JEQ Journal of Natural Resources and Life Sciences Education
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Published online 5 April 2007
Published in J Environ Qual 36:654-663 (2007)
DOI: 10.2134/jeq2006.0413
© 2007 American Society of Agronomy, Crop Science Society of America, and Soil Science Society of America
677 S. Segoe Rd., Madison, WI 53711 USA
This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Robinson, G. R.
Right arrow Articles by Sibrell, P. L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Robinson, G. R., Jr.
Right arrow Articles by Sibrell, P. L.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Robinson, G. R.
Right arrow Articles by Sibrell, P. L.
Related Collections
Right arrow Agricultural Pesticides
Right arrow Pesticides
Right arrow Heavy Metals

TECHNICAL REPORTS

Heavy Metals in the Environment

Assessment of Contamination from Arsenical Pesticide Use on Orchards in the Great Valley region, Virginia and West Virginia, USA

Gilpin R. Robinson, Jr.a,*, Peter Larkinsa, Carol J. Boughtonb, Bradley W. Reeda and Philip L. Sibrellb

a USGS, 954 National Center, Reston, VA USA 20192
b USGS, 11649 Leetown Road, Kearneysville, WV USA 25430

* Corresponding author (grobinso{at}usgs.gov)

Received for publication September 28, 2006.

   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Lead arsenate pesticides were widely used in apple orchards from 1925 to 1955. Soils from historic orchards in four counties in Virginia and West Virginia contained elevated concentrations of As and Pb, consistent with an arsenical pesticide source. Arsenic concentrations in approximately 50% of the orchard site soils and approximately 1% of reference site soils exceed the USEPA Preliminary Remediation Goal (PRG) screening guideline of 22 mg kg–1 for As in residential soil, defined on the basis of combined chronic exposure risk. Approximately 5% of orchard site soils exceed the USEPA PRG for Pb of 400 mg kg–1 in residential soil; no reference site soils sampled exceed this value. A variety of statistical methods were used to characterize the occurrence, distribution, and dispersion of arsenical pesticide residues in soils, stream sediments, and ground waters relative to landscape features and likely background conditions. Concentrations of Zn, Pb, and Cu were most strongly associated with high developed land density and population density, whereas elevated concentrations of As were weakly correlated with high orchard density, consistent with a pesticide residue source. Arsenic concentrations in ground water wells in the region are generally <0.005 mg L–1. There was no spatial association between As concentrations in ground water and proximity to orchards. Arsenic had limited mobility into ground water from surface soils contaminated with arsenical pesticide residues at concentrations typically found in orchards.

Abbreviations: GIS, Geographic Information System • ICP–MS, inductively coupled plasma mass spectrometry • LRL, laboratory reporting level • NLCD, National Land Cover Data Set • NWIS, USGS National Water Information System • PRG, USEPA Preliminary Remediation Goal


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
ARSENICAL pesticide use was extensive and widespread in agricultural applications from the 1920s to the late 1950s, and largely ceased agricultural use by the early 1960s in the USA (Fig. 1). During this time period, lead arsenate was the most extensively used arsenical pesticide (Peryea, 1998), particularly in apple (Malus sylvestris Mill.) orchards. Other metal-bearing pesticides and fungicides, such as copper acetoarsenite (C4H6As6Cu4O16, Paris Green), Bordeaux Blue (a mixture of copper sulfate pentahydrate, CuSO4·5H2O, and calcium hydroxide, Ca(OH)2), and phenylmercuric chloride (C6H5ClHg) were used to a lesser degree in orchards (Nardin, 1971; Peryea, 1998; Shepard, 1939; Veneman et al., 1983).


Figure 1
View larger version (16K):
[in this window]
[in a new window]

  		Fig. 1. Estimated use of lead arsenate as an agricultural pesticide in the United States during the 20th century. Arsenical pesticide use was most extensive and widespread in agricultural applications from the 1920s to the late 1950s, and largely ceased its agricultural use by the early 1960s in the nation. Data compiled from information in Buckingham and Brooks (2001) and Peryea (1998).

 
During the time of intense arsenical pesticide use, federal and state pesticide laws did not require farmers to keep accurate records of the quantity, location, and type of pesticides used on their property, thus the quantity and distribution of past arsenical pesticide use is not known in the region. Based on estimates from other areas (D'Angelo et al., 1996), cumulative application over the period of arsenical pesticide use may have been as much as 22.4 g m–2 of As and 100 g m–2 of Pb in orchard areas.

