| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Soil and Water Science Dep., Univ. of Florida, Gainesville, FL 32611
b Statistics Dep., Univ. of Florida, Gainesville, FL 32611
c Florida Power and Light, 700 Universe Boulevard, Juno Beach, FL 33408
* Corresponding author (Lqma{at}ufl.edu)
Received for publication October 8, 2001.
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Abbreviations: AM, arithmetic mean • GM, geometric mean • MDL, method detection limit • SCTL, soil cleanup target level
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Unlike in natural areas, concentrations of arsenic in urban areas vary considerably over short distances (Chirenje et al., 2001). Urban soils are significantly more heterogeneous than undisturbed soils (Chirenje et al., 2003; Craul, 1985; Davies et al., 1987), with human activity playing a predominant role in the development and modification of these soils (Barrett, 1987). Craul (1985) defined an urban soil as "a soil material having a nonagricultural, usually manmade, surface layer more than 50 cm thick that has been produced by mixing or filling of the land surface in urban and suburban areas." The extent of human activity (vertical mixing, compaction, use of fill, etc.) varies from one urban area to another, as well as among land-use types (Craul, 1985; Thornton, 1987). Therefore, land-use types form a good basis for classification when characterizing arsenic distribution in urban soils. High variation in arsenic concentrations in natural areas containing arsenic-bearing mineral deposits has been shown, but these are rare (Chen et al., 1999).
Soil arsenic concentrations in undisturbed areas range between 0.1 and 40 mg kg-1 worldwide, with an arithmetic mean concentration of 5 to 6 mg kg-1 (Kabata-Pendias and Pendias, 1992). Arsenic background concentrations in Florida nonurban soils, including disturbed or anthropogenically influenced soils, vary from 0.01 to 61.1 mg kg-1, with a geometric mean (GM) of 0.27 mg kg-1 (Chen et al., 1999). Relatively little information is available on background concentrations of arsenic in urban soils.
Florida is the fifth most urbanized state in the USA after New Jersey, Maryland, Massachusetts, and Connecticut. Currently, 11% of the total land area in Florida (total area: 14 258 000 ha) is considered urbanized (Nizeyimana et al., 2001) and this urbanization trend continues to increase. Gainesville and Miami are two of Florida's more than 700 cities. These two cities are situated in different parts of the state (Fig. 1) and, although they have approximately the same area, their population densities and economic bases are very different. They provide a diverse basis for the determination of the effects of human activity on arsenic levels in urban soils.
|
In contrast, Miami (Fig. 1) is a well-developed city, encompassing a large commercial district and very well-developed areas along the Miami-Dade County (population 2 253 000 in 2000) coastline. Geologically, Miami Limestone, a soft, oolitic limestone formation, is at or near the surface throughout Miami-Dade County. Most of the sites sampled in the city of Miami are comprised of soils classified as urban land, meaning that more than 85% of the surface is covered by parking lots, streets, large buildings, shopping centers, houses, and other structures (USDA, 1996). The urban land soil is mixed with Udorthent soils, nearly flat areas of extremely stony fill material (USDA, 1996). The fill material is usually a stony loam underlain by hard, porous, limestone bedrock (the reason many Miami sites could not be sampled below 15 cm). Often, a topsoil layer is applied to allow grass or ornamental plants to grow.
The city of Miami has a population of 370 000 in an area of about 91 km2, with a population density of 4 081 persons per square kilometer. However, the population in the surrounding 29 municipalities and the immediate hinterland brings the population in this region to about 935 000. The leading economic activities in Miami include construction, real estate, housing, recreation, motion picture and television filming, transportation, manufacturing, and cement production. According to the United States Census Bureau (2001), manufacturer shipments from Miami-Dade County were $8.5 billion in 1997 compared with $1 billion from Alachua.
This investigation was conducted to (i) compare the distribution of soil arsenic in two urban areas of equal size but different population density and industrial activity, and (ii) investigate the relationship among arsenic background concentrations, extent of human activity, and soil properties. Results of this research can be used as a benchmark when assessing anthropogenic and natural levels of arsenic in soils from elsewhere in Florida.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
As defined by the U.S. Census Bureau, an urbanized area comprises one or more central cores and adjacent densely settled surroundings (urban fringe) that together have a minimum of 50 000 people. The urban fringe generally consists of contiguous territory having a density of at least 1 000 people per square mile (386 people per square kilometer). Using this definition, both Gainesville and Miami are considered urbanized areas.
