- Email: [email protected]
Prediction of chemical residues in aquatic organisms for a field discharge situation
Chemosphere,Vol.29, No. 1, pp. 141-153, 1994 ElsevierScienceLtd Printedin GreatBritain
Pergamon 0045-6535(94)00148-0
PREDICTION OF CHEMICAL RESIDUES IN AQUATIC ORGANISMS FOR A FIELD DISCHARGE SITUATION
Lawrence P. Burkhard a', Barbara R. Sheedyb~, Dennis J. McCauleyc2
'U.S. Environmental Protection Agency, Environmental Research Laboratory-Duluth, 6201 Congdon Blvd., Duluth, Minnesota 55804 bAScI Corporation, 6201 Congdon Blvd., Duluth, Minnesota 55804 CBattelle-Great Lakes Environmental Center, 739 Hastings Street, Traverse City, Michigan 49684 (Roeeivexlin USA7 February1994;accepted10 March 1994) Abstract
A field study was performed which compared predicted and measured concentrations of chemicals in receiving water organisms from three sampling locations on Five Mile Creek, Birmingham, AI. Two point source discharges, both from coke manufacturing facilities, were included in the field site and five chemicals were studied, i.e., biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene. Composite samples of effluent, receiving water organisms, crayfish (Decapoda) and sunfish (Lepomis sp.), and stream and discharge flow data were collected in March and April 1990. For the crayfish and sunfish, the measured residues were within a factor of 5 for 80% (12 of 15) and 53% (8 of 15) of the residues predicted using EPA's draft procedure (US-EPA 1991b), respectively, and were within a factor of 5 for 60% (9 of 15) and 40% (6 of 15) of the residues predicted using EPA's procedure with a BCF set equal to the chemical's Kow (after adjustment for lipid content of the organism), respectively. The predicted residues tended to be larger than the measured residues and with increasing Kow, greater disagreement between the predicted and measured values was observed.
Introduction
Concern over chemical residues in fish and shellfish has been prominent throughout the country. Historically, the Environmental Protection Agency (EPA) has used chemical-specific effluent limitations to
1present Address: Integrated Laboratory Systems, 6201 Congdon Blvd., Duluth, Minnesota 55804 ZPresent Address: Great Lakes Environmental Center, 739 Hastings Street, Traverse City, Michigan 49684
141
142 prevent the formation of chemical residues in receiving water fishes and shellfishes (US-EPA 1991a). Effluent limitations can be derived only for chemicals with water quality criteria and unfortunately, the number of water quality criteria is small relative to the large number of pollutants discharged. EPA is developing guidance procedures which would permit the development of effluent limitations for bioconcentratable chemicals that are not controlled by water quality criteria (US-EPA 1991b). In this guidance, chemical residues in receiving water organisms are predicted by calculating the product of the BAF (bioaccumulation factor) and the concentration of the chemical in the receiving water (US-EPA 1991b). In point source discharge situations, receiving water concentrations are derived by using a simple dilution model with data normally collected by the discharger, i.e., discharge flows and concentration of the chemical in the effluent, and with total stream flow data from United States Geological Service (USGS) measurements on the stream. When available, measured BAFs are used in the above calculation. However, in most cases, BAFs are estimated from the BCF (bioconcentration factor) of the chemical because measured values are not available. The guidance procedure of EPA uses the equation of Veith and Kosian (1983) to estimate the BCF from the Kow (n-octanol/water partition coefficient) of the chemical and then corrects this BCF to a BAF by multiplying the BCF by a food chain multiplier (FM) term. The FM is dependent upon the Kow of the chemical and the trophic level for the organism (US-EPA 1991b). This correction accounts for accumulation of the chemical over and above that due to bioconcentration processes by the organism and the FM values were derived from a modeling exercise by Thomann (1989) using a four trophic level pelagic food chain. This report examines how well chemical concentrations in receiving water organisms can be predicted in a field discharge situation using the residue prediction procedure in the draft guidance (USEPA 1991b). In performing this evaluation, chemical concentrations in aquatic organisms for biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene were predicted and then compared to chemical concentrations in the organisms. In this evaluation, chemical concentrations were predicted in two ways: (a) by using EPA's draft procedure (US-EPA 1991b) and (b) by using a modification of EPA's draft procedure where the BCF was set equal to the Kow of the chemical (after adjustment for lipid content of the organism). The second method for predicting chemical concentrations was included in this evaluation because the model of Thomann (1989) which was used to derived the FMs assumes that the lipid normalized BCF is equal to the Kow of the chemical at zero growth and "equilibrium".
Experimental Methods Description of field site The field site was a 5.3 km stretch of Five Mile Creek near Birmingham, Alabama (Figure 1). Within the study area, Five Mile Creek receives discharges from two coking operations, Coke Plants 1 and 2, and runoff from a railroad maintenance facility and Coke Plant 1 grounds. Five Mile Creek originates within a
143
/" j/J
STATION 1
STATION 3
DISCHARG~ J
I LI~GS GAUGINGSTA110N NUMBER2467OOO LOCATED 14.4 i(M
DOWNSTREAMFROM STATION3
Miles
0
1
Km
1
i
/
Figure 1. Five Mile Creek study area. 26
1.4 Coke Plant 1 1.2
/',,,
Coke Plant 2 o. ....
I
_~ 1.0
%
I
24 I
t //
t#
I t
04
22
rt
rt
==
(9 O
O
o 0.8
20 o
o
.--~_~0.6 a
~
a
I
0.4
/
LU
0.2
0.0
""7
%
I
I
w
14
12 26
28
30
1
3
5
7 9 11 13 15 March/April 1990
17
19
21
23
25
Figure 2. Estimated daily in-stream waste concentrations 0WCs) for Coke Plants 1 and 2
144
residential and commercial area of Birmingham. An USGS stream flow gauge (number 2457000) was located 14.4 km downstream of the study site. Indigenous organisms were sampled at four stations in this study (Figure 1). Station 1 was a small reservoir caused by a low head dam. The Coke Plant 1 discharge pipe entered Five Mile Creek at the base of the dam downstream from this station. Station 2 was located 1.3 km downstream from the Coke Plant 1 effluent discharge and immediately upstream of the Coke Plant 2 discharge. Station 3 was located at the US 31 bridge, 3.2 km downstream from the Coke Plant 1 effluent discharge and 1.8 km downstream from the Coke Plant 2 effluent discharge. Station 4 was located 5.3 km downstream from the Coke Plant 1 effluent discharge and 3.9 km downstream from the Coke Plant 2 effluent discharge.
Sampling procedures A series of four, seven-day time integrated composite effluent samples from both Coke Plants 1 and 2 were collected using an ISCO * sampler and sample collection was initiated on March 26 and 27, 1990, respectively. The last seven-day composite samples were collected on April 23 and 24, 1990 at Coke Plants 1 and 2, respectively. Resident crayfish (Decapoda) and sunfish (Lepomis sp.) were collected on April 25 and 26, 1990 by electroshocking the creek at each station. Samples containing a minimum of 30 grams using whole organisms were assembled for each organism at each station. The only criterion for compositing the organisms was the minimum amount of mass. Compositing based on other physical characteristics such as sex, size, reproductive state, or age of the organism was not done. The number of organisms per sample varied from 1 to 5. All tissue samples were held frozen until analysis. Stream flows at two points on Five Mile Creek were measured at various times during the study by measuring stream velocity and depth at 30 cm (1 foot) intervals across the creek. Total stream flow was calculated by summing the flows from each interval. Stream flows were measured at the US 31 overpass (Station 3) and approximately 50 meters downstream of Station 1 (near Coke Plant 1).