Previous studies documented that elevated concentrations of As, Pb, and sometimes Cu occurred in the soils of former apple orchards (Jones and Hatch, 1937; Veneman et al., 1983; Renshaw et al., 2006) and that elevated As occurred in stream-bottom sediments from drainage basins containing orchards (McManus, 1976; Robinson and Ayuso, 2004; Renshaw et al., 2006). In minimally disturbed orchard soils, As and Pb are largely retained in the top 25 cm of the soil horizon (Peryea and Creger, 1994); intra-soil redistribution of these metals occurs with As showing greater mobility than Pb (Peryea and Creger, 1994), particularly in alkaline or phosphate-enriched soils (Peryea, 1991), but intra-soil migration appears to be limited in most settings (Veneman et al., 1983; Peryea, 1998). In the pH range of 4.5 to 7.5 typical of many soils, As and Pb are retained in soils and sediments through sorption to fine-grained Fe and Mn oxyhydroxide minerals (Padmanabham, 1983; Appel and Ma, 2002; Goldberg, 2002) and by remineralization as secondary mineral phases having limited solubility (Renshaw et al., 2006). The studies of Wauchope (1975), Woolson (1977), Duble et al. (1978), Richardson et al. (1978), and Renshaw et al. (2006) all indicate that erosion of contaminated soils is important in the dispersion of As and metals within drainage basins. Redevelopment of former orchard sites as residential properties is likely to increase the dispersal of sediment as airborne dust and runoff into waterways, potentially increasing the environmental exposure of humans and wildlife to As and xenobiotic metals. The significance of this exposure has not been adequately evaluated and the potential extent of contamination from arsenical pesticides has not been investigated in the region.

The objectives of this study were to sample soils and sediments and review water quality data to evaluate the environmental significance of arsenical pesticide residues in the karst environment of the northern Great Valley region, Virginia and West Virginia. The research questions addressed by this study include:

  1. Do the soils and stream-bottom sediments in and adjacent to orchard sites where arsenical pesticides were used contain elevated concentrations of As and metals relative to likely background conditions?
  2. Based on historic land use, agricultural census, and other data, what is the likely extent of potential contamination from arsenical pesticide residues in the region?
  3. Using risk-based concentration criteria, do the soils, sediments, and waters in the region pose a risk to human health or ecosystem function?
  4. Does redevelopment of former orchard sites as residential property increase the dispersal of contaminated sediment and increase the exposure of humans and wildlife to As and toxic metals?


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 
Study Area
Historic ground water quality data were reviewed and soils and stream-bottom sediments were collected from 2004 to 2006 to sample orchard, former orchard, and reference sites in Clarke and Frederick Counties, Virginia and Berkeley and Jefferson Counties, West Virginia. Orchard cultivation was extensive in this area from the 1920s to the 1960s during the time arsenical pesticides were commonly used on orchards. Approximately 5% of the study area was covered by orchard areas where arsenical pesticides were used. Orchard cultivation was most prevalent in areas of carbonate bedrock in the Valley and Ridge Province (Fig. 2A), where approximately 10% of the area of carbonate bedrock was once in orchard cultivation. Approximately 1.5% of the area of Valley and Ridge Province non-carbonate bedrock and 0.6% of Blue Ridge Province bedrock were once in orchard cultivation (Fig. 2A). Contamination from the past use of arsenical pesticides on orchard crops is of particular concern with regard to the rapid pace of residential development occurring on former agricultural lands in the region.


Figure 2
View larger version (53K):
[in this window]
[in a new window]

  		Fig. 2. Map of study area in Clarke and Frederick Counties, Virginia, and Berkeley and Jefferson Counties, West Virginia. (A) Location of public supply ground water well sites screened for As concentration in relation to bedrock geology. (B) Location of soil and stream-bottom sediment sample sites in relation to orchards inferred to have used arsenical pesticides and stream network.

 
No reliable and complete land cover data identifying areas under orchard cultivation during the time period of extensive use of arsenical pesticides between the 1920s and 1960s was available for the study area. A spatial dataset of orchard areas where arsenical pesticides were likely used was compiled using twenty-seven USGS 7.5 min series topographical maps, published between 1943 and 1972 at 1:24 000 scale, covering the study area. For each map area, the oldest 7.5 series topographic map available from the USGS map archive was used, going back only as far as the 1920s when use of arsenical pesticides started. Orchard areas on the topographic maps were traced and the maps were then scanned and geographically referenced using a raster editing program called ERDAS Imagine, producing rectified images. Orchard areas were digitized from these rectified images using ArcGIS v 9.1 to develop a spatial database that included orchard location polygons, polygon areas, source USGS topographic map, topographic map publication year, county, and state (Reed et al., 2006).

