Environ. Sci. Technol. 2005, 39, 5170-5176
Typical Dioxin Concentrations in Agriculture Soils of Washington State and Potential Sources D A V I D L . R O G O W S K I * ,† A N D WILLIAM YAKE Environmental Assessment Program, Washington State Department of Ecology, PO Box 47710, Olympia, Washington 98504-7710
Background or typical levels of polychlorinated dibenzop-dioxins and of polychlorinated dibenzo-p-furans in soils have become increasingly important in the regulatory community, as dioxins have come under increased scrutiny due to their toxicity and persistence. Knowing the typical levels of dioxin in soils is important to set regulatory levels, to prevent further contamination, and for setting cleanup levels. A random sample (n ) 54) of agricultural lands within Washington state revealed a typical concentration of “dioxin” of 0.14 ng/kg (toxicity equivalents, TEQ). For a comparison, residential urban areas, forested, and open areas were also investigated on a smaller scale (n ) 14, 8, and 8, respectively) with typical dioxin values of 4.1, 2.3, and 1.0 ng/kg (TEQs), respectively. A discriminant function analysis was used to examine soil dioxin profiles and associations with potential known dioxin sources in Washington state. Soil sample dioxin profiles could not be discriminated from those of biosolids and pentachlorophenol products.
Introduction Polychlorinated p-dioxins and polychlorinated dibenzofurans (PCCD/PCDFs) have been under increased scrutiny due to their toxicity and persistence. In response numerous governments and environmental groups have identified PCCD/ PCDFs as contaminants that should be reduced and regulated. The use of industrial byproducts as soil amendments and/or fertilizers has raised concerns about potential contaminants being applied to agricultural lands (1). Most of our intake of PCCD/PCDFs is derived from food consumption, particularly meat, milk, and fish; however vegetable crops can also contribute to a person’s intake, reflecting PCCD/PCDFs levels in the soil and/or air depending on the particular crop (2). As a result of these concerns, the Washington state legislature funded a study of “dioxins” in fertilizers, soil amendments, and soil. For simplicity, the term “dioxin(s)” is used here to refer to those dioxin/furan congeners that have chlorine atoms in the 2, 3, 7, and 8 positions of the molecule (results reported as toxicity equivalents, TEQs (3). A “soil amendment” is a regulatory term used to define various organic and inorganic materials added to soil to affect its physical properties. Background levels of dioxins in soils have been determined for a variety of countries, such as Germany (4), Austria (5), * Corresponding author phone: (217) 244-5123; fax: (217) 3336294; e-mail:
[email protected]. † Current address: Illinois Natural History Survey, Center for Aquatic Ecology and Conservation, 607 E. Peabody Dr., Champaign, IL 68120. 5170
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
and the United Kingdom (6, 7), but within the United States this information is lacking. Although there have been studies that have included background soil samples for comparisons, the bulk of these investigations were associated with known dioxin sources and areas of contamination in the United States. In addition, the majority of the studies were conducted in the eastern half of the continental United States of America. Historical industries and prevailing westerly winds may potentially contribute to higher dioxin levels in the eastern USA than what might be expected in western states. The primary objective of this study was to provide an initial assessment of typical dioxin concentrations of soils in Washington state, particularly agricultural soils. Agricultural soils were the main focus as a variety of micronutrients and fertilizers are derived from waste products and contain dioxins (1). To determine if use of these products might lead to elevated levels of dioxins in agricultural soils (and potentially in crops), knowledge of current levels in agricultural soils was necessary for comparison and possible regulation. We use the word “typical” in this study to avoid controversy associated with using a term such as background levels. Background levels implies a natural occurrence of these compounds, i.e., biogenic formation from natural organochlorine compounds, or from the burning of natural organic material (8-10). We were interested in current dioxin levels, regardless of contributions from natural occurrences or more recent ambient sources. Surface soils were collected from agricultural areas, residential urban areas, open (pasture, prairie), and forested areas. Discriminant function analyses (DFA) were used to investigate differences among groups and relationships to sources. A secondary objective was to compare PCDD/PCDF profiles from known dioxin sources within Washington state to the soil samples collected, using DFA. In DFA, categories or groups are decided a priori. A discriminant function analysis maximizes differences among groups while minimizing within group variance (11). Many investigators compare PCDD/PCDF profiles of sources and environmental samples using principal component analysis (PCA) (12-16, but see 17-18). In PCA, the objective is to reduce the original variation in multiple dimensions into a few principal components while preserving the overall variance structure. The data reduction algorithm/process used in PCA may actually obscure differences among groups, since principal components are based on representations of explained variation present in the original data set. Discriminant function analyses maximize differences among groups, and we use DFA to determine if there are significant dioxin profile differences among groups and to determine what factors contribute to those differences (11).