Analytical methods The weekly effluent composite samples were analyzed at two laboratories, Battelle-Columbus (B-C) and Environmental Research Laboratory-Duluth (ERL-D). The analysis procedure for the effluents at both labs consisted of spiking a known volume of effluent with d~o-biphenyl, d~0-phenanthrene, d~o-anthracene, d~0fluoranthene (B-C only), and d~0-pyrene, extracting the effluent with hexane, concentrating the extract to ca. 1.0 mL and analyzing the extract using gas chromatography/mass spectrometry (GC/MS). Tissue samples were analyzed at one laboratory, ERL-D, and were prepared for analysis by using a Waring blender. For each sample, an aliquot of the wet homogenized tissue was mixed with sodium sulfate to dry the sample, spiked with the deuterated compounds used for the effluent analyses, and then extracted using a 1:1 mixture of methylene chloride and hexane. The extract was taken to dryness and weighted. The extracts were redissolved using methylene chloride, and subjected to gel permeation chromatography and silica gel chromatography. The extract was concentrated to a volume less than 1 mL and analyzed using GC/MS. The
145 lipid content for each analysis was determined by dividing the mass of the extract, measured during the analysis, by the mass of tissue extracted. GC/MS analyses on the tissue and effluent extracts were performed by adding dlz-chrysene, the internal standard, to the extracts and then performing selected ion monitoring. Quantifications were performed using an internal standard method with a 4 or 5 point calibration curve using the M÷ ion for each chemical. Quantification standards contained the internal standard, and both deuterated and native forms of the five chemicals except for ERL-D's standards which did not contain dm-fluoranthene. For biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene (all quantified using the native form of the chemical), recovery corrections were made prior to reporting the data using the recoveries of dt0-biphenyl, d~o-phenanthrene, dloanthracene, dto-fluoranthene (B-C) or d~o-pyrene (ERL-D), and d~0-pyrene, respectively. For the effluent procedure, limits of detection for the five analytes were similar and ranged from 0.12 to 0.25/ag/L and from 0.005 to 0.010 lag/L for the samples from Coke Plants 1 and 2, respectively. Average recoveries for dto-biphenyl, d~0-phenanthrene, d~o-anthracene, d~o-fluoranthene, and dto-pyrene were 75, 80, 84, 79, and 86%, respectively, for the effluent procedure. For the tissue procedure, limits of detection for the five chemicals were similar and ranged from 0.15 to 0.40 pg/Kg of wet tissue. Average recoveries for d~o-biphenyl, d~o-phenanthrene, dlo-anthracene, and dm-pyrene were 53, 55, 54, and 62%, respectively, for the tissue procedure.
Results
In-stream waste concentrations (IWCs) fop" the effluents Stream flows for the site were estimated by constructing two regression equations (y=mx+b) where the independent variable was the flow measured at the USGS gauge downstream of the site and the dependent variable was the flow measured at Station 3 or 50 meters downstream of Station 1. Flows were measured at Station 3 and 50 meters downstream of Station 1 five or six times, and at the USGS gauge daily. Daily stream flows were estimated using the regression equations for Stations 1 and 3. In 1983, Mount et al. (1985) performed dye studies on Five Mile Creek to determine mixing characteristics for the discharges from both coking operations. These dye studies indicated that the effluents from Coke Plants I and 2 were completely mixed for the flows observed in this study at Stations 2 and 3, respectively. For each discharger, daily in-stream waste concentrations (IWCs) for the effluents from Coke Plants 1 and 2 were calculated by assuming complete mixing of the effluent with the stream and then dividing the measured discharge flow by the estimated stream flow for each day of the study. For Station 2, flows predicted for 50 m downstream of Station 1 and measured discharge flows for Coke Plant 1 were used. For Station 3, flows predicted for Station 3 and measured discharge flows for Coke Plant 2 were used. In Figure 2, the daily IWCs for each effluent are plotted and for both effluents, a gradual increase occurred during the study period. For the effluents from Coke Plants 1 and 2, the average of the daily IWCs (in percent)
146 (coefficients of variation, range of flows, and number of values averaged) were 0.644% (43.9%, 0.00-1.153%, 32) and 19.0% (15.8%, 13.9-24.0%, 32) at Stations 2 and 3, respectively.
Chemical concentrations in the effluents Replicate analyses were performed on each weekly composite for biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene. In Table 1, the weekly and four week mean concentrations for the five chemicals are reported for both effluents. Chemical concentrations in both effluents were relatively constant during the study. The median and average coefficients of variation for replicate analyses on each weekly composite were 20.5% and 37.6% for all chemicals, respectively.
Chemical concentrations in the receiving water organisms The crayfish and sunfish samples were analyzed for the five chemicals. Average chemical concentrations were calculated for each sampling station from the individual quantification results which were corrected to the mean lipid content for the species prior to averaging (Table 2). Therefore, the chemical concentrations in Table 2 are reported on a pg of chemical/Kg of wet tissue where the tissue had a lipid content set equal to the mean lipid content of the species. Duplicate analyses were performed on three tissue samples and good agreement was observed between the duplicate analyses, e.g., the median and average coefficient of variations were 12% and 23% for the 15 pairs of residue data, respectively. Analyses were also performed using different tissue composites for each sampling station and good agreement was observed among these analyses except for Station 3 (Table 2). One crayfish sample from Station 3, which was analyzed in duplicate, had concentrations for all five chemicals
Table 1. Chemical concentrationsa (ug/L) in weekly effluent composite samples from Coke Plants 1 and 2. Week I
Week 2
Week 3
Week 4
avgb
std dev
cv, %
1.88 15.6 6.68 21.4 15.7
0.34 15.5 7.22 21.4 17.4
0.56 15.1 6.40 20.3 14.9
0.91 1.68 1.33 1.95 2.28
162 11.1 20.8 9.62 15.3
Coke Plant 2 Effluente Biphenyl
0.04 0.03 0.02 0.23 0.12
0.04 0.07 0.02 0.23 0.12
0.06 0.20 0.02 0.28 0.13
0.04 0.08 0.02 0.23 0.12
0.03 0.08 0.00 0.04 0.01
71.9 104 0.0 15.8 10.7
Coke Plant 1 Effluentc Biphenyl Phenanthrene Anthracene Fluoranthene Pyrene
a Recovery and blank corrected. b avg = average, std dev = standard deviation, cv, % = coefficient of variation in percent. c Number of analyses performed for weeks 1, 2, 3, and 4 were 2, 2, 4, and 4, respectively. d Concentration of the chemical less than the limit of detection. e Number of analyses performed for weeks 1, 2, 3, and 4 were 4, 4, 6, and 4, respectively.
147 which were much higher than the other sample analyzed for that station, e.g., for phenanthrene, concentrations of 4200 pg/Kg as compared to 41.3 pg/Kg were determined. Furthermore, chemical concentrations determined for this sample were substantially larger than those measured for the other tissue samples from the site study area (Table 2). Additional tissue samples were also analyzed in this study (data not reported because lipid contents were not measured) and on a wet weight basis, the concentrations in these samples for Station 3 were within +35% of the measured concentrations for the lower concentration sample. Because this single sample was so different from the other samples from Station 3 as well as from all of the other tissue samples analyzed from the site study area, we believe that this sample was an outlier and not representative of resident organisms for this Station. Consequently, this sample was not included in the average concentrations reported in Table 2 for the crayfish and in the analysis of the data for the site study. The average chemical residues in the organisms at Stations 2, 3, and 4 were, in general, larger than those observed at Station 1. Significance testing by using a simple t-test with the pooled standard deviation for the crayfish and Dunnett's test for the sunfish revealed that most of the residues in the organisms at Station 2 were significantly greater (p_>0.05) than those at Station 1 (Table 2).