Ground Water Well Data
Data for As concentrations in ground water wells that penetrate bedrock in the study area were obtained from the USGS National Water Information System (NWIS) database and from state records on public supply wells collected for compliance with the Safe Drinking Water Act during the years of 1991 to 2006 and Virginia state regulations for the years 1975 to 1991. Sources of data include 24 ground water well records in the NWIS database, 26 public supply records collected by the Virginia Department of Health, Office of Drinking Water and 53 public supply records collected by the West Virginia Department of Health and Human Resources, Bureau for Public Health. Four of the NWIS sites were duplicates of West Virginia public supply wells. The criteria used for public supply well selection were similar to those described by Ayotte et al. (1999). Analytical results for source water (pre-treatment) were used when available. Water samples from the public supply wells were typically re-analyzed every 3 to 5 yr and the laboratory reporting level (LRL) for selected samples from the wells was variable over the sample period. For each well, the analytical results from the laboratory with the lowest LRL limit were selected for site evaluation. All selected analytical data for public supply wells in Virginia have a common highest LRL limit of 0.002 mg L–1 for As; all selected analytical data for public supply wells in West Virginia have a common highest LRL limit of 0.005 mg L–1. For comparison, the NWIS data have a common highest LRL limit of 0.004 mg L–1 As. The public supply well sample sites are shown in Fig. 2A in relation to the spatial distribution of orchard sites that likely used arsenical pesticides.

Soil and Sediment Sampling
Soils and stream-bottom sediments were sampled from 2004 to 2006 at orchard, former orchard, and reference sites in the study area. Soil and sediment sample sites are shown in Fig. 2B in relation to the spatial distribution of orchard sites that likely used arsenical pesticides. Most soil and sediment sites are located in areas of carbonate bedrock (Valley and Ridge carbonate rock in Fig. 2A). The statistical analysis of sample groups was restricted to sample sites located on carbonate bedrock, where arsenical pesticide use was most prevalent (Fig. 2A).

Representative samples of untilled soils (B horizon) were collected to depths of 15 to 25 cm (depending on depth of regolith) by soil auger and as grab samples. The sampled soils are Duffield-Ryder-Nollville and Hagerstown-Funkstown silt loam soil types derived from calcitic-limestone bedrock material (USDA-NRCS, 2006). Soil pH for these soil types range from 5.1 to 7.3 in the study area (USDA-NRCS, 2006). These B-horizon soils had organic carbon contents of less than 0.1 g kg–1. Stream-bottom sediments were collected from deposition sites in flowing streams with catchment areas on the order of 1 to 25 km2, representing local sediment input. The median catchment area of the stream-bottom sediment sample sites was 6.7 km2.

Soil and sediment samples were air-dried and homogenized. Soil samples were sieved to <200 µm (80 mesh). Sediment samples were disaggregated in deionized water, sieved to <200 µm, and then air-dried before analysis. Sieved soil and sediment samples were pulverized before analysis.

Element concentrations in soil and sediment samples were determined by inductively coupled plasma mass spectrometry (ICP–MS) following sample digestion in aqua regia by Activation Laboratories (ActLabs), Ancaster, ON, Canada (Ultratrace-2 method). Mercury was measured using cold vapor analysis techniques by ActLabs (Code 1G method). The ICP–MS and cold vapor analytical techniques are described in more detail in Taggart (2002). Duplicate samples were submitted at a rate of 5% of total samples. Data were accepted if the relative standard deviation was <15% for the duplicate samples at five times the limit of detection. Analytical methods, quality assurance and control methods, and laboratory accreditation information are described in Activation Laboratories (2006).

Soil Site Variables
The data for soil chemistry were grouped by historic land use status into (i) orchard and former orchard sites or (ii) reference sites with no evidence of previous orchard cultivation. A few of the soil sample sites in developed areas classified as former orchard land use appear to have been extensively landscaped and remediated during development. These samples were classified into a third group as disturbed land sites. These samples were excluded from the statistical analysis of the orchard and reference site soil groups, but they had As and Pb concentrations similar to reference site soils.

