Accumulation of polychlorinated biphenyls and polycyclic aromatic hydrocarbons in the snowpack of minnesota and lake superior

Seasonal snow cores were collected in late winter before snowmelt in northern and central Minnesota and at Eagle Harbor, Michigan on Lake Superior between 1982 and 1992.. Similarities be

Internat Assoc Great Lakes Res., 2000

Accumulation of Polychlorinated Biphenyls and Polycyclic Aromatic Hydrocarbons in the Snowpack of Minnesota and Lake Superior

Thomas P Franz† and Steven J Eisenreich *

Department of Environmental Sciences

Rutgers University

14 College Farm Rd.

New Brunswick, New Jersey 08901

ABSTRACT. The winter snowpack is a significant reservoir of polychlorinated biphenyls (PCBs) and polycyclic aromatic hydrocarbons (PAHs), and may be utilized as a surrogate receptor for assessing net atmospheric deposition Seasonal snow cores were collected in late winter before snowmelt in northern and central Minnesota and at Eagle Harbor, Michigan on Lake Superior between 1982 and 1992 Snow-pack concentrations of Σ-PCBs ranged from 1 to 14 ng/L with no significant decrease in concentrations from 1986 through 1992 Σ21 -PAH concentrations in 1989 and 1992 ranged from 35 to 3280 ng/L with significantly higher concentrations nearer urban areas Similarities between chemical accumulations in the snowpack and collection of integrated snowfall at Eagle Harbor support the hypothesis that dry depo-sition to accumulated snow is negligible at these remote locations Tributary discharges from spring snowmelt to Lake Superior in 1992 contributed 7 to 11 kg of Σ-PCBs and 220 to 350 kg of Σ21 -PAHs.

INDEX WORDS: PCBs, PAHs, snow, Lake Superior.

220

*Corresponding author: E-mail: eisenreich@envsci.rutgers.edu

† Present address: Metropolitan Council Environmental Services,

Re-search and Development, 2400 Childs Rd., St Paul, MN 55105

INTRODUCTION

Atmospheric transport and deposition distributes

chemical emissions from source regions to remote

environments causing toxicological concern for the

health of their biotic communities (Norstrom et al.

1988; Muir et al 1988, 1990; Bidleman et al 1989;

Hargrave et al 1992; Wania and Mackay 1993).

Methods for assessing atmospheric loadings include

mass balance modeling, the use of surrogate

recep-tors, such as lake sediment, peat and snow cores, as

well as direct measurements of precipitation and

dry deposition

Snow is an excellent tool for assessing

atmos-pheric deposition Snowpacks in northern temperate

and polar regions are a reservoir of accumulated

chemicals that have been deposited by wet and dry

processes over the winter Snow can account for 5

to 40% of annual precipitation within the Great

Lakes region (NCDC 1992) and about 75% of

Arc-tic precipitation (Gregor 1990) Research on

or-ganic chemicals in snow has confirmed the long

range transport of semivolatile organic compounds

(SOCs) to polar regions (Peel 1975, Risebrough et

al 1976, Tanabe et al 1983, McNeely and Gummer

1984, Hargrave et al 1988, Gregor and Gummer

1989, Patton et al 1989, Gregor 1990) However,

few studies have determined concentrations of SOCs in snow from the upper Great Lakes region (Murphy and Rzeszutko 1977, Swain 1978, Stra-chan and Huneault 1979, Murphy and Schinsky

1983, Rapaport et al 1985, Boom and Marsalek

1988) and its relative contribution compared to other inputs

Snow has been considered in mass balance mod-els for PAHs and PCBs at Siskiwit Lake, Isle Royale, in Lake Superior (McVeety and Hites 1988,

Swackhamer et al 1988) In Green Bay, the annual

wet flux (snow + rain) of PCBs was included in at-mospheric input calculations (Franz and Eisenreich

1993, Bierman et al 1993) However, in other mass

balance efforts, annual wet deposition fluxes are based on rain concentrations (Strachan and Eisenre-ich 1988, EisenreEisenre-ich and Strachan 1992) Omission

of snow as a separate input pathway to the Great Lakes is because of a lack of information on SOC concentrations in snow within the region

Snowpacks integrate various transport,

scaveng-ing, and deposition phenomena in addition to

vari-ous post-depositional diagenetic processes The

concentrations observed are the net result of

precip-itation, dry particle deposition, gas exchange, and

percolation Thus, the snowpack concentration

CSnowis given by:

(1) Wet and dry (gas + particle) depositional

processes, gaseous volatilization, and water

perco-lation are the principal pathways whereby SOCs

be-come first entrained and potentially lost in the

snowpack A complete description of the important

diagenetic procceses influencing accumulation is

given in Franz et al (1997).

