concentration results. Pulp and paper mills with and without chlorine were listed as point sources for 125 episodes (an episode is a fish sampling site). In 37 of these episodes, other point sources were identified, including one or more of the following: refinery (refinery using the catalytic reforming process), NPL site (a Superfund site), or other industry (an industrial discharge other than a paper mill or refinery). Given other sources, it is in fact a benefit to the exercise that predictions were lower than observations.

3. The model more closely predicts fish concentrations for the smaller receiving water bodies when the BSSAF is calibrated from 0.09 to 0.20. Considering that 0.09 was a value for 2,3,7,8-TCDD developed with data from Lake Ontario, a standing water body with principally historic and not ongoing 2,3,7,8-TCDD impacts, this setting is probably an inappropriate surrogate for ongoing discharges to a riverine situation. This would argue that a calibration is warranted.

The volatilization and near-field dispersion models of the soil source category were developed from well established theoretical principals, and were developed as part of an effort to assess the impact of soils contaminated with PCBs (Hwang, et al., 1986). The virtual point source dispersion model for far-field dispersion estimates is also based on well established theory (Turner, 1970). However, these models have not been field validated for soils contaminated with dioxin-like compounds. Ideally, the algorithms for estimating air concentrations would be validated using data on soil concentrations of dioxin-like compounds and concurrently measured concentrations in air above and downwind of the contaminated soil. A discussion of the Gaussian plume dispersion theories used in the COMPDEP model is given in Chapter 3, Section 3.4.1.

Some sense of the reasonableness of model values can be made by comparing predictions to measured concentrations in ambient air in urban environments. Reports of such concentrations are summarized in Section 4.7, Chapter 4 of Volume II, and Tables B.14-B.16, Appendix B of Volume II. Sources other than soil are likely to be the cause of levels measured in urban air environments. Also, the dispersion model is designed for situations where the contaminated soil is surrounded by relatively clean soil. As discussed in Section 7.2.3.1 above, off-site impacts have been noted in several sites of 2,3,7,8-TCDD contamination. These two points are made in order to establish a basis for comparing urban air concentration to concentrations predicted to occur from soils: one would expect urban air concentrations to be at least higher, if not higher by orders of magnitude, than soil emissions.

Tables B.14 and B.15 (Appendix B, Volume II) summarize average PCDD/PCDF congener-specific concentrations in urban air in the United States and in Europe. Results for two example compounds demonstrated in Chapter 5, 2,3,7,8-TCDD and 2,3,4,7,8-PCDF, are examined in this section. Observed concentrations of 2,3,7,8-TCDD were mostly non-detects with detection limits ranging from 0.01 to as high as 0.82 pg/m³. Occurrences were noted as high as 0.05 pg/m³ in Bridgeport, CT, and 0.004 pg/m³ in Wallingford, CT (both measurements as part of a study evaluating the impact of resource recovery facilities) and 0.06-0.08 pg/m³ in urban settings in Hamburg, Germany. In Stockholm, Sweden, occurrences in suburban, remote countryside, and coastal settings were listed at 0.0007, 0.0002, and 0.0001 pg/m³ respectively. Concentrations of 2,3,4,7,8-PCDF were detected in several reported studies. The range of 2,3,4,7,8-PCDF detections was 0.001-1.92 pg/m³. In the few reports where both compounds were detected, 2,3,4,7,8-PCDF was detected at 3 to 10 times higher concentration than 2,3,7,8-TCDD.

Eight air monitoring studies in the United States were used to arrive at a profile of air concentrations used for estimating background exposures to dioxin-like compounds through inhalation. These references were characterized as mostly urban and suburban, not background or rural. A summary of this compilation is in Volume II, Chapter 4, Section 4.7. and in Volume II Appendix Tables B-28 and B-29. The arithmetic mean concentrations (used for background exposure estimation) for 2,3,7,8-TCDD and 2,3,4,7,8-PCDF were 0.01 and 0.03 pg/m³, respectively. Section 7.2.3.9. below discusses the use of this compilation to craft a profile of air concentrations that might be typical of rural, background settings where cattle are raised. Evidence suggests that urban air concentrations are 4-6 times higher than rural air concentrations. If so, than 2,3,7,8-TCDD and 2,3,4,7,8-PCDF concentrations in a rural environment might 0.002 and 0.006 pg/m³.

The on-site source category was demonstrated using concentrations of 1.0 ng/kg (ppt) for each example compound. This low concentration was assigned based on reports by researchers who measured concentrations of dioxin-like compounds in what they described as "background" and "rural" soils - they found non-detects to concentrations in the low ppt level. Modeled air concentrations of the example compounds 2,3,7,8-TCDD and 2,3,4,7,8-PCDF resulting from this level in soil were in the 10^-5 pg/m³ range. For the stack emission source category demonstration, total (vapor + particle phases) concentrations of 2,3,7,8-TCDD simulated to arrive at points between 0.2 and 50 km were between 10^-7 to 10^-6 pg/m³. The off-site soil source category evaluated the impact of elevated soil concentrations to exposure sites that were located distant from the site of contamination. The example scenario demonstrating this source category had concentrations of this dioxin and furan congener set at 1 ppb, three orders of magnitude higher than the 1 ppt of the on-site source category demonstration scenarios. Air concentrations predicted in these example scenarios were in the 10^-3 pg/m³ range.

Only this air concentration from the off-site soil contamination is generally in line with urban air concentrations of 2,3,7,8-TCDD, and/or a hypothesized rural air environment. It is at least plausible that elevated concentrations in soil would result in air concentrations that are in the same range as found in urban environments. A model result that would have questioned the model validity would have been, for example, that air concentrations resulting from soils of high concentrations would greatly exceed, or be very much lower, than urban air concentrations. In the same vein, it is certainly reasonable that air concentrations resulting from a single stack emission with generally a low release rate of 2,3,7,8-TCDD should be much lower than urban air concentrations.

It is not that clear that emissions and resulting air concentrations above soils at background levels should be lower by up to 2 orders of magnitude lower than what is hypothesized to occur in background setting. The argument has been made in Volumes I and II of this assessment that emissions from tall industrial stacks, followed by long range transport, are the ultimate source of these compounds in rural environments where the food supply is produced. The question remains as to how much of the contaminant in rural air is due to annual emissions and long range transport versus emissions from the soil reservoir source. If the modeling of this assessment is correct, than soils contribute very little to rural air concentrations. However, other evidence developed in this assessment suggests that the soil release and dispersion algorithms of this assessment may be underestimating air concentrations. One piece of that evidence is discussed in the next section below. Plant/soil ratios, defined as the ratio of 2,3,7,8-TCDD concentration in plants divided by that in the soil, were found to be lower in model predictions as compared to literature values. Two possible hypotheses were offered below: 1) the model is underpredicting air concentrations resulting soil releases, and/or 2) plant:soil ratios derived in experiments are not only the result of soil related impacts, but also from distant sources of air-borne release and long range transport - i.e., the air reservoir is not solely explained by soil releases. One other possibility would be that the algorithms estimating air to plant transfers are not valid and estimating too low a transfer rate. However, the air to plant transfers algorithms were examined in the section further below, Section 7.2.3.9, describing an air-to-beef food chain validation exercise. There, air to plant transfers onto a leafy hay crop were examined with data and model was predicting hay concentrations right in line with observations.

In summary, three pieces of evidence suggest that the soil to air models, and/or the parameters values selected for this model, may be underestimating air concentrations. One is the comparison of predicted air concentrations for a background soil compared against air concentration data described above. The second is developed below where plant:soil ratios predicted by the model appear lower than measured under experimental conditions. Third, air-to-plant transfers appear to test well, leaving the soil-to-air algorithms questionable for predicting low plant:soil ratios.

There have been several studies which have measured plant concentrations of 2,3,7,8-TCDD for plants grown in soils with known concentrations of 2,3,7,8-TCDD, and more recently, studies with plant and soil concentrations for dioxin toxic equivalents or dioxin congener groups. One quantity that can be estimated from these studies is a plant:soil contaminant concentration ratio. The plant:soil ratio equals the concentration in the plant divided by the concentration in soil in which the plant is growing. Concentration ratios predicted to have occurred can be compared against those that have been measured in the various studies.

These ratio comparisons can be considered model validations, although none of the experimental or field conditions for the literature studies were duplicated in this exercise. The literature articles measuring soil and resulting plant concentrations of dioxin-like compounds are summarized in Table 7-6. This table also includes concentration ratios,

Table 7-6. Summary of plant concentration versus soil concentration data for 2,3,7,8-TCDD.

