EPA-600/3-84-101
ATMOSPHERIC TRANSPORT OF TOXAPHENE
TO LAKE MICHIGAN
by
C. P. Rice
Great Lakes Research Division
Institute of Science and Technology
The University of Michigan
Ann Arbor, Michigan 48109
P. J. Samson
Department of Atmospheric and Oceanic Sciences
The University of Michigan
Ann Arbor, Michigan 48109
G. Noguchi
Great Lakes Research Division
Institute of Science and Technology
The University of Michigan
Ann Arbor, Michigan 48109
February 1984
-------
ABSTRACT
Atmospheric levels of toxaphene were monitored during the summer and fall
of 1981 at 4 locations: Greenville, Mississippi, St. Louis, Missouri, Bridg-
man, Michigan, and Beaver Island, Michigan. Each collection was conducted by
continuously sampling air during the first two weeks of the months of August,
September, October, and November. The collected toxaphene was analyzed on a
capillary equipped electron capture gas chromatograph. The average concen-
trations over the entire sampling period for each site were 7.39 ng/m3 in
Greenville, 1.18 ng/m3 in St. Louis, and 0.27 ng/m3 for Lake Michigan (Bridg-
man and Beaver Island combined). The summer versus fall fluctuations in amount
of toxaphene at each site was 0.44 ng/m3 versus 0.26 ng/m3 for Bridgman;
1.73 ng/m3 versus 0.63 ng/m3 for St. Louis; and 9.05 ng/m3 versus 4.34 ng/m3
for Greenville. The maximum monthly average occurred in September for all of
these collections. Diagnostic modeling to describe possible air transport of
toxaphene showed that at all receptor locations the air transport corridor for
toxaphene was associated with southerly winds. The strength of this corridor
increased from northern to southern measurement sites. A flux estimate for
toxaphene deposition to the lake surface ranged from 3,360 to 6,720 kg/yr.
1 • . .
iii
-------
CONTENTS
Abstract ill
List of Figures vi
List of Tables viii
Introduction 1
Methods 5
Air Collection 5
Extraction 12
Chromatographic Analysis 12
Quality Assurance • • • 15
Modeling 19
Results and Discussion 21
Modeling of Atmospheric Transport 27
Estimates of Atmospheric Flux of Toxaphene to Lake Michigan 39
References 43
Appendix A - Tests For Silicic Acid Fractionation of Toxaphene Al
Appendix B - Letter to Houston Wells, Wells Laboratories, Describing Bl
Study of Toxaphene Formulations Used In Greenville,
Mississippi
-------
FIGURES
Number Page
1 Locations of the four sites where air was sampled for toxaphene
in the summer and fall of 1981 11
2 Capillary chromatograms of toxaphene-containing air, rain and
fish samples and EPA toxaphene standard. The sculpin sample
was analyzed on another project and is provided for comparison
of the atmospheric data with biological residues.
a) Air sample from Beaver Island (28 August to 31 August 1981)
containing 0.054 ng/nH toxaphene; b) Rain sample from
Beaver Island (28 September to 2 October 1981) containing
31.58 ng/L toxaphene; c) Sculpin composite sample from Lake
Michigan (17 June 1982) containing 0.718 ng/gm toxaphene;
d) toxaphene standard (4 yL of 118 ng/mL). G/C conditions -
50 m fused silica column coated with SE-54 and temperature-
programmed from 100 to 240° C at l°/min 16
3 Linearity plot of the electron capture detector response to
toxaphene. Plotted are the observed areas for the toxaphene
standards versus their concentration (118, 222, and 444 ppb). .. 17
4 Simultaneous determinations of toxaphene in air measured
at four sites lying along a general south-to-north transect
from Greenville, Mississippi, through St. Louis, Missouri,
and ending in the Lake Michigan area (Bridgman and
Beaver Island, Michigan). The vertical axes are in log
units and the horizontal axes are calendar dates 24
5 Toxaphene in air measured at Beaver Island, Michigan, from
28 August to 4 October 1981. Vertical axis is in log
units and the horizontal axis is in calendar dates 25
6 Unweighted probability of contribution during the sampling periods
in units of 10~8/km2 in Bridgman, Mi. (approximate center of
concentric rings) . 28
7 Weighted probability of contribution during the sampling periods
in units of 10~8/km2 in Bridgman, Mi 29
8 Difference between Figures 6 and 7 31
9 Unweighted probability of contribution during the sampling periods
in units of 10~8/km2 in St. Louis, Mo. (approximate center of
concentric rings) 32
10 Weighted probability of contribution during the sampling periods
in units of 10~8/km2 in St. Louis, Mo 33
vi
-------
11 Difference between Figures 9 and 10 34
12 Unweighted probability of contribution during the sampling periods
in units of 10~°/km^ in Greenville, Miss, (approximate center
of concentric rings) 35
13 Weighted probability of contribution during the sampling periods
in units of 10~8/km2 in Greenville, Miss 36
14 Difference between Figures 12 and 13 37
vii
-------
TABLES
Number Page
1 Amount of toxaphene in air samples from Bridgman, Michigan;
Beaver Island, Michigan; St. Louis, Missouri; and
Greenville , Mississippi . 7
Duplicate results 18
Trend in toxaphene concentration at sampling sites for the
summer and fall of 1981 26
viii
-------
INTRODUCTION
Occurrence of toxaphene in the Great Lakes ecosystem is possibly one of
the most perplexing pollution problems to have ever occurred in this region.
Detectable levels of toxaphene were first evident in fish collected in 1974
from Lake Michigan and reported by the U.S. Fish and Wildlife Service (Rappe
et al. 1979). Subsequent to this, additional data have been reported which
show toxaphene as a widespread contaminant of Great Lakes fish (Rice and
Evans, in press). However, the use of toxaphene is primarily concentrated in
the southern states with few known applications of this pesticide in the Great
Lakes region. The sources of toxaphene input to the Great Lakes have not yet
been determined. Difficulty in analyzing environmental samples for toxaphene
has prevented thorough examination of the occurrence and distribution of
toxaphene in the environment.
Like PCS, toxaphene is not just one compound. In fact, it is a mixture
of at least 180 separate chemical compounds (Holmstead et al. 1974). This
complex chemical makeup of toxaphene is the primary reason that its
environmental fate is so poorly understood. Conventional packed column
separation with electron capture detection is not adequate to precisely
characterize toxaphene. Fortunately, capillary chromatography techniques have
allowed better resolution of the component peaks and have improved the
confidence with which toxaphene is identified in samples. However, selective
weathering of the original toxaphene peaks is believed to greatly alter the
ratio of peaks from the parent material (Zell and Ballschmiter 1980), as well
as create new peaks which are derived from the parent toxaphene but cannot be
used for matching and quantification in the usual sense. In a study of the
1
-------
global distribution of toxaphene, Zell and Ballschmiter (1980) presented a
technique using capillary separation and analysis of fish residues for
chlorinated hydrocarbons which dealt with many of the analytical difficulties
presented by toxaphene. In this method, fractionation of the extracts with
florisil chromatography was carried out to reduce interferences from PCBs.
This type of approach has also been successfully applied by others. Ribick
et al. (1982) have analyzed fish tissues for toxaphene using silica gel
fractionation and capillary chromatography while others have relied on silica
or florisil separation and packed column analyses (Schmitt et al. 1981,
Bidleman et al. 1978, Devault et al. 1982). All of these techniques are based
on chromatographic methods which rely on retention time matching as the basis
for identification. However, these techniques do not provide absolute
chemical confirmation. To date, exact identification of only 11 of the 177
constituents in standard toxaphene has been made (Korte et al. 1979). Exact
chemical identification of environmental residues, however, is difficult under
the best conditions because levels are usually low and many inferences have to
be made. Furthermore, toxaphene poses additional problems because standard
electron impact mass spectrometry is relatively insensitive to the components
of toxaphene. Positive chemical ionization techniques with mass spectrometry
have been used to identify some toxaphene components in residues from fish
from the Great Lakes (Ribick et al. 1982) and this is the only verification to
date that absolutely confirms the presence of toxaphene in Great Lakes fish.
Because of the analytical difficulties in measuring toxaphene in environ-
mental matrices, this study is as much an analytical methods study as it is a
-study of environmental distribution. As already mentioned, the perplexing
question about toxaphene in the Great Lakes is "How is it getting there?"
-------
The choices are limited as uses of this insecticide in the Great Lakes area
u I--'"
* ^,*
are few, with the preponderant uses occurring in the cotton and soybean •-•;'
growing areas of the United States, e.g. the Cotton Belt. Some use has been
reported for the Great Plains States, however for 1972 the total of 0.5
million Ibs. used by all the states from the region was only 1% of the Cotton
Belt usage. In 1980, South Dakota used about 0.14 million Ibs. of toxaphene
which would represent about 1% of estimated 1980 U.S. usage (Rice and Evans,
in press). No evidence for accidental spillage has been uncovered, and
manufacturing is not located in any of the Great Lakes states. One proposal
contended that pulp bleaching may inadvertently produce toxaphene-like
materials in its waste liquor. However, controlled laboratory tests under
optimal conditions for chlorine substitutions of pulping waste liquors could
not produce toxaphene-like compounds with greater than three chlorine atoms
(presented by David Stalling, U.S. Fish and Wildlife Service, at EPA hearing
for cancellation of toxaphene, July 1982); toxaphene is composed mainly of six
to nine chlorine-containing compounds. To many scientists, the only logical
choice left seems to be atmospheric transport. Direct observations of this,
however, are limited.