Experimental Section Sampling Site Selection. Soil samples (84) were distributed among our land use categories: agricultural, residential urban, forest, and open (Table 1) across the state (Figure 1). As public interest centered on the use of micronutrients and fertilizers contaminated with dioxins, the bulk of the soil samples selected for analysis were allocated to agriculture lands (n ) 54), while 30 soil samples were allocated to urban, forested, and open areas. For this study, agricultural land does not include sites used for ornamentals, floriculture, greenhouses, turf grass, silviculture, Christmas tree farms, rangeland, or pastures. To represent the diversity of Washington agricultural soils, sampling sites were randomly distributed by county and crop type. The state’s agricultural acreage was represented by 5284 10.1021/es047945r CCC: $30.25
2005 American Chemical Society Published on Web 06/16/2005
FIGURE 1. Distribution and number of soil samples in Washington state by county.
TABLE 1. Number of Surface Soil Samples Allocated by Land Use for Dioxin Analysis and Approximate Land Use Area in Washington State (104 875 000 hectares) subgroup
samples
total
hectares (1 000)
grazed land prairies managed state and federal parks Seattle area Tacoma tri-Cities (Richland, Kennewick, Pasco) Spokane
4 4 4 4 9 2 2
8
7 590
8
46 976
14
3 064
54 84
16 061 84 856
land use open forest urban
agricultural total
1 54 84
Soil Sampling Procedures. A sampling unit of 0.4-hectare (one-acre) was selected; this was the largest practical unit that allowed for representative composite sampling across all land uses. Areas of potential dioxin contamination were avoided such as roads, railroad tracks, treated wood utility poles, or fences. Areas of significant erosion were also avoided. Each sample consisted of a composite of 10 samples collected within the sampling unit. The initial sample was collected at a starting point, with nine additional samples collected at the end of a radius originating from the starting point (center), extending a distance of 36 m, and rotated at equal intervals of 40°. The surface layer of organic vegetative material was removed, and a sample was collected from a depth of 0-5 cm below the surface layer. Analytical Methods. Samples were analyzed for the 2,3,7,8-substituted PCDD/PCDF congeners and tetra- through octahomologue totals at MAXIM Technologies Inc./Pace Analytical, using high-resolution gas chromatography/mass spectrometery (EPA Method 8290), with enhancements derived from Method 1613B. The target detection limit was 0.1 ng/kg, however, detection limits varied depending on the physical state of the sample (e.g., moisture, organic content). Total organic carbon (TOC) analysis was conducted at the Washington State Department of Ecology’s Manchester laboratory using an induction furnace method (23). Washington State Department of Ecology, Manchester Environmental Laboratory personnel reviewed all results for qualitative and quantitative accuracy, validity, and usefulness following appropriate guidelines and EPA Region 10 standard operating procedures for the validation of PCDD/PCDF.
Data Analyses units, each unit representing 400 hectares (1000 acres). These units were each assigned a specific crop type and county. These assignments were proportional to the number of harvested agricultural acres in the state associated with specific crops in specific counties (19-21). Units (54) were randomly selected from the data set. We randomly selected growers who met the crop/county criteria, were larger than 4 hectares, and did not apply biosolids to their lands using U.S. Department of Agriculture (USDA) Farm Service Agency (FSA) lists. Fourteen soil samples were allocated to urban areas, as defined and identified by the U.S. Census Bureau (22). Samples were randomly allocated based on acreage to the 10 urban areas identified within Washington. Sites in the “greater Seattle area” were randomly allocated from a data set of approximately 300 parks and schoolyards. Sites residing within Seattle city limits were excluded due to problems encountered with obtaining sample permits. Urban sample sites were located in residential areas within urban boundaries. “Forested sites” were defined as areas with an extensive canopy composed primarily of mature trees. Sites were chosen based on spatial distribution (four samples each, from east and west sides of the state) and ability to gain site access. Washington state is bisected from north to south by the Cascade Mountain range; this is traditionally used to classify the state into the East and West. An even mixture of private and government-managed sites used for timber harvesting, and undisturbed areas such as parks and wilderness areas were chosen. For this study, “open areas” were defined as historically nonforested, nonagricultural lands located outside of large urban areas. Four samples were from grazed land and four from nongrazed open areas (prairies) in wildlife refuges or national parks. As with the forested sites, our open area samples were evenly split between the east and west sides of the state.