Plant/Soil Contaminant

Concentrations Ratio Reference and Comments

I. Below-Ground Vegetations

54-167 ppt/ .01-.17 Wipf, et al., 1982; results are for 2,3,7,8-TCDD and greenhouse carrots grown in Seveso

1-5 ppb contaminated soil; the 54 ppt concentration listed was for carrot peels and inner portions; the 167 ppt listed includes the 54 ppt plus additional residues found in wash water and can be described as "unwashed" concentration; 96% of 167 ppt unwashed concentration includes that found in wash water (67%) and peels (29%).

0.8-9.2 ppb/ .24-1.73 Coccusi, et al., 1979; results are for 2,3,7,8-TCDD and carrots, potatoes, narcissus, and onions

2.7-8.3 ppb grown on contaminated soil the spring following the Seveso contamination; aerial plant part ratios were 0.25-0.40 - underground part ratios were 0.23-1.73; residues in contaminated plants were found to dissipate when contaminated plants transplanted to unpolluted soils; results show higher ratios than the Wipf, et al. (1982) noted above; results were expressed in fresh plant weight and fresh soil basis; very high ratios and plant impacts render these data suspect.

156-1807 ppt/ 1.00-2.40 Facchetti, et al., 1986; results are for 2,3,7,8-TCDD and bean and maize roots grown in indoor

160-752 ppt greenhouse pots and outdoor pots; unclear whether plant concentrations are fresh or dry weights. Data considered highly suspect due to very high ratios found and also reporting 16 and 37 ppt in roots when "blank" soil had 1.5 ppt (ratios of 10.7 and 24.7).

735 ppt/ 1.8 Young, 1983; results are for 2,3,7,8-TCDD and roots of grass and broadleaf plants at Eglin Air

411 ppt Force Base; unclear whether root concentrations are fresh or dry weight.

0.5-40.2 ppt/ .001-.3 Hulster and Marschner, 1991; results at right are for unpeeled potato tubers, in TEQ and dry

2-6000 ppt weight basis. Plant:soil ratio decreased as soil concentrations increased; highest ratios were at the 2.4 ppt low soil concentration. Peeled tuber concentration stayed below 0.5 ppt over all soil concentrations, indicating insignificant within plant translocation. Plant concentrations given in dry weight basis.

(continued on next page)

Table 7-6. (cont'd)

Plant/Soil Contaminant

Concentrations Ratio Reference and Comments

0.2-6.0 ppt/ .00001-.009 Hulster and Marschner, 1993a. Results are for potato tubers, peeled and unpeeled, and for potato

328-12,800 ppt shoots, results for TEQ and in dry matter terms. Concentrations for peeled potato tubers stayed consistently less than 0.5 ppt, despite soil concentrations, while shoots and unpeeled tubers increased as concentration increased. Plant:soil ratios remained relatively constant for tubers and shoots with soil concentration increases, leading authors to conclude that a soil/plant relationship exists for plants growing in the soil. Less transfer was noted for higher chlorination.

0.35, 0.96/ 0.02-0.07 Muller, et al., 1993a. Two plant/soil concentrations are for carrots in soil concentrations of 5 and

5, 56 ppt 56 ppt TEQ; carrot concentrations in dry matter and TEQ terms. Ratios decrease as concentrations increase; most of the concentration was in the peels.

 

 

II. Above-Ground Vegetations

(9-42 ppt)/ .0009-.0042 Wipf, et al., 1982; analysis of apples, pears, plums, figs, peaches, and apricots grown in

10 ppb Seveso, Italy year following contamination; apples, pears, and peaches showed >95% of whole fruit concentrations listed here was in the peels; analysis of vegetative samples in less contaminated areas showed non-detections at 1 ppt detection limit; reference was unclear as to whether reported concentrations in fruit was based on fresh or dry weight.

(8-9 ppt)/ .0008 Wipf, et al., 1982; concentrations listed were those found in sheaths of corn grown year following

10 ppb following Seveso contamination; none found in cobs and kernels at 1 ppt detection limit.

 

 

 

 

(continued on next page)

Table 7-6. (cont'd)

Plant/Soil Contaminant

Concentrations Ratio Reference and Comments

(1-63 ppt)/ 0.003-0.35 Sacchi, et al., 1986; data was for: "aerial parts" of bean and maize plants, tritiated TCDD

(12-3300 ppt) TCDD amended soil with concentrations ranging as noted, taken at different intervals including 7, 34 and 57 days (one test), 17, 34, and 57 days (another test), 8 and 77 days, and 8 and 49 days, and in tests where soil was and was not amended with peat. Results showed increasing plant concentrations with increasing soil concentrations, but the ratio of plant to soil concentrations was inversely related to increasing soil concentrations (lowest ratios at highest soil concentrations). Soils without peat had higher ratios than soils with peat. Plant concentrations were fresh weight basis; high plant impact and trend for increasing impact over time renders these results suspect.

ND (DL=1 ppb)/ <0.017 Isensee and Jones, 1971; results are for mature oat and soybean tops, and oat grain and the

60 ppb bean of soybean, in soil treated with [¹⁴C]TCDD to achieve a concentration of 60 ppb - no residues of TCDD were found; ratios of 0.14 and 0.28 were found for 2,4,-dichlorophenol (DCP) in oat and soybean tops, and 0.20 for 2,7-dichlorodibenzo-p-dioxin (DCDD) in oat tops; trace amounts of DCP and DCDD were found in the bean of soybean.

(10-270 ppt)/ .02-0.66 Young, 1983; data was for 2,3,7,8-TCDD and above ground plant parts of perennial grasses and

411 ppt broadleaf plants grown on 2,4,5,-T treated soils. Unclear whether plant concentrations are fresh or dry weight basis. Soil concentration was average over 3 depth increments to 15 cm. Crown near soil surface at 270 ppt and 0.66 ratio was highest; plant tops had ratios of 0.02-0.17.

0.3, 0.1 ppt/ 0.00003, Muller, et al, 1993a. Result at right are for whole pear (0.3) and whole apple (0.1) dry weight

8750, 5215 ppt 0.00002 concentrations (article presented TEQs for two pears from one tree which were averaged, and one apple, and for fresh weight; dry weight was estimated assuming 12% dry matter in pears/apples) and the average concentration over 70 cm (article supplied concentrations for the 0-30 and 30-70 cm depths). Article also provided peel and pulp results and results for congener groups. Article concluded: soil levels were not correlated to fruit concentrations and therefore fruits were impacted by airborne contamination, and that concentrations were higher in peel than in pulp.

(continued on next page)

Table 7-6. (cont'd)

Plant/Soil Contaminant

Concentrations Ratio Reference and Comments

0.1-0.6 ppt/ 0.00002- Hulster and Marschner (1993a). Results are for inner and outer leaves of lettuce, expressed as dry

326-5752 ppt 0.0008 matter, and in TEQs. Results indicate a drop in ratio as soil concentration increases, and unexpected small differences between inner and outer leaves.

4-38 ppt/ .001- Hulster and Marschner (1993a). Results are for hay, dry matter, and TEQs. Results indicate a drop

326-12,800 ppt .01 in ratio as soil concentrations increase.

<1 ppt/ .0001- Hulster and Marschner (1993a). Results are for grass and herbs, dry matter, and TEQs. Results

326-5752 ppt .0003 indicate a drop in ratio as soil concentrations increase. For above three entries, results are also given for congener groups. Authors conclude that: little correlation between soil and above ground plant concentrations, and that contamination is by atmospheric deposition.

<.01, .04/ <0.002 Muller, et al., 1993b. Results are for peas at soil concentrations of 5 and 56 ppt; pea

5, 56 ppt concentrations in TEQ and dry weight. Results for pods indicated more impact with ratios at 0.002-0.026. Ratios decreased as soil concentration increased.

0.32, 0.21 ppt/ .004- Muller, et al., 1993b. Results are for lettuce at soil concentrations of 5 and 56 ppt; lettuce

5, 56 ppt .064 concentrations in TEQ and dry weight. Little difference seen between inner and outer leaves, which was unexpected - outer leaves expected to be more impacted. Ratios decreased as soil concentration increased.

0.5-22.6 ppt/ .14- Hulster and Marschner, 1993b. Results are for zucchini fruit at two soil concentrations of 0.4 and

0.4, 148 ppt 2.5 148 ppt TEQ, fruit results are TEQ and dry weight. Results contradict conventional wisdom that above ground vegetation impact is from air only and mainly an outer surface phenomena; zucchini contamination was uniform throughout plant and plant:soil ratios highest ever found for above ground bulky fruits.

(continued on next page)

Table 7-6. (cont'd)

Plant/Soil Contaminant

Concentrations Ratio Reference and Comments

0.6 ppt/148 ppt .004 Hulster and Marschner, 1993b. Results are for cucumber grown in soil at 147 ppt TEQ; cucumber results in TEQ and dry weight. Results are more in line with most other studies for above ground bulky fruit plant:soil ratios.