Table 2. Average chemical concentrations' (ug/Kg of wet tissue) in crayfish and sunfish samples. Station 1
Station 2
Crayfishb 2.43% lipid content~ Biphenyl 0.84 (0.03) a Phenanthrene 36.8 (8.32) Anthracene 1.90 (0.60) Fluoranthene 19.0 (4.10) Pyrene 10.7 (0.49)
18.4 (4.92) e 190 (160) 26.0 (4.92) ~ 126 (6.59) e 109 (9.80) e
Sunfishb 4.23% lipid content~ Biphenyl 6.20 (1.37) Phenanthrene 56.8 (6.83) Anthracene 4.38 (1.59) Fluoranthene 16.6 (2.16) Pyrene 8.66 (2.03)
7.66 (4.55) 61.9 (19.4) 10.9 (4.38) g 27.8 (12.5) 8.77 (1.44)
Station 3
2.69 41.3 13.1 70.4 44.2
(--)~ (--)f (--)f (--)f (--)f
10.1 (0.66) 58.4 (8.70) 15.8 (3.22) g 39.2 (9.59) g 13.0 (1.91)
Station 4
4.70 44.2 13.3 77.9 56.8
(2.59) (18.7) (6.15) (30.7) (22.0)
4.81 24.2 9.35 22.7 13.8
(0.43) (3.92) (0.96) (2.87) (8.46)
a The average chemical concentrations were calculated from individual quantification results which were recovery and blank corrected, and corrected to the mean lipid content of the species prior to averaging. b For Stations 1, 2, 3, and 4, the number of samples analyzed (and the number of organisms in each sample) were 2 (1,2), 2 (3,4), 2 (3, the outlier 3), and 2 (2,4) for the crayfish and 2 (1,3), 3 (1,1,1), 2 (1,1), and 3 (1,2,5) for the sunfish, respectively. c Percent lipid content of the reported values and mean lipid content of the species. d average (standard deviation). e Significantly greater than residue levels for Station l, one way analysis of variance, 95% confidence level. f Outlier not used in calculating average residue concentration. The average concentrations including the outlier were 98.9 (136), 2120 (2940), 245 (230), 1470 (1980), and 715 (949) for biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene, respectively. g Significantly greater that residue levels for Station 1, Dunnett's test, 95% confidence level.
148
Prediction of Chemical Residues in Receiving Water Organisms In EPA's method for predicting chemical residues (referred to as method A), the BCF-Kow relationship of Veith and Kosian (1983) is used to calculate a BCF from the Kow of the chemical. Their equation is: log BCF = 0.79 • log Kow - 0.40
n = 112, rz = 0.86, 7.6% lipid content
In the second method for predicting chemical residues (referred to as method B), BCFs were predicted from the Kow of the chemical by assuming that chemical's BCF was equal to the chemical's Kow after correction for the species lipid content. This correction was done by multiplying the Kow for the chemical by the lipid fraction of the species. For methods A and B, the predicted BCFs were adjusted to BAFs by multiplying the BCF by the FM term for the chemical. In this study, trophic level three FMs were used because trophic level three values are applicable to small fishes and arthropods (US-EPA 1991b). In Table 3, the KowS, FMs, BCFs, and BAFs are summarized for the five chemicals under investigation for methods A and B. The derived BAFs were then used in the following equations to predict the chemical residues (Ct) for the crayfish and sunfish at Stations 2, 3, and 4 (Table 4): C 2 --- C 1 °
(1 - F1) + BAF • (Effl ° F1)
C~=C~.(1-F0.(1-F2)+BAF°(Eft,.F~+Eff~.FI.(1-F
0)
C4 = C 1 ° ( 1 - F ~ ) . (1-Fz) + B A F ° ( E f t : . F : + E f f ~ o F 1 . ( 1 - F 2 ) ) where F~ = IWCi / 100; IWC~ is the in-stream waste concentration (in percent) for Coke Plants 1 and 2; Eff~ is the concentration of the chemical in the effluents from Coke Plants 1 and 2 (Table 1); and Cj is the concentration of the chemical in the crayfish or sunfish samples for Stations 1, 2, 3, and 4 (Table 2).
Discussion
Comparison of the Predicted and Measured Chemical Residues For the crayfish, the measured chemical concentrations were within a factor of 5 for 80% (12 of 15) of the chemical residues predicted using EPA's procedure (method A) and 60% (9 of 15) of the chemical residues predicted using method B. In general, the predicted residues tended to be larger than the measured residues at all three sampling stations (Figure 3). For the sunfish, the measured chemical concentrations were within a factor of 5 for 53% (8 of 15) of the chemical residues predicted using method A and 40% (6 of 15) of the chemical residues predicted using method B. The predicted residues tended to be larger than the measured residues and with increasing Kow, greater disagreement between the predicted and measured values was observed (Figure 4). When a chemical is metabolized in an aquatic organism, the measured chemical residue will be lower than that predicted because the methods used here for predicting BAFs are based upon non-metabolizable chemicals, e.g. chlorinated organics (Kleinow et al. 1987). Research has shown that polycyclic aromatic hydrocarbons are metabolized by vertebrates, such as fish, and are slowly (if at all) metabolized by invertebrates, such as arthropods (James 1989). Furthermore, studies have shown that unsubstituted polycyclic aromatic hydrocarbons have rates of metabolism that depend upon molecular size. In general, larger
149 Table 3. BAFs predicted using methods A and B.
Log K,,w
FM"
Method Ab predicted predicted BCF BAF
Method B predicted predicted BCF BAF
Crayfish 2.43% lipid content~ Biphenyl 4.00 d Phenanthrene 4.56 ~ Anthracene 4.45 e Fluoranthene 5.16 d Pyrene 4.88 e
1.0 1.3 1.1 2.8 1.8
184 510 417 1520 912
184 662 459 4250 1640
243 882 685 3510 1843
243 1150 754 9830 3320
Sunfish 4.23% lipid content~ Biphenyl 4.00 Phenanthrene 4.56 Anthracene 4.45 Fluoranthene 5.16 Pyrene 4.88
1.0 1.3 1.1 2.8 1.8
320 887 726 2640 1590
320 1153 799 7400 2860
423 1530 1190 6110 3210
423 1990 1310 17100 5780
a
c
e
All FM values are for trophic level 3 organisms. EPA's draft procedure, BAF = BCF • FM. The BCF is from the equation of Veith and Kosian (1983). Percent lipid content of the predicted BCFs and BAFs, and the mean lipid content of the species. De Bruijn et al. 1989. Hansch and Leo 1979.
Table 4.
Predicted chemical residues (pg/Kg of wet tissue) in crayfish and sunfish.
Chemical
Predicted Residues for Station 2 Method A Method B
Predicted Residues for Station 3 and 4 Method A Method B
Crayfish 2.43% lipid content ~ Biphenyl 1.51 Phenanthrene 101 Anthracene 20.8 Fluoranthene 574 Pyrene 168
1.73 148 32.9 1300 329
2.62 91.8 18.6 651 174
3.25 137 29.5 1490 342
Sunfish 4.23% lipid content" Biphenyl Phenanthrene Anthracene Fluoranthene Pyrene
7.71 251 58.4 2260 563
8.37 154 33.2 1120 294
9.46 233 52.3 2570 588
7.33 169 37.3 984 283
Percent lipid content of the reported values and the mean lipid content of the species.
