Soil contamination and bioaccumulation of inorganics on petrochemical sites

Petrochemical-contaminated sites

Cotton rats and soils were collected and analyzed over a three-year period from 12 petrochemical-contaminated sites and 12 matched reference sites. Each year, four contaminated and matched reference sites were sampled. Reference sites were chosen in the vicinity of the each petrochemical waste site and were selected to match the vegetation on the waste site. Contaminated sites and their matched reference site all consisted of disturbed terrestrial ecosystems with predominantly early seral stage plant species. The most prominent plant species on these heavily disturbed prairies were johnsongrass (Sorghum halapense L.), little bluestem (Schizachyrium scoparium Nash), big bluestem (Andropogon gerardii Vitman), brome (Bromus spp.), and bermuda grass (Cynodon dactylon L.). Reference sites showed no visible evidence of previous petrochemical contamination. Waste sites were classified into three categories, landfarm, pond levee, or tar pit, based on petrochemical waste disposal practices. Sites A through E were landfarms in which oily sludges had been incorporated into soil. Sites F through H were pond levees constructed to contain oil sludges in storage pits. Sites J through L were tar pits that were subjected to periodical overflow of material from oil sludge pits (Table 1). All petrochemical waste sites were privately owned facilities in Oklahoma and supported resident populations of cotton rats.

Collection and analysis of soils

Surface soils (<2-cm depth) were collected from trapping grids on the petrochemical sites and the reference sites. Petrochemical sites were divided into six subsites and a composite sample consisting of six subsamples was collected from each subsite. Because they had less chemical variability, only two composite soil samples composed of six subsamples were collected from each reference site. All soils were stored and transported in sealed acid-washed glass jars.

Soil properties (pH, organic carbon content, texture, and electrical conductivity) were measured on air-dried, sieved (<2 mm) samples (Table 1). Soil pH was determined in a 1:2 soil: 0.01 M CaCl2 suspension [14]. Soil organic carbon was determined by automated dry combustion [15]. Soil texture was determined by the hydrometer method [16]. Electrical conductivity was measured in a 1:5 soil:deionized water extract [17].

Metals (Ba, Cd, Co, Cr, Cu, Ni, Pb, Sr, Ti, V, Zn) in soil were determined by acid wet digestion according to U.S. Environmental Protection Agency Method 3050 [18]. Metals in acid digests were measured by inductively coupled plasma (ICP) atomic emission spectroscopy. Total F content and potentially bioavailable F of soils were determined. Fusion with NaOH was used to measure total F content of soil [19]. The fused soil sample was neutralized with HCl, diluted to a volume of 100 ml, and F was determined using an Orion F-combination electrode (Orion 960900, Beverly, MA, USA). A weak acid extraction (0.03 M HCl, pH 1.5) followed by potentiometric determination was used to measure the potentially bioavailable F [20]. Blanks and spike recoveries were used for quality assurance in the analysis of total and bioavailable F.

Collection of animals and bone preparation

A total of 24 adult cotton rats were collected from each petrochemical and reference site during summer and winter trapping periods using Sherman live traps baited with rolled oats. After capture, cotton rats were housed overnight and were sacrificed the next morning by exsanguination. Two humeri from each cotton rat were cleaned of excess tissue with a scalpel and scissors, freeze-dried, weighed, and placed in petroleum ether for 96 h with daily changes to remove fat [12]. Skulls were removed and fixed in formalin for later evaluation of incisors for evidence of dental fluorosis.

Bone metal and bone F analysis

Bones were acid-digested using a method adapted from Andrews et al. [3, 4] for metal analysis. Each pair of bones (100 mg) was refluxed on a hot plate at 95°C with 5.0 ml of concentrated trace-metal HNO₃ for 1.0 h. Digested material was diluted with deionized distilled water to 10.0 ml and analyzed for Ba, Cr, Pb, Sr, Zn, and Ti using ICP atomic emission spectroscopy. Blanks and spike recoveries for each metal were used for quality assurance.

