lakes is reported in Table 7. In the above calculations, the deposition estimate is primarily sensitive to lake surface area (range of 19 000-82 100 km2) and precipitation intensity ( J = 78-87 cm/y) with other parameters constant for each lake. With this in mind and the small data base available, estimates are probably accurate within a factor of 2-5. We have estimated that dry deposition is 1.5-5 times the wet deposition for the organics selected. In some respects, this may be a result of our uncertainty in estimating dry deposition. However, deposition estimates for PCBs to Lakes Superior and Michigan using a more theoretical approach in38 Environmental Science & Technology
_."...^.
or*:,*:> ,
dicate dry deposition exceeds wet deposition. For a general case, the ratio of wet to dry deposition (R)can be given by:
whereAt=3.15 X 107s,andJAt,is the annual precipitation. For J = 80 cm/y, W = 5 X 104, and V,, = 0.3 cm/s, the ratio of wet to dry is about 0.4, or dry inputs exceed wet inputs by 2.5 times. Airborne PCBs in the basin Atmospheric concentrations. PCBs are a class of compounds produced by the complete or partial chlorination of the biphenyl molecule, which results in mixtures that exhibit properties of chemical and thermal stability, low volatility, and high dielectric constant. In the U.S., PCBs were manufactured under the trade name Aroclor as coolant/dielectric fluids for transformers and capacitors, as heattransfer fluids, and as protective coatings when low flammability was required. They have also been used in plastics, pesticide formulations, lubricants, dedusting agents, adhesives, inks, paints, and caulking compounds. U S . production from 1930 to 1975 was about 570 X lo6 kg, with an additional 1.4 X lo6 kg imported. As of 1975, -340 X lo6 kg were in service-the remainder were in environmental reservoirs: mobile, 68 X IO6 kg; degraded or incinerated, 25 X IO6 kg; disposed of in landfills and dumps, 130 X lo6 kg ( 3 6 ) . PCBs exhibit lipophilic and hydro-
phobic properties in water and therefore accumulate in lipid layers of biota. The ability of PCBs to codistill, volatilize from landfills, and resist degradation at low incinerating temperatures makes atmospheric transport the primary mode of global distribution. In 1979, PCB production in the U.S. was prohibited; disposal of materials contaminated by PCBs was regulated; and the use of materials still in service was restricted. However, PCB production and consumption from 1930 to 1975 resulted in their accumulation in every level of the Great Lakes foodchain. This greatly restricted commercial and recreational fishing and caused great concern for human health.
Given the large number of atmospheric PCB measurements reported in the literature, we were able to separate air and precipitation concentrations into different environmental classifications (Table 8). In this classification, rural represents nonurban, agricultural, or forested areas not adjacent to marine systems; remote signifies continental background or pole sites; and Great Lakes represent specific areas for which measurements are available. As anticipated, PCB concentrations in urban areas are higher than other regions, with rural and Great Lakes values being similar. The relatively large fluxes in rural and Great Lakes regions are partially a result of their proximity to urban/ industrial centers. The ratio of urban to remote fluxes is -40. The spatial distribution of PCBs in air and precipitation in the Great Lakes is shown in Figure 2. Based on these data, the present-day concentration of PCBs in precipitation is -20-50 ng/L, with the highest values reported in the Lake Superior and Lake Michigan Basins. The data shown in Figure 2 and Table 9 for the above lakes indicate that the over-lake PCB concentrations in vapor plus particle phases are -1-2 ng/m3, in contrast to the average PCB concentration in urban air of -7 ng/m3. Surprisingly, the over-lake concentrations in Lake Superior (1.0-1.5
ng/m’) are similar to those found in 1977-78 over southern Lake Michigan, which is much nearer the industrialized centers of Chicago and Gary. More recent data from 1979 suggests PCB levels are indeed higher. Although there are insufficient data to establish a trend, the range and mean of PCB concentrations in air over Lake Superior are significantly lower in 1979 than in 1978. Atmospheric deposition. The high PCB concentrations in Great Lakes fish ( 1 ) and precipitation (10, 1 1 ) even in pristine areas suggested to many that atmospheric inputs may be responsible. Therefore, studies of the input of airborne PCBs to the Great Lakes were undertaken to establish the mass entering the system compared to other sources. Estimating the atmospheric deposition of PCBs to the Great Lakes depends on quantifying wet and dry inputs. Atmospheric deposition to Lakes Superior and Michigan have been estimated by Eisenreich et al. (16) and Doskey and Andren (13), respectively, by measuring PCB concentrations in air over water and applying the following: Airborne particulate PCBs represent 10-15% of the total airborne concentration. V . and W are selected for submicrometer particles. There is gas-phase control for PCB
