Page 2 of 9

7.2.3.2. Soil concentrations and concurrent concentrations in bottom sediments and fish

The Connecticut Department of Environmental Protection (CDEP, 1992) established a program in 1986 for monitoring TCDD, TCDF, and other dioxin-like isomers of comparable toxicity in several environmental matrices near resource recovery facilities (RRFs). Matrices monitored include ambient air, residues and leachate from the ash disposal sites, surficial soils, surface water surficial bottom sediments, and whole fish. The purpose of the program is to evaluate the impact of RRFs. Eight locations were evaluated through 1990, with one location serving as a baseline or reference site. Of the seven remaining locations, RRFs began operation in 1983 (1 RRF), 1987 (3), 1988 (1), and 1990 (2). This section will examine the soil, sediment, and fish data from that program.

The soil concentrations throughout all eight sites might be characterized as typical of background concentrations mainly because the concentrations of 2,3,7,8-TCDD measured through 1990 averaged 0.56 ppt (n = 77; assuming non-detects were 1/2 detection limit), with roughly a 50% non-detect rate (at a detection limit which has varied by data set, but has been around 0.1 ppt). In studies measuring soil concentrations of 2,3,7,8-TCDD in background or rural settings, either none was found, or concentrations were found in the low ppt range - this seems to also characterize the Connecticut data. In a statistical analysis of the data collected through 1988, the average soil concentration for 2,3,7,8-TCDD as 0.44 ppt (n = 42; CDEP, 1992; same procedures for estimating average concentrations), which is lower than the 0.56 ppt concentration for all samples through 1990.

Concentrations of 2,3,7,8-TCDF taken after 1988, however, are lower than those taken in 1987 and 1988: the average including samples through 1988 was 8.20 ppt (n = 41; CDEP, 1992); while through 1990 was 6.77 (n = 77; CDEP, 1992). This simple examination of averages over time does not seem to indicate statistically significant change, if any. As well, in a statistical analysis of the data (principal component analysis of the concentration levels of all isomers in soil, fish, and sediment to attempt to identify stratification of the data by year) for four of the RRFs through 1990, MRI (1992) concluded that the RRFs had no apparent effect on the levels of the dioxin-like compounds in the three matrices.

The purpose of developing the argument that levels in soil are low and perhaps typical of background conditions is to be able to compare the soil to sediment, and sediment to fish ratios that arise from this data with those that were generated in example Scenarios 1 and 2 in Chapter 5. Those scenarios demonstrated the on-site scenario, with basin-wide soil concentrations of 2,3,7,8-TCDD of 1 ppt. As in all the comparison of literature data with model results in this section, this is not a validation exercise.

Information and results from the CDEP program for the soil, sediment, and fish matrices are presented in Tables 7-2 through 7-4. The data and supporting documentation was supplied by CDEP (1992). Table 7-2 provides a summary of the eight sites currently in the CDEP program. One "reference" or "control" site includes two areas, which for 1988 was at Union, Connecticut, and for 1990 was at Stafford, Connecticut. No nearby potential sources of dioxin release (industrial, commercial) were identified for these two reference sites. One of the sites, the Hartford site, is near the Connecticut River. All water bodies sampled were coves with direct links to the river. Industrial and commercial enterprises which use the river are speculated to have resulted in the generally higher fish concentrations noted in the Hartford site, as compared to the other sites. Twenty-one water bodies have been sampled, including harbors, channels, impoundments, reservoirs, coves, ponds, rivers, and lakes.

Six species of fish have been sampled, including carp, channel and white catfish, white sucker, brown bullhead, and yellow perch. All but the yellow perch are bottom feeders. The yellow perch was sampled mostly when a sufficient sample of bottom feeder could not be obtained. Samples of bottom feeders were sought because it was felt that they would have the highest tissue concentrations due to their association with bottom sediments, and therefore be the best markers for impact and change over time (C. Fredette, CDEP, personal communication). The soil sampling program was not extensive; samples were only taken near ambient air monitoring stations, and only 77 samples were taken through 1990.

It certainly cannot be claimed that the samples are statistically representative of soils which drain into the water bodies. However, given the consistency in concentrations noted and their low values, the supposition is made that concentrations are adequately representative of soils which impact the water bodies. Maps of the sites were obtained from CDEP to evaluate the distance from the soil sampling sites to the nearest water bodies. Nearly all soil sampling sites were within 3 miles of the nearest water body, and most were near to and less than one mile away.