7.5 ppt/148 ppt .05 Hulster and Marschner, 1993b. Results are for pumpkin grown in soil at 148 ppt TEQ; pumpkin results in TEQ and dry weight. Results not as dramatic as for zucchini, but plant concentrations are ratio are still high.

0.4-1.9 ppt/ .0003- Hulster and Marschner, 1991. Results are for lettuce, in TEQ and dry weight. Experiments were

2.4-6000 ppt .3 conducted outdoors with soil covered by a water permeable polypropylene fleece. Plant concentrations showed little variation with large increases in soil concentration, and given the soil covering, this would strongly indicate little root to shoot translocation and that lettuce concentrations were the result of air to plant transfer.s

 

and separates sections for above and below ground vegetations.

In measuring both the soil and the plant concentration, several of the early literature articles, particularly those from Seveso (Wipf, et al., 1982; and Coccusi, et al., 1979) presumed that the soil in which the plant was growing was the ultimate source for the 2,3,7,8-TCDD contamination of above ground plant parts, if not from direct uptake than from deposition of suspended particles. However, recent research has concluded that the contamination of above ground plant parts is due principally to air-to-plant transfers (Hulster and Marschner, 1993a; Muller, et al, 1993a; Muller, et al., 1993b; Welsh-Paush, et al (1993); and others). These cited research efforts have concluded that there is no consistent relationship between soil concentrations of dioxin-like compounds and above ground vegetative concentrations of these compounds, which has led the researchers to conclude that air-to-plant transfers explain plant concentrations (a recent report did strongly imply a direct soil/plant for dioxin-like compounds for at least one family of above ground vegetables, the cucumber family (Hulster and Marschner, 1993b); this will be discussed below). This fact, coupled with the fact that sources of airborne contamination by dioxins include both distant sources and soil releases, make it difficult to compare literature reports of plant:soil contamination concentrations with those predicted by the soil contamination modeling of this assessment.

Recall that the "on-site" soil source modeling presumes that air concentrations and depositions to which the plant are exposed originate only from the soil in which the plant is growing. One would expect that the modeled plant:soil ratio for above ground plant parts would be lower than plant:soil ratios measured in field settings, since the field measured ratios are influenced by more than just the soil releases into the air.

On the other hand, the literature is consistent in concluding that soil provides the source for underground soil to root transfers. For this reason, Table 7-6 and the following discussions distinguish between above and below-ground vegetations.

The following plant:soil contaminant concentration ratios were estimated for the two scenarios demonstrating the on-site source category in Chapter 5, Scenarios 1 and 2: below ground vegetables - 7x10^-3 (dry weight basis, assuming vegetables are 15% dry matter), above ground vegetables/fruit - 7x10^-5 (dry weight basis, assuming vegetables/fruits are 15% dry matter), grass - 6x10^-3 (dry weight), and feed 3x10^-3 (dry weight). Some observations from experimental results found in the literature, and comparison with the results of the model, are:

1) The largest body of consistently developed experimental data on soil-plant relationships of dioxin-like compounds comes from a research group in Germany who have published numerous articles for different vegetations and experimental conditions in the 1990s (Hulster and Marschner, 1991; Hulster and Marschner, 1993a,b; Muller, et al., 1993a,b). Some of the earlier literature showed much higher impacts to vegetations than measured by these German researchers (Coccusi, et al., 1979; Facchetti, et al., 1986; Young, 1983), which in the judgement of the authors of this EPA assessment, renders them suspect. One early report, that of Wipf, et al. (1982), does show results consistent with the German research. The observations following will focus mainly on this research from Germany.

2) Experimental results for both above and below ground vegetations suggest that plant:soil ratios decrease as soil concentration increases. For below ground vegetations, this suggests that the movement into plants is not a passive and unimpeded process occurring with transpiration water, for if it were, plant:soil ratios would be constant as concentration increases. For above ground vegetations, the observations given above that air-to-plant transfers and not soil-to-plant transfers better explain plant concentrations, and that air concentrations include soil releases as well as long term transport, leads one to conclude that a consistent relationship between soil concentrations and plant concentrations is not to be expected. An explanation for this trend for below ground vegetative trends could not be found.

The models of this assessment - soil to below ground vegetation, soil to air to above ground vegetation, and air to above ground vegetation - cannot duplicate these observed trends, that is, the models will not show a decrease in plant:soil ratios as soil concentration increases. Above and below ground vegetation concentrations are a linear function of a biotransfer factor and an appropriate media concentration - air, soil water. For particle depositions, no transfer parameters are used, but plant concentrations are a linear function of model inputs, including deposition rates, plant interceptions and yield, and a plant washoff factor. Therefore, plant concentrations will be a linear function of soil concentrations for the soil source categories.

3) Plant:soil ratios for below ground vegetables for soil concentrations in the low ppt range would appear to be in the 10^-1 to 10^-2 range (Muller, et al, 1993; Hulster and Marschner, 1991), in contrast to the 0.007 predicted by the model. Much higher ratios were found in the earlier studies (Coccusi, et al., 1979; Facchetti, et al., 1986; Yount, 1983), which earlier had been speculated as being questionable. One earlier study, that of Wipf, et al. (1982), does report ratios similar to these later studies, as noted above. At higher soil concentrations in the sub to low ppb range, plant soil ratios are more in the 10^-4 to 10^-3 range (Hulster and Marschner, 1993a; Hulster and Marschner, 1991), even lower than the modeled 0.007 ratio.

4) The results for above ground bulky vegetations, fruits and vegetables, indicate plant:soil ratios that are lower than plant:soil ratios for bulky below ground vegetations, for comparable soil concentrations. The evidence for this observation is best found in the Hulster and Marschner (1993a) concurrent experiments for potatoes and pears/apples, as well as the earlier work of Wipf, et al. (1982) for several fruits and carrots. The same trend is also found for the grass results for 2,3,7,8-TCDD given in Young (1983). This trend is also duplicated by the models, which showed two orders of magnitude difference in below ground as compared to above ground vegetations. The plant:soil modeled ratio of 7*10^-5 is similar to ratios found when the soil concentration was in the hundreds to thousands of ppt (Hulster and Marschner, 1993a). However, other data, particularly for leafy vegetations such as hay, grass, and lettuce, and for lower soil concentrations, indicate a soil:plant ratio of 10^-3 to 10^-2. Two possible explanations are offered for this trend: 1) above ground vegetations in experiments are likely to be impacted by not only soil releases, but distant sources of release, and/or 2) the models could be underpredicting air concentrations resulting from soil releases.

4) Several of the articles, both from the German work and the earlier work, noted that most of the concentration was in the outer portions of the below and above-ground vegetations, and not the inner portions. Despite significant increases in soil concentration from the ppt to the ppb range, inner potato tuber concentrations remained constant (Hulster and Marschner, 1991, 1993a). This evidence was the principal justification for the use of the empirical adjustment factors termed VG for soil to below ground transfers, VG_bg, and vapor-phase air transfers to bulky above ground vegetations, VG_ag. The chemical-specific empirical transfers factors for both of these transfers were developed in laboratory experiments with several chemicals using thin vegetations - solution phase transfers to barley roots for below ground vegetation concentrations, and vapor phase transfers to azalea leaves for vapor phase transfers. For the dioxin-like compounds, direct use of these transfer factors would be most appropriate for the outer few millimeters, perhaps, of below and above ground bulky vegetations. The assignment of a VG of 0.01 for bulky above and below ground vegetations was based on an outer surface volume to whole plant volume ratio for a common vegetation such as a carrot or an apple. A VG of 1.00 was used for grass, since that is a thin vegetation.

Further evidence for the above ground VG came from a recent study by McCrady (1994), who measured the uptake rate constants of vapor-phase 2,3,7,8-TCDD to several vegetations including grass and azalea leaves, kale, pepper, spruce needles, apple, and tomato. The uptake rate for the apple divided by the uptake rate for the grass leaf was 0.02 (where uptake rates were from air to whole vegetation on a dry weight basis). For the tomato and pepper, the same ratios were 0.03 and 0.08. The VG_ag was 0.01 for fruits and vegetables in this assessment. McCrady (1994) then went on to normalize his uptake rates on a surface area basis instead of a mass basis; i.e., air to vegetative surface area uptake rate instead of an air to vegetative mass uptake rate. Then, the uptake rates were substantially more similar, with the ratio of the apple uptake rate to the grass being 1.6 instead of 0.02; i.e., the apple uptake rate was 1.6 times higher than that of grass, instead of 1/50 as much when estimated on an air to dry weight mass basis. The ratios for tomato and pepper were 1.2 and 2.2, respectively. Therefore, since the B_vpa in this assessment is an air to plant mass transfer, the McCrady experiments would appear to justify the use of an above-ground VG of a magnitude less than 0.10.