Soils from Gainesville
The Gainesville study served as a pilot study for developing a comprehensive sampling protocol for other cities. In this pilot study, the number of samples collected was based on soil heterogeneity and determined with the following equation:
 | [1] |
is the value of the Student's t test for a given confidence interval (1.96 for the 95% confidence interval), and R is the accepted variability in mean estimation (usually 10–20% depending on the scale and budget of the project). The method detection limit (MDL) for the campus study was 0.43 mg kg-1 and the concentrations of arsenic in all samples were greater than the MDL. A value of 20% was used for R, and the minimum number of samples needed for Gainesville was determined to be 130. Three land-use types were selected for sampling in urban areas: residential, commercial, and public land sites. These types were chosen because, together, they cover the largest area in most urban areas. Differentiating the samples from these three land-use classes enabled us to test for differences among them. The number of categories selected within these three land uses depends on the depth of detail required in the final sampling.
Five categories were chosen from the three land uses in Gainesville (i.e., residential right-of-way, residential yards, public buildings, public parks, and commercial areas). Forty surface soil samples (0–20 cm) were collected in May and June 2000 from each category, resulting in a total of 200 samples. One out of every five samples taken from each category was duplicated (for comparison of reproducibility), bringing the total number of samples to 240. However, at least three cores were taken and composited at each of the sampling sites. The sites for sample collection were randomly selected within each category of land use with a set of strict exclusion criteria to avoid any potentially contaminated areas. Chirenje et al. (2001) discuss both the randomization process and the exclusion criteria.
Soils from Miami
No significant differences were observed in arsenic concentrations between soils in residential yards and residential right-of-ways in the pilot study in Gainesville. Thus, samples from residential right-of-ways were used to represent residential soil, reducing the number of land-use categories to four. It must be noted that although results from Gainesville suggest that right-of-way samples can be used in place of yard samples, this may not be true for other cities. Nevertheless, right-of-way samples are recommended not only because they are more practical and easier to sample, but also because they are just as representative of residential areas (spatially) as yard samples.
After the Gainesville study, it was also determined that the focus of such background studies should produce a good estimate of the overall distribution of arsenic in each stratum (category) without primarily focusing on the central tendency of each stratum. Therefore, the precision target for the number of samples required would be set on an upper percentile of the concentration distribution. This assures that the body of the distribution would be well represented while at the same time assuring a high probability that the tail of its distribution would be represented as well. Conover (1980) described a method for calculating the minimum number of samples needed for a given percentile of a distribution to be exceeded by the maximum observed sample value at a given confidence level. For example, the sample size needed to assure exceedence of the upper 95th percentile with 95% confidence is 59. These sample sizes would need to be applied to each stratum to assure adequate estimation of the stratum distribution. Based on these computations, it was recommended that 60 randomly selected samples be obtained in each urban area stratum for future studies, yielding a total of 240 samples from four categories per city. Thus, 60 surface soil samples were collected in January and February 2001 from four land-use categories in the Miami study: residential areas, commercial areas, public parks, and public buildings. Although the numbers of samples collected in each land-use category were different (40 in Gainesville versus 60 in Miami), sufficient samples were collected from each city to enable intercity comparisons.
Sample Preparation and Trace Element Analysis
All collected samples were air-dried and screened through a 2-mm sieve, and the <2-mm fraction was retained for analysis. Samples were digested in a microwave digester with USEPA Method 3051a, which is comparable to USEPA Method 3050, a hotplate digestion method (USEPA, 1995, 1996). In summary, 1 g of soil was weighed into a 120-mL Teflon tube and digested in 9 mL concentrated HNO3 in a CEM (Matthews, NC) MDS-2000 microwave digester. The soils in this study were very sandy (<10% clay), hence the HNO3 digestion solution was considered a strong enough extractant for total arsenic based on previous studies by Chen et al. (1999). The resulting solution was filtered through Whatman (Maidstone, UK) #42 filter paper and diluted to 100 mL. Arsenic concentrations in the digestates were determined with a SIMAA 6000 graphite furnace atomic absorption spectrophotometer (GFAAS) (PerkinElmer, Wellesley, MA), with USEPA Method 7060A (USEPA, 1995). A standard reference material (SRM 2709 Montana soil) of the National Institute of Standards and Technology (NIST) was used to check the extraction efficiency of the digestion method. Spikes, duplicates, and reagent blanks were also used as a part of our quality assurance–quality control (QA/QC). Twenty percent of all samples analyzed were QA/QC samples. Digestion sets showing a relative standard difference of more than 20% from the known values (for standards and spikes) were repeated.