There is a good deal of experimental evidence to suggest that chemicals
like toxaphene, PCB, and DDT can be transported via the air for thousands of
miles (Bidleman and Olney 1974, Seba and Prospero 1971, Risebrough et al.
1968). These data are strengthened by results from radioactive fallout
studies and by the recent concern over acid rain and its atmospheric link to
industrial centers.
There are literature data to support the contention that high amounts of
toxaphene can be lost due to volatility (Nash et al. 1977, Sieber et al.
-------
1979). Long-range transport of toxaphene by way of the air is well documented
both by direct observation of toxaphene in air at remote sites (Bidleman and
Olney 1975, Bidleman et al . 1981) and by inference from finding the material
in organisms at sites remote from the use of this material: Great Lakes fish
(Rappe et al . 1979); Antarctic cod from the South Pacific and an Arctic char
from the Tyroleon Alps (Zell and Ballschmiter 1980); and an Arctic char from a
lake in southern Sweden (Jansson et al . 1979).
Use of air mass trajectory analyses coupled with pollutant measurements
is a relatively new procedure for confirming sources of these materials. Pack
et al . (1977) used a trajectory model to map fluorocarbon transport in Europe.
Others have performed long range mapping to follow ozone transport along the
eastern coastline (Wolff et al. 1977). Trajectory modeling was used by Rice
and Olney (1978) to confirm that toxaphene transport to Bermuda could be tied
to the southeastern region of the U.S. This same type of trajectory modeling
was utilized in this study in order to determine if a transport link to the
cotton-growing regions (South Central to Southeast U.S.) could be established
~
which coincided with high episodes of toxaphene occurring over Lake Michigan.
r 1
J-*' t
Measurements of toxaphene in air, rain and surface water of the Great
Lakes have been reported (Rice and Evans, in press). Therefore, the potential
for this route appears possible. The purpose of this study was to examine
atmospheric transport of toxaphene to the Great Lakes from the high-use areas
in the southern United States. Establishing a reliable analytical method
became one of the most critical requirements to achieving this goal.
-------
METHODS
AIR COLLECTION
Toxaphene in the air was collected by passing air through an air scrub-
bing system composed of a glass fiber filter [only with the Hi-Vol® (General
Metals Works, Inc., Village of Cleves, Ohio) collectors], and a series of soft
polyurethane foam plugs. Two types of collectors were used in this study.
A Hi-Vol® collector was used for sampling air for toxaphene where concentra-
tions were anticipated to be low to moderate, e.g. at the Lake Michigan sampl-
ing locations on Beaver Island and at Bridgman, Michigan, and in St. Louis,
Missouri. A low volume air sampling device was used in Greenville,
Mississippi, where levels were anticipated to be high. The high volume
sampler was a standard Hi-Vol® high volume sampling system modified by adding
a stainless steel extension tube (23 cm long and 9 cm dia.) behind the filter
holder (10 x 8") such that two to three 9-cm diameter x 6.5 cm long
polyurethane foam plugs could be installed. The air first passes through a
Gelman type "A" glass fiber filter rated at an exclusion size for aerosols in
air of 1 m. The bulk of the toxaphene is expected to be collected as a gas
on the first foam plug. The second plug in the series was placed as a backup
to check for possible breakthrough of toxaphene and to also preclude back flow
of contamination from the pumping apparatus. The pump was a General Metal
Works model GMWL-2000 blower.
The entire sample holder was carefully cleaned prior to use by scrubbing
with soap and water, and solvent rinsing with pesticide grade acetone and
methylene chloride. Between each sample collection the collection assembly
was additionally rinsed with pesticide grade solvents. The filter and plugs
were carefully cleaned prior to use to free them of possible contaminants.
-------
The procedure for cleaning filters involved 450°C ignition in a muffle furnace
for 4 hours and storage in individual aluminum foil containers. The plugs
were carefully cleaned prior to use according to the procedures of Billings
and Bidleman (1980), and periodically checked to guarantee their cleanliness
prior to use.
Polyurethane foam plugs were also used for low volume collection.
These plugs measured 7 cm long x 3.5 cm diameter and were installed in glass
holders (10.5 cm long x 3.5 cm diameter). No prefliters were used with these
collectors and two plugs were placed in series. These plugs were also
precleaned according to Billings and Bidleman (1980).
The flow for the Hi-Vol® collection was maintained at 0.55 to 1.1 m3/min
and the volumes for collection ranged from 1,161 to 4,100 m3 (Table 1).
The flow for the low-volume collection ranged from 7 to 10 L/min for the short
day collections and 2 to 5 L/min for the monthly average collections. A Cast
pump was used to provide the vacuum for the low-volume collector and a Gilmont
flow meter (size #3, Cat. #F1300) was used to monitor the flow by taking
readings at 1 to 2-day intervals throughout the collection. A Sprague/Textron
Gas Meter was used to measure the monthly air volumes collected. These units
were calibrated at the University of Michigan Air Resources Laboratory prior
to employment. To monitor the airflow of the Hi-Vol® collections, a
Marshalltown type flow gauge was used at Bridgman and Beaver Island. Use of
these gauges involved observation of the gauge reading 2 to 3 times during the
3-day sampling period and deriving an average reading for the total period of
collection. With the St. Louis collection, a continuous disc chart recorder
monitored the gas flow. All of the General Metal Works pumps and gauges for
-------
r— 1
•*» CVt
Z 0.
itf >-H
O to
r-l tO
•C ^™^
u to
I-H tO
2
»
Z -
< H
2 J
Q rH
H-4 ^
(X Z
CQ Ed
2 CS
o a
K
ELI Q
z
to <
fc*1
kj **"
£X rH
2 K
< D
to O
to
cs to
< 2
z to
f^ H~4
EC o
z a
Ed
I •
n tn
LI* CH
< to
X
0 "
E- Z
Cu O
O. .
*~'
EH U
Z rH
D 2
i .
< Q
Z
• • ^J
rH tO
00 '"'
j a
CQ CO
<; ^»
EH <
CQ
Z 2
Ed \
rc o
a*— ?•
SH
x •
o u
EH Z
0
C_)
O
Z
5
U Q to
E- to in
< w o
2 D cu
s; «= =*
< ~— '
u
a
ClJ
K 2 '—
i— i 3ro
< iJ 2
O ^
CL)
y
rH
1 Eu
CO O
£H
<
P
pQ
2
rH
1 Z
DO O
EH
P
Z
K! O
lJ I-H
04 EH
2 <
V) O
J
CMTj-rMcr>r--covr;cr(rHr-'«or^coo
COOO'«DrHVO-HOOl»DT}'rHOin
rH
* * « * * *
^J'O>l^'^1^l^lO>CTl^'rHrHrHCMVD
cocorocococococoro^}'^)<«3'cocM
\, \> ^x ^^-^ v\ ""v. ^^ "\ "\ ~\ '\ >x ^^ "\
CTirHv£vOrOCMr--r»-CO^ \^Q ,_| ps» ^i rA f^ v^ f>^ ^J* Cn Cn
inino^irHCMr--^|vorHcnvocM^«
Tj'roinrHCMr^OrHcricrvrHcor-'in
co ro co co co vo ^ co co CM ^* ^ ^ ro
ininooooin«!foo>inrov£iin
OrHinCMrOCOCOrHrHOr> i r~!
•
1— 1
2
*
•
to
rH
K
>
u
CQ
r^ ^o
«s- in
* *
o o
V£> V£
^x^v^
en r-
I—I rH
o r-
o r^
rH in
«3< CO
o o
CO CO
in in
_j -_j
^^ ^^
i i
CM in
0 0
\,^ —
CO CO
Of-}
^_J
o o
ro co
in in
_j _^ j
^^ r^
i i
O CM
ro o
^X^v
r^ co
O(~*
\~~>
•
o
2
*
to
I-H
3
O
u
EH*
-------
13
0)
3
C
•H
•P
C
O
o
B
rH
w
EH
Z Z
a \
S O
a. z
x ^
o u
EH Z
O
z
U Q W
EH Ed 1/3
< in O
2 D Q-
:£*•==*«=
a.