150 30
10
..g O3
"~ ._o
3
[]
Sta~on2, me,rod A
[]
Sta~on2, method B
[]
Stal~on3, rne~3odA
[]
Station3, melhod B
[]
Station4, method A
[]
Station4, method B
"10
P
¢1
"¢= 0.3 0 0
rr
Dw
0.1
0.03
|
I
I
I
I
Biphenyl
Phenanthrene
Anthrancene
Fluoranthrene
Pyrene
Figure 3. Ratios of observed to predicted chemical residues for crayfish. The dotted, dashed, and solid lines represent ratios of 1/3 and 3, 1/5 and 5, and 1/10 and 10, respectively.
"g_ O3
=...
0.3
.m
0.1 "0 ¢,
;=.
Or) ,.Q 0
0.03
0 n"
[]
Station2, methodA
[]
Station2, method B
[]
Sta~on3, method A
[]
Station3, method B Station 4, method A
0.01 []
Station4, method B I
I
I
I
I
Biphenyl
Phenanthrene
Anthrancene
Fluoranthrene
Pyrene
Figure 4. Ratios of observed to predicted chemical residues for sunfish. The dotted, dashed, and solid lines represent ratios of 1/3 and 3, 1/5 and 5, and 1/10 and 10, respectively.
151 unsubstituted polycyclic aromatic hydrocarbons tend to have faster rates of metabolism than smaller polycyclic aromatic hydrocarbons (Varanasi et al. 1989), and thus, for the five chemicals of interest, rates of metabolism should follow the general order of biphenyl < phenanthrene -_-anthracene < fluoranthene -_-pyrene. The predicted residues were, in general, larger than the measured residues for both species. In addition, with increasing size of the chemical, greater differences among the measured and predicted residues were observed for the crayfish and sunfish, and these differences were much greater for the sunfish. These results are consistent with the metabolic behavior of the organisms analyzed and suggest that metabolism might be the cause of the observed differences between the measured and predicted residues. The predicted residues were in better agreement with the measured residues for method A than for method B. This result was somewhat anticipated since the predicted BCFs for method A are based upon experimental BCF data which incorporates the effects of metabolism in the measurement. In contrast, method B does not include any effects due to metabolism. Additional considerations in comparing the measured and predicted chemical residues Methods A and B implicitly assume that (a) the concentrations of the chemicals in the stream were constant during the study, (b) the chemical residues in the organisms were at steady-state with the receiving water, (c) the estimated stream and measured discharge flows accounted for all of the stream flows at Stations 2, 3, and 4, (d) losses due to degradation, volatilization, photolysis, and sedimentation of the chemicals in the stream were small, (e) the chemicals discharged into the stream were completely bioavailable, and (f) the chemical residues in the organisms from the upstream sampling station, Station 1, accounted for all sources of the chemicals other than that from the effluents. In selecting the study site, a "simple" site was chosen to minimize many of these implicit assumptions, i.e., the flow regime of the site was not complex, and chemical concentrations in the effluents discharged to the receiving water were relatively constant. However, not all of these implicit assumptions were true during the study. When the implicit assumptions a, b, c, d, and/or e are not true, the measured chemical residues would be smaller than the predicted residues because the actual exposure conditions due to the discharges would be lower than those based upon simple dilution. When other sources of these chemicals, e.g., runoff, exist within the study area, violation of implicit assumption f, the measured residues should be larger than the predicted residues because the actual exposure conditions would be higher than those based upon simple dilution of the discharges. Since the measured residues are smaller than the predicted residues, violations of the implicit assumptions a, b, c, d, and/or e are possible. All of the above implicit assumptions are a source of uncertainty in the predicted residues. Many of the implicit assumptions, i.e., c, d, e, and f, and their associated uncertainties could have been eliminated if ambient water concentrations had been determined in the study. Furthermore, the uncertainties associated with the estimation of the stream flows from the USGS flow data, 14.4 km downstream of the site, would have been eliminated.
152 Another variable which affects the accuracy and precision of measured residues are the number of tissue samples analyzed and the compositing technique used for these samples. In this study, a limited number of samples were analyzed at each station. In addition, the compositing technique for these samples did not consider the physical characteristics of the individuals in the samples, e.g., size, sex and age. The uncertainty associated with the average chemical concentrations in the organisms could have been smaller in this study if larger numbers of sample analyses had been performed using composite samples with larger and consistent numbers of organisms of similar physical characteristics.
Acknowledgements. Support for this work was provided by USEPA Office of Water, Permits Division. Technical assistance was provided by Correne Jenson, Phillip Marquis, Christine Berini, Jennifer Johnson and Kevin Hogfeldt (AScI Corporation); Mark Yancey, Dale Foster, Laura Hemon-Kenny, Terry Nobles, Steve Summer, and William Clement (Battelle-Columbus); John Shuey, Trae Forgette, and Mick DeGraeve (BattelleGreat Lakes Environmental Center); Michael Griffin (State of Alabama); and Nelson Thomas and Donald Mount (USEPA, Environmental Research Laboratory-Duluth).
References
De Bruijn J, Busser F, Seinen W, Hermans J (1989) Determination of octanol/water partition coefficients for hydrophobic organic chemicals with the "slow-stirring" method. Environ Toxic Chem 8:499-512 Hansch C, Leo A (1979) Substitutent constants for correlation analysis in chemistry and biology. John Wiley & Sons, New York, NY James MO (1989) Biotransformation and disposition of PAH in aquatic invertebrates, In: Varanasi, U (ed) Metabolism of polycyclic aromatic hydrocarbons in the aquatic environment. CRC: Boca Raton, Florida, Chap 3, pp 69-92 Kleinow KM, Melancon MJ, Lech JJ (1987) Biotransformation and induction: implications for toxicity, bioaccumulation and monitoring of environmental xenobiotics in fish. Environ Health Perspect 71:105119 Mount DI, Steen AE, Norberg-King TJ (1985) Validity of effluent and ambient toxicity testing for predicting biological impact on Five Mile Creek, Birmingham, Alabama. EPA/600/8-85/015. US-Environmental Protection Agency, Environmental Research Laboratory-Duluth, Minnesota Thomann RV (1989) Bioaccumulation model of organic chemical distribution in aquatic food chains. Environ Sci Technol 23:699-707 US-Environmental Protection Agency (1991a) Technical support document for water quality-based toxics control. EPA/505/2-90-001. Office of Water, Washington, DC US-Environmental Protection Agency (1991b) Assessment and control of bioconcentratable contaminants in surface waters. Public comment draft (Fed Regist 1991, 56, 13150-13151). Washington, DC
Office of Water,
153 Varanasi U, Stein JE, Nishimoto M (1989) Biotransformation and disposition of polycyclic aromatic hydrocarbon (PAH) in fish. In: Varanasi U (ed) Metabolism of polycyclic aromatic hydrocarbons in the aquatic environment. CRC: Boca Raton, Florida, Chap 4, pp 93-150 Veith GD, Kosian P (1983) Estimating bioconcentration potential from octanol/water partition coefficients. In: MacKay D, Paterson S, Eisenreich SJ, Simmons MS (eds) Physical behavior of PCBs in the Great Lakes. Ann Arbor Science, Ann Arbor, Chap 15, pp 269-282
Pergamon 0045-6535(94)00148-0
PREDICTION OF CHEMICAL RESIDUES IN AQUATIC ORGANISMS FOR A FIELD DISCHARGE SITUATION
Lawrence P. Burkhard a', Barbara R. Sheedyb~, Dennis J. McCauleyc2
'U.S. Environmental Protection Agency, Environmental Research Laboratory-Duluth, 6201 Congdon Blvd., Duluth, Minnesota 55804 bAScI Corporation, 6201 Congdon Blvd., Duluth, Minnesota 55804 CBattelle-Great Lakes Environmental Center, 739 Hastings Street, Traverse City, Michigan 49684 (Roeeivexlin USA7 February1994;accepted10 March 1994) Abstract
A field study was performed which compared predicted and measured concentrations of chemicals in receiving water organisms from three sampling locations on Five Mile Creek, Birmingham, AI. Two point source discharges, both from coke manufacturing facilities, were included in the field site and five chemicals were studied, i.e., biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene. Composite samples of effluent, receiving water organisms, crayfish (Decapoda) and sunfish (Lepomis sp.), and stream and discharge flow data were collected in March and April 1990. For the crayfish and sunfish, the measured residues were within a factor of 5 for 80% (12 of 15) and 53% (8 of 15) of the residues predicted using EPA's draft procedure (US-EPA 1991b), respectively, and were within a factor of 5 for 60% (9 of 15) and 40% (6 of 15) of the residues predicted using EPA's procedure with a BCF set equal to the chemical's Kow (after adjustment for lipid content of the organism), respectively. The predicted residues tended to be larger than the measured residues and with increasing Kow, greater disagreement between the predicted and measured values was observed.