In addition, soil properties that have been shown to affect arsenic concentrations (pH, clay content, total organic carbon, and total Fe and Al) were also measured. The pH was determined with an Accumet Model 20 pH/conductivity meter (Fisher Scientific, Pittsburgh, PA) and the concentrations of Na and Ca were measured on a Varian (Walnut Creek, CA) 2380 graphite furnace atomic absorption spectrophotometer. The total organic carbon was determined on a Shimadzu (Kyoto, Japan) TOC 5050 total organic carbon analyzer. The concentrations of Cd, Cl, Fe, and Mn were determined with a Thermo Jarrell Ash 61E inductively coupled plasma atomic emission spectrophotometer (ICP–AES) (Thermo Elemental, Franklin, MA). Particle size analysis was done with Stoke's law of sedimentation. Fifty grams of soil were weighed into a 1000-mL cylinder and hydrometer readings were taken at predetermined intervals to estimate the amount of sand and silt particles that had settled out.
Data Analyses
All element concentrations are presented on a dry matter basis. Both arithmetic and geometric means (AM and GM) were used to describe the central tendency and variation of the data. The AM is calculated as the sum of the arsenic concentrations divided by the number of samples in the data set (n) and the GM is calculated as the nth root of the product of the arsenic concentrations. Baseline concentrations of arsenic were calculated with GM/GSD2 (where GSD is geometric standard deviation) and GM x GSD2 of the samples, which include approximately 95% of the sample population (Dudka, 1992; Chen et al., 1999). Although the data in this study approached the normal distribution after log transformation, the dataset was also investigated for the presence of mixtures of populations. This was achieved through the use of quantile–quantile (QQ) plots. Such investigation facilitated the distinction of natural from anthropogenic background levels, and the distinction of natural and anthropogenic background levels from those of affected (possibly contaminated) areas (the three possible populations present in most urban areas).
All statistical analyses were performed with the Statistical Analysis System (SAS Institute, 2000). Quantile–quantile plots were used to identify and eliminate outliers from the dataset. These outliers represented samples with abnormally high arsenic concentrations that could not be attributed to the background levels. The Shapiro–Wilk and Kolmogorov–Smirnov tests were used to test for normality with SAS. Because arsenic concentrations showed a lognormal distribution, the data were log-transformed before analysis to meet the requirements of normality. However, the original values are still provided for comparison with transformed data. Calculations for all the descriptive statistics were done after eliminating outliers and censoring the concentrations that were less than the MDL (a value equal to half the MDL was used). The MDL was reduced from 0.43 mg kg-1 in the Gainesville study to 0.28 mg kg-1 in the Miami study. All graphs, used to test for poly-populations, were prepared in SAS before eliminating outliers from the dataset.
Spatial analyses were performed with Spatial Analyst tools in Arcview Geographical Information Systems (GIS) software (Environmental Systems Research Institute, 2002). Pathfinder (Trimble, 2002) was used to geoprocess the differential Global Positioning System (GPS) unit-logged positions and transform them into forms that could be read by Arcview. These images were used to assess spatial distribution, and graphically display the analytical results from the study on a digital map (not shown).