5) A recent experiment by the Hulster and Marschner (1993b) on vegetations of the cucumber family contradicted the conventional wisdom that direct soil to root to above ground plant impact would not occur for the dioxin-like compounds. Their results were most striking for zucchini, which showed uniform plant concentrations from inner to outer portions of the zucchini fruit, and the highest whole fruit concentrations and plant:soil ratios they had ever measured, despite careful experimental conditions which physically isolated the fruit from the soil. Pumpkins also showed high plant contamination and plant:soil ratios, with more expected plant concentrations measured for the cucumber. No explanation was offered for these results. It was assumed for this exposure assessment that the fruits and vegetables for human consumption, and the grasses, hay, and other vegetations animals consume, would not follow this pattern.

A principal conclusion that can be drawn from this examination is that the plant:soil contaminant concentration ratios developed by the soil contamination models of this assessment may be lower by perhaps an order of magnitude or more than measured ratios at lower soil concentrations, in the low ppt range, whereas they may be more in line and even higher when soil concentrations are the hundreds of ppt to the ppb range. This trend appears to hold for both above and below ground vegetations. This difference in the comparison of modeled and observed ratios as the concentration changes is because the data shows that plant:soil ratios decrease as soil concentrations increase. This cannot be duplicated by the model since the plant concentrations are a linear function of the source strength terms - the soil, soil water, or air concentrations and deposition. An explanation for this observed trend could not be found. The observation that plant:soil ratios for above ground vegetations are higher in the literature at lower soil concentrations (and more typical of background rather than heavily contaminated soils) as compared to the modeled ratios, has to be carefully considered. Two explanations are offered. For experiments conducted outdoors, the source of air reservoirs of dioxin-like compounds are the soil in which the plant is growing as well as from distant sources and long-term transport. Also, it is possible that the model is underpredicting air concentrations and hence underpredicting air to plant transfers.

The premise of this modeling exercise to test the beef food chain model for dioxin-like compounds is that air-borne reservoirs of these compounds in rural environments are the "source term" explaining concentrations found in beef. Further, this exercise probably would not qualify as a validation exercise in the traditional sense. Most environmental model validation exercises rely on data obtained from a single site. This exercise instead develops a representative rural air concentration profile and attempts to model a profile of average beef concentrations.

The model structure, from air to beef, is shown in Figure 7-3. The algorithms for these components, and assignment of model parameters, were described in Chapter 4, and are very briefly summarized here. The "observed" source, or independent, term in this modeling exercise are the air-borne concentrations of dioxin-like compounds shown at the top of Figure 7-3, and the "predicted", or dependent, results are the concentrations in

whole beef shown at the bottom of this figure. Both these quantities are developed from reported United States measurements. Section 7.2.3.9.1 below describes the generation of these concentration profiles. Section 7.2.3.9.2 summarizes model algorithms and parameter assignments. Section 7.2.3.9.3. presents the results and discussions from this exercise.

7.2.3.9.1. Air and beef concentrations

Very little data are available worldwide on air concentrations of individual dioxin-like congeners in a rural setting. This is the kind of air concentration data that would be needed for this exercise. An evaluation of ambient air monitoring studies in the United States conducted for Volume II of this assessment showed that nearly all of the data was from urban or suburban settings. The purpose of this compilation was to determine an ambient air concentration suitable for estimating inhalation exposures to dioxin-like compounds. Measurements which were attributed to a nearby identifiable source, such as an incinerator, were not considered for this effort. From several studies around the country, a total of 84 air samples were available, from which a mean TEQ level of 0.095 pg/m³ was determined. Further detail on this compilation can be found in Chapter 4 of Volume II.

There are a few references which do have congener-specific data which might be characterized as rural. One is outside of United States in Sweden (Broman, et al., 1991). Air samples were taken in four areas, ranging from the Stockholm urban area to the open coastal area of the Baltic Sea. Results indicate lower TEQs when going from the urbanized area to the remote areas. The Stockholm city center was 0.024 pg TEQ/m³, a "suburb" was 0.013 pg TEQ/m³, a "countryside remote" area was 0.0044 pg TEQ/m³, and an "open coastal" area was 0.0026 pg TEQ/m³. Twenty-five PCDD/F concentrations were listed at the fg/m³ level (i.e., 0.001 pg/m³).

The only reference found for the United States with congener specific data for an area described as rural was from Ohio (Edgarton, et al., 1989). Six sites were tested, one of which might be considered rural. The data contained many non-detects, with detection limits between 0.033 to 0.82 pg/m³, although most non-detects had detection limits less than 0.3 pg/m³. The following TEQ concentrations were derived only from the positive listings: two sites in Akron - 0.077 and 0.079 pg TEQ/m³, two sites in Columbus - 0.092 and 0.179 pg TEQ/m³, a site near a highway - 0.065 pg TEQ/m³, and a rural site in a town called Waldo - 0.045 pg TEQ/m³. Like the data from Sweden, one can see a trend for lower concentrations in the Waldo site as compared to the sites in Columbus and Akron.

Other references did contain other pertinent data, such as total concentrations, TEQ concentrations, or congener group concentrations, in rural and urban settings. Eitzer and Hites (1989) took data from Bloomington, Indiana and a remote area in Wisconsin known as Trout Lake. TEQ concentrations were not given, but total congener group concentrations were reported. The sum of congener group concentrations, or total concentrations of dioxins and furans, equaled 2.2 pg/m³ for Bloomington, and 0.51 pg/m³ for Trout Lake. This 0.51 pg/m³ total concentration is similar to the total concentration found in the "countryside remote" area in Sweden discussed above, which is 0.41 pg/m³ (TEQ concentration was 0.0044 pg/m³, as noted above).

In an evaluation of air, soil, sediment, and fish in Elk River, Minnesota, a rural setting, again total congener concentrations in the air were reported (Reed, et al., 1990). Concentrations for three sites and for two sampling dates, one in the winter and one in the summer, were available. Two of the three sites were in rural settings and the third was near a refuse derived fuel incinerator. Total concentrations for the two rural sites in winter and in summer were 2.29 and 2.91 pg/m³ in winter, and 0.58 and 0.38 pg/m³ in summer. For the third site near the incinerator, winter and summer concentrations were 15.2 and 0.35 pg/m³, respectively. The average of the four data points for rural settings was 1.54 pg/m³, while the average of the two data points near the incinerator was 7.78 pg/m³.

Finally, Maisel and Hunt (1990) list TEQ concentrations only for monitoring networks including: a Connecticut coastal location described as urban (measurements described as "wintertime"), a southern California urban setting ("annualized"), and a central Minnesota rural setting ("annualized"). While not identifying it as such, this central Minnesota setting could be the one described above in Elk River, Minnesota. The TEQ concentrations for the two urban and one rural setting were: 0.092, 0.091, and 0.021 pg TEQ/m³.

Key points from this literature summary are:

1. Congener specific profiles for rural settings in the United States are generally not available. Based on several studies encompassing 84 data points with specific congener concentrations which best represent urban/suburban settings, but are not near identified emission sources, a mean TEQ air concentration of 0.095 pg/m³ is estimated.

2. Studies are available which do provide side by side data on urban and rural settings, although the literature references only list congener group concentrations or total TEQ concentrations (with the exception of the Edgarton, et al. (1989) described above). What this summary shows is that rural air concentrations of dioxin-like compounds appear to be 4-6 times lower than in urban settings, and that a TEQ concentration for rural settings appear to range from 0.004 to 0.04 pg/m³.

In order to develop a profile of air concentrations that will be considered representative of rural settings, what will be done, therefore, is to take the profile of congener-specific air concentrations for urban/suburban settings leading to a TEQ concentration of 0.095 pg/m³, and divide each concentration by 5. The resulting TEQ concentration is 0.019 pg/m³. The total concentration of PCDD/Fs in this rural profile equals 1.09 pg/m³. A uniform division by five for all congeners essentially assumes that the ultimate sources for an urban and a rural profile of air concentrations are the same. The specific concentrations used are shown in Table 7-7.

A review of data on concentrations of dioxin-like compounds in beef showed that very limited data was available worldwide, much less United States. Only three studies contained congener-specific data of dioxins and furans in beef in United States. In one study beef samples were composited with veal and the results described as beef/veal. The three studies only encompassed 14 samples. These studies include one conducted by the California Air Resources Board (CARB; Stanley and Bauer, 1989), the results of background analysis from a study conducted by the National Coalition for Air and Stream Improvement (NCASI; the study described in Lafleur, et al., 1990) and a survey of foods conducted in New York by Schecter et al. (1993).