Bone F content was determined by dry ashing in porcelain crucibles at 550°C overnight [21]. Ashed bone was dissolved in 0.25 M HCl. The acid digest was neutralized with NaOH. Sample solution (5 ml) was combined with 5.0 ml of TISAB II buffer (Orion 940909) to adjust ionic strength and inhibit complexation of F by Fe and Al [22]. Fluoride was measured using an Orion F-combination electrode. Calibration standards were prepared in a similar manner using certified 100 mg F/L standard (Orion 940907). Bone F content is expressed as mg F/kg on a freeze-dried basis. Blanks, standard reference material (National Institute of Standards and Technology bone meal SRM 1486), and spike recoveries were used for quality assurance.

Scoring of teeth for dental lesions

Scoring of incisors was performed to document gross morphologic lesions commonly referred to as dental fluorosis using a system previously described for mammals [9, 23]. All cotton rats were scored by two different analysts for confirmation and results were averaged.

Statistical analysis

Soil data were analyzed as a randomized complete block design using PROC GLM [24]. Data were transformed using the natural logarithm function to adjust for heterogeneity of variance. The means of each contaminant of each petrochemical site were compared to the combined mean of all reference sites for each contaminant using Duncan's multiple range test to determine inorganic contamination of petrochemical sites.

Tissue data were analyzed using PROC MIXED [24] as a split block arrangement in a randomized block design with subsampling where sites were considered blocks, treatments were the main factor, and season as the split-unit factor. A log transformation of data was used to control the heterogeneity of variance. Analysis of the simple effects of treatment (controlling for season) was performed using the SLICE option with the LSMEANS statement [24] when the treatment by season interaction was significant. All reference sites were combined and Duncan's multiple range test was used to determine contamination of petrochemical sites. Pearson's linear correlation coefficients were calculated using PROC CORR [24] to evaluate the relationship between mean metal content in bone and in soil for each study site. Fisher's exact test was used to analyze the relationship between bone F content and bone incisor lesion score. Incisor scores ≥3 were categorized as high and those <3 as low. Bone F was categorized as follows: bone F < 1,000 mg/kg, low; 1,000 mg/kg ≤ bone F < 3,000 mg/kg, medium; and bone F ≥ 3,000 mg/kg, high. A 2 × 3 contingency table of incisor score and bone F was analyzed using PROC FREQ [24]. A Fisher's exact test was performed to test whether the percentages of high incisor scores were equal for cotton rats in the three bone F categories.

Extent of soil contamination

Metal concentrations in soil were elevated on petrochemical sites compared to reference sites: Cd (p = 0.016), Cr (p = 0.003), Cu (p = 0.002), Ni (p = 0.005), Pb (p = 0.0002), Sr (p = 0.006.), Ti (p = 0.025), V (p = 0.018), and Zn (p = 0.0001). Quality assurance and quality control laboratory procedures showed that 90 to 100% of metals were recovered for analysis of standard reference soil material (CRM020–050, RTC Corporation, Laramie, WY, USA) and spike recoveries. Mean soil contents of all metals except Ti on reference sites were similar to values reported for baseline soils of Oklahoma (Table 2). Titanium contents of baseline soils reported by Ka-bata-Pendias and Pendias [25] were summarized from studies that used wet digestion of soil with hydrofluoric acid (HF). Most soil Ti occurs as TiO2, which is only dissolved by using acid digestion with HF. Soil Ti levels from petrochemical and reference sites in our study were determined by a wet chemical digestion [18] that does not incorporate HF. Therefore, soil Ti values measured in our study are lower than soil Ti levels measured by HF digestion. The number of petrochemical-contaminated sites (in parentheses) that had elevated levels of metal in soil as compared to the mean of all reference sites were Ba (3), Cd (2), Co (3), Cr (9), Cu (8), Ni (7), Pb (9), Sr (6), Ti (5), V (5), and Zn (12) (Table 3). Soils from six or more of the petrochemical-contaminated sites were elevated in Cr, Cu, Ni, Pb, Sr, and Zn. Chromium in petrochemical-contaminated soil ranged from 2- to 100-fold greater than the overall mean for reference sites. Similarly, petrochemical-contaminated soils were elevated in Cu (2- to 85-fold), Ni (1.5-to 3-fold), Pb (5- to 140-fold) Sr (2- to 20-fold), and Zn (2-to 26-fold). The type of metal contamination did not show a trend between soils from landfarms, pond levees, and tar pits. A land treatment unit where petrochemical waste had been applied for more than 30 years had elevated mean Cr (280 mg/kg), mean Pb (130 mg/kg), mean Ni (110 mg/kg), and mean Zn (235 mg/kg) [26]. In general, most petrochemical-contaminated sites in our study had higher levels of soil Pb (Table 3) than reported by Loehr et al [26].