|
RESULTS AND DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
Second, sampling methods (based on the given objectives) and sample distribution (measured by skewness) determine how the background concentrations are calculated. For normally distributed discrete samples, the background level is calculated with the arithmetic mean (AM) of the sample population and the arithmetic standard deviation (ASD; Halmes et al., 1998). The data in our case were lognormally distributed. In such cases, the 95% upper confidence limit (UCL) of the mean is calculated with the H statistic as follows:
 | [2] |
and H are the H statistics from tables provided by Land (1975) for the UCLs (Gilbert, 1987). The UCL depends on the variables xy, n, and , the chosen confidence limit. The final important consideration in background concentration studies is the randomization procedure. Data analyses in these studies depend on the type of population distribution. Hence, care should be taken to avoid experimenter-induced skewness, multiple populations, or an excessive number of outliers, so that the nature of the population distribution would only be explained by its intrinsic characteristics and not by the experimental design. Singh et al. (1997) provide an excellent discussion on the treatment of environmental data from lognormal distributions using nonparametric statistical procedures. In general, the jackknife and bootstrap procedures as discussed by Efron (1982) and Miller (1974) are recommended for studies similar to the current study. Neither method requires assumptions about the distribution (they work for both normal and lognormal distributions). In this study, outliers were eliminated before all the descriptive statistics were calculated. However, outliers were not eliminated when distribution graphs were plotted.
Table 1 summarizes the means, concentration ranges, and other relevant descriptive statistics for soil arsenic concentrations and Fig. 2 shows their distribution and cumulative frequency in the land-use categories analyzed. As discussed earlier, soil arsenic distributions in all land-use classes were positively skewed, hence deviating from the "normality" assumption. This is not unexpected in background studies of trace elements (Gilbert, 1987), so the data were log-transformed before analyses. The next step was to eliminate outliers from the sample population. Outliers (in this case, areas suspected to have high arsenic concentrations due to some form of contamination) were excluded because, although log-transformation dampens variation, outliers still shift the mean and inflate the variance. Background data from a database established by Ma et al. (1997) and Chen et al. (1999) for Florida soils were used to determine typical background levels and quantile–quantile (QQ) plots were used to identify the outliers (Fig. 3) . The eliminated sample points included four values from the public buildings category (107, 79, 37, and 36 mg kg-1 As) and one value from a commercial area (656 mg kg-1 As, data not shown) for Gainesville samples, and two values from residential areas (112 and 37.8 mg As kg-1 soil) and one value from a public building site (47.9 mg As kg-1 soil, data not shown) for Miami samples.
|
|
|
All 60 samples collected from Miami residential areas had arsenic concentrations greater than 0.8 mg kg-1, as did 98% of the samples collected from Miami public parks. Almost a third of all samples collected in Miami had arsenic concentrations greater than the Florida SCTL for commercial areas, 3.7 mg kg-1 (Table 2). Most of these samples came from residential areas (48%) and public buildings (28%). Only 10% of soil samples from commercial areas had arsenic levels greater than the commercial SCTL. These results are in stark contrast to Gainesville, where approximately 29% of all samples were greater than the Florida SCTL for residential areas and only 4% (Table 2) were greater than the SCTL of 3.7 mg kg-1 for commercial areas (67% of these exceeded samples came from the commercial areas; data not shown). In fact, 90% of the samples from Gainesville had arsenic concentrations less than 1.4 mg kg-1 (Fig. 2).
|
|
Soil Arsenic Distribution in Different Land Uses
As described earlier, soils from Miami had higher arsenic concentrations than the Gainesville soils in all four land-use categories (Table 1). In terms of arsenic concentration rank, Miami residential areas had the highest arsenic concentration (GM = 3.72 mg kg-1; Table 1), although this was not significantly higher than the arsenic concentration in Miami public parks (GM = 3.49 mg kg-1; Table 1). Soils from Miami public buildings had the third highest mean arsenic concentrations (GM = 2.49 mg kg-1; Table 1) followed by Miami commercial areas (GM = 1.93 mg kg-1; Table 1), although these two categories were not significantly different from each other ( = 0.05). Gainesville commercial, residential areas, and public buildings (GM = 0.63, 0.46, and 0.34 mg kg-1, respectively; Table 1) all had higher concentrations than those of Gainesville public parks (GM = 0.23 mg kg-1; Table 1), although those three categories (commercial, residential, and public buildings) were not significantly different from each other ( = 0.05).