These were the data used to estimate background exposures to dioxins in beef in Chapter 5 of Volume II. The total TEQ for beef and veal was calculated by using one-half the detection limits reported by the researchers to represent the concentration of nondetectable CDD/F congeners in the samples. Using this methodology, the TEQ concentration was estimated to be 0.48 ng/kg (ppt) for beef and veal on a wet weight basis. If nondetectable concentrations are assumed to be zero, the estimated TEQ for beef and veal is 0.29 ppt. The average whole beef congener-specific concentrations assuming non-detects were one-half the detection limit are to be used to represent beef concentrations, and they are shown in Table 7-7. All studies reported concentrations as lipid-based concentrations. Where lipid fractions were not supplied, 19% lipid content for beef was assumed to estimate whole beef concentrations.

It is important to note that the United States samples came from commercial food outlets (grocery stores, e.g.). This fact will be used to imply that the data represents beef cattle that went through a feedlot fattening process prior to slaughter. As will be discussed below, this has implications regarding final concentrations.

Table 7-7. Observed air and beef concentrations, and fate parameters for individual dioxin and furan congeners.

Parameters for B_vpa Parameters for Vapor/Particle Partitioning Observed Data

Compound H log Kow B_vpa T_m, K VP_s, atm Vapor/Particle BCF air, pg/m³ beef, ppt

2378-TCDD 1.6*10^-5 6.64 1.0*10⁵ 578 9.7*10^-13 0.55/0.45 4.32 0.002 0.03

12378-PeCDD 2.6*10^-6 6.64 6.3*10⁵ 513 1.3*10^-12 0.26/0.74 4.16 0.006 0.22

123478-HxCDD 1.2*10^-5 7.79 2.3*10⁶ 547 1.3*10^-13 0.07/0.93 2.02 0.005 0.26

123789-HxCDD 1.2*10^-5 7.79 6.9*10⁵ 516 6.5*10^-14 0.02/0.98 2.24 0.007 0.84

123678-HxCDD 1.2*10^-5 7.30 6.9*10⁵ 558 4.7*10^-14 0.04/0.96 1.74 0.010 0.21

1234678-HpCDD 7.5*10^-6 8.20 1.0*10⁷ 538 4.2*10^-14 0.02/0.98 0.36 0.116 1.92

OctaCDD 7.0*10^-9 7.59 2.4*10⁹ 598 1.1*10^-15 0.00/1.00 0.52 0.586 2.91

2378-TCDF 8.6*10^-6 6.53 1.5*10⁵ 500 1.2*10^-11 0.71/0.29 0.94 0.023 0.06

23478-PeCDF 6.2*10^-6 6.92 5.3*10⁵ 469 4.3*10^-12 0.30/0.70 3.10 0.010 0.04

12378-PeCDF 6.2*10^-6 6.79 3.8*10⁵ 499 3.6*10^-12 0.42/0.58 0.73 0.006 0.21

123478-HxCDF 1.4*10^-5 7.30 5.9*10⁵ 499 3.2*10^-13 0.06/0.94 2.34 0.012 0.51

123678-HxCDF 6.1*10^-6 7.30 1.4*10⁶ 506 2.9*10^-13 0.06/0.94 2.00 0.012 0.06

123789-HxCDF 1.0*10^-5 7.30 8.3*10⁵ 520 3.7*10^-13 0.11/0.89 2.00^* 0.003 0.06

234678-HxCDF 1.0*10^-5 7.30 8.3*10⁵ 512 2.6*10^-13 0.07/0.93 1.78 0.009 0.07

1234678-HpCDF 5.3*10^-5 7.90 6.8*10⁵ 509 1.8*10^-13 0.04/0.96 0.41 0.042 0.40

1234789-HpCDF 5.3*10^-5 7.90 6.8*10⁵ 495 1.4*10^-13 0.03/0.98 0.99 0.006 0.13

OctaCDF 1.9*10^-6 8.80 1.7*10⁸ 532 4.9*10^-15 0.00/1.00 0.20 0.034 0.22

Column headings are:

H: Henry's Constant, atm-m³-mole T_m: Melting point temperature, K

log Kow: log octanol water partition coefficient Ps: Crystalline solid vapor pressure, atm^-1

B_vpa: air-to-leaf transfer factor, unitless Vapor/Particle: fraction of total reservoir in vapor and particle phases

BCF: beef biotransfer factor, unitless air: total reservoir of congener in air, pg/m³

beef: whole beef observed concentrations, ng/kg

All parameters associated with individual congeners are shown in Table 7-7, and all parameters not specific to the congeners are shown in Table 7-8. Following now are summaries of the algorithms and key assumptions of this exercise. Many of them have been described in earlier chapters of this Volume and Volume II, and are not repeated here.

As shown in Figure 7-3, this is the first key step in this modeling exercise. Chapter 3 of this volume described air monitoring studies which reported the partitioning of dioxins into a particle and a vapor phase. Arguments were presented as to why these studies would likely overestimate the portion in the vapor phase. Because of this, a theoretical model for estimating the fraction of total concentration in the particulate and vapor phases was recommended for use in this assessment. The model of Bidleman (1988) was presented and discussed in Chapter 3, and will be used here as well. Table 7-7 presents the vapor and particle fractions assumed in this assessment, based on the Bidleman model.

Chapter 4 described the wet and dry particle deposition algorithm used for this assessment. The dry deposition algorithm and the key parameter assignment of a dry deposition velocity of 0.2 cm/sec will be used for this exercise without change. However, the wet deposition algorithm described in that chapter includes assignment of an annual rainfall amount with a washout factor. This is more appropriate for a site-specific application, and because this exercise is based on a "representative" rural profile of air concentrations and an average beef concentration profile derived from three locations in the United States, a simplification of the wet deposition algorithm is used in this exercise. This simplifications is based on the measurements made by Koester and Hites (1992). They measured wet deposition of total dioxins at two sites in Indianapolis and Bloomington, Indiana, and generally found wet deposition to be comparable to dry deposition. Specifically, the estimated annual wet deposition of dioxins at Indianapolis was equal to 0.7 times dry deposition, while at Bloomington, wet deposition was 1.3 times dry deposition. Therefore, it will be assumed that wet deposition equals dry deposition in this exercise. Crop yields and interceptions which were used for the demonstration scenarios of Chapter 5 are used for the deposition algorithms here as well.

Table 7-8. Model parameters used for all dioxin-like congeners.

Parameter Description Value

I. For Vapor/Particle Partitioning

C constant to estimate sorbed fraction

in Equation (?), atm-cm 1.7*10^-4

T ambient air temperature, ° K 298.1

D S_f/R entropy of fusion/universal gas constant, unitless 6.79

S_T average total surface area of aerosol particles

relative to average total volume of air, cm²/cm³ 3.5*10^-6

V_T average total volume of aerosol particles

per volume of air, cm³/cm³ 3*10^-11

II. Particle Depositions

k_w first-order plant weathering constant, yr^-1 18.01

k_s first-order soil dissipation constant, yr^-1 0.0693

Y_g yield of grass, kg/m² 0.15

I_g interception fraction of grass 0.35

Y_h/s yield of hay/silage/grain 0.63

I_h/s interception fraction of hay/silage/grain 0.62

V_d velocity of particle deposition, m/sec 0.002

M mass of mixing soil, kg/m² 10

R_w retention of wet deposition on vegetations, fraction 0.30

III. Vapor Transfers

VG_gr empirical correction factor for grass, unitless 1.00

VG_h/s empirical correction factor for hay/silage/grain, unitless 0.50

IV. Bioconcentration

B_s bioavailability of contaminant on the soil vehicle

relative to the vegetative vehicle, unitless 0.65

DF_s cattle soil diet fraction 0.04

DF_g cattle grass diet fraction 0.48

DF_h/s cattle hay/silage/grain diet fraction 0.48

V. Other

fat content of beef 0.19

concentration reduction due to feedlot fattening 0.50

assumption: wet deposition equals dry deposition

The soil deposition algorithm remains unchanged from the structure and parameter assignments described in Chapter 4 and demonstrated in Chapter 5.

The key parameters for this algorithm include the air-to-leaf transfer factor, the B_vpa, and the empirical adjustment parameter, VG, which reduces vapor transfers considering the difference in the thin azalea and grass leaves used in experiments to derive the B_vpa and the bulky and protected vegetations of the cattle diet, such as silages as grains. The values of these parameters are the same ones used in Chapter 5.

The bioconcentration model includes assignment of the congener-specific bioconcentration factor, BCF, and the soil bioavailability parameter, B_s. The parameter assignments for these parameters are the ones which were developed in Chapter 4, used for the demonstration scenarios of Chapter 5, and shown on Tables 7-7 and 7-8.