Table 7-3 lists the frequency of non-detects for all data through 1990, and incomplete information on detection limits. For determining average concentrations in sampled matrices, non-detects were assumed to equal 1/2 the detection limit. The detection limits for these matrices varied over time with different data sets. The detection limits noted were those cited for 1987 and 1988 data (from a draft Monitoring Progress Report supplied by CDEP). That report did not list detection limits for three matrices noted. In parenthesis is noted the lowest concentration in the data sets, which would correspond to 1/2 the detection limit at the time the non-detect was measured. The purpose of presenting this data is simply to argue that assuming 1/2 the detection limit for computing averages will not greatly impact the averages. This can be demonstrated for the one matrix where this is most likely to be a concern - soil concentrations of 2,3,7,8-TCDD where a 50% non-detect rate was noted. If half the samples were assigned a value of 0.0 instead of perhaps 0.07 ppt (half the noted detection limit of 0.13 ppt), than the overall average would drop from 0.56 ppt to 0.52 ppt.

table Table 7-2 Description of soil, sediment, and fish sampling program of dioxin-like compounds undertaken by the Connecticut Department of Environmental Protection . table Table 7-3 Frequency of nondetects and detection limits for soil, and fish, for three congeners in the Connecticut Department of Environmental Protection data set .
expand table Table V3 7-2 expand table Table V3 7-3

Table 7-4 summarizes the key results from the CDEP data. The Csed:Csoil ratio is the ratio of sediment concentration to soil concentration for the eight sites for 2,3,7,8-TCDD - these are not organic carbon normalized concentration ratios. The second ratio noted is called the BSAF ratio, because it is defined in the same way that the Biota Sediment Accumulation Factor is defined: the ratio of the lipid-normalized whole fish tissue concentration over the organic carbon normalized bottom sediment concentration.

The BSAF is used to estimate fish tissue concentrations from bottom sediment concentrations in this assessment. The fish lipid contents and organic carbon contents for each site were supplied by CDEP (1992). The BSAFs for the entire data set and the four concentrations are based on averages of fish lipid and organic carbon contents from the entire data set.

Key observations from the demonstration scenarios and the results of the CDEP program are:

1) Demonstration scenarios 1 and 2 in Chapter 5 estimated the impact from basin-wide soil concentrations of 1 ppt of the example compounds. The difference in the scenarios was in exposure patterns and exposure site characteristics - the impacts to surface water sediments and fish were the same in both scenarios. The estimated concentration of 2,3,7,8-TCDD and 2,3,4,7,8-TCDF in bottom sediments was 2.79 and 2.85 ppt, respectively.

The small difference was due to a larger organic carbon partition coefficient assumed for 2,3,4,7,8-TCDF. Sediment to soil ratios for these two compounds are, therefore, 2.79 and 2.85. These compare to the overall 3.86 and 1.58 estimated in the Connecticut data set for these two compounds. The ratios for 2,3,7,8-TCDF and for total toxic equivalents were 2.58 and 2.69. The average of these TCDD and TCDF ratios is 2.68. This tends to support but does not validate the model's approach.

table Table 7-4 Results for Connecticut Department of Environmental Protection sampling, including soil, sediment and fish concentrations, and key concentration ratios of sediment to soil and the Biota Sediment Accumulation Factor (BSAF) ratio.
One of the key model parameters in the soil to sediment algorithm which is uncertain and has been questioned is the soil enrichment ratio. It was assigned a value of 3.0, which means that concentrations in soil eroding from the field are three times higher than concentrations on the field. This value was questioned in Section 6.3.3.3, Chapter 6, where it was shown that its assigned value of 3.0 could lead to unrealistically high off-site soil concentrations. However, if it is assigned a lower value, that the sediment:soil ratios would also be lower - an enrichment ratio of 1.0 would lead to sediment:soil ratios less than 1.0. The close match of sediment:soil ratios with an enrichment ratio of 3.0 does not validate the model's approach to evaluating surface water impacts from low basin-wide soil concentrations, however.
expand table Table V3 7-4

The model assumes that all surface water impacts are from erosion of basin soils. In fact, sediment concentrations in water bodies are also a function of direct atmospheric depositions onto water bodies and other direct, industrial related, discharges into water bodies. Such depositions may originate from sources other than soil contamination - such as air emissions from cars or industry. The impact of industrial sources to the sediments in the CDEP data is unclear. As noted, evidence collected so far does not indicate an impact from incinerator emissions.

The Hartford site has been cited as being impacted with industrial use of the nearby Connecticut River. However, the sediment concentrations of the water bodies at this site are not higher than other sites - in fact, the sediment concentrations from the Bridgeport, Bristol, Preston, and even the background Union/Stafford sites are comparable or higher. In any case, deposition of air-borne contaminants are likely to impact bottom sediments to some degree, and the soil contamination models of this assessment do not include such an impact (the stack emission source category does include this impact for emissions reaching water bodies).