This study was initially conducted to assess the

similarities between atmospheric deposition and

ac-cumulations of PCBs and chlorinated pesticides in

rural/remote peat bogs of North America (Rapaport

et al 1985, Rapaport and Eisenreich 1988) Snow

cores were collected from 1982 to 1985 in northern

Minnesota during this phase of the study In 1986

and 1989, samples were taken to continue the chronological record and to compare PCB

concen-trations in snow to those in rain (Franz et al 1991,

Franz and Eisenreich 1993) Field investigations in

1992 evaluated diagenetic processes within the snowpack (Franz 1994) and determined snow scav-enging of atmospheric SOCs (Franz and Eisenreich 1998) The objective of this paper is to summarize the 1982 to 1992 snow data and to report the con-centrations and regional variability of PCBs and PAHs in annual snowpacks Precipitation data from the Integrated Atmospheric Deposition Network

(IADN) site at Eagle Harbor (Gatz et al 1994, Hoff

et al 1996) are compared to snowpack

concentra-tions to evaluate the importance of dry deposition And finally, snowmelt contributions to tributary loadings to Lake Superior during the spring snowmelt are estimated

EXPERIMENTAL Site Description

Snow was collected at four sites in Minnesota and at Eagle Harbor, Michigan (Fig 1) near the end

of winter before snowmelt Table 1 lists the

Wet dry Adsorption Volatilization Percolation

Precipitation Evaporation Percolation

tions of the sites and characteristics of the snow

cores Within the Marcell Experimental Forest in

northern Minnesota, snow was collected in an open

meadow of 0.4 hectare surrounded by forest and

maintained as a National Atmospheric Deposition

Network (NADP) site At the Lake Itasca State

For-est, sampling occurred in a meadow located near

the shore of Lake Itasca, headwaters of the

Missis-sippi River The Cedar Creek Natural History Area

is located about 50 km NW of Minneapolis/St Paul

in an agricultural region Sampling occurred in a

grassy field used as an atmospheric monitoring site

The Gray Freshwater Biological Institute (GFBI) is

located approximately 35 km west of

Minneapo-lis/St Paul in a suburban setting The IADN Eagle

Harbor, Michigan site is near the northwest tip of the Keweenaw Peninsula Samples were taken within 50 m of Lake Superior

Sampling Protocol

All equipment was washed with Alconox and rinsed with tap water, Milli-Q® water (Millipore), acetone and hexane, or methanol and dichloro-methane and wrapped in aluminum foil prior to transport to the field In the autumn of each year (1982 to 1986 samples), 1-m2sheets of 3 mil plastic were secured on the ground at each sampling loca-tion In subsequent years, no plastic sheeting was used because it was deemed unnecessary In late winter, a one-square-meter area was inscribed on

Water Number of Precipitation Equivalent Snow Accumulation during Percent of Location Number # Year Surface Area Snow Depth Depth Density Days Accumulation Precipitation

Marcell State

(Lat 47° 32 ′ N,

2 1991–92 1.05 47.5 11.3 ± 0.7 0.24 ± 0.01 114 10.8 105 ± 6 Cedar Creek Natural

History Area, MN 3 1985–86 1.25 41 13.6 ± 0.3 0.34 ± 0.2 114 13.1 104 ± 2 (Lat 45° 19 ′ N,

Long 93° 17 ′ W) 2 1988–89 1.25 14.5–23.5 a 6.8 ± 1.4 0.36 ± 0.01 102 10.5 109 ± 30 Lake Itasca State

(Lat 47° 13 ′ N,

Long 95° 12 ′ W)

(Lat 44° 57 ′ N,

Long 93° 39 ′ W)

(Lat 47° 28 ′ N, (1/7/92)