transfer across the air-water intulaaUu (13). Dissolved, water-column concentrations equal -2 ng/L. Given these data and appropriate constants for calculation of wet and dry deposition, total inputs to Lake Michigan (13) were -9000 kg/y (-160 @g/m2 y) where wet inputs represented -56% of the total. Similar estimates for Lake Superior (16) were -6600-8300 kg/y (80-lo0 pg/m2. y), where wet inputs represented 30% of the total. The differences in the areal loading rates are due to differences in over-lake concentrations and distribution between isomeric mixtures in air and water. The relative importance of the atmospheric pathway for PCB input to the Great Lakes can be surmised only if reliable input estimates are available. The atmospheric inputs of PCBs to Lake Michigan have been estimated as 2500-9000 kg/y with a central value of 5000-6000 kg/y. These values may be compared to 7500 kg/y in 1977 ( l o ) , of which more than 80% came from the atmosphere. Even taking the low estimate of Murphy et al. (15). the atmospheric pathway may still contribute -60% of the total input. Once all sources are identified and PCB inputs are quantified, the question still remains as to what is the fate of these compounds. Eisenreich et al. (16) attempted to answer these important
.
volume 15. Number 1, January 1981 37
questions by quantifying the sources and sinks for PCBs in Lake Superior. the most pristine and sensitive of the lakes. Considering the sources and sinks shown in Table 10, atmospheric deposition represented greater than 85% of the total input to the lake. with tributary inputs contributing most of the remainder. Losses to the sediment were estimated by measuring fluxes in different sedimentation regimes in the lake. Tributary outflow represents only 10% of sedimentary losses based on a whole lake concentration of 2 ng/L. Given the persistent character of PCBs in the aquatic environment, degradation is probably minimal. Losses due to aerosolization and volaltilization, although difficult to quantify. are thought to be negligible. Therefore, net input of PCBs to the lake is -6300-8000 kg/y. Assuming these are accurate within a factor of two, the largest sink for PCB loss appears to be the water column: the water concentration is therefore increasingat therateof0.l-0.5 ng/L.y. All of this assumes a steady-state situation for PCB cycling in the lake, which is probably not true. What is obvious, however, is that atmospheric deposition represents the principal source to the upper lakes. and a significant source to the lower lakes. The PCB concentration in air, precipitation. water. and biota are highest in the upper lakes and result from many factors, including source strength. meteorological patterns. and lake susceptibility. Considering the magnitude of the environmental reservoir available. the general susceptibility of the Great Lakes to deleterious effects and the resistance of many organics to chemical and biological degradation, PCBs, and other trace organics will continue to be a serious environmental and health problem in thcGreat Lakes Basin for decades to come. We must attempt to identify existing atmospheric sources and eliminate them. and to design more effective means to dispose of toxic organics. This task will be most difficult for airborne trace organics. Acknowledgments
The preparation of this paper was assisted greatly by critical discussion of its content and/or contributionof unpublished material by A. Andren, T. Bidleman. J. Kramer. T. Murphy, C. Rice. D.Bondy, and J. Galloway. Financial support was provided in part by the Science Advisory Board ofthe International Joint Commission. and the Large Lakes Research Station of the US. Environmental Protection Agency. This article was read and commcnted on for appropriateness and suitability as an 38
Environmental Science 8 Technology
ES& T feature article by James P. Lodge. consultant in atmospheric chemistry. Boulder. Colo.
References (1)
Dellino. J. J. Enriron. Sci. Txhnol, 1979,
IJ. 1462. (2) Konasewich. D.: Traversy. W.: Znr. 11. S t a m Report on Orgnnic and Heavy Metal Conraminants in the Lakes lirie. Mich.. Huron. and Supcrior Basins. lntcrnationnl Joint Commission. 197X. Windsor. 0nt;irio. (3) Kaiscr. K . 1. E. Enriron. Sci. 'lchnol. 1978.12. 520-28. (4) International Join! Commission. Report o l the Water Oualilv . , Board. 1979. Windsor. Ontario. ( 5 ) Winchester. J. W.: Nilong. ti. D. Worrr. Airond Soil Poll. 1971, 1. 5 0 ~ 6 4 . ( 6 ) Gall, D.F.. Warer. AirmrdSoil Poll, 1975,
' ,IO-