Because the soil contamination models do not include air-borne depositions, it is possible that the enrichment ratio of 3.0 serves to artificially increase modeled bottom sediment concentration. In any case, the CDEP data does appear to indicate that bottom sediment concentrations exceed surface soil concentrations by more than a factor of 2.0 in environmental settings that mostly do not appear to be impacted by industrial activities.

2) The overall BSAF ratios for the three dioxin compounds and the TEQ ratio, ranging from 0.24-0.86, are higher than the BSAF used in the demonstration scenario of 0.09 for 2,3,7,8-TCDD and 2,3,4,7,8-PCDF. Higher BSAF in the CDEP data are expected because the fish species sampled were bottom feeders, except for the yellow perch. The selected BSAF of 0.09 is mainly supported by Lake Ontario data (EPA, 1990a), which was on brown and lake trout, smallmouth bass, and white and yellow perch, all column feeders.

Bottom feeders are expected to have more exposure to the contaminants because of their direct contact with sediments. This implies that use of the BSAF for site-specific assessments should consider the dietary pattern of exposed individuals. If a significant portion of local fish consumption includes bottom feeders (such as catfish), then perhaps a BSAF greater the 0.09 used for the demonstration scenarios is warranted.

3) Of the six sites for which BSAFs were individually determined for 2,3,7,8-TCDD, the highest BSAF was from the Hartford site at 0.97. The claim is not made that it is substantially or significantly different from BSAFs at the other sites - it is simply a point of interest for comment. The Hartford site has been previously identified as likely to have been impacted by activity on Connecticut River - all the fish are taken from coves directly connected to the river. Although the bottom sediment concentrations at this site are not different from other sites, one hypothesis is that the water column is more impacted for this site as compared to other sites.

In Chapter 4, Section 4.3.4.1, a key issue identified for the validity of the BSAF approach is the issue of past versus ongoing contamination. The same issue was discussed in Section 4.6. of Chapter 4, on the effluent discharge source category. Generally, the hypothesis offered was that fish are likely to be more exposed with ongoing impacts to the water body as compared to a situation where impacts were principally historical. The effluent discharge source category is a case of ongoing impacts.

The argument presented in Section 4.6. of Chapter was that the BSSAF (biota suspended sediment accumulation factor) should be greater in numerical value than a BSAF whose value is derived from data on a water body whose impacts have been primarily historical. This was the case for the assignment of a 0.09 for the BSAF, which was based on data on column feeders in Lake Ontario, a lake whose impact has been speculated as primarily historical. Although the numerical difference between the Hartford BSAF, at 0.97, and the next largest BSAF at Bristol, at 0.78, is not that large, perhaps that difference is due to the fact that the fish are more exposed at Hartford due to ongoing impacts from the Connecticut River.

In summary, this section has evaluated data supplied by the Connecticut Department of Environmental Protection on fish, sediment, and soil data. It is the only data set that could be found where soil and sediment data were concurrently taken in areas evaluated as (mostly) not impacted by industrial activity. An examination of the sediment to surface soil concentration ratios, showing them generally to be above 2.0 with an average for all data points and dioxin-like congeners of 2.68, supports the soil contamination model of this assessment for estimating sediment impacts from uniform basin-wide soil concentrations, which showed sediment to surface soil concentration ratios of 2.8 for 2,3,7,8-TCDD and 2,3,4,7,8-PCDF.

The BSAFs determined from the CDEP data are higher than the BSAFs used in the demonstration scenarios of this assessment. This was likely due to use of bottom feeders for fish concentration of the CDEP - bottom feeders are likely to have more exposure to dioxin-like compounds in water bodies than column feeders due to their association with contaminated bottom sediments.

7.2.3.3. Other bottom sediment concentration data

Assuming elevated sediment concentrations are a function of elevated surface soil concentrations is reasonable when the only source of water body contamination is soil contamination. However, comparing soil and sediment concentrations would not be appropriate if sediments and water were impacted by industrial discharges, which has often been cited as the cause for sediment and water impacts (see Bopp, et al., 1991; Norwood, et al., 1989; e.g.). Sediment concentrations of note have also been found in surface water bodies near urban settings, with car and industrial stack emissions cited as likely causes (Gotz and Schumacher, 1990; Rappe and Kjeller, 1987).