The TEQ of dioxin congeners (the different forms of dioxins and furans present) is calculated using toxicity equivalency factors (TEF) reported by EPA (3). The EPA TEFs differ from the World Health Organization (WHO) TEFs (24-25) in only 3 compounds. Only one compound differs substantially; 1,2,3,7,8-PentaCDD is 0.5 for EPA vs 1 for WHO, while the EPA TEF for OCDD and OCDF are both 0.001 vs 0.0001 for WHO. Unless otherwise specified, TEQ values reported here assume that when a specific congener or homologue is not detected (ND) in a sample, its concentration is zero (ND ) 0). Statistical analyses were conducted using JMP (Release 5.01.a; SAS Institute Inc., Cary, NC, 1989-2002), or Systat (Release 7.0.1. SPSS Inc. Chicago IL, 1997). Source Discrimination. A DFA was used to examine relationships between the PCDD/PCDFs profile of soil samples and potential dioxin sources (Table 2). Sample sizes were unequal so we used a priori probabilities, and to examine how well soil samples were classified, we used a jackknifed classification matrix (26). Thirty-five dioxin source sample results and data from nine fertilizer products (containing dioxins) were obtained for this evaluation. This analysis focuses on source data obtained from facilities within Washington state. However some source information from Washington state was not available so additional data was obtained from literature (5, 14, 28-29). Source data for Washington state was obtained from the Washington State Dioxin Source Assessment (31). Fertilizer information was obtained from Rogowski et al. (1). Only fertilizer materials containing dioxin at levels above 5 pptr TEQ were included. Because of the limited source data available from Washington, some of the data includes multiple measurements from the same source under different operating conditions and at different times. Most of the data was fairly representative of normal operating conditions, with the exception of emission data from an activated carbon regenerating facility (CameronYakima). Data from Cameron-Yakima were from a “moderate worst case” test scenario (31), and emission data from the VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5171
TABLE 2. Washington State Dioxin Sources and Additional Dioxin Data Used in Analyses data source (ref)
samples
medium
process
location
Rappe & Oberg 1997 (28) Duarte-Davidson et al. 1996 (29) Hagenmaier & Brunner 1987 (30)
1a 1a 4
biosolid biosolid PCP(2), PCP-Na (2)
sewage sludge sewage sludge products
NCSS, USA United Kingdom Germany
Washington State Sources (31) VA Medical center Kennewick Hospital Northwest Hospital Olivine Corp. Spokane Municipal Incinerator Fort Lewis Incinerators # 1,2, & 3 Fort Lewis Incinerator Tacoma City Light - steam plant Recomp-TRC
2 1 2 1 3 3 1 2 1
air emissions bottom ash air emissions air emissions air emissions & fly ash air emissions fly ash air emissions air emissions
Seattle Kennewick, closed Seattle Bellingham Spokane Tacoma Tacoma Tacoma, closed Bellingham
Kaiser Aluminum & Chemical Corp.
1
air emissions
Cameron-Yakima Inc.
2
air emissions
Holnam, Cement Kiln Fort James Pulp and Paper Simpson Kraft Mill
6 1 1
air emissions fly ash effluent
incinerator (hospital) incinerator (hospital) incinerator (hospital) incinerator (municipal) incinerator (municipal) incinerator (municipal) incinerator (municipal) incinerator (municipal) incinerator (municipal and medical waste) Aluminum Rolling Mill, aluminum remelt furnace incinerator (multihearth and rotary kiln) incinerator (cement kiln) hog fuel boiler wastewater
2 1 1 1 1 1 1
micronutrient fertilizer fertilizer fertilizer micronutrient wood ash micronutrient
1 1
raw material is tire dust brass ingot dust
Washington State Fertilizer Products (1) Frit F-503G, #1, #2 Fort James Pulp and Paper, NutriLime McLendon Weed and Feed 15-5-5 NuLife All-Purpose Trace Elements Bay Zinc K061 Kimberly Clark Bay Zinc 18% (zinc) Blu-Min Bay Zinc LHM Bay Zinc Liquid a
Trentwood Yakima, closed Seattle Camas Tacoma
product, steel mill flue dust product, hog fuel boiler fly ash product product source material potential product product derived from steel mill flue dust product product
Median value.
TABLE 3. Mean Soil Dioxin (ng/kg TEQ), Total Organic Carbon (TOC), and Dioxin Results Normalized to TOC for Washington State by Land Usea land use forest open urban agriculture
TEQ total total TEQ (ND ) dioxin TOC TEQ/TOC dioxin/ (ND ) 0) 0.5) (ND ) 0.5) (percent) (ND ) 0) TOC 2.3 1.0 4.1 0.14
3.5 1.9 5.8 0.99
220 260 610 42
18.5 10.2 3.98 1.45
17.9 12.1 89.0 11.1
1 700 3 600 13 000 3 000
Agricultural lands had the lowest dioxin concentrations of the four land uses (Table 3). Agricultural dioxin values in this study (median and mean 0.054, 0.14 ng/kg, respectively) were similar to the median TEQ level of 0.097 pg/g, reported for agricultural soils in Australia (WHO98-TEQs using zero for nondetected congeners) (33). However, our dioxin levels were lower than agricultural soil dioxin levels in Austria (5), Russia (34), and Sweden (35) (Figure 2). It is not entirely
a TEQ ) toxicity equivalents; ND ) nondetects. TEQs were calculated based on giving values of zero for nondetects (ND ) 0) or a value of half of the detection limit (ND ) 0.5).
cement kiln represents a variety of conditions with different combinations of fuel (31). Only source data reporting complete results for the 10 dioxin/furan homologues were used in the DFA (these data are presented in Rogowski et al. 1999 (1)). The 17 2,3,7,8PCDD/PCDF substituted congeners were not used for the discriminant analysis due to the infrequent detection of those specific congeners in soil and source samples. For homologues that were not detected, a concentration equal to half the value of the detection limit was assumed. To remove biases due to concentration differences among samples, the percentage of each homologue to total PCDD/PCDF concentration was used in the discriminant analysis. Percentage results were transformed prior to analysis to approximate a near-normal distribution using an arcsine transformation developed by Freeman and Tukey (32).