Stream-Bottom Sediment Site Variables
Five landscape variables were defined for the stream-bottom sediment sample sites and used as explanatory factors describing the observed chemical variation in stream-bottom sediments. The "Orchard" variable classifies stream-bottom sediment sites according to the proportion of their resident catchment basin covered by orchard status land. The "Popden" variable measures the population density of the census tract enclosing the catchment area. The "Dev_land" variable measures the proportion of the resident catchment basin of the sediment site covered by developed land and was also used as a proxy for the probability of metal contamination related to industrial or transportation sources. The "Rowcrop" variable measures a row crop agricultural index classified according to the percentage of the resident catchment basin covered by agricultural row crops. The "Forest" variable measures a forested land index classified according to the percentage of the catchment area covered by untilled forest and woodlands, and includes untilled land in orchard cultivation in 1992.

Catchment basins for the stream-bottom sediment sites included all land drained upstream and upslope from the sample site location. These catchment basins were defined using a Geographic Information System (GIS). A digital elevation model, or DEM, of the study area at 10 m resolution (NED 1/3 Arc Second; USGS, 1999) was used in conjunction with the point delineation feature in the ArcHydro Tools extension of ArcGIS v 9.1 (Maidment, 2000) to process the most likely locations of catchment basins. A vector polygon outlining the catchment area and its area measurement were defined for 55 of the 57 sediment sample sites located on carbonate bedrock; the data used to define the catchment basins was insufficient to adequately define a polygon and area for two sediment sample sites. The catchment basin information for each site was compared with the mapped orchard areas where arsenical pesticides were likely used, as well as land use and land cover information on developed land, forest cover, and row crop cultivation in the area to measure the fractional area of the catchment basin covered by the landscape variable features. The land use and land cover data used in this analysis are from the 1992 National Land Cover Data Set (NLCD) (Vogelmann et al., 2001). The land classes defined in this dataset include: water, residential, commercial+industrial+transportation, forest, pasture+hay, and row crops. The developed land variable (Dev_land) used in this study combined the residential and commercial+industrial+transportation NLCD categories. The population density variable was calculated using the 2000 Census population data by census tract area (U.S. Census Bureau, 2000).

A reconnaissance comparison of the 1992 NLCD classifications with more recent aerial imagery available via GoogleEarth indicated that the 1992 NLCD classification is generally consistent with present day landscape conditions, although it is evident that some agricultural land is being replaced by residential development.

Ground Water Well Site Variables
Four landscape variables were defined for the ground water well sites. The "Geology Type" variable, or the bedrock geologic setting for each well, was classified into three groups: (i) Blue Ridge Province rocks, (ii) Valley and Ridge Province carbonate rocks, and (iii) Valley and Ridge Province non-carbonate rocks (Fig. 2A). The "Distance to Orchard" variable is the minimum distance of the well from an orchard area that likely used arsenical pesticides. The "Or_buffer" variable is the percentage of a 250-m buffer area surrounding the well site that was covered by orchards that used arsenical pesticides. This variable is a measure of the intensity of arsenical pesticide use in the local vicinity of the well site. The "Or_bsn" variable, a measure of the proportion of the drainage basin catchment area enclosing the well site that was covered by orchards that likely used arsenical pesticides, measures the intensity of arsenical pesticide use over a somewhat larger drainage basin area, averaging approximately 7 km2.

Statistical Methods
Normality tests were made on the geochemical data for soils and sediments. Most constituents analyzed for this report were neither normally nor lognormally distributed, but a log transformation generally improved the normality of the data. Nonparametric methods, which do not require assumptions about the distributions of the data, were used for many of the statistical tests in this study.

Nonparametric Kruskal–Wallis tests on the ranks of the soil chemistry data were used to determine whether element concentrations occurred at significantly different frequencies in sample subgroups as defined by site status (orchard or reference site). The null hypothesis for this test is that there is no significant difference between the ranks of the concentrations of a chemical constituent among the site groups. For all hypothesis tests in this study, rejection of the null hypothesis requires that the attained significance level (p), the probability that the observed differences are due to chance rather than the factor tested, is less than 0.05 (95% confidence level). The SAS General Linear Models (GLM) Type III sum of squares procedure (SAS Institute, Inc., 1999) was used to perform the nonparametric tests because the number of samples was not the same for all of the category groups (Helsel and Hirsch, 2002).

Rank-order correlations (Spearman's {rho}) were calculated to measure the direction and strength of association between pairs of geochemical variables (soils and stream-bottom sediments) and between geochemical and landscape variables (stream-bottom sediments).