Long 87° 52 ′ W) 2 (3/21/92)

a Samples taken in meadow with small hillocks and depressions, some drifting snow Depths highly irregular and listed for each sample.

b Top section of snowpack represents snowfall accumulation from 7 Jan to 21 March 92.

c Bottom section is replicate sample of January snowpack that accumulated snow 23 Nov 91 to 7 Jan 92.

the snow surface, quartered, and the snow on two

sides removed to ground level to allow access to the

entire core Duplicate snow cores were collected in

0.25 m2 quadrants in 110 L anodized aluminum

cans and covered with an aluminum foil-lined lid

Samples for dissolved organic carbon (DOC) and

suspended particulate matter (SPM) were collected

using a 6.5 cm i.d plexiglass tube and kept frozen

in plastic bags Monthly IADN wet-only

precipita-tion samples were taken at Eagle Harbor as

de-scribed by Sweet et al (1993), Hoff et al (1996),

and Hillery et al (1998).

In the laboratory, the snow containers were

weighed to determine water volumes Between

1982 and 1989, the snow was allowed to melt for 2

to 5 days within a walk-in refrigerator at 4°C

Melted snow was then passed through an XAD-2

resin (Sigma Chemical Co.) column (glass cartridge

2.5 cm i.d × 20 cm) using a peristaltic pump at

flow rates of 100 to 200 mL/min Particulate matter

was trapped on glass wool plugs holding the resin

within the column The empty snow container was

rinsed with either acetone (1982 to 1986), or

methanol and dichloromethane (1989) to collect

ad-hered particles and compounds sorbed to the

con-tainer walls These rinses were later added to the

extract

In 1992, the snowmelt was maintained at 4°C and

filtered using a submersible pump, a stainless steel

filter head, and precleaned 293 mm diameter glass

fiber filters (GFFs) (Schleicher and Schuell No

25) The filtrate, collected in precleaned 65 L

stain-less steel tanks, was passed through a XAD-2 resin

column as described The snow cans were rinsed

with 2 L of Milli-Q water and filtered with the

re-maining sample No solvent rinse of the cans was

performed This method allowed the determination

of both dissolved (XAD-2) and particulate (GFF)

fractions within the snowmelt

Subsamples for DOC and SPM were transfered to

2 L glass beakers and allowed to thaw at room

tem-perature while covered with aluminum foil

Ap-proximately 250 to 750 mL of the melt water was

filtered through a 0.4 µm Nuclepore filter for

sus-pended particulate matter (SPM) analysis The

re-mainder was filtered through 47 mm GFFs with the

filtrate collected in polyethylene bottles for DOC

analysis

Analytical Procedure

Although sampling and analysis occurred over a

decade, similar analytical procedures were

em-ployed with minor variations Basically, the proce-dure consisted of 24 hr sequential Soxhlet extrac-tions of the XAD resin and GFFs using acetone and hexane, or methanol and dichloromethane Surro-gate standards of mirex (1982 through 1986), or PCB congener #166 (2,3,4,4

′,5,6-hexachloro-biphenyl) and d12 -chrysene (1989 and 1992 sam-ples) were added to the resin in the Soxhlet prior to extraction to evaluate analytical recoveries The ex-tracts and rinses were back-extracted with Milli-Q water to remove water soluble solvents, concen-trated in a Kuderna-Danish apparatus with a solvent switch to hexane, cleaned and fractionated using a Florisil or alumina/silica column, concentrated in a Kuderna-Danish apparatus, and reduced with N2 gas to final volume Internal quantification stan-dards (2,4,6-trichlorobiphenyl, IUPAC #30 and

and deuterated PAHs d10 anthracene, d12 benzo(a)anthracene, d12 benzo(a)pyrene and d12 benzo(g,h,i)perylene) were added prior to final vol-ume reduction in 1989 and 1992 samples The con-centrated extracts were analyzed on either an Hewlett-Packard (HP) 5840A or HP-5890 GC with

63Ni electron capture detector (PCBs) or HP-5890

GC with an HP-5970 mass selective detector (PAHs) Selective ion monitoring and retention times were used to identify the PAH compounds using a 30m DB-5 (J & W Scientific), 0.32 mm i.d., 0.25 µm film thick glass capillary column Helium was the carrier gas with a linear velocity of about

33 cm/sec Injection was splitless with an initial column temperature of 50°C held for 1 minute, then ramped at 25°C/min to 125°C and then at 10°C/min

to 290°C and held for 10 min Injection port and GC-MS interface temperatures were 290°C and 300°C, respectively Electron multiplier voltage was either 1,800 or 2,000 emv Compounds were quantified using either external (1982 to 1986) or internal standards (1989 and 1992) Details of ana-lytical methods and GC-ECD instrumental

condi-tions for PCBs are described in Rapaport et al.