05 S —
< J 2
0 —
DP
2
HH
E-, [u
1 tu
C£l O
H
Q
i— i
I Z
63 O
E*
Q
H O
ij H^
a E-
2 <
< U
in O
a
in t-H rH
* # * * *
rHOrOrHr-li— lOi— lOrHi— IOO
ooooooooomooo
i—li—Ir-lrHi-trHrHi-ltHrHi—liHfM
I I I I I I I I I I I I I
COr-HCOrH.rOVOr-I^J'r-COCOrHfN
OrHfMPOOOOOOOOrHrH
OOCOCDCDON^OOOOrHi-Hr-l
OOOOOOiHrHiHrHi—IrHi—I
OOOOOOOOOOOOO
ro ro c^ C3 ^5 c** ro ro ro co ro ro ro
i i i i i i i i i i i i i
oocNtspoorgoooooi—i
COCOCOOOOOOO^OOOrHiHr-H
OOOOOOOr-lrtrHtHi-t^H
(M
0)
3
C
o
u
cocococooocooococoooooo
«
*
*
*
oooooomooooooo
oomoooorHooooooo
I I I I I I I I I I I I I I
fMincoiHCDi-icovoas^j'r^ocoin
OOOrHrslrOCNOCNOOrHrHO
CDCOOOCOCOCOCOCTiCTiOOOOr—I
OOOOOOOOOi—It-lrHi—(r-»
oooooooooooooo
ro PO ro O ro ro ro co O O O O O O
i—tr-ti—liHOi-Hi-HOCMCNCMtNt—Ir-i
I I I I I I I I I I I I I I
POOOOfMtVinOOCNOOrHO
OOOOOOOOOOi-Hr-Hi-H^
cn
i—i
2
Z
u
Cd
a
a
-------
t
r.
c
"^
c
0
rj
,_
CL
pc
EH
[£»•>
Z Z
X rjj
f)j 2*
< •—»
O 0
EH Z
0
CJ
o
X *^
CJ Q W5
E" Ed CO
< V) O
2 D CU
•j£ =»fe =*C
T}<
rH rH rH (N
*
*
0000
CO •* •* ^3
^J" CM CM rH
VO CO CO O
CM
O O O O
O O O O
0 O 0 0
rH rH rH rH
1 1 1 1
rH rH rH l— 1
rH rH rH rH
rH rH rH rH
O O O O
o o o o
0 O 0 O
rH rH rH iH
1 1 1 1
ID rH *J< CM
O i— 1 rH CM
^X^^N^^N^ V^
rH rH iH O
rH rH rH rH
0)
in
•i-<
*
u
0)
j=
4J
O
...
in
0)
rH
a
E
CO
in
0)
in
0)
jj
l_i
O
^^ J
Ed
D
a
i
"a
a
in
co
^
rrj
CO
•0
C
CO
4J •
in 73
a>
d) in
E 3
jj in
CO
c »
o
4-> C
C 0)
0) rH
4-> (0
0) J=
l-l JJ
CO
V 4=
> a
•rH CO
JJ C
CO O
rH U
ft) O
U rH
.c
0) U
.c o
EH J->
* u
o
•
0)
0)
^^
a
E
(0
in
c
0
•I-l
CO
i-l
D
'O
en
c
o
J
*
*
-------
the Hi-Vol® samples also were calibrated on the University of Michigan Air
Resources Laboratory gas meter prior to use in the field.
The air collectors were operated at locations which were selected to
describe a possible transport pathway for toxaphene from Greenville, Missis-
sippi, a region known for high use of toxaphene in the past, to Lake Michigan.
Three locations, Greenville, Mississippi, St. Louis, Missouri, and Bridgman,
Michigan (Figure 1) were sampled simultaneously during four 2-week time
intervals through the summer and fall of 1981. Samples were also collected
from Beaver Island, in northeastern Lake Michigan, during the second and third
collection period. Rain was collected from Beaver Island during the last
sampling interval and also in April before air collection began in the spring.
Several attempts were made to collect rain at Bridgman. However, the sample
collectors were vandalized in each instance and no rain samples were obtained.
The rain collectors were simple bucket collectors (total fallout collectors)
which were outfitted with screen wire rings on the top edges to discourage
birds from landing on them. The buckets were carefully washed with soap and
water and rinsed with pesticide grade methylene chloride just prior to each
use. For storage, the contents of the buckets were poured into pre-cleaned
brown glass 1-gal bottles. The interior of the buckets were rinsed with
methylene chloride and these rinses were added to the sample bottles.
Additional methylene chloride (total 200 mL/bottle) was added for
stabilization in storage.
Periodic blanks were taken at each of the sampling locations. These in-
volved all the steps of the setup for collection except for turning on the
pumps. These samples were then broken down and returned to the laboratory for
extraction and analysis as if they were actual samples.
10
-------
00
LO
0)
e
f
4_)
B
•H
01
B
J3
rt
o
M
O
•8
oo
CU
(-1
CO
M-l
o
M-l
ai
ca
c
o
O
a
bO
•H
LO
11
-------
EXTRACTION
To extract the toxaphene residues from the filters and foam plugs,
each was cycled with pesticide grade petroleum ether for 12 hr in a Soxhlet
assembly. Following petroleum ether extraction, extracts were concentrated to
approximately 5 to 10 mL and stored until analysis. The rain samples were
extracted by mixing the sample with methylene chloride (1:3 v:v solvent to
water) on a ball-mill roller for 3 hrs. Much thought went into how to work up
the samples in a way that would maximize the identification of the toxaphene
expected in the sample. Our major interests were to study any weathering
which might be apparent and to avoid lengthy handling procedures which might
reduce our recoveries. Nitration as a cleanup (Klein and Link 1970) was
considered but discarded because some chemical alteration by the procedure had
been reported (Ribick et al. 1982). Silicic acid (Bidleman et al. 1978) was
repeatedly tested, however reproducibility was variable and we felt the
results couldn't be trusted. Some of these tests are reported in Appendix A.
We finally settled on a simple treatment of the extracts with sulfuric acid,
followed by direct injection and analysis by electron capture gas chroma-
tography.
CHROMATOGRAPHIC ANALYSIS
Analyses of the extracts involved concentration of each sample to 1 to
2 mL, acid treatment with an equal volume of concentrated sulfuric acid,
transfer of the organic phase, and then an initial prescreening of each sample
by packed column electron capture gas chromatography. Based on these initial
prescreenings, the volume was adjusted for final qualitative and quantitative
analyses by the capillary GC. For most of these final capillary injections,
12
-------
a Varian 8000 auto-sampler was used. The injection volume was 4 yL and the
solvent was petroleum ether or hexane. The injection method was splitless
with a back flush delay of 0.75 minutes. A toxaphene standard was usually
injected for each of four samples injected. The operating conditions for the
capillary instrument were as follows: the column was a (0.20 mm I.D. by 50 m
long) Hewlett Packard fused silica capillary column coated with SE-54, the
carrier gas was hydrogen flowing at a linear velocity of 35 cm/min, and
nitrogen was used as the makeup gas flowing at 30 mL/min. The temperature
program was 100° to 240°C at the rate of l°C/min, and the other heated zones
were 220°C injector and 320°C detector. Data handling for the capillary
instrument was carried out using a Perkin-Elmer Sigma 10 data system with a
basic programming upgrade. To ensure proper data slicing, i.e., discrete area
slicing in small enough intervals to collect the capillary output, the Perkin
Elmer standard slicing option was overridden by forcing the instrument to
operate at its maximum slicing time of 0.13 sec/scan at 10 minute intervals.
This was determined by trial and error to give the best data treatment to the
entire spectrum of each 140-minute sample run. The method for processing the
data was set to quantitate according to the external standard method. The
retention times were calculated as relative retention times (RRT) to
octachloronaphthalene which was used as the reference peak. DDE was used as a
reference peak for some of the samples (refer to Table 1). These standards
were added to each sample just prior to GC analyses. Relative retention time
of the reference peak was set to 10 so as to get an additional decimal place
listing from the relative retention time output of the Perkin Elmer data
system (i.e., the Sigma 10 only prints three figures after the decimal), and
achieve greater precision for peak matching.
13
-------
To select a standard which most consistently matched the peaks that were
being measured in the Greenville air samples, we examined three different
toxaphene formulations used in the Greenville area (Hercules BFC 90-100,
Central American 90-100, and Drexel 616 Form Chem). These were compared to
our EPA Research Triangle Park standard (Lot #B610). From this study we de-
cided that the EPA standard gave the best matches (Appendix B).
In order to establish an appropriate representation of toxaphene to be
used for comparison with samples, standards were screened and peaks were
chosen based on the following criteria. Only peaks that were well resolved
and could be reproducibly identified between standard runs were considered.
In order to achieve reproducibility in peak identification, i.e., RRT match-
ing, while minimizing the inclusion of close eluting peaks within the peak
window, an RRT tolerance of +_ 0.008 was chosen. Furthermore, it was
considered important that an even distribution of peaks over the total elution
time for toxaphene be used. Approximately 90 peaks were selected based on the
above criteria. The selected peaks were then screened against expected
interferences. For this study Aroclors 1242 and 1254 were especially
important in the Lake Michigan and St. Louis samples, and the pesticides DDT,
DDE, and chlordane were important interferences in Greenville. Those peaks
which were matched with the various interfering components were removed from
the peak table. During the course of our analyses, the number of peaks which
met all screening criteria for standards ranged from 26 to 60 (c.f. Table 1).
These peaks were then used to represent toxaphene for comparison with samples.
Toxaphene identification in the samples was accomplished by screening
sample peaks for matches with the standard peak tables described above.