Total F in soil (p = 0.001) and HCl-extractable F (p = 0.002) levels were elevated on the petrochemical-contaminated sites as compared to reference sites. Total F was elevated on seven sites and HCl-extractable F was elevated on 9 of the 12 petrochemical-contaminated sites (Table 3). Laboratory quality assurance and quality control procedures showed that the fusion procedure recovered 95 to 105% of F spikes added to soil. Total soil F of reference sites was similar to total F in baseline soil, which ranges from 10 to 400 mg/kg (Table 2). The HCl-extractable F was 4- to 25-fold greater on the sites with elevated F compared to overall mean of reference sites; total F was 10- to 60-fold greater. In a detailed investigation of a landfarm, Schroder et al [10] found elevated levels of both total F (mean of 1,954 mg/kg) and HCl-extractable F (mean of 326 mg/kg) in soil that received application of oily sludges containing HF. All five landfarms (sites A through E) had elevated levels of both forms of F, suggesting that F in soil is more prevalent on landfarms than on other types of petrochemical sites.

Bone metal and F content

Preliminary studies showed that six metals and F listed in Table 4 had a tendency to accumulate in bone. Other metals did not accumulate in bone and are not reported in Table 4. The overall mean content of Pb in bone was elevated (p = 0.003) for cotton rats collected from petrochemical-contaminated sites compared to reference sites (Table 4). Quality assurance and quality control laboratory procedures showed that 90 to 100% of metals were recovered for analysis of standard reference bone material (National Institute of Standards and Technology SRM 1486) and spike recoveries. A significant treatment by season interaction was found for Pb content of bone (p = 0.0175). Lead concentrations in bone were higher in winter (mean of 21.5 mg/kg) than summer (mean of 10 mg/ kg, p = 0.0003). The cotton rats trapped in the winter were older than those trapped in the summer [27]. In part, the higher bone Pb in the winter versus summer may be due to longer exposure periods. Baseline levels of Pb in bone of rodents typically range from <2 to 3 mg/kg [28], similar to the bone Pb of 1.5 mg/kg of cotton rats collected from reference sites in our study. The number of sites on which the metal level in bone was elevated (in parentheses) as compared to the mean of reference sites were Ba (1), Cr (6), Pb (8), Sr (4), and Zn (1) (Table 4). Chromium, Pb, and Sr were the most prevalent contaminants in bone of cotton rats collected from petrochemical-contaminated sites. Chromium content of bone on contaminated sites was only about twofold greater than reference sites. Cotton rats from reference sites had Cr in bone similar to Cr in bone (0.1–10 mg/kg) from other small mammal studies [29]. Taylor and Parr [30] observed that Cr in bone (0.46 mg/kg) from cotton rats on a polluted site downwind of an airborne Cr source was higher than Cr in bone (0.16 mg/kg) from an uncontaminated reference site. Levels of Cr in bone in our study are less than the 4 mg/kg bone that Eisler [31] considered indicative of Cr contamination in bone. The elevated concentrations of Pb in bone were approximately 2- to 42-fold greater than the overall mean of Pb in cotton rats collected from the reference sites (Table 4). Chronic exposure to Pb may result in renal dysfunction, reduced growth rate, and reproductive impairment [32]. The most common exposure route for Pb is through ingestion and more than 90% of the lead in small mammals is found in bone tissue [33]. Bioaccumulation of Pb in bone (352 mg/kg) in areas with elevated levels of Pb in soil (8,430 mg/kg) has been documented in wood mice (Apodemus sylvaticus) inhabiting lead-zinc mining sites [6]. Similarly, mean bone Pb levels of 189 mg/kg of wood mice inhabiting a lead-zinc mining site with soil containing 14,010 mg/kg were reported [7]. Much lower levels of Pb contamination were associated with petrochemical-contaminated sites compared to lead-zinc mining sites. Consequently, cotton rats inhabiting petrochemical-contaminated sites did not accumulate Pb to the degree observed in small mammals on metal mining and smelting sites.