(1985); Rapaport and Eisenreich (1988) (1982

through 1985 samples); Franz et al (1991) (1986

samples); and Franz and Eisenreich (1993) (1989 and 1992 samples)

Nuclepore filters (SPM) were dried overnight at 50°C and placed in a dessicator prior to weighing

on a Perkin Elmer Model AD-2 microbalance Dis-solved organic carbon (DOC) was measured by IR following either persulfate-enhanced UV digestion

in a Dohrmann DC-80 Carbon Analyzer or

combus-tion at 750°C in an Ionics Model 555 Total Organic

Carbon Analyzer

Quality Control/Quality Assurance

Quality control and assurance (QA/QC) details

are described elsewhere (Rapaport et al 1985,

Ra-paport 1985, RaRa-paport and Eisenreich 1988, Franz

et al 1991, Franz and Eisenreich 1993, Franz

1994) Briefly, instrument detection limits (defined

as 3x signal:noise ratio) ranged from 0.001 to 0.2

ng for PCB congeners (0.7 to 10 ng for Σ-PCBs)

and from 0.01 to 0.1 ng for individual PAHs

Ma-trix blanks accounted for ~10 to 20% of the sample

mass for PCBs and ~5 to 10% for PAHs

Break-through of dissolved SOCs was evaluated by two

XAD columns in series The primary column

recov-ered an average of 82 ± 12% (n = 5) of Σ-PCBs and

97 ± 3% of individual PAHs Annual average

surro-gate recoveries ranged from 71 to 108% for mirex

or PCB congener #166 and from 74 to 89% for the

PAH surrogate d12-chrysene

Data for 1982 to 1985 are not corrected for

surro-gate recoveries or blanks Samples in 1986 were

corrected for the recovery of mirex and the average

mass from XAD Blanks Similarly, all PCB results

in 1989 and 1992 were blank corrected after being

adjusted for the recovery of surrogate PCB con-gener #166 The PAH results (1989 and 1992) are blank corrected but not adjusted for surrogate recovery

RESULTS AND DISCUSSION

Table 1 lists the location of the snow cores and their characteristics Snow events at Cedar Creek and Itasca in 1989 and at Marcell and GFBI in

1992, had densities of 0.12 to 0.18 g/cm3 (Franz 1994) Seasonal snow cores exhibited densities ranging from 0.16 to 0.37 g/cm3 Cores from north-ern Minnesota, which experience few days with above-freezing temperatures during winter, had densities of 0.16 to 0.25 g/cm3, compared to central Minnesota cores with densities of 0.34 to 0.36 g/cm3 which experienced some melting These den-sities are similar to the 0.38 ± 0.03 g/cm3density in cores from Canada (Strachan and Huneault 1979); 0.3 to 0.4 g/cm3 in Canadian Arctic snow cores (McNeely and Gummer 1984) and the 0.25 to 0.41 g/cm3 in cores from Sault Ste Marie, Ontario (Boom and Marsalek 1988) The Eagle Harbor cores exhibited densities of 0.32 to 0.37 g/cm3, sim-ilar to central Minnesota