Results and Discussion
FIGURE 2. Dioxin concentrations (TEQ) of soils in Washington state by land use compared to soils in Spain (12, 36), Germany (2), Austria (5), Russia (34), and Sweden (35). German samples and Austrian plowland are plotted as median values.
A summary of the dioxin TEQ data is listed in Table 3. Complete analytical results for the 17 congeners of concern for dioxins and furans can be found in Rogowski et al. 1999 (1). Dioxins were found in surface soils throughout Washington state, including those from remote wilderness areas.
clear why there is a difference in dioxin levels from Washington state and Australia compared to Europe and Russia, perhaps it is a result of a longer time period of anthropogenic impacts. The difference in agricultural soils
5172
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
FIGURE 3. Dioxin/furan homologue profiles of soil samples from Washington state normalized to sum total. from other land uses can probably be attributed in part to the actions of harvesting and tilling. Tilling may dilute dioxin levels within soil surface layers, and subsequent erosion of topsoil (wind and/or water runoff) may also result in lower levels of dioxins in surface soils. There was a significant correlation between log TOC and log TEQs for the entire data set (r ) 0.66, p < 0.0005). Dioxin concentrations (log TEQs) were positively correlated with log TOC in agricultural soils (r ) 0.43, p ) 0.001), as well as in urban soils (r ) 0.78, p ) 0.001). There was not a significant correlation of log TOC and log TEQs in soils from open (p ) 0.15) or forested (p ) 0.39) areas. The lack of significant correlation in the latter two cases may be due to small sample sizes (n ) 8). Normalizing TEQs with total organic content resulted in similar results to the previous TEQ comparisons. Urban soils had the highest levels and agricultural soils had the lowest concentration of dioxins (Table 3). Our findings are consistent with other studies, showing higher levels of dioxins in urban lands compared to rural lands in Austria (5), Britain (6), and Spain (12, 36) (Figure 2). The comparative elevation of dioxin concentrations in urban areas may be due to proximity to combustion processes and other sources that are generally concentrated in urban areas (4, 6, 36). The two highest dioxin values measured (19 and 9.5 ng/kg TEQ) were soil samples from urban areas within the city of Tacoma. These two samples were collected from residential areas; however, the two sites appear to be closer to industrial areas (potential sources) than the other urban sites sampled. Tacoma has several hazardous waste cleanup sites with confirmed dioxin contamination (31). Possible historical sources of dioxin in Tacoma include emissions from wood waste boilers, smelters, pulp, and paper mills, municipal incinerators, and other incinerators. Land Use Dioxin Profile Comparisons. PCDD/PCDF homologue profiles can vary with land use (Figure 3). From the profile plot (Figure 3) it appears that forest soils had higher levels of TCDF, HpCDD, and HxCDD. A DFA revealed a significant difference in dioxin profiles among land uses (Wilk’s Lambda ) 0.230, F[30,209] ) 4.531, p < 0.00005). A plot of the soil sample scores on the first two canonical axes is presented in Figure 4. The closer samples are to one another the more similar their dioxin profiles. Nonoverlapping circles have significantly different dioxin profiles; samples can be discriminated and correctly classified based on their dioxin profiles. The first two canonical axes account for 0.786 and 0.155 proportion of the variation in the original data set, respectively. Groups were discriminated based on levels of OCDD, OCDF, and PeCDF for the first canonical axis (with standardized canonical coefficients of 1.317, 1.183, and 0.790, respectively) and OCDD, HxDF, and PeCDD for the second canonical axis (with standardized canonical coefficients of -1.08, -0.861, and -0.753 respectively).