The final key areas in this model are the assumptions concerning cattle exposure to dioxin-like compounds through their diet. A related key issue is the impact of feedlot fattening on final beef concentrations. The general diet profile used for the demonstration scenarios for beef concentration estimations in Chapter 5 is used here as well. This included an assumption of equal proportions of pasture grass and non-grass feed such as hay, silage, or grain, and a small amount of incidental soil. As discussed in Chapter 4, a 4% soil ingestion rate was assumed, leaving 48% each for pasture grass and the second category of cattle vegetation intake, abbreviated hay/silage/grain. Chapter 4 also discussed the impact of feedlot fattening. The demonstration scenario of Chapter 5 did not include feedlot fattening since the scenario was one of a farmer home slaughtering for personal consumption. For this exercise, however, it is likely that the commercial beef samples from which the "observed" concentration profile was derived came from cattle which had undergone a period of feedlot fattening. Chapter 4 summarized modeling efforts which attempted to characterize the impact of a period of fattening assuming residue-free intake for a period of 120 days. Based on their results, these modeling efforts hypothesized that such a diet regime would reduce fat concentrations by one-half. This will be the assumption used here as well; beef concentrations estimated using all the modeling described above will be halved as a final step in the modeling process.

A final comparison of predicted versus observed whole beef concentrations is shown in Table 7-9. Total TEQ concentrations compare favorably, with observed total TEQ at 0.48 ppt and predicted TEQ at 0.36 ppt. The congeners of most toxicity also had the best match of predicted and observed concentrations: 2,3,7,8-TCDD - 0.03 ppt observed and 0.03 ppt predicted; 1,2,3,7,8-PCDD - 0.22 observed and 0.27 ppt predicted; 2,3,4,7,8-PCDF - 0.21 ppt observed and 0.17 ppt predicted. The largest discrepancies, an order of magnitude and more, were for two of the HxCDDs and for all HpCDD/Fs and OCDD/Fs. The total concentrations did not compare as well as the TEQ concentrations, with observed total whole beef concentration of 8.15 ppt and predicted at 2.13 ppt.

As a way of further examining these results, limited examinations are now presented on the two key components of this food chain model - the air to vegetation algorithm, and the air to soil algorithms.

One data set in the literature allows some limited comparisons between model predictions and observations of vegetation concentrations. This data was from a rural setting in Elk River, Minnesota (Reed, et al., 1990). This site was mentioned in the section above describing the derivation of the rural air concentration profile. The reference listed air concentrations by congener grouping for a rural setting (2 air sampling sites) and near an incinerator (1 site). It was noted that the average annual air concentrations near the incinerator was about 5 times higher than the average annual air concentration at the two rural sampling stations. The total PCDD/F air concentration in the rural setting was estimated at 1.54 pg/m³. The corresponding TEQ concentration cannot be estimated without knowing the concentration of the congeners with non-zero toxicity. Therefore, a comparison to the crafted 0.019 pg/m³ concentration for the rural setting in this paper cannot be made. However, a data set earlier described from Sweden (Broman, et al., 1990), listed a total concentration of 0.42 pg/m³ and a corresponding TEQ concentration of 0.004 pg/m³ for a rural Swedish countryside. This ratio of 100 between total and TEQ concentrations indicates that the Elk River total concentration of 1.54 pg/m³ may translate to a TEQ concentration around 0.015 pg/m³, which would be consistent with the 0.019 pg TEQ/m³ developed in this paper.

This study also took samples of vegetations in this rural setting, including two hay and two corn samples. The limits of detection for these vegetation samples varied

Table 7-9. Results of validation exercise showing observed and predicted concentrations of dioxin-like compounds in whole beef.

Observed whole beef Predicted whole beef

Compound concentrations, ng/kg¹ concentrations, ng/kg

2378-TCDD 0.03 0.03

12378-PCDD 0.22 0.27

123478-HxCDD 0.26 0.10

123678-HxCDD 0.84 0.03

123789-HxCDD 0.21 0.04

1234678-HpCDD 1.92 0.29

OCDD 2.91 0.29

2378-TCDF 0.06 0.46

12378-PCDF 0.04 0.07

23478-PCDF 0.21 0.17

123478-HxCDF 0.51 0.08

123678-HxCDF 0.06 0.13

123789-HxCDF 0.06 0.04

234678-HxCDF 0.07 0.07

1234678-HpCDF 0.4 0.04

1234789-HpCDF 0.13 0.01

OCDF 0.22 0.01

TOTAL CONCENTRATION 8.15 2.13

TEQ CONCENTRATION 0.48 0.36

between 0.31 and 6.5 ppt on a congener-specific and site-specific basis. With vegetation concentrations predicted to be in this range generally, the data therefore cannot be rigorously informative. The congener found with the highest concentration is OCDD, found at 72 (site 1) and 170 (site 2) ppt in two corn samples, and 270 (site 1) and 300 (site 2) ppt in two hay samples. In addition to this higher finding in the hay samples, generally more positives were detected in hay rather in corn. This is consistent with discussions in this paper indicating that vegetation concentrations of dioxin-like compounds is a surface phenomena with little within plant translocation. Hay, in this observation, is considered a leafy vegetation, whereas corn is considered a bulky vegetation.

Table 7-10 lists the average congener specific hay concentrations observed in Elk River (the average of two hay samples, with non-detects counted as 0.0 when one of the two samples had a positive, and just listed as ND when both hay samples showed non-detects) compared against the model's predicted concentrations in grass. This is felt to be a valid comparison. It assumes that hay alone is reasonably similar to grass in that both are "leafy" vegetations and would be modeled similarly in the framework of this paper.

What is now available to interpret and analyze are the predicted and observed beef concentrations, the predicted and observed leafy vegetation concentrations, and further model trends. Several observations are now summarized based on these analyses:

1) Given the range of the detection limit, 0.31-6.5 ppt for the hay sampling, the model's predictions of grass concentrations are generally consistent with observations, with the exception of the OCDD and OCDF concentrations. It is noted that the second highest congener observation of 30 ppt of 1,2,3,4,6,7,8-HpCDD is matched by the model's prediction of 20.7 ppt for 1,2,3,4,6,7,8-HpCDD.

2) The analysis of the OCDD and OCDF results for hay is very telling. First, it is noted that the crafted rural air concentrations of these two congeners matches very well with the observed air concentrations at this Elk River site: OCDD observed at 0.5 pg/m³ and crafted at 0.57 pg/m³; and OCDF observed at 0.09 pg/m³ and crafted at 0.034 pg/m³ (note: the observed concentrations for OCDD/F congeners is the average of four listed concentrations of OCDD/F congeners in Reed, et al. (1990) - rural sites 1 and 2 and winter and summer listings). Since the crafted air concentrations match well with the observed air concentrations, one would hope that the vegetative concentrations also match. An analysis of why they did not indicates the importance of vapor phase contributions to vegetative concentrations. According to the application of the Bidleman (1988) approach for estimating the bound fraction, f , in the air, both these congeners were assigned a f of 1.00. In fact, using the OCDD/F vapor pressures and melting points, these f values were both 0.998. If one allows for the possibility that f for OCDD/F could be less than one, and calibrates f for OCDD/F for this exercise, one can show that small reductions in f result in better predictions of both grass and beef concentrations. Recall that the observed "grass" concentrations are, in fact, the hay concentrations found at Elk River, Minnesota,

Table 7-10. Comparison of concentrations of dioxin-like compounds found in hay in a rural setting with model predictions of grass concentrations.

Observed hay Predicted grass

Compound concentration, ng/kg¹ concentrations, ng/kg

2378-TCDD ND 0.1

12378-PCDD ND 0.9

123478-HxCDD ND 0.7

123678-HxCDD 1.2 0.2

123789-HxCDD ND 0.2

1234678-HpCDD 30 21.0

OCDD 285 6.0

2378-TCDF ND 7.2

12378-PCDF ND 1.4

23478-PCDF ND 0.8

123478-HxCDF ND 0.5

123678-HxCDF ND 0.9

123789-HxCDF ND 0.3

234678-HxCDF ND 0.5

1234678-HpCDF 5.4 1.4

1234789-HpCDF ND 0.1

OCDF 7.5 0.4

Assumption/

Method Approach Rationale Uncertainty Comments

Lifetime 70 yrs Standard EPA assumption Actuary data indicate Not a major source of uncertainty

and based on data that lifetime may

may be increasing

Body Weight 70 kg adult Standard EPA assumption Not much uncertainty Not a major source of uncertainty

and based on data

Exposure 9 & 20 yrs Based on assumptions Can vary for popula- Assuming non-farming families

duration for central and high tions in rural settings; are more transient than farming

end exposure scenarios. also important to con- families is probably reasonable,

9 years based on data sider how long exposure although data is unavailable

for time spent in one has been occurring for to verify that assumption.

residence; rural farming retrospective site- Considering exposure durations

families assumed to specific assessments to be in the range of 10-20 yrs

live in one location or how long exposure rather than 70 years is felt

longer than non-farming may occur for prospec- to be more appropriate. Con-

families in rural tive assessments. Source sideration should be given to source

settings strength dissipation not strength dissipation over time.

a consideration for effluent

discharge or stack emission

sources assuming emissions

occur unabated over duration;

data and mass balances

exercises indicate soil

concentrations remain constant

for a 9-20 year time frame.