The high background arsenic concentrations observed in this study are not unique to Florida. In a study to determine arsenic baseline concentrations in Denver, Colorado, Folkes and Kuehster (2001) observed extremely high baseline concentrations in the suburban areas of Denver. For example, residential areas had a GM of approximately 6 mg kg-1, which is significantly higher than that observed for either Miami or Gainesville. Other samples collected from the greater Denver area were also significantly higher than those in Miami and Gainesville (GM of Denver urban soils approximately 7 mg kg-1). However, the rural background concentrations of arsenic in Colorado were also significantly higher than those of Florida soils (GM = 3.7 vs. 0.28 mg kg-1, respectively). These elevated concentrations in rural, agricultural, and wilderness areas are largely due to natural factors, for example, parent materials.
Other researchers have also observed elevated arsenic concentrations in urban areas (Murphy and Aucott, 1998; Rasmussen et al., 2001; Tiller, 1992; Tripathi et al., 1997). Rasmussen et al. (2001) showed that garden soils from households in Ottawa, Canada, had arsenic concentrations of approximately 3 mg kg-1 compared with a GM of approximately 5 in housedusts, and Murphy and Aucott (1998) attributed the high arsenic concentrations in residential areas to historical land use (former heavily sprayed orchards) in New Jersey. Recognizing the importance of historical land use, Tiller (1992) examined the history of sampling points and made an effort to avoid areas that were likely to be contaminated in Australian urban areas. Nonetheless, arsenic concentration ranges of <1 to 8 mg kg-1 were recorded. Historical records of all sites were used in eliminating affected sites in our study. Bak et al. (1997) went a step further and calculated the contribution of each natural and anthropogenic activity toward the total arsenic concentration in the soil and concluded that sludge application contributed the highest amount of arsenic annually to the soil.
Soil Arsenic Distribution Characteristics
The complexity of urban soils often leads to distinct patterns in arsenic distribution. Portier (2001) discusses the implications of using various statistical techniques on the final outcome of soil elemental background studies. Ultimately, these techniques depend on the interpretation of the elemental distributions. The premise in this study was that arsenic distribution in urban areas is likely to encompass at least three populations of concentrations that may or may not be easily distinguishable. These include: (i) natural background, (ii) a diffuse anthropogenic influence or "anthropogenic" background, and (iii) localized point sources (impacted). Natural arsenic concentrations do not necessarily correspond to very low values, but rather arsenic concentrations that do not reflect any significant anthropogenic influence (Portier, 2001). Anthropogenic arsenic refers to arsenic mostly due to nonpoint sources as a result of human activity. Impacted areas refer to sites that have elevated arsenic concentrations as a result of a specific activity (point source). The results of this study were used to test how well the observed data confirmed this mixed-source hypothesis.
Probability plots for both the transformed and untransformed data for Miami and Gainesville are shown in Fig. 3. The distributions of arsenic in both cities came closer to meeting an assumption of normality after log transformation. However, the plot for Gainesville still showed three distinct populations: natural, anthropogenic, and potentially contaminated soils (Fig. 3D). The same pattern was not as easily discernible in Miami (Fig. 3B, nearly straight line), possibly because the natural background was higher (Chen et al., 1999) and the affected areas did not exhibit considerably higher arsenic concentrations than the nonaffected areas as was the case in Gainesville. Transforming such a population brings it closer to a log-normal distribution than a well spread out population. Lower end censoring (a value of half the MDL was used in place of nondetects) in the Gainesville dataset also contributed to the shape of the curve because of the large proportion (40%) of samples that were less than the MDL (Fig. 3, extreme lower left tail of curve).
Plots of untransformed data for residential, commercial, parks, and public buildings showed highly skewed distributions for both Gainesville and Miami (data not shown). The plots of the log-transformed data for the four land-use categories in Gainesville and Miami are shown in Fig. 4 and 5 , respectively. Some or a combination of the three parts of the distribution (natural, anthropogenic, and potentially affected) were discernible in public buildings and commercial areas in Gainesville (Fig. 4B,C). Although at least two separate components were evident in the curves for residential areas and parks, they do not necessarily fit the predefined parts of the distribution. There are reasons for this: (i) the concentrations observed at the lower tail of the curve do not necessarily represent natural concentrations as a group, but samples that fell below the MDL, and (ii) the samples in the mid-portion and third part of the curve actually represent both the natural background and anthropogenic influence. It must be noted that there are some locations with naturally high concentrations of arsenic and others may have high arsenic due to anthropogenic input. The most efficient way to distinguish between the two sources is to determine the correlation between arsenic concentrations with natural soil properties (discussed in the next section).