FIGURE 4. Canonical score plots of individual soil samples from a discriminant analysis based on 10 dioxin/furan homologues. Circles represent 95% confidence limits of the mean (centroid) for the four land use categories. Canonical axes one and two represent 0.786 and 0.155 of the variation, respectively. Insert: contributions (standardized canonical coefficients) from homologues to the formation of the canonical axes. Thirty-five percent of the soil samples (n ) 29) in the jackknifed classification matrix were incorrectly classified. Seventy-eight percent of the agriculture samples were correctly classified, while only 25, 38, and 57% of urban, forest, and open samples were correctly classified, respectively. Soil groups were differentiated primarily as a result of differences in OCDD/Fs, PeCDF, and HxCDD in the first canonical axis and by OCDD, HxCDF, and HpCDD differences in the second canonical axis. Dioxin soil profiles of agricultural and forested land were effectively different; there were no misclassification of samples between these two land uses. Among urban, open, and forest soil samples, there was quite a bit of overlap and misclassification. Over half the forest samples were misclassified to either urban or open land use, thus for the subsequent DFA soil and source comparison, soil samples were classified into two groups, an agriculture group and a group composed of the remaining land uses (urban, forest, open). Source Discrimination. A preliminary DFA was conducted to determine appropriate dioxin source categories to use for subsequent comparisons with soil sample results. The preliminary DFA revealed a close association between biosolids and PCP; thus these two source types were grouped together. Other classification categories were wood (pulp/ paper mill), metal associated processes, municipal incinerators, cement kilns, and medical incinerators (Figure 5). Canonical plots for the DFA of the dioxin source profiles are presented in Figure 6. There were significant differences in dioxin profiles among source samples (Wilk’s Lambda ) 0.00219, F[50,135] ) 7.673, p < 0.00005), with the first two canonical axes accounting for 0.701 and 0.145 of the original variation, respectively. In the jackknife classification matrix 15 samples (34%) were incorrectly classified. Dioxin profiles of the biosolids/PCP sample group were differentiated quite well among all source categories, with all biosolids/PCP samples being correctly classified, and no other source samples incorrectly identified as belonging to the biosolids/ PCP group (Figure 6). Groups were discriminated based on levels of TeCDF, PeCDD, HpDD, and OCDD for the first canonical axis (with standardized canonical coefficients of 1.985, 1.682, 1.697, and 1.092, respectively) and TeCDF, HpDD, and OCDD for the second canonical axis (with VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5173
FIGURE 5. Dioxin/furan homologue profiles of dioxin source samples normalized to sum total. FIGURE 7. Canonical score plots of dioxin source and Washington soil samples from a discriminant analysis based on the 10 dioxin/ furan homologues. Circles represent 95% confidence limits of the mean (centroid) of the 9 groups. Canonical axes one and two represent 0.609 and 0.175 of the variation, respectively. Insert: contributions (standardized canonical coefficients) from homologues to the formation of the canonical axes.
FIGURE 6. Canonical score plots of dioxin source samples from a discriminant analysis based on the 10 dioxin/furan homologues. Circles represent 95% confidence limits of the mean (centroid) of the six source groups. Canonical axes one and two represent 0.701 and 0.142 of the variation, respectively. standardized canonical coefficients of 1.094, 1.305, and 1.26, respectively). Dioxin profiles of pulp and paper mills (wood products) were similar, even though samples were from different facilities and different media (effluent, fly ash, and a wood ash sample). Nutrilime, a fertilizer product consisting of pulp mill fly ash, was included with the pulp and paper mill samples. Wood samples were dominated by tetra to penta CDDs and CDFs, as well as overall low values of PCDFs compared to PCDDs, with the exception of TCDF (Figure 5). Metal-associated materials had similar dioxinprofiles. The metal cluster included emission samples from an aluminum remelt furnace, a sample from an activated carbon regenerating facility, and fertilizer-type product samples (Frit products, Bay Zinc products, NuLife All Purpose Trace Elements, and McLendon Weed and Feed). It is believed that fertilizer products included with this group have material components that are derived from steel mill flue dust (31). These samples had a common dioxin source profile and are all associated with metal processing or combustion of metalcontaining materials. Samples that were from incinerators (hospitals, municipal, and cement kiln) showed a greater variation in the DFA plot than other source groups (Figure 6). Few of the actual samples from these groups were within their 96% confidence intervals. Slight differences among the various combustion sources were observed. Samples from hospitals tended to have dioxin profiles dominated by penta to hepta CDDs and CDFs. Samples from a cement kiln incinerator were characterized by having a higher proportion of the low and higher chlorinated congeners groups. Soil and Source Discrimination. For the soil and source comparison, data from both soil and source samples were 5174
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
included for use in discriminant function analysis. On the basis of the previous soil DFA analysis, soil samples were categorized into two groups, an agricultural group and soil group (urban, open, forest). There was a significant difference among groups (Wilk’s Lambda ) 0.0151403, F[70,654] ) 9.826, p < 0.00005). Canonical plots of the samples and their means are presented in Figure 7. Groups were discriminated based on levels of OCDD, OCDF, and PeCDF for the first canonical axis (with standardized canonical coefficients of 1.072, 0.716, and 0.664, respectively) and OCDD, HpCDD, and HxCDDD for the second canonical axis (with standardized canonical coefficients of 0.828, 0.777, and 0.626, respectively). In the Jackknife classification matrix, 28% of the samples were incorrectly classified. The most interesting result was that agricultural soil samples were not clearly discriminated from the biosolid/PCP source samples. Of the 54 agricultural soil samples, 5 were classified to the biosolid/PCP group and 4 to the general soil group (urban, open, forest). Only one of the six biosolid/PCP samples was incorrectly classified; it was classified as belonging to the agricultural group. Agricultural lands where biosolids were spread were specifically avoided in this study. Thus, it is not believed that biosolids were the source for dioxins in agricultural soils. Stormwater inflow and street runoff into combined sewer lines may account in part for associations observed between biosolids and soil dioxin profiles. Dioxin in both soil and biosolids may be a result of atmospheric deposition. PCPs are potentially a primary source of dioxin in biosolids, specifically, contaminants in cotton clothing produced from raw materials that have been treated with PCP (37, 38). Biosolids/PCP profiles were characterized by relatively higher-level chlorinated homologues, the hepta and octa CDDs and CDFs (Figure 6). Other PCDD/PCDFs sources had a more even distribution of homologues than both soil and biosolids/PCP sample profiles. It is possible that dioxins in agricultural soils and biosolids are a result of natural processes. Samples free from anthropogenic impacts have been reported to have similar profiles, dominated by hepta and octa CDDs (8, 39). By use of homologue total concentrations, we lose some discriminatory sensitivity as dioxin sources and environmental samples can have similar homologue totals but may differ in specific congeners and isomers (8, 39).