Factor Analysis
R-mode factor analysis (varimax model, orthogonal solution, standardized data) was used to test if the observed variation of As and metals in soils and sediments were correlated with each other and with landscape variables for the sample sites. The factor analysis dataset for soils and sediments used log-normalized geochemical data; the log transformation of the positively skewed data distributions (Table 1) improved the normality of these data in the dataset. The highest factor loading value for each constituent was used to group the association of these constituents with the factor groups that most strongly influenced their distribution.


View this table:
[in this window]
[in a new window]

  		Table 1. Summary of geochemical statistics for Al, Fe, Cu, Zn, As, Pb, and Hg concentrations in soil samples, characterized by location on orchard or reference sites. Attained significance levels (p value) for Mann–Whitney (MW) tests on element concentrations grouped by site show the probability that the observed differences are due to chance rather than the factor tested.{dagger}

 

   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
CONCLUSIONS
 REFERENCES
 
Soil Data
Summary statistics for Al, Fe, Cu, Zn, As, and Hg concentrations in 183 soil samples, characterized by sample site into orchard (102 samples) and reference site (81 samples) groups, are listed in Table 1. Concentrations of Al and Fe were similar between the orchard and reference site groups; however, soils sampled from orchard and former orchard sites where arsenical pesticides were likely used extensively typically had elevated As and heavy metal concentrations relative to reference soils (Table 1). A scatter plot of soil As and Pb concentrations, differentiated by orchard and reference group status, shows differing data trends and group differences between the orchard and reference site settings (Fig. 3). The orchard soils have a strong linear trend on the log-scale Pb-As plot and the trend of orchard soil data generally falls between and along the trends expected to result from contamination with arsenical pesticide residues. The two forms of lead arsenate commonly used as pesticides (USEPA, 1986) have Pb/As mass ratios of 2.8 (acid lead orthoarsenate PbHAsO4) and 4.6 (basic lead orthoarsenate Pb4(PbOH)(AsO4)3), and these trend lines are shown in Fig. 3 in relation to the soil data. Regression of the Pb and As data for orchard site soils yields a linear equation,

Formula 1[1]
similar to the trend for the basic lead orthoarsenate and consistent with the expected arsenical pesticide trend. Regression of the Pb and As data for the reference sites yields the linear equation,

Formula 2[2]
that is inconsistent with the expected arsenical pesticide trend and poorly describes the data. Residues from arsenical pesticide and herbicide applications, used in non-orchard applications for pest and weed control (Peryea, 1998), may have influenced As and Pb concentrations in some soil samples in the reference site category.


Figure 3
View larger version (18K):
[in this window]
[in a new window]

  		Fig. 3. Scatter plot of soil As and Pb concentrations, characterized by location on orchard or reference sites. Arsenic and Pb trend lines for the two prevalent forms of lead arsenate pesticide are shown as solid lines.

 
The USEPA Risk-Based Concentration Table defines the Preliminary Remediation Goal (PRG) screening guidelines for residential soils. These screening guidelines, based on estimates of cancer and chronic health risks estimated to result from human exposure to contaminated soil, were used as site-screening tools and are not legally enforceable standards (USEPA, 2004a, 2004b). Arsenic concentrations in approximately 50% of orchard site soils exceed the USEPA PRG guideline of 22 mg kg–1 for As (USEPA, 2004a) in residential soil, defined on the basis of combined chronic exposure risk; approximately 1% of reference site soils exceeded the As screening guideline. Approximately 5% of orchard site soils exceeded the USEPA PRG for Pb of 400 mg kg–1 in residential soil (USEPA, 2004b); no reference site soils sampled exceeded this threshold value. All orchard and reference site soils sampled contained Cu, Zn, and Hg concentrations that were less than the PRG guidelines for residential soils.

Under the soil pH range of 5.1 to 7.3, typical for our soil sample sites, As and base metals, such as Pb and Cu, have been found to be retained in soils primarily through sorption to fine-grained Al, Fe, and Mn oxyhydroxide minerals that occur as grain coatings, films, and clay minerals (Rose et al., 1979; Yan-Chu, 1994; Goldberg, 2002). Rank-order correlation (Spearman's {rho}) was used to determine the direction and strength of association of element concentrations in the soils, split by sample group. In reference soils, all metals, with the exception of Pb, correlate moderately to strongly with Al and Fe concentrations, and each other, consistent with their association with Fe and Al oxyhydroxide minerals in the soils (Table 2). In orchard soils, Fe and Al correlate strongly, Pb and As correlate strongly, and the other metals (with the exception of Zn) correlate best with Pb and As, consistent with an association with arsenical pesticide residues, rather than Fe and Al concentrations (Table 2).