Introduction
Concern over chemical residues in fish and shellfish has been prominent throughout the country. Historically, the Environmental Protection Agency (EPA) has used chemical-specific effluent limitations to
1present Address: Integrated Laboratory Systems, 6201 Congdon Blvd., Duluth, Minnesota 55804 ZPresent Address: Great Lakes Environmental Center, 739 Hastings Street, Traverse City, Michigan 49684
141
142 prevent the formation of chemical residues in receiving water fishes and shellfishes (US-EPA 1991a). Effluent limitations can be derived only for chemicals with water quality criteria and unfortunately, the number of water quality criteria is small relative to the large number of pollutants discharged. EPA is developing guidance procedures which would permit the development of effluent limitations for bioconcentratable chemicals that are not controlled by water quality criteria (US-EPA 1991b). In this guidance, chemical residues in receiving water organisms are predicted by calculating the product of the BAF (bioaccumulation factor) and the concentration of the chemical in the receiving water (US-EPA 1991b). In point source discharge situations, receiving water concentrations are derived by using a simple dilution model with data normally collected by the discharger, i.e., discharge flows and concentration of the chemical in the effluent, and with total stream flow data from United States Geological Service (USGS) measurements on the stream. When available, measured BAFs are used in the above calculation. However, in most cases, BAFs are estimated from the BCF (bioconcentration factor) of the chemical because measured values are not available. The guidance procedure of EPA uses the equation of Veith and Kosian (1983) to estimate the BCF from the Kow (n-octanol/water partition coefficient) of the chemical and then corrects this BCF to a BAF by multiplying the BCF by a food chain multiplier (FM) term. The FM is dependent upon the Kow of the chemical and the trophic level for the organism (US-EPA 1991b). This correction accounts for accumulation of the chemical over and above that due to bioconcentration processes by the organism and the FM values were derived from a modeling exercise by Thomann (1989) using a four trophic level pelagic food chain. This report examines how well chemical concentrations in receiving water organisms can be predicted in a field discharge situation using the residue prediction procedure in the draft guidance (USEPA 1991b). In performing this evaluation, chemical concentrations in aquatic organisms for biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene were predicted and then compared to chemical concentrations in the organisms. In this evaluation, chemical concentrations were predicted in two ways: (a) by using EPA's draft procedure (US-EPA 1991b) and (b) by using a modification of EPA's draft procedure where the BCF was set equal to the Kow of the chemical (after adjustment for lipid content of the organism). The second method for predicting chemical concentrations was included in this evaluation because the model of Thomann (1989) which was used to derived the FMs assumes that the lipid normalized BCF is equal to the Kow of the chemical at zero growth and "equilibrium".
Experimental Methods Description of field site The field site was a 5.3 km stretch of Five Mile Creek near Birmingham, Alabama (Figure 1). Within the study area, Five Mile Creek receives discharges from two coking operations, Coke Plants 1 and 2, and runoff from a railroad maintenance facility and Coke Plant 1 grounds. Five Mile Creek originates within a
143
/" j/J
STATION 1
STATION 3
DISCHARG~ J
I LI~GS GAUGINGSTA110N NUMBER2467OOO LOCATED 14.4 i(M
DOWNSTREAMFROM STATION3
Miles
0
1
Km
1
i
/
Figure 1. Five Mile Creek study area. 26
1.4 Coke Plant 1 1.2
/',,,
Coke Plant 2 o. ....
I
_~ 1.0
%
I
24 I
t //
t#
I t
04
22
rt
rt
==
(9 O
O
o 0.8
20 o
o
.--~_~0.6 a
~
a
I
0.4
/
LU
0.2
0.0
""7
%
I
I
w
14
12 26
28
30
1
3
5
7 9 11 13 15 March/April 1990
17
19
21
23
25
Figure 2. Estimated daily in-stream waste concentrations 0WCs) for Coke Plants 1 and 2
144
residential and commercial area of Birmingham. An USGS stream flow gauge (number 2457000) was located 14.4 km downstream of the study site. Indigenous organisms were sampled at four stations in this study (Figure 1). Station 1 was a small reservoir caused by a low head dam. The Coke Plant 1 discharge pipe entered Five Mile Creek at the base of the dam downstream from this station. Station 2 was located 1.3 km downstream from the Coke Plant 1 effluent discharge and immediately upstream of the Coke Plant 2 discharge. Station 3 was located at the US 31 bridge, 3.2 km downstream from the Coke Plant 1 effluent discharge and 1.8 km downstream from the Coke Plant 2 effluent discharge. Station 4 was located 5.3 km downstream from the Coke Plant 1 effluent discharge and 3.9 km downstream from the Coke Plant 2 effluent discharge.
Sampling procedures A series of four, seven-day time integrated composite effluent samples from both Coke Plants 1 and 2 were collected using an ISCO * sampler and sample collection was initiated on March 26 and 27, 1990, respectively. The last seven-day composite samples were collected on April 23 and 24, 1990 at Coke Plants 1 and 2, respectively. Resident crayfish (Decapoda) and sunfish (Lepomis sp.) were collected on April 25 and 26, 1990 by electroshocking the creek at each station. Samples containing a minimum of 30 grams using whole organisms were assembled for each organism at each station. The only criterion for compositing the organisms was the minimum amount of mass. Compositing based on other physical characteristics such as sex, size, reproductive state, or age of the organism was not done. The number of organisms per sample varied from 1 to 5. All tissue samples were held frozen until analysis. Stream flows at two points on Five Mile Creek were measured at various times during the study by measuring stream velocity and depth at 30 cm (1 foot) intervals across the creek. Total stream flow was calculated by summing the flows from each interval. Stream flows were measured at the US 31 overpass (Station 3) and approximately 50 meters downstream of Station 1 (near Coke Plant 1).