With the possible exception of agricultural soils having a similar dioxinprofile as the biosolids/PCP source group, soil samples were not clearly linked to any specific dioxin source evaluated. The relative concentration of the higher chlorinated homologues was greater in soil samples compared to source samples. This difference in homologue profiles can probably be attributed to decreasing degradation rates with increasing chlorination (40), evaporation of lower chlorinated PCDD/Fs, or biogenic formation of higher chlorinated compounds. Dioxin in soils have been exposed to atmospheric weathering (photolysis and oxidation) resulting in profiles with relatively high concentrations of HpCDDs and OCDDs. Dioxins in biosolids have also been exposed to degradation processes, whereas other source samples had little opportunity for degradation. Not all sources of dioxins were accounted for in Washington state. Dioxin sources may not be limited to Washington state. Dioxins can be transported long distances once released into the atmosphere (27, 41, 42). Samples on the remote western coast of Washington (Olympic National Park) had detectable levels of dioxins that were not appreciably lower than samples located east of them. A potential dioxin source that has not been quantified is fine particulate matter that has migrated from Asia to North America (43, 44). In addition, combustion of fuel and refuse by vessels in transit along the coast could also contribute to the atmospheric deposition of dioxins in Washington state soils.
Acknowledgments The authors are grateful to the Washington State Department of Ecology for their support of this work. We thank S. Golding, P. Marti, J. Summers, K. Sinclair, N. Glenn, G. Hoyle-Dodson, S. Treccani, R. Nalley, B. Stone, T. Yelton, C. Stonick, L. Tabor, and D. Serdar for their help in collecting samples. In addition we would like to thank the Washington State Fertilizer Technical Advisory Group, the numerous reviewers who helped to bring this study to completion, as well as the Illinois Natural History Survey.
Literature Cited (1) Rogowski, D.; Golding, S.; Bowhay, D.; Singleton, S. Final report, Screening survey for metals and dioxins in fertilizers, soil amendments, and soils in Washington State. Washington State Department of Ecology, WDOE 99-309, 1999. (2) Malisch, R. Update of PCDD/PCDF-Intake from food in Germany. Chemosphere 1998, 37, 1687-1698. (3) Interim Procedures for estimating risks associated with exposure to mixtures of chlorinated dibenzo-p-dioxins and dibenzofurans (CDDs and CDFs) and 1989 update; U.S. EPA: Washington, DC, EPA/625/3-89/016, 1989. (4) Rotard, W.; Christmann, W.; Knoth, W. Background levels of PCDD/F in soils of Germany. Chemosphere 1994, 29, 21932200. (5) Boos, R.; Himsl, A.; Worst, F.; Prey, T.; Scheidl, K.; Sperka, G.; Gla¨ser, O. Determination of PCDDs and PCDFs in soil samples from Salzburg, Austria. Chemosphere 1992, 25, 283-293. (6) Creaser, C. S.; Fernandes, A. R.; Al-Haddad, A.; Harrad, S. J.; Homer, R. B.; Skett, P. W.; Cox, E. A. Survey of background levels of PCDDS & PCDFs in UK soils. Chemosphere 1989, 18, 767-776. (7) Creaser, C. S.; Fernandes, A. R.; Harrad, S. J.; Cox, E. A. Levels and sources of PCDDs and PCDFs in urban British soils. Chemosphere 1990, 21, 931-938. (8) Rappe, C.; Bergek, S.; Fiedler, H.; Cooper, K. R. PCDD and PCDF contamination in catfish feed from Arkansas, USA. Chemosphere 1998, 36, 2705-2720. (9) Kjeller, L.-O.; Jones, K. C.; Johnston, A. E.; Rappe, C. Increases in the polychlorinated dibenzo-p-dioxin and -furan content of soils and vegetation since the 1840s. Environ. Sci. Technol. 2001, 25, 1619-1627. (10) Alcock, R. E.; Jones, K. C. Dioxins in the environment: a review of trend data. Environ. Sci. Technol. 1996, 30, 3133-3143.