The criteria for peak matching in the samples was identical to those used for
14
-------
standard comparison. However, those peaks that were matched in the samples
were further screened for possible errors in baseline treatment, and unknown
interferences. Peaks occurring in the sample that were disproportionately
high in area (relative to other matched peaks) were considered interfered with
and were excluded from quantitation. This latter screening presupposes that
the ratio of toxaphene peaks in the sample is similar to the standard.
Lacking proof of this, we selectively eliminated peaks only in the most
extreme case, e.g., when the (interfered) peaks accounted for greater than
approximately 40 percent of the sum of the area for all of the matched peaks.
Quantitation of toxaphene in the samples was based on the ratio of the
sum of the areas for peaks matched (and screened) in the sample to the sum of
the areas of the corresponding peaks in the standard. This value was then
multiplied by the standard concentration and sample volume to determine the
total ng of toxaphene. The toxaphene concentration was the quotient of total
nanogram of toxaphene determined to be in the sample divided by the volume of
air (m-*) collected (Table 1). Representative chromatograms of samples and
standards containing toxaphene are shown in Figure 2.
QUALITY ASSURANCE
Blanks were found to contain no recognizable toxaphene patterns.
However, some spurious peaks did match the relative retention times for
toxaphene and upon quantitation averaged 23.1 ng. Therefore, the limit of
detection for the high volume samples was 0.06 ng/m^ and for the low volume
samples it was 0.6 ng/nH. Linearity checks for electron capture response to
toxaphene were performed over the standard ranges of 118 to 444 ppb and,
generally, linearity was good for the total mass of peaks (Figure 3).
15
-------
a
wm
' wdftliMWiT
M!m Ife i! I
Figure 2. Capillary chromatograms of toxaphene-containing air, rain
and fish samples and EPA toxaphene standard. The sculpin sample was
analyzed on another project and is provided for comparison of the
atmospheric data with biological residues, a) Air sample from
Beaver Island (28 August to 31 August 1981) containing 0.054 ng/m^
toxaphene; b) Rain sample from Beaver Island (28 September to
2 October 1981) containing 31.58 ng/L toxaphene;
c) Sculpin composite sample from Lake Michigan (17 June 1982) containing
0.718 ng/gm toxaphene; d) toxaphene standard (4 yL of 118 ng/mL).
G/C conditions -50 m fused silica column coated with SE-54
and temperature-programmed from 100 to 240° C at l°/min.
16
-------
I
(£>
ro
ro
oo
*•
10 ^
o o*
II
>
u
o
o
10
o
o
_Q
Q.
CL
82
ro h-
<
(T
LU
O
Z
O
o
o
o
C\J LL)
Z
LU
X
Q_
<
X
O
H-
O
O
i
O
ro
c\J
00
i
CM
I
CD
o
• -H
0) 4-1
d CB
0) U,
o
o
^> l-l
•H
0) (U
2 43
d u
o
P. CO
03 3
I) CO
S-i VJ
(1)
>
O
HI
CD 01
r-l J2
CU 4J
CN
CM
CN
QO
0)
•I-1 U
O cfl
a)
CO
0)
d
3 0)
W) 4J
SilNH V3dV
17
-------
Consistent linearity was not observed for the individual peaks, however. This
may bias our results as different assortments of selected peaks are used.
A thorough test for this was beyond the scope of this study. However, we
observed the following: In general, the ECD response of smaller, earlier
eluting peaks tended to exhibit greater linearity within the range of
concentrations tested than did larger, later eluting peaks. In addition, the
intercepts of the linearity plots for the smaller peaks tended to be negative
while the intercepts for the larger peaks were generally positive. One
conclusion from this treatment was that using a good spread of peaks over the
range of those possible would cancel out this opposing linearity, therefore
our peak selection procedure should have minimized this problem.
Duplicates were run at Beaver Island and reasonable replicate performance
was observed, see Table 2.
TABLE 2. DUPLICATE RESULTS
Beaver Island
Sample Date
8/28
8/31
9/28
Toxaphene
Concentration
(ng/m3)
0.054 0.023
0.2335 0.2773
0.0237 0.0309
Relative %
Difference of
Duplicates
43
8
13
x = 21
Averaging the results for the three duplicate sets gave a value of 21 for the
average relative percent differences of three duplicates.
A performance test was conducted using standard solutions of toxaphene
and Aroclor 1254 provided by the Grosse lie EPA laboratory in January of 1982.
18
-------
MODELING
The atmospheric deposition of toxaphene into the Great Lakes is
speculated to be the result of transport from regions far removed from the
area. To assess this claim, diagnostic modeling tools have been developed and
applied to the measured values of toxaphene for three of the sampling sites.
The diagnostic techniques used in this study take into account the potential
long-range transport of toxaphene and can include enroute processes of dry and
wet deposition, chemical transformation, and dispersion. The areal
probability of contribution to the toxaphene levels at a receptor is
calculated in two ways. In the first, the areal probability field resulting
from the ensemble of individual trajectories arriving at the receptor during
the hours of sampling is calculated. This is the probability of contribution
due to "natural" phenomena. It would represent the spatial distribution of
contribution if emissions were universally homogeneous. In the second method,
the individual trajectory probabilities are weighted proportionally to the
resulting toxaphene concentrations. If there is systematic transport of the
toxaphene to the receptor from a particular area or areas, the two fields will
be dissimilar. On the other hand, if there is no clear "corridor" associated
with the transport of toxaphene, then there will be little difference between
the weighted and unweighted contribution probability fields.
The assessment of potential long-range transport of atmospheric
contaminants requires an estimation of the trajectory of the air prior to
being sampled. In this study, the model of Heffter (1980) was used to
calculate the upwind trajectories. The model estimates the height of the
transporting layer by scanning temperature data from rawinsonde ascents ,
looking for stable layers 300 m or more above the surface. The bottom of the
19
-------
transport layer is defined as the top of the layer of surface-induced wind
shear. Typical transport depths in the Midwest United States are about
1,600 m during the summer months. Once the trajectory to the receptor has
been estimated, the dispersion of material along the route is estimated based
on the wind shear through the mixed-layer. Estimates of the rate of
dispersion have been reported by Samson (1980), Samson and Moody (1980), and
Draxler and Taylor (1982). In this work it is assumed that the dispersion of
the contribution probability is linearly proportional to time. The area which
could have contributed to the concentration at the receptor is determined
through the integration of normally-distributed, two-dimensional "puffs"
growing upwind of the receptor. Deposition of the contaminant en route to the
receptor will reduce the potential for far upwind sources to contribute to the
sample. The reduction, probably due to dry deposition, has been expressed by
use of a dry deposition velocity, V
-------
to the measurement locations would affect the potential of upwind source
regions to contribute, but little is known about wet deposition of the
toxaphene. Differential distributions of precipitation could have an impact
upon even the relative difference in probability, but quantitative estimates
of precipitation rates for 1981 are not yet available from the National
Climatic Center. While this limits the interpretation of results, the
comparison of weighted versus unweighted contribution fields still provides a
useful first test of the hypothesis that the toxaphene is being transported to
the receptors from source fields in the southern United States.
,* ,
RESULTS AND DISCUSSION
The concentrations of toxaphene measured in the air collected at the four
sampling sites are listed in Table 1. The levels were highest in Greenville,
Mississippi (average of 7.39 ng/rn-^ - monthly collections not included),
followed by St. Louis, Missouri (1.18 ng/rn-^), and lowest in Bridgman (0.35
ng/nH), and Beaver Island, Michigan (0.09 ng/rn^). The listing for peak
matching in Table 1 provides an estimate of the similarity of the material
measured to the standard. These can also be calculated as percentage matches
which allows direct comparisons to be made between locations. The mean
percent matches for each of the sampled regions merging Beaver Island and
Bridgman for the Lake Michigan segment were as follows: 36% for Lake
Michigan, 38% for St. Louis, and 51% for Greenville. One interpretation of
this might be that the toxaphene in Greenville is closer to the source than
the material measured in air at St. Louis or near Lake Michigan. And that
distance from the source causes changes in the composition of the toxaphene
mixture (often expressed as "weathering"). Another interpretation might be
21
-------
that the lower match percents are merely a function of the smaller amount of
material in the samples at the more northern sites. Regression analyses on
the amount of toxaphene measured versus the percentage of peak matched with
each sample subset (Greenville, Bridgman, St. Louis, and Beaver Island),
however, were not found to be significant. Therefore, the degree of
alteration of the toxaphene mixture does, in fact, appear to be related to the
distance from the likeliest source regions which we propose are in the
Southern United States.
Table 1 also lists the total volume of air collected for each sample.
Regression analyses of this parameter versus the concentration measured were
carried out for each sample subset. No significant correlations were found
for any of the high volume collections taken in the northern sampling sites.
However a positive correlation was found for the Greenville site, i.e., the
greater the air volume sampled, the lower the total nanogram amount.
Even when the monthly collections, which tended to underestimate the real
concentrations, were left out of the calculations, the correlations were still
significant. This finding led to further tests to see if breakthrough of the
toxaphene could be observed on the low volume backup plugs. Recall that the
Greenville sampling plugs were smaller than the Hi-Vol® plugs. No toxaphene,
however, was found on the backup plugs, therefore breakthrough did not appear
to be occurring. This was consistent with extensive tests of polyurethane
foam collection efficiency for trapping of toxaphene vapors (Rice et al. 1977,
Billings and Bidleman 1980).