The overall mean content of F in bone of cotton rats was elevated 5- to 23-fold (p = 0.004) on petrochemical-contaminated sites compared to reference sites (Table 4). Quality assurance and quality control laboratory procedures showed that 90 to 100% of F were recovered for analysis of standard reference bone material (National Institute of Standards and Technology SRM 1486) and spike recoveries. A significant treatment by season interaction was found for total F content (p = 0.0377) in bone, where F levels in winter (mean of 1,930 mg/kg) were greater than F levels in summer (mean of 788 mg/kg, p = 0.0001). In part, the higher bone F in the older rats trapped in the winter versus in the younger cotton rats trapped in the summer may be due to longer exposure periods. Total F concentrations in bone of cotton rats from reference sites were similar to levels reported in other small mammal studies [34]. Elevated levels of F in bone of small mammals have been associated with elevated levels of F in soil [3, 4, 35, 36]. The results of this study are similar to those of Schroder [10], who reported elevated levels of F in bone (mean of 1,515 mg/kg) of cotton rats collected from a landfarm to which oily sludges that contained HF had been applied.

Dental lesions

Dental lesions have been noted for various species of small mammals collected from F-contaminated sites [8, 37, 38]. Schroder [11] reported that ∼80% of the cotton rats collected from a landfarm contaminated with F exhibited dental lesions, with an average severity score ≥3. Of the seven petrochemical sites with elevated levels of total F in soil, the prevalence of dental fluorosis was ∼50%, with an average severity score ≥3. The majority (>99%) of the cotton rats collected from the reference sites in this study did not have dental lesions. Severity of dental lesions varied with site and ranged from a score of 1 (slight striation in lower incisor) to a score of 5 (white chalky lower and upper incisors). The prevalence of dental fluorosis was approximately 50% higher in winter than in summer. Dental lesions (severity score ≥ 3) were more prevalent on sites A, C, D, and L than on other contaminated sites. However, more than 50% of the cotton rats from sites B, E, and H had lesions (Fig. 1). Regression analysis revealed a strong relationship (p = 0.0001) between incisor score and F content in bone of cotton rats. However, a more detailed analysis using Fisher's exact test indicated that the severity of dental fluorosis was not accurately predicted across the concentration range of F in bone. The Fisher's exact test revealed that 5, 52, and 78% of cotton rats had a high severity score ≥3 when bone F was < 1,000, 1,000 to 3,000, and >3,000 mg/kg, respectively. Thus, bone F was an accurate predictor of the severity of fluorosis when bone F was low (< 1,000 mg/kg) or high (>3,000 mg/kg) but not intermediate (1,000–3,000 mg/kg).

Relationship between bone and soil concentrations

A strong relationship was found between bone F and HCl-extractable F in soil (r = 0.70, p = 0.02; Fig. 2) and bone F and total F in soil (r = 0.85, p = 0.001; Fig. 3). Although elevated levels of metal were found in both soils and bone from the petrochemical-contaminated sites, strong relationships (p < 0.21) between these were not found. However, cotton rats from petrochemical-contaminated soils with elevated Pb (>60 mg/kg) had bone Pb greater than cotton rats from sites with low soil Pb.

Other studies have reported accumulation of Pb in bone (189–672 mg/kg) of wood mice on sites with Pb-contaminated soils [33]. Accumulation of bone Pb in small mammals was prevalent on smelter sites where soil Pb ranged from 130 to 14,010 mg/kg soil [39]. Soil Pb was positively correlated with bone of wood mice (0.05 < p < 0.10) and field voles (p < 0.10) [39]. A strong relationship between soil Pb and bone Pb was not found in our study (r = 0.36, p = 0.25). Perhaps the lack of correlation of soil Pb and bone Pb could be attributed to the lower levels of soil Pb in our study. The soil Pb in our study was 10-fold less than in the study of Shore [39].