Overall: Of these three parameters, the exposure duration is the most uncertain. The estimates of 9 and 20 years were made in this assessment for non-farming residents in rural settings, and farming residents in rural settings. These values were based on assumptions of time living at one residence. A critical assumption of a constant soil concentration for contaminated soil sites should be carefully considered for site-specific assessments. Data on degradation indicates very slow rates of degradation, and only photolysis as a possible degradation mechanism, which would not impact residues below the surface. Dissipation other than degradation should be considered, but a mass balance exercise indicates that it would take 90 years to dissipate a reservoir of contaminants extending 6 inches into the soil.

in areas described as "background" in the 600 ha site to soil concentrations in the contaminated areas was 1/8 to 1/2 as much (depending on how the contaminated area soil concentration was interpreted) as the ratio modeled in the off-site demonstration scenario. This might imply that the model overpredicts off-site soil impacts, except that the "background" areas in the Dow Site appear substantially further away than the 150 meters in the off-site demonstration scenario. Also, data was unavailable to determine the erodibility of soil at the Dow Site, which along with other site-specific information, may have allowed for a more precise test of the algorithms of this assessment. Still, a key finding in the sensitivity analysis exercises was that the erosion algorithms may be overestimating off-site impacts. If so, the overestimation is most likely the result of assuming an "enrichment ratio" of 3 for soil erosion (the concentration on eroded soil divided by the concentration of in-situ soil). No information is available on estimating how much of an overestimation may have resulted, and this finding is not a definite conclusion. If the algorithm overestimated the impact from soil erosion, it is unlikely that overestimation exceeded the factor of 3 attributed to the enrichment ratio. Other sensitivity analyses exercises indicated that different parameters values for individual parameters result in roughly an order of magnitude difference in soil concentration estimation around the concentration which was estimated using all parameters assumed for the demonstration scenarios in Chapter 5.

In addition to the enrichment ratio, the depth of mixing is an uncertain parameter. This is a theoretical parameter for which little data is available. Others have also assumed depths of mixing of 1 cm for analogous applications. Evidence from radioactive fallout suggests depths no deeper than 5 cm. Sensitivity analysis on the erosion algorithms showed that assuming a depth of 1 cm instead of 5 cm would have increased soil concentrations by a factor of 2.5, while decreasing the mixing depth to 10 cm decreases soil concentrations by 60%.

No information could be found in the literature which could be used to evaluate the algorithm for soil concentrations resulting from particulate depositions from stack emissions as modeled by the COMPDEP model. However, evaluation of the air-to-soil algorithm of the stack emission source category suggests that the model may be underpredicting soil concentrations, possibly by about an order of magnitude. Soil concentrations are, of course, not an issue for the on-site soil source category, where that concentration is a principal input and not an estimated value.

Another issue is whether children should be assumed to be exposed to tilled soils, tilled by home gardening, farming, etc., or untilled soils. It is feasible that children would be exposed to tilled soils in farming or home garden settings. If the soil was impacted by stack emission depositions or erosion from a nearby site of soil contamination, then tilling would reduce soil concentrations. However, it is more reasonable to assume that they generally play outside in areas that are not mechanically tilled.

The estimated soil ingestion quantity is based on field measurements, using trace elements, of soil ingested by relatively small groups of children over brief periods. Methodological issues in these studies remain to be addressed. In particular, ingestion estimates may have been lower if dietary intake of the trace elements was taken into account. Research is underway to refine soil ingestion estimates obtained through trace element measurements. Given the available data, 0.2 g/day is used as a typical value for soil ingestion in young children. Due to the behavior known as pica, some children are known to be high ingesters of various non-food materials. Estimates of pica ingestion of soil by children have ranged as high as 5 g/day. Although no quantitative data on soil ingestion are available for children known to exhibit pica, the use of the high-end estimate of 0.8 g/day may better reflect such behavior.

Soil ingestion exposure estimates also depend on the duration of the period over which children are assumed to ingest soil. Data on soil ingestion by age are not available, and the estimate that significant ingestion occurs between ages 2 and 6 is broadly supportable on behavioral grounds.

No measurement data are available on soil ingestion in infants (0-2 yrs. old) or in older children or adults, and no ingestion is assumed for these groups. While some soil ingestion will occur in these groups, e.g., through contact of soiled hands with food, it is plausible that such ingestion is of a lesser degree than occurs in early childhood. If Hawley's (1985) estimate that an adult ingests an average 0.060 g/d of soil is used, after accounting for differences in exposure duration (9-20 yrs versus 5 yr) and body weight (70 kg versus 17 kg), the adult soil ingestion exposure is close to the estimated exposure for children (at 0.2 g/d). The high end example scenarios in Chapter 9 assumed that the exposed family was involved in farming operations. One implication is that individuals on the farm would be working closely with the soil, which may result in some soil or dust ingestion (dust ingestion is distinct from the particulate inhalation exposure pathway). The other implication is that, should this be the case, they would be in contact with tilled soil, whose concentration is 20 times less than the no-till soil for which children are assumed to be exposed.

Considering these uncertainties, the soil ingestion exposure estimates presented for children are plausible. Further consideration may be warranted for considering adult soil ingestion, particularly in farming situations. Uncertainties associated with the soil ingestion pathway are summarized in Table 7-14.

Estimates of dermal exposure to soil rely largely on four factors unique to this pathway: exposed skin area, soil adherence, frequency of soil contact and fraction of contaminant absorbed. The uncertainty in these three terms are discussed below.

Before that discussion, a brief note is made on uncertainties associated with soil concentrations. Discussions above on the soil ingestion pathway addressed uncertainties associated with soil concentrations which result from migration of residues from a distant source to the site of exposure. Distant sources in this assessment include off-site soil contamination and stack emissions. Discussions in the soil ingestion pathway section above pertain to this exposure pathway and are not repeated here. However, there is one key difference in the soil dermal and soil ingestion pathways. Soil ingestion exposures are assumed to occur only from surficial soil layers and from untilled soils, which translates to the 5-cm (soil contamination source categories) and 1-cm (stack emission source category) mixing depth for both the "central" (residential) and "high end" (farming properties) scenarios. Soil dermal contact for the high end scenario assumes many dermal contact events, 350 per year, that is based on farming activities; the soil concentrations pertinent for this behavior, therefore, are tilled soil concentrations. On the other hand, only 40 dermal contact events per year, which may correspond to some gardening or other contact, is assumed for the central scenario. The soil concentration used in these scenarios is the untilled soil concentrations.

The uncertainty in the assumed value for exposed skin area reflects primarily population variability. As reported in EPA (1992a), relatively accurate measurements have yielded a good data base on total skin area. Thus the uncertainty in this factor is derived more from the assumptions of how much of the total skin area is exposed. EPA (1992a) recommends approaching this issue by determining the coverage of normal apparel in the exposed population and assuming exposure is limited to the uncovered skin. As discussed in EPA (1992a), this assumption could lead to underestimates of exposure since studies have shown that some exposure can occur under clothing, especially in the case of vapors or fine particulates. A default assumption of 25% uncovered is recommended

Table 7-14. Uncertainties associated with the soil ingestion pathway.

Assumption/

Method Approach Rationale Uncertainty Comments

Erosion/deposi- The 1 & 5-cm depths There is no research Exercises suggest No data could be

tion results in are "no-till" depths; to refine the mixing use of an enrichment found to more fully

a 1 & 5-cm cont- 20 cm is assumed as depth parameter; ratio for eroding evaluate the erosion

aminated layer the tilled depth; others have assumed contaminants may algorithm; the result

to which children are exposed similar depths for lead to higher of 0.28 ppb concentration

children are to no-till concentra- analogous applications; concentrations at resulting from 1 ppb

exposed tions exposure sites than nearby contamination

are warranted, although may be high; the use of

by no more than a factor an enrichment ratio of 3

of 3; exercises on the was speculated as being the

stack emission source parameter of most uncertainty.

suggest that the impact For stack emissions, the

of depositing particles dissipation half-life of 10 yrs.

may be underestimated was speculated as the model

by an order of magnitude parameter most likely to be

or more. leading to underpredictions.