Since breakthrough was not occurring with the low-volume collectors,
there appears to be no obvious explanation for the low values found for the
monthly collections in Greenville. Possibly, with these high volumes (90 to
22
-------
200 m^ of air) and the relatively high levels of toxaphene being collected,
re-release of toxaphene from the collection plugs was occurring. Other
factors which were also different for the monthly low-volume collection were
the lower air flow used, i.e., 2 to 4 L/min vs. 5 to 10 L/min for the short
duration low-vols, and also the length of time the samples were left out, 20
to 30 days versus 2 to 3 days. Any one or all of these factors may have
contributed to the difference in results between the monthly collections and
the shorter interval Greenville collections. Tests for collection efficiency
of toxaphene under low-volume/long duration sampling conditions would be
recommended if further work using this method is planned. Comparison of the
low-volume collector (short duration) with simultaneous high-volume
collections were performed in a previous study (C. P. Rice and C. E. Olney,
unpublished results), and the two systems compared favorably. Therefore,
we do not feel there is any problem in the short duration results.
Filter retention of toxaphene appeared to be less than 5% of the total
measured toxaphene for the Hi-Vol® collections. A qualification is needed
here, however, in that the error range for these measurements were relatively
high. The peak matching percentage for most of the toxaphene identifications
on the filters were 15% or less, i.e. 4 to 5 peaks of the 26 to 40 possible.
Because of the low values for the filters and the high uncertainty connected
with these results, only the first plug analyses were used in reporting the
toxaphene amounts measured.
Figures 4 and 5 present the results listed in Table 1 in graphical form.
It is apparent that for each of the locations an increase occurred in amount
of toxaphene from the first sample period in early August to early September,
whereupon a gradual decline in concentration was observed for the October and
23
-------
20 OH
ro 10 0^
200
n
8/1
GREENVILLE
n
9/1
ST LOUIS
20-1
10-
o
s
I
n
i '•
0/1
BRIDGMAN
1
9/1
10/1
ll/l
Figure 4. Simultaneous determinations of toxaphene in air measured at
four sites lying along a general south-to-north transect from Greenville,
Mississippi, through St. Louis, Missouri, and ending in the Lake Michigan
area (Bridgman and Beaver Island, Michigan). The vertical axes are in
log units and the horizontal axes are calendar dates.
24
-------
2.0-
ro
1.0-
BEAVER ISLAND
O
cn
UJ
o
z
o
o
UJ
z
LU
Q_
O.I -
.01-
j
II
8/1
9/1 10/1
SAMPLING DATE
I I/I
Figure 5. Toxaphene in air measured at Beaver Island, Michigan,
from 28 August to 4 October 1981. Vertical axis is in log units
and the horizontal axis is in calendar dates.
25
-------
November sampling periods. Table 3 presents these findings in average
concentrations for each of these discrete sampling periods.
This trend for toxaphene levels in air observed in our data was similar
to the trend for relative levels of atmospheric/toxaphene observed by Arthur
et al. (1976) in the Mississippi Delta in 1972-1974. In Arthur et al.'s
st'dy, the maximum average toxaphene occurred in August/ September periods:
1,540 ng/m3 - Aug. '72; 269 ng/m3 - Sept. '73; 903 ng/m3 - Aug. '74. These
average concentrations were considerably higher than measured by us. However,
toxaphene usage is reported to be much less now than in the early 1970s (Larry
Lane, Mississippi State University, Agricultural Experiment Station, Personal
Comm.).
TABLE 3. TREND IN TOXAPHENE CONCENTRATION AT SAMPLING SITES
FOR THE SUMMER AND FALL OF 1981
Average toxaphene concentration by location (ng/m3)
Sampling
Interval
August
September
October
November
Beaver Island ,
Michigan
0.15
0.03
Bridgman,
Michigan
0.26
0.62
0.30
0.23
St. Louis,
Missouri
1.31
2.16
0.69
0.56
Greenville ,
Mississippi
6.61
11.40
6.04
2.64
In looking at Figure 3, one might expect some similarities in concentration
patterns between the three locations, especially if an atmospheric link is
predicted. For example, the first interval bears a close resemblance in general
pattern between St. Louis and Greenville. There appears to be a tendency to
observe possible shifts by one sampling period of a high level of toxaphene
observed at one location to an appearance of this material as a high level in a
26
-------
more northerly location a few days later (notice such a connection for St. Louis
to Bridgman for the first sampling period, Figure 4). Keep in mind that the
scale for the Bridgman figure is a factor of ten less than the two other sites
depicted.
One rain sample was collected from Beaver Island, Michigan, while air
measurements were being made from 28 September to 2 October 1981. The sample
contained 31.6 ng/L of toxaphene with 14 peaks matched out of 37 possible in the
peak table. Another rain sample collected on 1 April 1981, also from Beaver
Island, was found to contain 70.2 ng/L with 5 out of 37 peaks matched.
MODELING OF ATMOSPHERIC TRANSPORT
One goal of this study was to use these data as a basis for proposing
some boundary limits for toxaphene transport and deposition to Lake Michigan.
We chose an atmospheric transport model to assist in these estimates.
The "natural potential" for contribution to toxaphene concentrations at
the Bridgman, Michigan, site is shown in Figure 6. This figure shows the
probability of contribution in units of 10~° km~^. It is the area which would
contribute to the concentration of toxaphene if emissions were homogeneous and
constant over the grid. The plot shows that the highest occurrence of winds
during the sampling periods was from the southwest of the receptor. Figure 7
shows the result of weighting each trajectory by the concentration measured at
the time of arrival. It is assumed that each trajectory arriving through the
sample period contributed equally to the sample loading of toxaphene. The
results indicate that the basic structure of the contribution field is not
substantially varied from the spatial pattern found for the unweighted case in
Figure 6. The similarity of the two fields can be discerned only by
27
-------
CO
4J
•H
§
CO ^v
13 CO
O bo
•H d
M -H
O
bO -H
C VJ
•H 4J
1-8
H g
co o
o
cu
J= M-l
•"-> O
bO M
c cu
3 CU
T3 o
C cu
O 4J
•^ ee
S
o
CJ
•S
CO
tJ PQ
P<
T3 t-l
CU
4JCN
f e
s
CU
t-i
bO
•H
28
-------
o
CO
§
C
•H
O
•H
i-i
CD
O.
bO
8
CO
0)
.e
bo 2
5
•H
O ^i
o *•».
oo
O O
5*""1
•s
fl
o
(-1
p<
JS
bt>
•H
r-
01
no
29
-------
calculating the difference between the weighted and unweighted fields as shown
in Figure 8. This plot suggests that there is a slight spatial bias (or
corridor) for toxaphene, with higher concentrations associated with winds from
the south to southeast, but the magnitude of the difference in the weighted
and unweighted fields is small, being an order of magnitude lower than the
magnitude of the probability of contribution field (Fig. 6).
Figures 9 and 10 show the unweighted and weighted fields, respectively,
for the receptor at St. Louis, Missouri, based on the 15 samples collected
there in 1981. These two fields again appear to be relatively similar, but
the difference of the weighted and unweighted fields, plotted in Figure 11,
has a markedly higher amplitude than was apparent at the Bridgman, Michigan,
site. Here the analyses suggest a more well-defined corridor of higher
concentrations from the south, supportive of the hypothesis that the toxaphene
was being transported from regions of application in the southern United
States.
The unweighted probability field for Greenville, Mississippi, is plotted
in Figure 12. The distribution of wind flow to the site on the days of
sampling is more isotropic here than at either Bridgman or St. Louis, as
indicated by the relative symmetry of the probability field about the
receptor. The weighted probability field, shown in Figure 13, does not appear
to be dissimilar to the unweighted field, but its difference from Figure 12,
plotted as Figure 14, indicates that a relatively large bias is exhibited by
the data. The corridor for higher concentrations of toxaphene at the
Greenville site is associated with winds originating along the Gulf Coast.
Winds from the north are systematically associated with lower than average
concentrations.
30
-------
vO
w
O)
t-i
3
bO
C
01
OJ
fi
0)
O
C
0)
M
01
14-1
U-l
oo
bO
•H
LO
31
-------
u-i
o
3
P
0)
a
is
•S I
I?
»S
e
o
g
o
•8
o
I'o
cr\
0)
i-i
3
32
-------
CO
4-1
•ft
§
CO
T3
O
•H
S-l
01
O.
bO
C
CX
e
cd
CO
bO S
C
•H •
SJ Cfl
3 T4
^J 3
O
O B
o ^
UJCX5
O I
O
•3
&
O
u
ex
T3
0)
4-J
bC
01
!-i
3
33
-------
to
bC
•H
-------
4-1
o
03
4J
c
•H
CO
tJ
O
be
C
O
T-!