Analytical methods The weekly effluent composite samples were analyzed at two laboratories, Battelle-Columbus (B-C) and Environmental Research Laboratory-Duluth (ERL-D). The analysis procedure for the effluents at both labs consisted of spiking a known volume of effluent with d~o-biphenyl, d~0-phenanthrene, d~o-anthracene, d~0fluoranthene (B-C only), and d~0-pyrene, extracting the effluent with hexane, concentrating the extract to ca. 1.0 mL and analyzing the extract using gas chromatography/mass spectrometry (GC/MS). Tissue samples were analyzed at one laboratory, ERL-D, and were prepared for analysis by using a Waring blender. For each sample, an aliquot of the wet homogenized tissue was mixed with sodium sulfate to dry the sample, spiked with the deuterated compounds used for the effluent analyses, and then extracted using a 1:1 mixture of methylene chloride and hexane. The extract was taken to dryness and weighted. The extracts were redissolved using methylene chloride, and subjected to gel permeation chromatography and silica gel chromatography. The extract was concentrated to a volume less than 1 mL and analyzed using GC/MS. The
145 lipid content for each analysis was determined by dividing the mass of the extract, measured during the analysis, by the mass of tissue extracted. GC/MS analyses on the tissue and effluent extracts were performed by adding dlz-chrysene, the internal standard, to the extracts and then performing selected ion monitoring. Quantifications were performed using an internal standard method with a 4 or 5 point calibration curve using the M÷ ion for each chemical. Quantification standards contained the internal standard, and both deuterated and native forms of the five chemicals except for ERL-D's standards which did not contain dm-fluoranthene. For biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene (all quantified using the native form of the chemical), recovery corrections were made prior to reporting the data using the recoveries of dt0-biphenyl, d~o-phenanthrene, dloanthracene, dto-fluoranthene (B-C) or d~o-pyrene (ERL-D), and d~0-pyrene, respectively. For the effluent procedure, limits of detection for the five analytes were similar and ranged from 0.12 to 0.25/ag/L and from 0.005 to 0.010 lag/L for the samples from Coke Plants 1 and 2, respectively. Average recoveries for dto-biphenyl, d~0-phenanthrene, d~o-anthracene, d~o-fluoranthene, and dto-pyrene were 75, 80, 84, 79, and 86%, respectively, for the effluent procedure. For the tissue procedure, limits of detection for the five chemicals were similar and ranged from 0.15 to 0.40 pg/Kg of wet tissue. Average recoveries for d~o-biphenyl, d~o-phenanthrene, dlo-anthracene, and dm-pyrene were 53, 55, 54, and 62%, respectively, for the tissue procedure.
Results
In-stream waste concentrations (IWCs) fop" the effluents Stream flows for the site were estimated by constructing two regression equations (y=mx+b) where the independent variable was the flow measured at the USGS gauge downstream of the site and the dependent variable was the flow measured at Station 3 or 50 meters downstream of Station 1. Flows were measured at Station 3 and 50 meters downstream of Station 1 five or six times, and at the USGS gauge daily. Daily stream flows were estimated using the regression equations for Stations 1 and 3. In 1983, Mount et al. (1985) performed dye studies on Five Mile Creek to determine mixing characteristics for the discharges from both coking operations. These dye studies indicated that the effluents from Coke Plants I and 2 were completely mixed for the flows observed in this study at Stations 2 and 3, respectively. For each discharger, daily in-stream waste concentrations (IWCs) for the effluents from Coke Plants 1 and 2 were calculated by assuming complete mixing of the effluent with the stream and then dividing the measured discharge flow by the estimated stream flow for each day of the study. For Station 2, flows predicted for 50 m downstream of Station 1 and measured discharge flows for Coke Plant 1 were used. For Station 3, flows predicted for Station 3 and measured discharge flows for Coke Plant 2 were used. In Figure 2, the daily IWCs for each effluent are plotted and for both effluents, a gradual increase occurred during the study period. For the effluents from Coke Plants 1 and 2, the average of the daily IWCs (in percent)
146 (coefficients of variation, range of flows, and number of values averaged) were 0.644% (43.9%, 0.00-1.153%, 32) and 19.0% (15.8%, 13.9-24.0%, 32) at Stations 2 and 3, respectively.
Chemical concentrations in the effluents Replicate analyses were performed on each weekly composite for biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene. In Table 1, the weekly and four week mean concentrations for the five chemicals are reported for both effluents. Chemical concentrations in both effluents were relatively constant during the study. The median and average coefficients of variation for replicate analyses on each weekly composite were 20.5% and 37.6% for all chemicals, respectively.
Chemical concentrations in the receiving water organisms The crayfish and sunfish samples were analyzed for the five chemicals. Average chemical concentrations were calculated for each sampling station from the individual quantification results which were corrected to the mean lipid content for the species prior to averaging (Table 2). Therefore, the chemical concentrations in Table 2 are reported on a pg of chemical/Kg of wet tissue where the tissue had a lipid content set equal to the mean lipid content of the species. Duplicate analyses were performed on three tissue samples and good agreement was observed between the duplicate analyses, e.g., the median and average coefficient of variations were 12% and 23% for the 15 pairs of residue data, respectively. Analyses were also performed using different tissue composites for each sampling station and good agreement was observed among these analyses except for Station 3 (Table 2). One crayfish sample from Station 3, which was analyzed in duplicate, had concentrations for all five chemicals
Table 1. Chemical concentrationsa (ug/L) in weekly effluent composite samples from Coke Plants 1 and 2. Week I
Week 2
Week 3
Week 4
avgb
std dev
cv, %
1.88 15.6 6.68 21.4 15.7
0.34 15.5 7.22 21.4 17.4
0.56 15.1 6.40 20.3 14.9
0.91 1.68 1.33 1.95 2.28
162 11.1 20.8 9.62 15.3
Coke Plant 2 Effluente Biphenyl
0.04 0.03 0.02 0.23 0.12
0.04 0.07 0.02 0.23 0.12
0.06 0.20 0.02 0.28 0.13
0.04 0.08 0.02 0.23 0.12
0.03 0.08 0.00 0.04 0.01
71.9 104 0.0 15.8 10.7
Coke Plant 1 Effluentc Biphenyl Phenanthrene Anthracene Fluoranthene Pyrene
a Recovery and blank corrected. b avg = average, std dev = standard deviation, cv, % = coefficient of variation in percent. c Number of analyses performed for weeks 1, 2, 3, and 4 were 2, 2, 4, and 4, respectively. d Concentration of the chemical less than the limit of detection. e Number of analyses performed for weeks 1, 2, 3, and 4 were 4, 4, 6, and 4, respectively.
147 which were much higher than the other sample analyzed for that station, e.g., for phenanthrene, concentrations of 4200 pg/Kg as compared to 41.3 pg/Kg were determined. Furthermore, chemical concentrations determined for this sample were substantially larger than those measured for the other tissue samples from the site study area (Table 2). Additional tissue samples were also analyzed in this study (data not reported because lipid contents were not measured) and on a wet weight basis, the concentrations in these samples for Station 3 were within +35% of the measured concentrations for the lower concentration sample. Because this single sample was so different from the other samples from Station 3 as well as from all of the other tissue samples analyzed from the site study area, we believe that this sample was an outlier and not representative of resident organisms for this Station. Consequently, this sample was not included in the average concentrations reported in Table 2 for the crayfish and in the analysis of the data for the site study. The average chemical residues in the organisms at Stations 2, 3, and 4 were, in general, larger than those observed at Station 1. Significance testing by using a simple t-test with the pooled standard deviation for the crayfish and Dunnett's test for the sunfish revealed that most of the residues in the organisms at Station 2 were significantly greater (p_>0.05) than those at Station 1 (Table 2).