View this table:
[in this window]
[in a new window]

  		Table 2. Summary of Spearman's {rho} rank-correlation coefficients for paired element concentrations in soil samples, categorized according to orchard or reference site status. Orchard site soil correlations are shown in the upper right-hand portion of the matrix; reference site soil correlations are shown in the lower left-hand portion of the matrix. Significant correlations, the probability that the observed correlations are due to the relation tested rather than due to chance, at {alpha} = 0.05 (95% confidence level), are shown without parentheses.

 
Factor Analysis of Soil Data
Factor analysis of the soil data for Al, Fe, Cu, Zn, As, Pb, and Hg concentrations, split by orchard and reference groups, further illustrates the different elemental associations in these sample groups (Table 3). For these reference site soils, the strongest association occurs between Al, Fe, Zn, Cu, and Hg concentrations, and a second weaker association between As and Pb. The factor analysis results for the reference site soil data are consistent with an association of Zn, Cu, and Hg with Al and Fe oxyhydroxide minerals in the soil, as the soil concentrations of these metals vary in proportion to the abundance of Fe and Al contained in oxyhydroxide minerals in the soil formed during weathering. The weaker association between As and Pb in these reference soils might indicate a small degree of contamination from arsenical pesticide residues on sites classified as part of the reference group.


View this table:
[in this window]
[in a new window]

  		Table 3. Factor analysis results for Al, Fe, Cu, Zn, As, Pb, and Hg concentrations in soil samples, split into orchard and reference site groups. The largest factor loading for each variable is in italicized font for each factor.

 
In orchard site soils, the strongest geochemical association occurs between Pb, As, Hg, and Cu, with a second weaker association between Fe, Al, and Zn. The factor analysis results for the orchard site soils are consistent with an association of elevated Pb, As, Hg, and Cu concentrations occurring in areas of contamination from arsenical pesticide and other metal-based agricultural chemicals used in orchards. The second weaker association between Fe, Al, and Zn likely reflects the association of Zn with Al and Fe oxyhydroxide minerals in the soil, consistent with the reference soil results.

Stream-Bottom Sediment Data in Relation to Landscape Variables
Summary statistics for Ca, Al, Fe, Cu, Zn, As, and Pb concentrations in 57 stream-bottom sediment samples and land use attributes for the sample sites used in the factor analysis and correlation methods are listed in Table 4. Calcium has a wide range of variation in the stream sediment draining areas of carbonate bedrock, from less than 0.1 weight percent to more than 2.0 g kg–1 (Table 4). Some of these stream-bottom sediments contained as much as 40% of calcite mud by weight. Population density in 2000 by census tract for the sediment sample sites ranged from 19 to 173 people km–2. The "Orchard" variable, a measure of the proportion of the stream-bottom sediment sample site catchment basin covered by orchard and former orchard areas where arsenical pesticides were likely used extensively, ranges from 0 to more than 50%.


View this table:
[in this window]
[in a new window]

  		Table 4. Summary of statistics for Ca, Al, Fe, Cu, Zn, As, and Pb concentrations in stream-bottom sediment samples and the landscape variables for the catchment areas of the sediment sample sites. The "Forest" category includes untilled land in orchard status in 1992. Sediment quality guidelines for freshwater ecosystems are from MacDonald et al. (2000): TEC values are the threshold value of metal concentrations in sediments below which adverse effects in sediment-dwelling organisms are not expected to occur; PEC values are the threshold value of metal concentrations in sediments above which adverse effects on sediment-dwelling organisms are likely to be observed.

 
The direction and strength of association between element concentrations in the stream-bottom sediments are shown by rank-order correlation (Spearman's {rho}, Table 5). Among the major elements, Al and Fe are strongly positively correlated and Ca is strongly negatively correlated with both Al and Fe. In these small-order streams developed on carbonate bedrock, the fine-grained sediment varies from carbonate-rich muds to clay and Fe-hydroxyoxide-rich muds. The positive correlation of sediment Ca concentrations with increasing row crop cultivation, and positive correlation of sediment Fe and Al concentrations having increasing forest and untilled orchard cover (Table 5), is likely related to both higher rates of soil loss from row crop cultivation areas relative to untilled and forested land areas and the presence of higher Ca concentrations in tilled soils relative to untilled soils. We did not sample tilled agricultural soils, but speculate that Ca concentrations may be elevated in these soils due to use of agricultural lime and to the mixing of topsoil with high-Ca limestone regolith during plowing. Metal concentrations are strongly correlated with Fe and Al concentrations (Table 5), consistent with the retention of heavy metals in these sediments through sorption to fine-grained clay and iron and manganese oxyhydroxide minerals.