Rappe, et al. (1989) collected samples from the Baltic Sea, which were described as background samples. They note that the pattern of tetra-CDF congener concentrations found in the Baltic Sea were typical of what they termed the "incineration patterns" - air and air particulate concentrations that were attributed to sources such as incineration, car exhausts, steel mills, etc. On the other hand, sediment samples collected between 4 and 30 km downstream from a pulp mill revealed a congener pattern typical of bleaching mills. The stack emission and effluent discharge source categories provide separate models for water body impacts. The capability of the effluent discharge model is estimating fish tissue concentrations is examined in Section 7.2.3.6 below. The remainder of this section examines some of the data available which is not attributed to industrial or urban sources.

Except for the CDEP data described in Section 7.2.3.2 above, data was not found linking sediment concentrations to soil concentrations, in an urban or more pertinent to this assessment, a rural setting. Some sampling has occurred in areas described as rural or background. Sediment sampling in Lake Orono in Central Minnesota in such a setting found no tetra- and penta-CDDs, although occurrences of total hexa-CDDs were found in the low ng/kg (ppt) level, occurrences of hepta-CDDs to a high of 110 ppt, and total OCDD concentrations ranged from 490-600 ppt for three samples (Reed, et al., 1990).

A report on sampling of several estuaries in Eastern United States included a "reference" or relatively clean site, central Long Island Sound. There were no occurrences of 2,3,7,8-TCDD, although 2,3,7,8-TCDF was found at 15 ppt in this clean site. Other sites had identified industrial source inputs and higher noted concentrations (Norwood, et al., 1990). 2,3,7,8-TCDD was extensively found in sediments of Lake Ontario (EPA, 1990a). The average of samples from all depths of sediment collection from 49 stations including 55 samples was 68 ppt.

The average of 30 surficial sediment samples was 110 ppt. A modeling exercise implied that an annual load of 2.1 kg/year into Lake Ontario corresponds to a concentration of 110 ppt. One identified source was the Hyde Park Landfill, located about 2000 feet from the Niagara River, which drains into Lake Ontario. Between 1954 and 1975, an estimated 0.7 to 1.6 tons of 2,3,7,8-TCDD were deposited in the landfill. A principal conclusion from the modeling exercise, however, was that a characterization of historical loadings of 2,3,7,8-TCDD into the lake was not available and would be necessary to evaluate the contributions by the Hyde Park Landfill.

7.2.3.4. Data on water concentrations of dioxin-like compounds

Tables B-3 and B-4, Appendix B of Volume II, summarize available data on surface water concentrations of the PCDDs and PCDFs. The results on this table are not directly amenable to comparison because the sources of contamination were unspecified except to note that, in some studies, a portion of the sampling occurred for water bodies known to be impacted by industrial discharges. The 104-mill pulp and paper mill study, which measured discharges into of 2,3,7,8-TCDD and 2,3,7,8-TCDF into surface water bodies, was the only such study currently available which measured impacts to surface water bodies. Section 7.2.3.6 below discusses the use of this data to evaluate the effluent discharge source category. This study did not measure water concentrations, and no other studies could be found which measured both source strength and resulting surface water concentrations.

Nonetheless, the data on water concentrations of dioxin-like compounds does indicate that occurrences of PCDDs and PCDFs are generally not-detected or in the low pg/L (ppq) range; detection limits were generally at or near 1 pg/L. The one exception to this is occurrences in tens to hundreds of pg/L range for PCDFs in one of twenty community water systems sampled in New York (Meyer, et al., 1989). Concentrations exceeding 200 pg/L were found in the hepta- and octa-CDFs; concentrations between 2 and 85 pg/L were found in the tetra to hexa-CDFs for this impacted water system.

The highest water concentration estimated in the demonstration scenarios in this assessment was 0.2 pg/L for the off-site soil demonstration scenario #6, for 2,3,7,8-TCDD. Soil concentrations were 1 ppb in the bounded area of soil contamination in this scenarios. Water concentrations for 2,3,4,7,8-PCDF and 2,3,3',4,4',5,5'-HPCB were both 0.1 and 0.02 pg/L for this scenario, respectively. For Scenarios 1 and 2, where watershed soil concentrations were set at 1 ng/kg (ppt), surface water concentrations were roughly one order of magnitude lower at 10-2 pg/L (ppq) range for the example TCDD and PCDF, and 10-3 ppq for the example PCB. For Scenarios 4 and 5 demonstrating stack emission depositions and where watershed soil concentrations were estimated to be in the 10-4 ppt range, surface water concentrations were the in the 10-6 ppq range. For Scenario 6, which demonstrated the effluent discharge source category, surface water concentrations were comparable to the on-site soil demonstration scenarios.