Based on daily precipitation records at nearby

Eagle Harbor, MI 1/7/92 (2) 1991–92 1.8 ± 0.5 5.6 ± 1.4 1.5 ± 0.3

(a) Suspended particulate matter in snow

(b) Dissolved organic carbon in snow

(c) Volume weighted mean of January and March top snow cores

(d) Volume weighted mean of March top and bottom snow cores

National Weather Service sites (NCDC 1992), the

water retention efficiency of the snowpack relative

to the amount of precipitation that occurred during

the accumulation period was calculated The

per-cent of precipitation sampled in the snow core

(Table 1) is defined as the water equivalent snow

depth relative to the amount of recorded

precipita-tion Deviations from unity are attributable to

subli-mation, percolation of water to the ground surface,

snow drifting, and snowfall variability between the

snowpack sampling site and the snowfall recording

site Water loss by sublimation was not significant

during the winter Snow cores retained an average

of 91 ± 10% of the precipitation that occurred

dur-ing snowpack accumulation Obvious exceptions

occurred in suburban MN (GFBI) and Eagle

Harbor

The seasonal snow core at GFBI exhibited low

water recovery (67 ± 2%) that was attributed to

sig-nificant snow melt Melting may not sigsig-nificantly

increase the density of a snowpack if some water

percolates out of the core The measured density

then reflects the packing density of the remaining

snow cover

Cores at Eagle Harbor were obtained on 7

Janu-ary and 21 March 1992 to examine temporal

changes in the snowpack Low water recovery

(45%) was noted in the bottom section of the March

core, a replicate sample of the January core This

section had the same density as in January (0.33 ±

0.1 g/cm3), but half the water content The January

and March cores were taken within 5 m of each

other and were visually similar with no obvious

in-dication of melting

PCB Snow Concentrations

The concentration of total PCBs (Σ-PCBs) in

sea-sonal snow cores from 1982 to 1992 ranged from

0.8 to 14 ng/L (Table 2) The coefficient of

varia-tion (RSD) amongst several sets of replicate cores

averaged 41 ± 22% With the exception of Σ-PCB

concentrations of ~10 ng/L in 1983 to 1985,

con-centrations were about 1 to 2 ng/L, similar to the

values in Great Lakes rain (Hoff et al 1996)

Win-ter deposition in Win-terms of concentrations of

atmos-pheric PCBs has not diminished significantly since

1986, a behavior reminiscent of atmospheric PCBs

(Hillery et al 1997) It is now known that

atmos-pheric PCBs measured at some IADN sites are

de-creasing with a half-life of about 3 to 6 years

(Hillery et al 1997; Simcik et al 1999)

Interest-ingly, IADN Lake Superior data at Eagle Harbor do

not show any statistical decrease This agrees with measurements of atmospheric PCBs over and near Lake Superior which have not decreased

apprecia-bly (Baker and Eisenreich 1990, Hornbuckle et al.

1994, Hillery et al 1997) Also, with the exception

of Marcell in 1983 to 1985, there is no clear spatial variation among the sites suggesting a well-mixed atmospheric source signal The mean Σ-PCB

con-centrations are equivalent among the sites during any one year (p < 0.05) Samples collected within

50 km of the Minneapolis/St Paul metropolitan area at Cedar Creek and at suburban GFBI have ap-proximately the same concentrations as those from remote northern Minnesota (Marcell and Lake Itasca) and at Eagle Harbor on Lake Superior The range of PCB concentrations in snow are similar to other values within the Great Lakes region (Table 3) and are similar to rain concentrations (Strachan

1990, Franz and Eisenreich 1993, Gatz et al 1994, Hoff et al 1996, Hillery et al 1998) The

volume-weighted mean (VWM) concentration of Σ-PCBs in

snowpack at Eagle Harbor in March was 1.7 ng/L The wet-only VWM Σ-PCB concentration from

De-cember through mid-March in IADN precipitation

samples was 2.0 ng/L (Gatz et al 1994) In the

1992 snowpack, 47 to 80% of Σ-PCBs were in the

particulate phase The di- and tri-chlorinated con-geners were primarily in the dissolved phase (< 50% particulate), while the higher chlorinated congeners were predominantly in the particle phase

PCB Snow Accumulations

The mean concentration of PCBs in the snow-pack and the water equivalent depth were used to calculate the winter accumulation (Fig 2) Winter accumulation of Σ-PCBs ranged from 0.13 to 1.0

µg/m2 No significant differences (p < 0.05) were found among the sampling sites in the snowpack deposition in 1982 and 1983 and from 1986 through

1992 Thus, no temporal or spatial differences in the regional deposition of PCBs is evident even at suburban sites (Cedar Creek and GFBI) within 50

km of Minneapolis/St Paul This suggests a nearly uniform atmospheric source signal throughout the region in winter with the accumulation of PCBs ranging from 0.2 to 0.4 µg/m2since 1986