Child's inges- Ingestion rate assumed The range selected was Field study methodology Pica children have been

tion rate (2-6 to vary from 0.2-0.8 g/d. primarily based on the not fully validated. estimated to ingest higher

years old). results of two field Data from several sources quantities (5 g/d).

studies of soil ingestion indicate this range of

in children. values for small children.

Ingestion rate Ingestion assumed to Mouthing tendencies Hawley estimates inadver- Adults may inadvertently

for other ages occur only during strongest and under- tent ingestion may be 60 ingest soil during gardening

ages 2-6. standing of personal m g/d for adults, which and yard work; farmers may

hygiene low during would lead to an have a non-trivial soil ingestion

childhood exposure pattern similar pattern

to that of children

Overall: Soil ingestion for older children and adults were not considered, which may have underestimated lifetime soil ingestion exposures by a factor of two. The other major area of uncertainty for this pathway is for the scenarios where the source of contamination is located distant from the site of exposure, including areas of high soil contamination, the off-site soil source category, and the stack emission source category. Analysis of results from demonstration of the off-site soil source category suggests that the 0.28 ppb soil concentration (within a 5-cm layer) to which children are exposed, and which resulted from the 1 ppb nearby (150 m) soil contaminated site, may be high. On the other hand, analysis suggests that the soil concentration in a 1-cm layer resulting from depositing particles, the air-to-soil algorithm of the stack emission source category, may be underestimating soil concentrations by an order of magnitude or more. The uncertain parameters in these algorithms are the enrichment ratio (for the erosion algorithm only) the depth of mixing (for both the erosion and deposition algorithms), and the mixing zone depths. The mixing zone depth for untilled situation is particularly uncertain - the assumed depths of 1 and 5 cms for the stack emission and off-site soil contamination sources are supported by data on radioactive fallout, and is similar to 1 to 5 cm depths others have assumed. Pica soil ingestion patterns were not evaluated in this assessment; the ingestion rates considering this appear reasonable.

corresponding to short sleeved shirt, short pants, shoes, and socks. Thus the key uncertainty issue concerns the variability in clothing behavior of the exposed population. In this document the 25% assumption was adopted for residents and 5% was judged more reasonable for farmers who are more likely to wear long pants and long sleeved shirts for field work. Although clothing coverage is likely to vary over the year and with personal habits, these assumptions are judged to be reasonable averages and unlikely to introduce more than a factor of two uncertainty.

The potential for soil adherence probably varies little across the population, but few actual measurements have been made. Thus the uncertainty in these estimates reflect primarily the lack of measurement data rather than population variability. Site variability is probably important as well since soil properties such as moisture content, clay content and particle size distribution are likely to affect adherence. EPA (1992a) reports four studies which estimated soil adherence on hands under both laboratory and field conditions. Data from these studies were analyzed to obtain a central estimate of 0.2 mg/cm² and a high end estimate of 1.0 mg/cm². The uncertainty in these estimates are derived from unknown efficiencies in the collection methods, relatively small number of subjects, assumption that hand measurements apply to other parts of the body and assumption that child measurements apply to adults as well. The central default value 0.2 mg/cm² was adopted here for the residents and the high end value of 1.0 mg/cm² was adopted for farmers. The uncertainties in this estimate could combine to produce either under or over estimates and may vary by as much as a factor of 5 on the basis of the ratio of the high end to central estimates.

Exposure frequency to soil reflects largely personal habits and thus the uncertainty is primarily based on population variability. Seasonal and climate conditions can also affect this behavior introducing site variability as well. EPA (1992a) suggests a central frequency of 40 days/yr corresponding to someone who does yard work, gardens or plays outdoors on most weekends and a high end estimate of 350 days/yr corresponding to a farmer or serious gardener in a warm climate. These recommendations were based on judgement rather than actual survey data. In this document, 40 days/year was selected for the residential scenarios and 350 days/yr for the farmer. The lack of survey data to support these estimates introduces uncertainty, but the values are judged to be reasonable and to create relatively little uncertainty.

The dermal absorption fraction of compounds varies widely across chemicals, whereas skin properties that affect absorption, i.e. thickness and composition vary little across the population. Thus the uncertainty in this factor is derived primarily from measurement error rather than population variability. Soil properties, such as organic carbon content, can also affect the extent of dermal absorption and thus create site variability as well. EPA (1992a) reports two studies which measured dermal absorption of 2,3,7,8-TCDD from soil. Testing included human skin in vitro, rat skin in vitro and rat skin in vivo. On the basis of these tests, a range of 0.1 - 3.0% was recommended in EPA (1992a). Dermal absorption testing, especially for soils, is a relatively new field and many uncertainty issues are involved. These include extrapolation of animal tests to humans, extrapolation of in vitro to in vivo conditions, and extrapolation of experimental conditions to expected exposure conditions. Extrapolation of the tests on 2,3,7,8-TCDD to the other dioxin like compounds (which have not been tested) introduces further uncertainties. A dermal absorption fraction of 3.0% was adopted here for application to all the dioxin like compounds. Based on the observed range of values for 2,3,7,8-TCDD this assumption may lead to overestimates of a factor of 30. Considering all possible uncertainties, under estimates are also possible, though judged less likely.

In summary, dermal exposure estimations rely on a number of parameters whose values are not well established. Although it is difficult to estimate the overall uncertainty with this pathway, it is judged to be plus or minus one to two orders of magnitude. A summary of the uncertainties associated with the dermal absorption pathway is given in Table 7-15.

The strong sorptive tendencies of the dioxin-like compounds result in very low water concentrations. Monitoring for PCDDs and PCDFs mostly have not found these compounds at a detection limit around 1 pg/L (ppq), and when found, have generally been very near this concentration. The one exception is an upstate New York community water system, where tetra through octa-CDFs were found at concentrations ranging from 2 pg/L (tetra) to over 200 pg/L (octa). The surface water concentrations predicted by the algorithms of this assessment for all source categories are 10^-2 pg/L and lower, which is consistent with the sparse monitoring data. Although there was no data found that could be directly applicable to the source categories, it does not appear that the models estimating water concentrations will introduce significant uncertainty into water ingestion exposure estimates.

The classically assumed water ingestion rate of 2.0 L/day was examined in EPA

Table 7-15. Uncertainties associated with the dermal exposure pathway.

Assumption/

Method Approach Rationale Uncertainty Comments

Soil Same issues as soil ingestion pathway for migration of residues from a distant source to site of exposure; see discussions

Concentrations in Section 7.3.2 and Table 7-14 above

Use of tilled only one conc. for behavior parameters tilled concentrations are For real sites, monitoring

vs. untilled on-site category; used for farmers assume lower than non-tilled is best way to resolve

soil "tilled" conc. for high dermal exposure results concentrations; much un- this uncertainty.

concentrations end farming scenarios, from farming in tilled certainty associated with

"non-tilled" for non- for non-farmers, difference between tilled

farmers in central yard work during summer and untilled mixing zone

scenarios for all assumes dermal contact depths; non-farmers working

other sources occurs in non-tilled in home gardens also are

soils exposed to tilled

concentrations

Contact 0.2 mg/cm²-event for non- corresponds to central measurement data may have soil properties could

rate farming residents; and high end values of experimental error or not be affect adherence

1 mg/cm²-event measurement data; representative; uncer-

for farmers supported by EPA, 1992a tain + factor of 5

Contact 40 for residents; Based on judgement; personal behavior patterns climatic conditions intro-

frequency 350 for farm families supported by EPA could differ but uncertainty duce site variability

(1992a) judged to be small

Surface 5% (1000 cm²) for farming Based on total body Clothing assumptions Studies have shown that

area adults and 25% (5000 cm²) surface area data and based on judgement fine particulates can

for non-farming adults clothing assumptions; rather than survey data; deposit under clothing

supported by EPA(1992a) uncertainty judged to be

plus or minus factor of

2.

Absorption 0.03 for all dioxin- Based on EPA (1992a) Experimental procedures Soil properties may also

fraction like compounds which gave a range of very uncertain, may over affect absorption

0.001 to 0.03; range estimate by factor of 30

supported by 2 different

studies in rat and human

Overall: The high uncertainty in estimates of soil adherence and absorption fraction make the overall uncertainty in the exposure estimates highly uncertain, judged to be plus or minus 1 to 2 orders of magnitude.

(1989). The conclusion was that this estimate is more appropriately described as an upper percentile consumption rate for adults, and recommended 1.4 L/day for use as an average. This value was used for water ingestion in the central scenarios. EPA (1989) cautions that data on consumption rate for sensitive subpopulations such as manual laborers are unavailable. As such, the 1.4 L/day rate for individuals in farming families who work the