4-1
3
o
c
c
CJ
1+-I
o
lJ
OJ
4J
flj
o
0)
4-1
«
a
a
to
IB
.a
cfl
2
T3
O
3?
o
W)
35
-------
CO
4J
•H
§
0)
•o
o
•H
M
8,
c
•H
rH
cx
HI
x:
1
c
O,
'oo
O
1-1
0)
4J
£1
bO
•H
SO
LO
36
-------
to
OJ
e
0)
-------
The analysis of transport corridors for Beaver Island, Michigan samples
was not conducted because only four samples were available for that site.
It has been assumed that each trajectory during the sample period was
equally responsible for the observed concentration. However, without better
temporal resolution in the sampling, it is difficult to assess the uncertainty
inherent in this assumption. For comparison, the transport of atmospheric
trace elements is highly episodic. The dry deposition of trace elements,
assuming the process is proportional to the ambient concentration, is highest
during specific episodes of roughly 3 or 4 days duration (cf. Husain and
Samson 1980). The significance of such episodic contribution could be lost in
this analysis if the episode were split by two or more long-period samples.
The diagnostic methodology developed for this study is designed to be employed
with a larger data base. The sample size available in this study limits the
use of ensemble analyses.
Nonetheless, the results of this diagnostic analysis of the measured
/&
* . •* toxaphene concentrations shows a consistent pattern. At all measurment
i> •
\-y.t locations the toxaphene corridor was associated with southerly winds. The
preferred corridors of transport of higher concentrations increased from
northern to southern measurement sites, presumably in response to the larger
range of concentrations in the south. This analysis is not capable of proving
which source region(s) contributed to the observed concentrations, in part
because so little is known about the magnitude of the source strengths.
However, the methodology developed for this analysis is capable of identifying
the probable corridors of transport which deliver the material to the
receptor. The method is well suited for analysis of data which, because of
low concentrations in the atmosphere, must be collected over long time periods
38
-------
(greater than a day). With a sufficient number of samples, intra-period
fluctuations in contribution to the sample will be removed and meaningful
transport characteristics will be discerned.
ESTIMATES OF ATMOSPHERIC FLUX OF TOXAPHENE TO LAKE MICHIGAN
For this estimate the concentration over Lake Michigan was used as the
immediate source for input. Our data provide a good estimate of this amount
for August through November of 1981. Using the formulations proposed by
Eisenreich et al. (1981) for estimating atmospheric flux of trace organics to
the Great Lakes, total atmospheric loading to Lake Michigan for 1981 was
estimated. For wet and dry flux, the general equations presented by
Eisenreich et al. (1981) were:
Fw = Cr/s • J • SA (1)
Fd = cv/P * Vd ' SA (2)
Fw and F^ are wet and dry deposition, respectively; Cr/s = concentration in
rain and snow; Cv/p = concentration in vapor and particulate; J = annual
precipitation; V
-------
1/2 of this amount. Therefore, as a whole lake estimate, these two
concentrations were averaged, i.e.,
[0.35 ng/m3 + 0.35/2 ng/m3] = 0.26 ng/m3.
2
No direct observations were available for estimating a winter concentration
(December to May) and no observation of input as snow has ever been reported.
Therefore, to approximate what the air levels might have been, comparisons of
summer to winter levels of toxaphene determined for air in South Carolina
(Harder et al. 1980), and Bermuda (C. P. Rice, unpublished data) were
available. These data indicated that approximately 2 to 4-fold decreases in
atmospheric levels in winter could occur. Because of wind transport into Lake
Michigan during winter months, was expected to be dominated by northern cold
fronts, the lower estimate (1/4) was selected. Therefore, the toxaphene
concentration estimated for the 6 winter months (December to May 1980-81)
preceding the data presented here was 0.26 ng/m3/4 or 0.065 ng/m3. An average
yearly concentration for Lake Michigan air, northern and southern plus winter
and summer, was calculated as
0.26 + 0.065
= 0.16 ng/m3
2
This air concentration was next used to derive an expected yearly rain
concentration. For this, a washout ratio was needed. The washout ratio from
our one rain sample came out to 1,355
40
-------
ng toxaphene/kg rain
WR . (3)
ng toxaphene/kg air
31 ng/kg rain (28 Sept-1 Oct)
WR =
0.027 ng/(m3 =1.18 kg) Air (28 Sept-1 Oct)
WR =1,355
This greatly exceeded Bidleman's (Bidleman and Christensen 1979) highest
observed value for washout ratios calculated from rainfall input of toxaphene in
South Carolina (i.e., <16 to 861), and therefore appeared to be too high.
As a compromise we chose a range of 500 to 1,000 for the likely range of washout
ratio to be used in our calculations. Carrying out the calculation, rain
concentrations were estimated to range from 68 to 136 ng/kg, e.g.,
cone rain
500 to 1,000 =
0.16 ng/(m3 = 1.18 kg)
= 68 to 136 ng toxaphene/kg rain
Using these values in Eisenreich's formulation (Equation 1), the fol-
lowing was calculated:
Flux wet = Cr/s • J • SA (68 ng/L)
= (103 kg/m3) (68 x 10"12 kg tox/kg rain) (0.74 m/yr) (5.9 x 1010 m2)
= 2,969 kg tox/yr at a rain concentration of 68 ng toxaphene/kg rain
= 5,938 kg tox/yr at a rain concentration of 136 ng toxaphene/kg rain
41
-------
Dry Flux Calculation
For this calculation, the deposition velocity of <0.12 - 0.24 cm/sec was
taken from literature values presented by Bidleman and Christensen (1979) and
Harder et al. (1982). The calculation for a 0.12 cm/sec was as follows:
Fd = Cv/p • Vd - SA
Fd = (602 • 24 • 365 sec/yr) (0.16 x 10~12 kg/m3)
(0.12 x 10~2 m/sec) (5.9 x 1010m2)
= 357 kg toxaphene/yr @ Vd = 0.12 cm/sec
= 715 kg toxaphene/yr @ Vd = 0.24 cm/sec
Comparing the relative input from the two processes, wet deposition
appears to predominate, i.e., 3,000 to 6,000 kg/yr wet, versus 350 to 720
kg/yr for dry deposition.
A total flux would be 3,360 to 6,720 kg/yr for toxaphene from all
atmospheric sources. These are similar to the estimated total input for PCB
of 4,700 kg/yr for Lake Michigan (Murphy et al. 1982). However, Murphy's
estimates indicated the largest input for PCBs are dry deposition processes,
while our results indicate that rainfall is the predominant vector for
toxaphene input. The greater likelihood of rainfall deposition dominating for
toxaphene over other organochlorines (PCB, DDT, and chlordane) was also
observed by Harder et al. 1982. It should finally be pointed out that the
above calculations are based on a very limited data set, especially for the
rainfall and dryfall estimates. Thus, the flux values which are derived must
be considered speculative at this time. However, recent analytical data on
residue levels in biota of the lakes suggest that the current levels of
toxaphene are indeed similar to those of the PCBs (Rice and Evans, in press).
42
-------
REFERENCES
Arthur, R.D., J.D. Cain, and B.F. Barrentine. 1976. Atmospheric levels of
pesticides in the Mississippi Delta. Bull. Environ. Contain. Toxicol.
15: 129-134.
Bidleman, T.F., and E.J. Christensen. 1979. Atmospheric removal processes
for high molecular weight organochlorine compounds. J. Geophys. Res. 84;
7857-7862.
Bidleman, T.F., and C.E. Olney. 1974. Chlorinated hydrocarbons in the
Sargasso Sea atmosphere and surface water. Science 183: 516-518.
Bidleman, T.F., and C.E. Olney. 1975. Long range transport of toxaphene
insecticide in the western North Atlantic atmosphere. Nature 257: 475.
Bidleman, T.F., J.R. Matthews, C.E. Olney, and C.P. Rice. 1978. Separation
of polychlorinated biphenyls, chlordane, and p,p'-DDT from toxaphene by
silicic acid column chromatography. J. Assoc. Offic. Anal. Chem. 61:
820-828.
Bidleman, T.F., E.J. Christensen, W.N. Billings, and R. Leonard. 1981.
Atmospheric transport of organochlorines in the North Atlantic.
J. Marine Res. 39: 443-464.
Billings, W.N., and T.F. Bidleman. 1980. Field comparison of polyurethane
foam and Tenax-GC resin for high volume air sampling of chlorinated
hydrocarbons. Env. Sci. Technol. 14: 679-683.
DeVault, D., R.J. Bowden, J.C. Clark, and J. Weishaar. 1982. Results of
contaminant analysis of fall run coho salmon, 1980. Presented at 25th
Conference on Great Lakes Research, International Association for Great
Lakes Research, Sault Ste. Marie, Ontario.
43
-------
Draxler, R.R., and A.D. Taylor. 1982. Horizontal dispersion parameters for
long-range transport models. J. Appl. Meteor. 21: 367-372.
Eisenreich, S.J., B.B. Looney, and J.D. Thornton. 1981. Airborne contami-
nants in the Great Lakes ecosystem. Environ. Sci. Technol. 15: 30-38.
Harder, H.W., E.J. Christensen, J.R. Matthews, and T.F. Bidleman. 1980.