Table 2. Average chemical concentrations' (ug/Kg of wet tissue) in crayfish and sunfish samples. Station 1
Station 2
Crayfishb 2.43% lipid content~ Biphenyl 0.84 (0.03) a Phenanthrene 36.8 (8.32) Anthracene 1.90 (0.60) Fluoranthene 19.0 (4.10) Pyrene 10.7 (0.49)
18.4 (4.92) e 190 (160) 26.0 (4.92) ~ 126 (6.59) e 109 (9.80) e
Sunfishb 4.23% lipid content~ Biphenyl 6.20 (1.37) Phenanthrene 56.8 (6.83) Anthracene 4.38 (1.59) Fluoranthene 16.6 (2.16) Pyrene 8.66 (2.03)
7.66 (4.55) 61.9 (19.4) 10.9 (4.38) g 27.8 (12.5) 8.77 (1.44)
Station 3
2.69 41.3 13.1 70.4 44.2
(--)~ (--)f (--)f (--)f (--)f
10.1 (0.66) 58.4 (8.70) 15.8 (3.22) g 39.2 (9.59) g 13.0 (1.91)
Station 4
4.70 44.2 13.3 77.9 56.8
(2.59) (18.7) (6.15) (30.7) (22.0)
4.81 24.2 9.35 22.7 13.8
(0.43) (3.92) (0.96) (2.87) (8.46)
a The average chemical concentrations were calculated from individual quantification results which were recovery and blank corrected, and corrected to the mean lipid content of the species prior to averaging. b For Stations 1, 2, 3, and 4, the number of samples analyzed (and the number of organisms in each sample) were 2 (1,2), 2 (3,4), 2 (3, the outlier 3), and 2 (2,4) for the crayfish and 2 (1,3), 3 (1,1,1), 2 (1,1), and 3 (1,2,5) for the sunfish, respectively. c Percent lipid content of the reported values and mean lipid content of the species. d average (standard deviation). e Significantly greater than residue levels for Station l, one way analysis of variance, 95% confidence level. f Outlier not used in calculating average residue concentration. The average concentrations including the outlier were 98.9 (136), 2120 (2940), 245 (230), 1470 (1980), and 715 (949) for biphenyl, phenanthrene, anthracene, fluoranthene, and pyrene, respectively. g Significantly greater that residue levels for Station 1, Dunnett's test, 95% confidence level.
148
Prediction of Chemical Residues in Receiving Water Organisms In EPA's method for predicting chemical residues (referred to as method A), the BCF-Kow relationship of Veith and Kosian (1983) is used to calculate a BCF from the Kow of the chemical. Their equation is: log BCF = 0.79 • log Kow - 0.40
n = 112, rz = 0.86, 7.6% lipid content
In the second method for predicting chemical residues (referred to as method B), BCFs were predicted from the Kow of the chemical by assuming that chemical's BCF was equal to the chemical's Kow after correction for the species lipid content. This correction was done by multiplying the Kow for the chemical by the lipid fraction of the species. For methods A and B, the predicted BCFs were adjusted to BAFs by multiplying the BCF by the FM term for the chemical. In this study, trophic level three FMs were used because trophic level three values are applicable to small fishes and arthropods (US-EPA 1991b). In Table 3, the KowS, FMs, BCFs, and BAFs are summarized for the five chemicals under investigation for methods A and B. The derived BAFs were then used in the following equations to predict the chemical residues (Ct) for the crayfish and sunfish at Stations 2, 3, and 4 (Table 4): C 2 --- C 1 °
(1 - F1) + BAF • (Effl ° F1)
C~=C~.(1-F0.(1-F2)+BAF°(Eft,.F~+Eff~.FI.(1-F
0)
C4 = C 1 ° ( 1 - F ~ ) . (1-Fz) + B A F ° ( E f t : . F : + E f f ~ o F 1 . ( 1 - F 2 ) ) where F~ = IWCi / 100; IWC~ is the in-stream waste concentration (in percent) for Coke Plants 1 and 2; Eff~ is the concentration of the chemical in the effluents from Coke Plants 1 and 2 (Table 1); and Cj is the concentration of the chemical in the crayfish or sunfish samples for Stations 1, 2, 3, and 4 (Table 2).
Discussion
Comparison of the Predicted and Measured Chemical Residues For the crayfish, the measured chemical concentrations were within a factor of 5 for 80% (12 of 15) of the chemical residues predicted using EPA's procedure (method A) and 60% (9 of 15) of the chemical residues predicted using method B. In general, the predicted residues tended to be larger than the measured residues at all three sampling stations (Figure 3). For the sunfish, the measured chemical concentrations were within a factor of 5 for 53% (8 of 15) of the chemical residues predicted using method A and 40% (6 of 15) of the chemical residues predicted using method B. The predicted residues tended to be larger than the measured residues and with increasing Kow, greater disagreement between the predicted and measured values was observed (Figure 4). When a chemical is metabolized in an aquatic organism, the measured chemical residue will be lower than that predicted because the methods used here for predicting BAFs are based upon non-metabolizable chemicals, e.g. chlorinated organics (Kleinow et al. 1987). Research has shown that polycyclic aromatic hydrocarbons are metabolized by vertebrates, such as fish, and are slowly (if at all) metabolized by invertebrates, such as arthropods (James 1989). Furthermore, studies have shown that unsubstituted polycyclic aromatic hydrocarbons have rates of metabolism that depend upon molecular size. In general, larger
149 Table 3. BAFs predicted using methods A and B.
Log K,,w
FM"
Method Ab predicted predicted BCF BAF
Method B predicted predicted BCF BAF
Crayfish 2.43% lipid content~ Biphenyl 4.00 d Phenanthrene 4.56 ~ Anthracene 4.45 e Fluoranthene 5.16 d Pyrene 4.88 e
1.0 1.3 1.1 2.8 1.8
184 510 417 1520 912
184 662 459 4250 1640
243 882 685 3510 1843
243 1150 754 9830 3320
Sunfish 4.23% lipid content~ Biphenyl 4.00 Phenanthrene 4.56 Anthracene 4.45 Fluoranthene 5.16 Pyrene 4.88
1.0 1.3 1.1 2.8 1.8
320 887 726 2640 1590
320 1153 799 7400 2860
423 1530 1190 6110 3210
423 1990 1310 17100 5780
a
c
e
All FM values are for trophic level 3 organisms. EPA's draft procedure, BAF = BCF • FM. The BCF is from the equation of Veith and Kosian (1983). Percent lipid content of the predicted BCFs and BAFs, and the mean lipid content of the species. De Bruijn et al. 1989. Hansch and Leo 1979.
Table 4.
Predicted chemical residues (pg/Kg of wet tissue) in crayfish and sunfish.
Chemical
Predicted Residues for Station 2 Method A Method B
Predicted Residues for Station 3 and 4 Method A Method B
Crayfish 2.43% lipid content ~ Biphenyl 1.51 Phenanthrene 101 Anthracene 20.8 Fluoranthene 574 Pyrene 168
1.73 148 32.9 1300 329
2.62 91.8 18.6 651 174
3.25 137 29.5 1490 342
Sunfish 4.23% lipid content" Biphenyl Phenanthrene Anthracene Fluoranthene Pyrene
7.71 251 58.4 2260 563
8.37 154 33.2 1120 294
9.46 233 52.3 2570 588
7.33 169 37.3 984 283
Percent lipid content of the reported values and the mean lipid content of the species.