|
|
Arsenic distributions in soils from Miami residential areas and commercial areas seemed to fit only two distributions. The first part of the residential area curve (Fig. 5A) seems to represent a combination of natural and anthropogenic influence while the second portion represents possibly affected areas. The same can be inferred for soils from the commercial areas. Soils from public parks and buildings seemed to fit the predefined three parts of the distribution (Fig. 5C,D). This trend is not unusual for public buildings, where a mix of relatively undisturbed, disturbed, and affected soils can be found depending on the fill material, extent of development, and the location of the sample site. Miami public parks have a considerable amount of fill material, which comes from areas with varying concentrations of arsenic, hence the resulting distribution. As expected, the lower end of the distribution is dominant due to the preponderance of undisturbed areas.
Factors Influencing Soil Arsenic Concentrations
It is important to note that, after removing outliers, the concentration range of arsenic in this study was very narrow (MDL to <20 mg kg-1, with most samples having arsenic concentration less than 5 mg kg-1). Small changes in these low concentrations are often reflected as large relative changes, for example, a change of 1 mg kg-1 in a soil with background concentration of 1 mg kg-1 is reflected as 100% change while an increase of the same magnitude to a soil with a background concentration of 20 mg kg-1 is reflected as a 5% change. This has important implications on correlation coefficients of arsenic concentration with natural factors of soil formation.
Soil pH has been shown to affect the arsenic species present while soil organic matter (SOM) affects the binding and retention of arsenic in the soil (Rivero et al., 1998; Chen et al., 1999). Correlation analyses were performed on arsenic concentrations and soil pH and SOM in all land-use categories. Although previous observations by Ma et al. (1997) showed strong correlation between soil arsenic concentrations and both pH and SOM in nonurban areas, correlation coefficients were very low in all land-use categories for both pH and SOM in both cities (data not shown). There was a difference in soil pH of almost 1 unit between Miami and Gainesville soils (mean pH = 6.31 for Gainesville compared with 7.23 for Miami), which was statistically significant ( = 0.05). The average soil pH of rural soils around Gainesville is 5.6 (USDA, 1982) compared with 6.3 in the greater Gainesville urban area. The mean pH of soils from public parks in Gainesville was not significantly different from that of surrounding undisturbed areas ( = 0.05). However, the soil pH values at all the other disturbed areas within Gainesville city limits were significantly higher than in public parks and surrounding rural areas ( = 0.05). This suggests that the increase in pH was related to anthropogenic disturbances, possibly construction and the use of fill containing greater carbonate levels. Nonetheless, the correlation coefficients between arsenic concentrations (adjusted for land use) and pH were very low.
The situation in Miami was different for two reasons. First, the sediments of most of southern Florida are dominated by limestone and dolomite (30–94% CaCO3; Li, 2001). This has a significant effect on the soil pH of the undisturbed areas (pH approximately 7.2; Chen et al., 1999). The average pH in public parks (pH approximately 7.1) in Miami was not significantly different from the pH in undisturbed areas. The mean pHs from the three other categories analyzed (public buildings and residential and commercial areas) were also not significantly higher than that in public parks.
Numerous researchers have reported strong positive correlation between trace element concentrations and organic carbon (OC) and the silt + clay content of the soil (Wilcke et al., 1998; Aloupi and Angelidis, 2001). This study did not show such correlation. It must be noted that outliers, which often change data distribution patterns significantly, were not included in the analyses. Analyses in our study were only performed on urban surface soils, which are more strongly affected by anthropogenic activities than the agricultural soils studied by Wilcke et al. (1998) and Aloupi and Angelidis (2001). The increased variability in properties of urban soils also considerably influences data patterns (Folkes and Kuehster, 2001; Portier, 2001). Therefore, patterns that are more significant in natural soils are not as easily discernible in urban soils.