Rainfall input of toxaphene to a South Carolina estuary. Estuaries 3:
142-147.
Heffter, J.L. 1980. Air Resources Laboratories Atmospheric Transport and
Dispersion Model (ARL-ATAD). NOAA Tech. Memo., ERL ARL-81. 17 pp.
Holmstead, R.L., S. Khalifa, and J.E. Casida. 1974. Toxaphene composition
analyzed by combined gas chromatography-chemical ionization mass spec-
trometry. J. Agric. Food Chem. 22(6): 939-944.
Husain, L., and P.J. Samson. 1980. Long-range transport of trace elements.
J. Geophys. Res. 84: 1237-1240.
Jansson, B., R. Vaz, G. Blomkist, S. Jensen, and M. Olsson. 1979. Chlor-
inated terpenes and chlordane components found in fish, guillemot and
seal from Swedish waters. Chemosphere 4: 181-190.
Klein, A. K., and J. D. Link. 1970. Elimination of interferences in the
determination of toxaphene residues. J. Assoc. Official Anal. Chemists 53:
524-529.
Korte, F. , I. Scheonert, and H. Parlar. 1979. Toxaphene (camphechlor) ,
a special report. Internat. Union of Pure and Appl. Chem. 51: 1583-1601.
Murphy, T.J., G. Paolucci, A.W. Schinsky, M.L. Combs, and J.C. Pokojowczyk.
1982. Inputs of PCBs from the atmosphere to Lakes Huron and Michigan.
Report of USEPA Project R-805325. Duluth Environmental Research
Laboratory.
44
-------
Nash, R., M. Beall, Jr., and W. Harris. 1977. Toxaphene and 1,1,1 tri-
chloro-2,2-bis (p-chlorophenyl) ethane (DDT) losses from cotton in an
agroecosystem chamber. J. Agric. & Food Chem. 25(2): 336-341.
Pack, D.H., J.E. Lovelock, G. Cotton, and C. Curthoys. 1977. Halocarbon
behavior from a long time series. Atmospheric Environment 11: 329-344.
Rappe, C., D.L. Stalling, M. Ribick, and G. Dubay. 1979. Identification of
Chlorinated "Toxaphene Like" Compounds in Baltic Seal Fat and Lake
Michigan Fish Extracts by CI-GC/MS., Paper 102, Pesticide Section 177th
Nat. Meeting of the Amer. Chem. Soc., Honolulu, Hawaii.
Ribick, M.A., G.R. Dubay, J.D. Petty, D.L. Stalling, and C.J. Schmitt. 1982.
Toxaphene residues in fish: Identification, quantification, and con-
firmation at part per billion levels. Environ. Sci. Technol. 16(6): 310-
318.
Rice, C.P., and M.S. Evans, (in press). Toxaphene in the Great Lakes.
_In_ Toxic Contaminants in the Great Lakes, ed. J. Nriagu and M. Simmons.
New York: John Wiley & Sons.
Rice, C.P., and C.E. Olney. 1978. Air mass transport of toxaphene.
Abstract 176th American Chemical Society National Meeting,
Environmental Chemistry Division. Miami, Florida.
Rice, C.P., C.E. Olney, and T.F. Bidleman. 1977. Use of polyurethane foam
to collect trace amounts of chlorinated hydrocarbons and other organics
from air. World Meteorological Organization Special Environmental Report
#10, WMO-No. 460.
Risebrough, R., R.J. Huggett, J.J. Griffin, and E.D. Goldbert. 1968.
Pesticides: transatlantic movements in the Northeast trades.
Science 159: 1233-1235.
45
-------
Samson, P.J. 1980. Trajectory analysis of summertime sulfate concentrations
in the Northeastern United States. J. Appl. Meteor. 19: 1382-1394.
Samson, P.J., and J.L. Moody. 1981. Trajectories as two-dimensional proba-
bility fields. Proceedings llth NATO/CCMS Int. Tech. Meeting on Air
Poll. Modeling. Amsterdam, in Air Pollution Modeling and its Application
1^. C.D. Wispelaere, Ed. Plenum Press. New York. 43-54.
Schmitt, C.J., J.L. Ludke, and D.F. Walsh. 1981. Organochlorine residues
in fish: National Pesticide Monitoring Program. Pestic. Monit. J.
14: 136-206.
Seba, D.B., and J.M. Prospero. 1971. Pesticides in the lower atmosphere of
the northern equatorial Atlantic Ocean. Atmos. Environ. 5: 1043-1050.
Seiber, J.N., S.C. Madden, M.M. McChesney, and W.L. Wintertin. 1979.
Toxaphene dissipation from treated cotton field environments : component
residue behavior on leaves and in air, soil and sediments determined by
capillary GC. J. Agric. & Food Chem. 27(2): 284-291.
Slinn, W.G.N. 1982. Estimates for the long-range transport of air pollu-
tion. Water, Air, and Soil Pollution 18: 45-64.
Wolff, G.T., P.J. Lioy, R.E. Meyers, R.T. Cederwall, G.D. Wight, R.E. Pasceri,
and R.S. Taylor. 1977. Anatomy of two ozone transport episodes in the
Washington, D.C., to Boston, Mass., corridor. Env. Sci. Technol. 11:
506-510.
Zell, M., and K. Ballschmiter. 1980. Baseline studies of the global pollu-
tion. II. Global occurrence of hexachlorobenzene (HCB) and polychloro-
camphenes (toxaphene) (PCC) in biological samples. Fresenius Z. Anal.
Chem. 300: 387-402.
46
-------
APPENDIX A
TESTS FOR SILICIC ACID FRACTIONATION OF TOXAPHENE
In order to reduce the possibility of incorrect identification of gas
chromatographic peaks which could co-elute with toxaphene, methods were
investigated which were designed to selectively remove these interferences.
The most problematic interferences expected were PCBs. Silicic acid
chromatography has been reported to effectively separate toxaphene from PCBs
(Bidleman et al. 1978).
METHODS
The silicic acid method in its complete form as proposed by Bidleman
et al. (1978) was first tested. The water content of the original method was
modified to contain 2.8% water rather than 3.3%. Also, the elution pattern
was changed slightly so that separation of PCBs from DDT and chlordane was
optimized. The elution pattern was as follows:
Solvent
petroleum ether
Vol.
1- 10 mL
10- 40 mL
40-120 mL
20 mL
Fraction
1
2
3
4
Compounds
Expected
HCB
o>10% toxaphene,
PCBs, DDE
^30% toxaphene,
chlordane, DDT
60% toxaphene,
dieldrin, ODD
dichloromethane
To maximize toxaphene recovery for this study, we recombined fractions 3 and
4. Using this procedure recovery of toxaphene from column spikes was poor
(20%, see Table Al). We then further modified the collection sequence by
Al
-------
taking only 2 fractions, i.e. 0 to 40 mL for PCB, HCB, and DDE and then a
20 tnL methylene chloride sweep to secure the bulk of the toxaphene along with
DDT, ODD, dieldrin, and chlordanes (Table Al, samples 2 and 3). With the 2
fraction procedure, recovery was improved but only marginally acceptable, i.e.
76 and 98%.
TABLE Al. SILICIC ACID RECOVERY OF TOXAPHENE
Method Used
4-Fract ion
(F3 + F4 combined)
2-Fraction
(20 mL DLM)
2-Fraction
(20 mL DCM)
# of Peaks
Matched
12/31
29/33
21/33
Total Area Units
Spike Recovered Percentage
Standard Standard Recovered
11.795 2.37 20.1
12.23 9.33 76.3
8.24 8.13 98.0
Based on these results, we decided that the fractionation procedures was
not acceptable because recovery losses were too great. Such losses could not
be risked, especially for the low level samples from Bridgman and Beaver
Island. Furthermore, funds were not available to carry out these labor
intensive procedures on the large number of samples generated during this
study. Qualitatively, however, the method was a valuable confirmation tool.
We applied the two fraction silicic acid method to one of the Bridgraan air
samples. In Figure Al, the unfractionated air sample can be compared to the
first 40 mL eluate of the fractionated sample (the PCB fraction). An Aroclor
A2
-------
A3
-------
1254 standard is also included in this figure as a representative PCBs for
comparison with the fractionated sample. In Figure A2, the second fraction
(toxaphene fraction) is shown along with the unfractionated air sample and a
fractionated toxaphene standard.
It is clear that most of the peaks appearing in the unfractionated sample
were eluted in the first fraction and many of them matched the PCB standard.
This obviously reduced the potential for interference occurring in the second
fraction (i.e., toxaphene fraction). Furthermore, the chromatograms of the
fractionated toxaphene standard and the toxaphene fraction of the air sample
can be compared visually and they do bear some similarities. (it is reassur-
ing to get a visual impression that the matches are real and not just blindly
trusting the data system to do peak matching.) Figure A3 has been included to
indicate the relative distribution of peaks used for peak matching.