150 30
10
..g O3
"~ ._o
3
[]
Sta~on2, me,rod A
[]
Sta~on2, method B
[]
Stal~on3, rne~3odA
[]
Station3, melhod B
[]
Station4, method A
[]
Station4, method B
"10
P
¢1
"¢= 0.3 0 0
rr
Dw
0.1
0.03
|
I
I
I
I
Biphenyl
Phenanthrene
Anthrancene
Fluoranthrene
Pyrene
Figure 3. Ratios of observed to predicted chemical residues for crayfish. The dotted, dashed, and solid lines represent ratios of 1/3 and 3, 1/5 and 5, and 1/10 and 10, respectively.
"g_ O3
=...
0.3
.m
0.1 "0 ¢,
;=.
Or) ,.Q 0
0.03
0 n"
[]
Station2, methodA
[]
Station2, method B
[]
Sta~on3, method A
[]
Station3, method B Station 4, method A
0.01 []
Station4, method B I
I
I
I
I
Biphenyl
Phenanthrene
Anthrancene
Fluoranthrene
Pyrene
Figure 4. Ratios of observed to predicted chemical residues for sunfish. The dotted, dashed, and solid lines represent ratios of 1/3 and 3, 1/5 and 5, and 1/10 and 10, respectively.
151 unsubstituted polycyclic aromatic hydrocarbons tend to have faster rates of metabolism than smaller polycyclic aromatic hydrocarbons (Varanasi et al. 1989), and thus, for the five chemicals of interest, rates of metabolism should follow the general order of biphenyl < phenanthrene -_-anthracene < fluoranthene -_-pyrene. The predicted residues were, in general, larger than the measured residues for both species. In addition, with increasing size of the chemical, greater differences among the measured and predicted residues were observed for the crayfish and sunfish, and these differences were much greater for the sunfish. These results are consistent with the metabolic behavior of the organisms analyzed and suggest that metabolism might be the cause of the observed differences between the measured and predicted residues. The predicted residues were in better agreement with the measured residues for method A than for method B. This result was somewhat anticipated since the predicted BCFs for method A are based upon experimental BCF data which incorporates the effects of metabolism in the measurement. In contrast, method B does not include any effects due to metabolism. Additional considerations in comparing the measured and predicted chemical residues Methods A and B implicitly assume that (a) the concentrations of the chemicals in the stream were constant during the study, (b) the chemical residues in the organisms were at steady-state with the receiving water, (c) the estimated stream and measured discharge flows accounted for all of the stream flows at Stations 2, 3, and 4, (d) losses due to degradation, volatilization, photolysis, and sedimentation of the chemicals in the stream were small, (e) the chemicals discharged into the stream were completely bioavailable, and (f) the chemical residues in the organisms from the upstream sampling station, Station 1, accounted for all sources of the chemicals other than that from the effluents. In selecting the study site, a "simple" site was chosen to minimize many of these implicit assumptions, i.e., the flow regime of the site was not complex, and chemical concentrations in the effluents discharged to the receiving water were relatively constant. However, not all of these implicit assumptions were true during the study. When the implicit assumptions a, b, c, d, and/or e are not true, the measured chemical residues would be smaller than the predicted residues because the actual exposure conditions due to the discharges would be lower than those based upon simple dilution. When other sources of these chemicals, e.g., runoff, exist within the study area, violation of implicit assumption f, the measured residues should be larger than the predicted residues because the actual exposure conditions would be higher than those based upon simple dilution of the discharges. Since the measured residues are smaller than the predicted residues, violations of the implicit assumptions a, b, c, d, and/or e are possible. All of the above implicit assumptions are a source of uncertainty in the predicted residues. Many of the implicit assumptions, i.e., c, d, e, and f, and their associated uncertainties could have been eliminated if ambient water concentrations had been determined in the study. Furthermore, the uncertainties associated with the estimation of the stream flows from the USGS flow data, 14.4 km downstream of the site, would have been eliminated.
152 Another variable which affects the accuracy and precision of measured residues are the number of tissue samples analyzed and the compositing technique used for these samples. In this study, a limited number of samples were analyzed at each station. In addition, the compositing technique for these samples did not consider the physical characteristics of the individuals in the samples, e.g., size, sex and age. The uncertainty associated with the average chemical concentrations in the organisms could have been smaller in this study if larger numbers of sample analyses had been performed using composite samples with larger and consistent numbers of organisms of similar physical characteristics.
Acknowledgements. Support for this work was provided by USEPA Office of Water, Permits Division. Technical assistance was provided by Correne Jenson, Phillip Marquis, Christine Berini, Jennifer Johnson and Kevin Hogfeldt (AScI Corporation); Mark Yancey, Dale Foster, Laura Hemon-Kenny, Terry Nobles, Steve Summer, and William Clement (Battelle-Columbus); John Shuey, Trae Forgette, and Mick DeGraeve (BattelleGreat Lakes Environmental Center); Michael Griffin (State of Alabama); and Nelson Thomas and Donald Mount (USEPA, Environmental Research Laboratory-Duluth).
References
De Bruijn J, Busser F, Seinen W, Hermans J (1989) Determination of octanol/water partition coefficients for hydrophobic organic chemicals with the "slow-stirring" method. Environ Toxic Chem 8:499-512 Hansch C, Leo A (1979) Substitutent constants for correlation analysis in chemistry and biology. John Wiley & Sons, New York, NY James MO (1989) Biotransformation and disposition of PAH in aquatic invertebrates, In: Varanasi, U (ed) Metabolism of polycyclic aromatic hydrocarbons in the aquatic environment. CRC: Boca Raton, Florida, Chap 3, pp 69-92 Kleinow KM, Melancon MJ, Lech JJ (1987) Biotransformation and induction: implications for toxicity, bioaccumulation and monitoring of environmental xenobiotics in fish. Environ Health Perspect 71:105119 Mount DI, Steen AE, Norberg-King TJ (1985) Validity of effluent and ambient toxicity testing for predicting biological impact on Five Mile Creek, Birmingham, Alabama. EPA/600/8-85/015. US-Environmental Protection Agency, Environmental Research Laboratory-Duluth, Minnesota Thomann RV (1989) Bioaccumulation model of organic chemical distribution in aquatic food chains. Environ Sci Technol 23:699-707 US-Environmental Protection Agency (1991a) Technical support document for water quality-based toxics control. EPA/505/2-90-001. Office of Water, Washington, DC US-Environmental Protection Agency (1991b) Assessment and control of bioconcentratable contaminants in surface waters. Public comment draft (Fed Regist 1991, 56, 13150-13151). Washington, DC
Office of Water,
153 Varanasi U, Stein JE, Nishimoto M (1989) Biotransformation and disposition of polycyclic aromatic hydrocarbon (PAH) in fish. In: Varanasi U (ed) Metabolism of polycyclic aromatic hydrocarbons in the aquatic environment. CRC: Boca Raton, Florida, Chap 4, pp 93-150 Veith GD, Kosian P (1983) Estimating bioconcentration potential from octanol/water partition coefficients. In: MacKay D, Paterson S, Eisenreich SJ, Simmons MS (eds) Physical behavior of PCBs in the Great Lakes. Ann Arbor Science, Ann Arbor, Chap 15, pp 269-282