The apparent deposition of Σ-PCB reflected in

snow accumulations are generally less than other snow deposition estimates from the Great Lakes region—range: ~0.4 to 3.5 µg/m2 (Murphy and

Schinsky 1983, Swackhamer et al 1988, Franz and

Eisenreich 1993) Snow accumulation in 1992 is

similar to the upper range for Arctic winter

accu-mulation of 0.01 to 0.3 µg/m2 (Gregor 1991)

A direct comparison between rain and snow

load-ings can be made from precipitation studies at

Cedar Creek in 1986 (Franz et al 1991) and Green

Bay in 1989 and 1990 (Franz and Eisenreich 1993)

The PCB flux from rain at Cedar Creek was 1.4 ±

0.3 µg/m2/yr Snowfall during the 1985–86 winter

accounted for about 20% of annual precipitation at this site while contributing 0.18 ± 0.14 µg/m2/yr of PCBs Thus, snow contributed ~12% of the annual PCB flux At three sites near Green Bay, Lake Michigan, the mean Σ-PCB flux ranged from 1.0 to

2.0 µg/m2/yr for rain and 0.36 to 0.54 µg/m2/yr for snow (Franz and Eisenreich 1993) Thus snow was responsible for 22 to 27% of annual PCB loadings

PCB Concentration Year Location mean (range), ng/L Type of Snow Sample Reference

1975–76 Ontario, Canada 18–43a Snowpack Strachan and Huneault (1979) 1975–76 Chicago, IL 212 ± 97 Snow Events Murphy and Rzeszutko (1977) 1982–83 Isle Royale, Lake Superior 17 Snowpack Swackhamer et al (1988)

1982–85 Marcell, MN 7.6 ± 4.4 (1.4–13.6) Snowpack This Study

1985–86 Minnesota 1.6 ± 0.3 (1.3–1.9) Snowpack This Study

1988–89 Minnesota 1.7 ± 0.8 (0.8–2.8) Snowpack This Study

1989–90 Green Bay region, WI (1.4 – 5.1) Integrated Snow Events Franz and Eisenreich (1993) 1991–92 Eagle Harbor, MI 2.0 (1.3 – 2.6)c Integrated Snow Events Gatz et al 1994

1991–92 Minnesota & Michigan 1.8 ± 0.4 (1.3–2.3) Snowpacks This Study

aRange of means within various regions

bEvent began as rain, then turned to snow, half of the precipitation amount in each form

cVolume-weighted mean and range of wet-only precipitation between 12/3/91 to 3/17/92 for total PCBs for same con-geners as analyzed in this study

1993 The values given are the averages of two snow cores each.

from wet deposition to Green Bay while accounting

for 20 to 30% of annual precipitation in the region

Dry deposition of PCBs to the snowpack was

evaluated by comparing cumulative snowfall

depo-sition versus accumulation in the snowpack at

Eagle Harbor Cumulative wet deposition was

cal-culated from the measured monthly IADN Σ-PCB

concentrations and snowfall (water equivalent)

dur-ing the 1991-92 winter (Gatz et al 1994)

Cumula-tive snowpack accumulation is the sum of measured

Σ-PCB deposition in the Eagle Harbor snow core

taken in January and the top section of the March

core that integrated atmospheric inputs from 23

No-vember 1991 to 7 January 1992 and from 7 January

to 21 March 1992, respectively Cumulative snow

deposition is the sum of the integrated snowfall

measurements at the IADN site collected monthly

from 3 December 1991 through 19 March 1992 No

significant difference (p < 0.05) was observed

be-tween the cumulative snow deposition of PCBs

(0.32 ± 0.05 µg/m2) and accumulation within the

snowpack (0.40 ± 0.11 µg/m2) This suggests that

falling snowfall is the dominant source of PCBs in

snowpacks Thus if gaseous PCBs are sorbed to

ice/snow crystals, it likely happens during snowfall

Atmospheric concentrations of S-PCBs seldom

ex-ceeeds 60 pg/m3in the cold of winter

Comparison of Σ-PCB annual snow accumulation

of ~ 0.4 µg/m2 to other fluxes in Lake Superior is

informative The surface sediment accumulation

rate of Σ-PCBs is ~1 to 2 µg/m2/y (Jeremiason

et al 1994), and the atmospheric loading is

~1 µg/m2/y from wet and dry particle deposition

and ~5 µg/m2/y if PCB gas absorption is included

(Hoff et al 1996) Σ-PCB fluxes on settling

parti-cles in Lake Superior for this time period were

about 18 µg/m2/y (Jeremiason et al 1998)

How-ever benthic recycling ratios of ~20 lead to

ob-served sediment accumulation rates (Baker et al.