View this table:
[in this window]
[in a new window]

  		Table 5. Summary of Spearman's {rho} rank-correlation coefficients for paired element and landscape variables for stream-bottom sediment samples. Attained significance levels (p value) for Spearman's {rho} rank-correlation coefficients show the probability that the observed correlations are due to chance rather than the factors tested.

 
Spatial analysis of associations between metal concentrations and landscape variables is complicated by the variation of calcite muds in the sediment samples. This complication was addressed by normalizing metal concentrations by Fe, allowing subtle associations of metal variation in the stream-bottom sediments to be evaluated relative to landscape variables. Metal/iron ratios for Pb, Cu, Zn, and As in stream-bottom sediments show weak positive correlations with increasing population density and density of developed land in the sediment catchment area. These trends likely reflect anthropogenic sources associated with developed, industrial, and transportation areas, such as industrial contamination, airfall deposition from smokestack particles, runoff from roadways, and runoff from contaminated sites (including former orchard sites) during construction activity. Ratios of Pb, Cu, and Zn with Fe are not correlated with increasing orchard density; As/Fe ratios were weakly, but not significantly, correlated with increasing orchard density (Table 5). For the stream-bottom sediments in the study area, the dominant anthropogenic sources of metals appear to have been derived from developed and developing land.

Factor Analysis of Stream-Bottom Sediment Data
Factor analysis was used as another method to identify trends and correlations between the stream-bottom sediment data and landscape variables. Factor analysis was performed on Fe-normalized data for Cu, Zn, As, and Pb concentrations in stream-bottom sediments in conjunction with data on population density ("Popden" variable), the density of developed land ("Dev_land" variable), and the density of orchard status land ("Orchard" variable) for the catchment areas of the sediment sample sites (Table 6). These elements were expected to be associated both in terms of host minerals in the sediments (Fe-oxyhydroxides) and contributing sources, such as natural weathering of base metal sulfides in rocks and anthropogenic sources, including metal-bearing arsenical pesticides. Anthropogenic sources of As and base metals, excluding the use of arsenical pesticides, are expected to be prevalent in higher population density settings with more developed land. Two factors defined for the seven-variable dataset (Table 6) have standardized eigenvalues >1. The largest factor loading for each variable is highlighted in Table 6. The factor that accounts for the largest proportion of variability is a Cu-Zn-Pb-developed land-population density factor (factor 1, Table 6) that likely reflects urban contamination. Factor 2 is an As-orchard density factor (having a moderate Pb loading) that likely reflects metal sources related to arsenical pesticide contamination (factor 2, Table 6). The strong association between the Pb, population density, and developed land variables dominates the association between As and Pb that is likely related to lead-arsenate pesticide contamination.


View this table:
[in this window]
[in a new window]

  		Table 6. Factor analysis results for Cu/Fe, Zn/Fe, As/Fe, and Pb/Fe element ratios and landscape variables Dev_land, Popden, and Orchard for stream-bottom sediment samples. Dev_land, the fraction of the stream-bottom sediment site catchment basin area covered by developed land (buildings, roads, etc.) and Popden, the population density of the sample site census tract area in 2000, are considered variables measuring the potential for metal loading from industrial and transportation sources. Orchard, the fraction of the stream-bottom sediment catchment basin area covered by orchard status land, is considered a variable measuring the potential of metal loading from soils contaminated with arsenical pesticides. The largest factor loading for each variable is in italicized font for each factor.