(11) Legendre, P.; Legendre, L. Numerical Ecology; Elsevier Science B. V.: Amsterdam, 1998. (12) Jime´nez, B.; Eljarrat, E.; Herna´ndez, L. M.; Rivera, J.; Gonza´lez, M. J. Polychlorinated dibenzo-p-dioxins and dibenzofurnas in soils near a clinical waste incinerator in Madrid, Spain. Chemosphere 1996, 32, 1327-1348. (13) Jime´nez, B.; Eljarrat, E.; Herna´ndez, L. M.; Rivera, J.; Gonza´lez, M. J. Congener specific analysis of polychlorinated dibenzop-dioxins and dibenzofurans in crabs and sediments from the Venice and Orbetello Lagoons, Italy. Environ. Sci. Technol. 1998, 32, 3853-3861. (14) Bonn, B. A. Polychlorinated dibenzo-p-dioxins and dibenzofurans in the remote North Atlantic marine atmosphere. Environ. Sci. Technol. 1998, 32, 729-735. (15) Grundy S. L.; Bright, D. A.; Dushenko, W. T.; Dodd, M.; Englander, S.; Johnston, K.; Pier, D.; Reimer, K. J. Dioxin and furan signatures in northern Canadian soils: correlation to source signatures using multivariate unmixing techniques. Chemosphere 1997, 34, 1203-1219. (16) Sakurai, T. Dioxins in aquatic sediment and soil in the Kanto region of Japan: major sources and their contributions. Environ. Sci. Technol. 2003, 37, 3133-3140. (17) Peek, D. C.; Butcher, M. K.; Shields, W. J.; Yost, L. J.; Maloy, J. A. Discrimination of aerial deposition sources of polychlorinated dibenzo-p-dioxin and polychlorinated dibenzofuran downwind from a pulp mill near Ketchikan, Alaska. Environ. Sci. Technol. 2002, 36, 1671-1675. (18) Fattore, E.; Benfenati, E.; Mariani, G.; Fanelli, R. Patterns and sources of polychlorinated dibenzo-p-dioxins and dibenzofurans in sediments from the Venice Lagoon, Italy. Environ. Sci. Technol. 1997, 31, 1777-1784. (19) Washington Agricultural Statistics Service (WASS). Washington Agricultural Statistics 1996-1997; WASS: Washington State Department of Agriculture and U.S. Department of Agricultural Statistics Service; 1997. (20) U.S. Department of Agriculture (USDA). Census of Agriculture Volume 1: Part 47, Chapter 1, Washington State and County Data State Data. USDA- National Agricultural Statistics Service. U.S. Department of Commerce, 1995. (21) Schreiber; R. Washington Minor Crops: Washington State University Agricultural Publications MISC0181, 1995. (22) U.S. Department of Census. Urbanized area Boundary Map; U.S. Department of Census: Bureau of the Census, 1990, Washington State, 1991. (23) Tetra Tech, Inc. Recommended Protocols and Guidelines for Measuring Selected Environmental Variables in Puget Sound. Puget Sound Estuary Program (PSEP); Tetra Tech, Inc. for EPA Region 10, Seattle, Washington, 1996. (24) U.S. Environmental Protection Agency (EPA). Estimating Exposures to Dioxin-Like Compounds. Volume II: Properties Sources, Occurrences and Background Exposures. (External Review Draft); U.S. EPA: Washington, DC, EPA/600-6-88/005Cb, 1994. (25) Van den Berg, M.; Birnbaum, L.; Bosveld, A. T. C.; Brunstro¨m, B.; Cook, P.; Feeley, M.; Giesy, J. P.; Hanberg, A.; Hasegawa, R.; Kennedy, S. W.; Kubiak, T.; Larsen, J. C.; Leeuwen, F. X. R. v.; Liem, A. K. D.; Nolt, C.; Peterson, R. E.; Poellinger, L.; Safe, S.; Schrenk, D.; Tillitt, D.; Tysklind, M.; Younes, M.; Wærn, F.; Zacharewski, T. Toxic equivalency factors (TEFs) for PCBs, PCDDs, PCDFs for humans and wildlife. Environ. Health Persp. 1998, 106, 775-792. (26) Sokal, R. R.; Rohlf, F. J. Biometry; W. H. Freeman and Company: New York, 1995. (27) Baker J. I.; Hites, R. A. Polychlorinated dibenzo-p-dioxins and dibenzofurans in the remote North Atlantic marine atmosphere. Environ. Sci. Technol. 1999, 33, 14-20. (28) Rappe, C.; O ¨ berg, L. The US EPA National Sewage Sludge Survey (NSS) - A recalculation. Ed. R. Hites. Dioxin 1997, 17th International Symposium on Chlorinated Dioxins and Related Compounds. Organohalogen Compd. 