1991, Jeremiason et al 1998) Thus Σ-PCB snow

accumulation rates are comparable to assessed

at-mospheric deposition and surficial sediment

accu-mulation rates

PAH Snow Concentrations

Total PAHs (Σ21-PAHs) as the sum of 21

individ-ual PAHs ranged from 35 to 3,300 ng/L among

1989 and 1992 seasonal snow cores (Table 4)

Replicate variability of all individual PAHs

aver-aged 17 ± 13% Relatively low concentrations of

Σ21-PAHs (35 to 120 ng/L) were found at the

rural/remote sites Higher concentrations (Σ21-PAHs

230 to 3,280 ng/L) were found nearer the urban areas at Cedar Creek and GFBI

Table 5 compares the concentrations of PAHs in winter precipitation at a number of remote and urban locations Snowpack concentrations in Sault Ste Marie, Ontario at the eastern shore of Lake Su-perior (Boom and Marsalek 1988) are significantly higher than observed elsewhere and are attributable

to nearby steel manufacturing In Portland, Oregon

(Ligocki et al 1985a,b), PAH concentrations of the

lower molecular weight species (< Pyr) in winter rain are higher than Lake Superior snow concentra-tions, while the higher molecular weight PAHs are

in close agreement All other samples taken in the Lake Superior region are similar, although high concentrations of low molecular weight PAHs (< Pyr) were observed on Isle Royale (McVeety and Hites 1988)

Filtration of 1992 snow samples determined that

28 to 100% of PAHs were associated with particu-late matter Only the low MW PAHs acenaphthy-lene (Acy), acenaphthene (Ace), and fluorene (Flr) were found primarily in the snow filtrate The dom-inance of the particulate fraction of medium and high molecular weight PAHs in winter snowpack suggests that snow scavenging of soot particles is likely the primary atmospheric removal mechanism However, Schmitt (1982) suggested that the impor-tance of particle scavenging diminishes with dis-tance from urban sources as particulate emissions are efficiently washed out close to the source The data in Table 4 supports this observation such that the proportion of particulate PAHs in rural snow-packs (~80%) is somewhat less than found in sub-urban snow (~98%)

PAH Snow Accumulations

The winter accumulation of Σ21-PAHs ranged from 4.7 to 13 µg/m2at the remote sites in 1989 and

1992 and from 20 to 210 µg/m2 at the suburban sites Urban sources contribute significantly to the suburban snowpack Deposition is much less than estimated annual emissions of PAHs in the Great Lakes region which range from approximately 400

to 6,400 µg/m2/yr (Johnson et al 1992) At Eagle

Harbor, the deposition of PAHs in the January 1992 snowpack was similar to that calculated from the VWM concentration of the top and bottom sections

of the March snowpack This suggests that losses from meltwater percolation during this period equaled gains from subsequent snowfalls

A comparison of the cumulative IADN snowfall

Franz and Eisenreich

TABLE 4 Concentrations (ng/L) of PAHs in seasonal snow cores.

Marcell, MN Lake Itasca, MN Cedar Creek, MN Marcell, MN Eagle Harbor, MI a GFBI, MN

(a) Volume-weighted mean concentration of top and bottom core on 3/21/92.

(b) Standard deviation calculated from replicate coefficient of variation of January snow cores.

PCBs and P

Portland, OR a Isle Royale, LS b Sault Ste Marie, ONT c Arctic d Narragansett Bay, RI e Eagle Harbor, MI f Rural/Remote g Suburban, MN g

aLigocki et al 1985 a,b f IADN Data (Gatz et al 1994)

b McVeety and Hites 1988 g This study

c Boom and Marsalek 1988 h Sum of 5 methylphenanthrene isomers

dWelch et al 1991 i Sum of benzo(b+j+k)fluoranthene isomers

e Latimer (1994)