 
Ground Water Data
Summary statistics for As concentrations in water samples from bedrock ground water wells in relation to land use attributes for the well sites are listed in Table 7. Well depth and source water pH values were available for a subset of the wells. Arsenic concentrations in ground water wells in the region were low; generally <0.005 mg L–1. Approximately 95% of the wells had As concentrations below the highest LRL limit of 0.005 mg L–1. The few wells with As concentrations exceeding 0.005 mg L–1 were located in non-carbonate rocks of the Valley and Ridge Province, where orchard cultivation was not prevalent (Fig. 2A). The "Distance to Orchard" variable, a measure of the proximity of the well site to an orchard area that used arsenical pesticides, ranges from zero to more than 3 km; a value of zero means that the well is sited on a former orchard site. The "Or_buffer" variable, a measure of the intensity of arsenical pesticide use in the local vicinity of the well site, ranges from 0 to 84%. The "Or_bsn" variable, a measure of the proportion of the drainage basin catchment area enclosing the well site that was covered by orchards that used arsenical pesticides, ranges from 0 to 68%. There was no spatial association between As concentrations in the public supply bedrock wells and these landscape variables associated with arsenical pesticide use.


View this table:
[in this window]
[in a new window]

  		Table 7. Summary of statistics for As, pH, well depth, and landscape variables for public supply bedrock well source water samples. The Or_buffer landscape variable is the percentage of the 250-m buffer area surrounding the well site that was covered by orchards that used arsenical pesticides. The Or_bsn landscape variable is the percentage of the drainage basin catchment area enclosing the well site that was covered by orchards that used arsenical pesticides.

 

   CONCLUSIONS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
REFERENCES
 
Undisturbed soils from historic orchard sites under cultivation before the 1960s had elevated As and Pb concentrations relative to reference sites, consistent with trends expected from lead-arsenate pesticide contamination. These orchard soils also contained elevated Hg and Cu concentrations relative to reference sites, consistent with the co-use of other metal-based pesticides and mercury-based fungicides in orchards during this time period. Orchard and reference soils, and stream-bottom sediments in the area, show an association of As and heavy metals with both Fe and Al concentrations, consistent with their retention in soils and sediments by sorption to clay and fine-grained Fe-Mn oxyhydroxide minerals.

Bottom sediments collected from small flowing stream drainages downstream from both orchard and reference sites in the study area generally had low As and heavy metal concentrations, generally below the level at which adverse effects on sediment-dwelling organism are expected to occur. Sediment runoff from orchard sites contaminated with arsenical pesticide residues appears to be limited, even from historic orchard sites subsequently redeveloped as residential or industrial properties. However, sediment runoff from row crop agricultural areas was found to influence As, metal, and Ca concentrations in downstream sediments. Normalizing sediment metal concentrations by Fe allowed subtle associations of metal variation in the stream-bottom sediments to be evaluated relative to orchard density, developed land, and population density variables defined for the sediment sample sites. The strongest association of Zn, Pb, and Cu was with the developed land and population density variables, consistent with an anthropogenic source for these metals related to industrial and transportation activities. Arsenic was weakly correlated with the orchard density variable, consistent with an anthropogenic source related to arsenical pesticide residues from orchard soils.

Arsenic concentrations in ground water wells in the region were low, generally <0.005 mg L–1 As. There was no spatial association between As concentrations in ground water and proximity to orchards that used arsenical pesticides. Arsenic had limited mobility into ground water from surface soils contaminated with arsenical pesticide residues at the levels typically found in orchards in the Great Valley region, Virginia and West Virginia.


   ACKNOWLEDGMENTS
 
The authors thank Drs. William Cannon and Suzanne Nicholson (both at the USGS) and two anonymous reviewers who provided thoughtful comments and suggestions to improve the manuscript. In addition, the authors thank the private citizens in the study area that allowed site access for soil and sediment sampling. Doug Caldwell (Field Director) and Mark Perry of the Virginia Dep. of Health, Office of Drinking Water (Lexington Field Office), and William Toomey (Unit Manager), Bradley Reed (Supervising Engineer), Alan Marchun, Yvonne Wilson, and James Mitchell of the West Virginia Dep. of Health and Human Resources, Bureau for Public Health, Office of Environmental Health Services provided access to analytical and other data on public supply drinking water that was used in this study.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 CONCLUSIONS
 REFERENCES
 





This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF) Free
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Robinson, G. R.
Right arrow Articles by Sibrell, P. L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Robinson, G. R., Jr.
Right arrow Articles by Sibrell, P. L.
GeoRef
Right arrow GeoRef Citation
Agricola
Right arrow Articles by Robinson, G. R.
Right arrow Articles by Sibrell, P. L.
Related Collections
Right arrow Agricultural Pesticides
Right arrow Pesticides
Right arrow Heavy Metals

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
The SCI Journals Agronomy Journal Crop Science
Journal of Natural Resources
and Life Sciences Education		 Vadose Zone Journal
Soil Science Society of America Journal Journal of Plant Registrations The Plant Genome