The toxaphene concentration calculated for the Bridgman air sample using
unfractionated material, which was the general procedure adopted throughout
this study, was 0.665 ng/ra^. The concentration calculated using the
fractionated sample and fractionated standard was 0.582 ng/m . Therefore, the
two methods showed good agreement. Furthermore, the peak matching for the
unfractionated sample was 14 of 41 possible (34.1%), and for the fractionated
sample it was 12 of 31 (39%). Therefore, removing the interferences did
improve the matching as it should have. Another advantage of the fractionated
method is that since PCBs are removed by the fractionation procedure, the peak
selection criteria used for standards could have been altered so as to include
peaks previously excluded from the peak table due to interferences with PCBs.
This would have further expanded the peak table used for peak matching arid
most likely further improved the basis for toxaphene quantitation.
A4
-------
Ss
. *J o
2 -2
bC
•* « *
£2^
« 8 S .
£ t i. §
«*- J c --
•• OJ ±J
4) * £ R
!-i •£ a. c
a S- « o
iggs
•2- »
U <-> u
•JS u -
^ i-IS
5s" S
•^ s •* °
4J ~ U ^
o « « -*
« " C -4
ur-i ° »
lfc- 0 ^-
§E u M
^c y e
s- « -i
o o£
— ~* «
B e 0 -5
* m •* C
^ 2 -1 «
ec £ - «
5 o * «
W ° 86 V
8 = sj
1*11
— »i x
>» L, O O
— o>
— hi
3 OS
M O
A5
-------
0
QC
0)
III
z
UJ
X
Q.
<
X
o
CO
01
i-H
O.
I
•o
01
CO
"8
u
01
Ol
to
CO
01
ex
s
CU
01
o
01
p
01
IM
01
CO
CO
"8
Cfl
3
01
c
01
o.
CO
o
o
CO
u
o
T3
IM
CO
•o
CO
0)
C
01
re
x
o
CO
O
o /-.
CO TJ
l-i 01
U-l 4J
C U
3 01
•<~l
C C
CO -H
O 3.
E -a-
CO
M I
6C
O J
E 60
O C
QJ
C
01
tn
01
)-
D.
o>
i-
3
00
A6
-------
APPENDIX B
LETTER TO HOUSTON WELLS, WELLS LABORATORIES, DESCRIBING
STUDY OF TOXAPHENE FORMULATIONS
USED IN GREENVILLE, MISSISSIPPI
Bl
-------
Mr. R. Houston Wells
Wells Laboratories
RT 2 Box 714
Wilmont Rd.
Greenville, MS 38701
Dear Mr. Wells:
Enclosed you will find tabulated the results of our analytical comparison
between four sources of toxaphene and two air samples taken in the Greenville
area:
.
Toxaphene sources: '
- U.S. EPA Toxaphene Reference Standard
- Hercules l&C 90-100
- Central America 90-100
- Drexel 616 form chem.
Samples :
- 3011-2 collected 9/1/81 to 9/29/81 Greenville, MS (185 m3 of
air)
- 3018-2 collected 11/2/81 to 11/5/81 Greenville, MS (32_ m3 of
air)
Our comparisons were based on total peak matches (table 1), distribution of the
matched peaks throughout the chromatogram (table 2) and the relative abundance,
in terms of Area percent, for peaks within prescribed intervals throughout the
chromatogram( table 3). Seventy-six peaks were chosen to represent toxaphene for
our comparisons based on their relative abundance and non-interference with PCBs
and other pesticides. Peak tables were constructed for each of the toxaphene
sources using our previously analyzed EPA toxaphene standard. The analysis was
performed on a Varian 3700 Gas-liquid chromatograph equipped with a Hewlett
Packard 50 m. , fused silica, SE-54 capillary column and electron capture
detection. Runs were temperature programmed from 100 °C to 245 °C at 1°C per
minute. Injector and detector temperatures were 270°C and 320°C respectively.
The primary purpose of this comparison was to select a toxphene standard
that best matched our samples. For both total matches (table 1) and
distribution of those matches (table 2), the U.S. EPA toxaphene standard (tox.
std.) showed the greatest consistency between samples. Greater than 50% of the
peaks were matched, for both samples, at a tolerance of 0.004 e.g. relative
retention time 40.004. Those peaks that were matched appeared to be evenly
distributed throughout the chromatogram as well. However, in terms of
composition, greater than 50% of the total peak area was represented within the
first two intervals (peaks 1 through 20, table 3) within which only about 25% of
the matched peaks occured.
We were also interested in similarities between the four toxaphene sources.
Surprisingly, the peak matches between sources were relatively poor. Best
matches occured between tox. std. and Central America (53 matched peaks) and
between Hercules and Drexel (58 and 64 matched peaks), (table 1). Differences
between sources were also evident in the relative amount of material occuring
B2
-------
within the various peak intervals (table 3). Common to all sources however, was
the relatively sparse abundance of material in the last two peak intervals.
Based on the aforementioned results, we feel that the U.S. EPA Reference
Standard is best suited for analyzing samples collected in the Greenville area.
We plan on continuing its use as our laboratory standard for future analysis.
Singer
P.S. Notice that for sample 3018-2 in table 2 the highest peak match was for
the Central American standard (perhaps this indicates a wind pattern link to a
user of this specific brand during this sampling.) This was a short sampling
interval and could be more source specific than sample 3011 which was a 28 day
integrated sample. Table 3 does show a similarity between Hercules and the
Toxaphene standard at least in relative percentage of peak areas represented by
each 10 peak interval. The Drexel does appear to lack area for the later peaks
in this comparison (this may be construed as Drexel having less persistence
potential since these later peaks have heavier molecular weights and will be
less volatile). Notice that from our air samples the heavier peaks were also
lacking, thus confirming that the more volatile early GC eluters are enriched in
the air as they should be due to their greater volatility. The Central American
standard appears to be a good retention time match to our Toxaphene reference
standard e.g. table 1 with 53 for toxaphene compared with Central America and 53
with Central America compared to toxaphene. Also, it appears to have a similar
percentage area distribution over the 8-10 peak area intervals as was observed
for the toxaphene and Hercules standards.
B3
-------
Table 1. Number of Peak matches between standards and two Greenville samples
(3011-2 and 3018-2) at two tolerance levels. Total number of peaks chosen for
comparison • 76.
Standards Used for Comparison
Tolerance - 0.004
Tox. Std.
Hercules
Central
America
Drexel
3011-2
3018-2
Tox. Std.
(76)*
34
53
41
40
40
Hercules
35
(76)*
35
58
33
21
Central
America
53
31
(76)*
32
25
43
Drexel
45
64
37
(76)*
32
26
Tolerance
Tox.
Hercules
Central
America
Drexel
3011-2
3018-2
0.002
Std.
(76)*
10
31
19
26
23
11
(76)*
17
47
13
11
27
15
(76)*
12
10
30
19
50
15
(76)*
24
10
* Each peak table was adjusted to match 76 peaks when compared against itself
B4
-------
Table 2. Numbers of peaks matched between two Greenville samples and four
standards, peak intervals of 10 peaks per interval. Tolerance « 0.004.
Numbers of Peak Matches Within the Specified Peak Range.
Sample Standard Total 1-10 11-20 21-30 31-40 41-50 51-60 61-70 71-76
363
3011-2 Hercules 33 3 4 3 3 8 552
2 3 1
552
Tox. Std. 40 7 5 7 5 5 452
3018-2 Hercules 21 7 6 2 2 0 1 1 2
Central
America 43 6 3 6 7 9 570
Drexel 27 3 5 4 6 5 1 1 1
Total
Standard
Tox. Std.
Hercules
Central
America
Drexel
Total
Total
41
33
25
32
1-10
5
3
4
0
11-20
5
4
3
4
21-30
7
3
8
6
31-40
7
3
4
5
41-50
4
8
0
5
B5
-------
Table 3. Percentage of Total Peak area within Peak intervals. Tolerance
0.004.
Tox. Std.
Hercules
Central
America
Drexel
*3011-2
*3018-2
1-10
17.38
12.55
16.22
19.70
19,20
12.47
11-20
16.30
13.27
7.47
14.04
43.38
38.61
21-30
12.57
15.39
12.22
19.29
9.20
12.04
31-40
14.13
11.74
15.39
11.75
15.93
11.96
41-50
11.77
15.88
18.02
14.88
3.62
13.91
51-60
16.49
18.78
15.85
11.50
5.04
7.10
61-70
9.72
10.16
13.16
5.49
3.22
3.63
71-76
1.64
2.22
1.67
3.34
0.42
0.28
*Greenville samples - Identified using EPA Toxaphene Reference Standard
B6
-------
Table 4. Average Retention times and Relative Retention times (RRT x 10) for
peaks used on upper and lower limits of peak intervals referred to in tables 2
and 3. Reference peak was DDE (RT 70.70 min.)
Peak
Number
1
10
11
20
21
30
31
40
41
50
51
60
61
70
71
76
Retention
Time (min)
44.53
56.36
57.24
67.99
68.36
74.30
70.44
77.86
78.46
85.56
85.93
90.94
91.06
98.67
99.51
100.37
Relative Retention
Time (RRT x
6.30
7.972
8.100
9.617
9.670
10.509
10.533
11.014
11.098
12.103
12.154
12.864
12.884
13.958
14.076
15.048
10)
B7
------- |