Soil organic matter concentrations in rural areas surrounding Gainesville are approximately 1% (USDA, 1982; Chirenje, 2000). The corresponding soil organic matter concentrations in Gainesville urban soils averaged 2.4% (Table 3). The elevated soil organic matter content in Gainesville may be explained by the increased use of fill in swale areas and the carbon cycling from the lawn in these areas. The mineral soils in the areas surrounding Miami have 1 to 10% organic matter (USDA, 1996), which corresponded with the soil organic matter concentrations observed in Miami urban areas (Table 3). Additionally, there was considerably higher inorganic carbon concentration, that is, carbonate in Miami (Li, 2001), affecting pH and, consequently, arsenic retention in soil.
Soil texture was considered because its relationship to the binding capacity and weathering was expected to affect the soil arsenic content (Wilcke et al., 1998). Correlation coefficients for texture, however, were very low in this study (data not shown). Correlation of arsenic concentrations and concentrations of Fe and Al were also examined for both cities, but were found to be very low (data not shown). This may be explained by the high extent of anthropogenic disturbances in these areas (Baize and Sterckeman, 2001). Higher correlation may be detected if a stronger extracting solution (e.g., a mixture of HNO3 and HCl) is used in the digestion. Nonetheless, the preponderance of Fe and Al can be easily linked to soil-forming processes in undisturbed areas. This is not necessarily the case in urban areas where human influence constitutes a significant factor of soil formation and development, making it difficult to distinguish soil horizons according to their pedogenesis.
Correlation coefficients were also determined between arsenic concentration and population density and income level of population census tracts, but these were also very low (data not shown). Population density affects the level of anthropogenic effect, although specific land use may be a more reliable indicator of human influence. A trace metal monitoring study conducted in both urban and nonurban areas in Denmark (Bak et al., 1997) showed low correlation coefficients for soil texture (0.41), population density (0.00), and atmospheric deposition (0.02) with arsenic. Clay soils consistently had higher arsenic concentration than sandy soils in both Denmark and Holland (5.5 and 13 mg kg-1 and 3.1 and 5 mg kg-1, respectively). Bak et al. (1997) concluded that arsenic concentrations in these areas were more sensitive to soil factors (e.g., clay content) than anthropogenic activities, hence the low correlation with population density and atmospheric deposition. Land-use classification played a significant role in the current study, but not population density. Population density is an unreliable parameter to use in many urban settings due to the heterogeneity in densities in residential areas. The majority of cities in the world (especially large cities like Miami) have mixed classification in residential areas, making population density an unreliable parameter. This is supported by Kelly et al. (1996), who observed that land use had the highest effect on concentration of trace elements in two cities in England.
Statistical Parameters of Soil Arsenic Concentrations
Based on the GM, the 95th percentile concentration (95% of all data fall below this value) and the 95% upper confidence level (UCL) of the log-transformed data mean for each land use were calculated. As expected from the above discussion, the 95th percentile and the UCL for all four of the land uses in Miami were significantly higher than the corresponding values in Gainesville, with the exception of the UCL for commercial areas (Table 2). The highest 95th percentile value was 25.3 mg As kg-1 (Miami residential), which was almost 15 times higher than the 95th percentile value for Gainesville residential sites (1.74 mg kg-1; Table 2). The combined 95th percentile value for all the land-use categories for Miami is more than four times higher than for Gainesville (16.4 vs. 3.53 mg kg-1). These results underscore the greater variance in soil arsenic concentrations as well as the higher arsenic concentrations found in the Miami soils. However, it must be noted that the 95th percentile is sensitive to the number of samples. Hence, caution must be taken in interpreting these results as the number of samples collected in Gainesville was lower than that collected in Miami.
The overall geometric mean arsenic concentration from this study (after removing outliers) was considerably greater than that obtained by Chen et al. (1999) for nonurban soils of Florida (GM = 0.39 vs. 0.27 mg kg-1), suggesting that background arsenic concentrations in urban soils are higher than those in undisturbed areas. The differences between these two areas can be attributed mostly to increased anthropogenic activities (nonpoint source) in urban areas, adding arsenic to the soil, water, and air; these are discussed in a separate publication (Chirenje et al., 2003).
|
CONCLUSIONS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 | |||