1997, 32, 94. (29) Duarte-Davidson, R.; Sewart, A.; Alcock, R. E.; Cousins, I. T.; Jones; K. C. Exploring the balance between sources, deposition, and the environmental burden of PCDD/Fs in the U.K. terrestrial environment: and aid to identifying uncertainties and research needs. Environ. Sci. Technol. 1997, 31, 1-11. (30) Hagenmaier, H.; Lindig, C.; She, J. Correlation of environmental occurrence of polychlorinated dibezo-p-dioxins and dibenzofurans with possible sources. Chemosphere 1994, 29, 21632174. (31) Yake, W.; Singleton, S.; Erikson, K. Washington State Dioxin Source Assessment, Washington State Department of Ecology, 1988; WDOE 98-320. VOL. 39, NO. 14, 2005 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
5175
(32) Zar, J. Biostatistical Analysis; Prentice Hall: Englewood Cliffs, NJ, 1984. (33) Mu ¨ ller, J.; Muller, R.; Goudkamp, K.; Shaw, M.; Mortimer, M.; Haynes, D.; Burniston, D.; Symons, R.; Moore, M. Dioxins in Soil in Australia. Australian Government Department of the Environment and Heritage, Canberra National Dioxins Program Technical Report No. 5, 2003; http://www.deh.gov.au/industry/ chemicals/dioxins/report-5/#download. (34) Khizbullin, F. F.; Estrina, G. Y.; Mavrodieva, N. N.; Chaziev, F. S.; Kruglov, E. Y.; Amirova, Z. K. Pollution of agricultural landscapes of Bashkirtosan with the herbicide 2,4-D and dioxins. Meditsina truda i promyshlennaia ekologiia. 1997, 8, 26-31. (35) Matscheko, N.; Tysklind, M.; De Wit, C.; Bergek, S.; Andersson, R.; Sellstrom, U. Application of sewage sludge to arable landsoil concentrations of polybrominated diphenyl ethers and polychlorinated dibenzo-p-dioxins, dibenzofurans, and biphenyls, and their accumulation in earthworms. Environ. Toxicol. Chem. 2002, 21, 2515-2525. (36) Schuhmacher, M.; Granero, S.; Llobet, J. M.; de Kok, H. A. M.; Domingo, J. L. Assessment of baseline levels of PCDD/F in soils in the neighborhood of a new hazardous waste incinerator in Catalonia, Spain. Chemosphere 1997, 35, 1947-1958. (37) Horstmann, M.; Mclachlan, M. S. Concentrations of polychlorinated dibenzo-p-dioxins (PCDD) and dibenzofurans (PCDF) in urban runoff and household wastewater. Chemosphere 1995, 31, 2887-2896. (38) Jones, K. C.; Sewart, A. P. Dioxins and furans in sewage sludge: a review of their occurrence and sources in sludge and of their environmental fate, behavior, and significance in sludgeamended agricultural systems. Crit. Rev. Environ. Sci. Tec. 1997, 27, 1-86.
5176
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 39, NO. 14, 2005
(39) Gaus, C.; Brunskill, G. J.; Weber, R.; Pa¨pke, O.; Mu ¨ ller, J. F. Historical PCDD inputs and their source implications from dated sediment cores in Queensland (Australia). Environ. Sci. Technol. 2001, 35, 4597-4603. (40) Choi, W.; Hong, S. J.; Chang, Y.-S.; Cho, Y. Photocatalytic degradation of polychlorinated dibenzo-p-dioxins on TiO2 film under UV or solar light irradiation Environ. Sci. Technol. 2000, 34, 4810-4815. (41) Brzuzy, L. P.; Hites, R. A. Global mass balance for polychlorinated dibenzo-p-dioxins and dibenzofurans. Environ. Sci. Technol. 1996, 30, 1797-1804. (42) Tysklind, M.; Fa¨ngmark, I.; Marklund, S.; Lindskog, A.; Thaning, L.; Rappe, C. Atmospheric transport and transformation of polychlorinated-p-dioxins (PCDDs) and -furans (PCDFs). Environ. Sci. Technol. 1993, 27, 2190-2197. (43) Berntsen, T. K.; Karlsdottir, S.; Jaffe, D. A.; Berntsen, T. K.; Karlsdottir, S.; Jaffe, D. A. 1999. Influence of Asian emissions on the composition of air reaching the northwestern United States. Geophys. Res. Lett. 1999, 26, 2171-2174. (44) Jaffe, D.; McKendry, I.; Anderson, T.; Price, H. Six ‘new’ episodes of trans-Pacific transport of air pollutants. Atmos. Environ. 2003, 37, 391-404. (45) Agricultural soils in Washington State have low levels of dioxins (0.14 ng/kg TEQ) that do not appear to be directly attributed to known dioxin sources.
Received for review December 27, 2004. Revised manuscript received May 13, 2005. Accepted May 16, 2005. ES047945R
