Total concentrations of Pb in bulk soil in the 4 Former Cracking Unit samples ranged from 31 to 158 mg.kg−1 wet weight, with a mean of 92.8 mg.kg−1 (Table 1). Concentrations of organic Pb in these samples ranged from <0.25 to 1.0 mg.kg−1, indicating that the majority was inorganic Pb. On average, inorganic Pb in Former Cracking Unit samples was 99.4%. Concentrations of total Pb in the <250-μm soil fraction of Former Cracking Unit soil samples were greater than in bulk soil and ranged from 34.9 to 277 mg.kg−1, with a mean of 130 mg.kg−1.
Concentrations of total Pb measured in bulk soil samples from the 6 Former Refinery locations ranged from 105 to 58,300 mg.kg−1 wet weight (Table 1). Two of the 6 samples contained 35,100 and 58,300 mg.kg−1 Pb, while the remaining 4 samples contained between 105 and 238 mg.kg−1 Pb. However, concentrations in the <250-μm soil fraction of soil samples FR1 and FR2 were much lower than the bulk soil measurements, averaging only 9% of the total Pb concentration. The difference in concentrations in bulk soil as compared to the <250-μm soil fraction is due to the forms of inorganic Pb present, which is discussed in more detail in the following section. Concentrations of organic Pb in bulk soil were low, ranging from <0.25 to 4.6 mg.kg−1. As in Former Cracking Unit soil samples, inorganic Pb predominated in Former Refinery soils, averaging over 99% of the total Pb present.
In general, Pb, Cr, and other trace metals are usually associated with fine-grained particles in soils (Chaney et al. 1989; Horowitz 1991; Ruby et al. 1999), indicated by solid metal concentrations being generally higher in finer-sized fraction materials than in the bulk soil. In Former Cracking Unit samples, this same trend was observed. In samples FR1 and FR2, the concentrations of Pb in <250-μm fractions Pb is an order of magnitude lower than the bulk sample concentration (Table 1), likely because of the cementation of the samples by Pb carbonate minerals.
Inorganic Lead speciation
Two phases dominate the Pb mineralogy in the site soils: cerussite (Pb carbonate) and Pb-bearing iron oxide minerals (Table 2). Lead carbonate is the dominant phase in samples FR1 and FR2. This Pb likely is either from spillage of basic Pb carbonate or Pb oxide feedstock for tetraethyl Pb formulation or a weathering product of organic Pb compounds in soils. Most of this cerussite binds soil clasts together as cement. Given the high concentration of Pb in these samples and their nodular nature, this binding cerussite cement may account for the significantly greater concentrations of Pb in bulk samples relative to the <250-μm samples. In addition to cerussite, several samples contain Pb-rich calcite that is likely related to weathering of either organic or inorganic Pb in soil and subsequent calcite and/or caliche formation. Importantly, at neutral to slightly alkaline pH, Pb carbonate minerals are generally insoluble. However, at the acidic pH of the stomach, these minerals easily dissolve, and the Pb can be quite bioavailable. This high solubility under acidic conditions is the underlying reason for the high (>74%) bioavailability of Pb in samples FR1 and FR2 (Table 1).
Table Table 2.. Lead speciation in soil determined by electron microprobe analysis3
|Mineral||Relative lead mass (%)|
|Clay||1||0|| ||1.6||6.5|| ||0.3|
|Calcite||1||1|| ||7.5|| || ||5.7|
|Cerussite||92||95|| || || ||45.2|| |
|Mn oxide||1||1|| ||7.4|| || || |
|PbSiO4|| ||1|| || || || || |
|Fe sulfate||0|| ||1||1.5|| ||1.3||6.8|
|Anglesite|| || || || || ||2.5|| |
|Lead phosphate|| || || || ||0.1|| || |
|Brass|| || || ||1.0|| || || |
|Solder|| || || ||1.3|| || || |
The other major phases present in most samples were Pb-bearing iron oxides (Figure 2). Lead sorbs strongly to iron oxide compounds in soils (Dzombak and Morel 1990; Martinez-Villegas et al. 2004), so that weathered Pb from soluble phases is redistributed into iron oxide minerals in soils. In the 2 Former Cracking Unit samples analyzed (FC2 and FC4) and in 3 of the Former Refinery samples analyzed (FR3, FR4, FR6), the iron oxide-bound Pb dominates (Table 2). This sorbed Pb is bound even under relatively low pH conditions. Therefore, the lower observed bioavailability of these samples is consistent with the Pb speciation data.
Several minor Pb phases were also observed in soil samples (Table 2). These include Pb phases that are likely products of weathering (Pb bound to clay, Pb silicates, Fe-Pb sulfates, Pb phosphate) and industrial products containing Pb (brass and solder). The abundance of all these minerals is small enough that their contributions to total bioavailability are expected to be minor.
Concentrations of total Cr in the 4 Former Cracking Unit samples ranged from 18 to 494 mg.kg−1 wet weight, with a mean of 256 mg.kg−1 (Table 3). Concentrations of Cr(VI) ranged from <2 to 9 mg.kg−1 with a mean of 5.3 mg.kg−1. The concentrations of Cr(III) were calculated by subtracting the Cr(VI) from the total Cr concentration, yielding a range of 15 to 485 mg.kg−1 and a mean of 250.5 mg.kg−1; Cr(III) predominated in Former Cracking Unit samples, averaging 94.2% of the total Cr present.
Concentrations of total Cr ranged from 75 to 351 mg.kg−1 wet weight in the 3 Former Refinery samples, with a mean of 171 mg.kg−1 (Table 3). Concentrations of Cr(VI) in these same samples ranged from 4 to 8 mg.kg−1 with a mean of 6.3 mg.kg−1. The concentrations of Cr(III) were calculated as above and ranged from 71 to 343 mg.kg−1, with a mean of 165 mg.kg−1; Cr(III) prevailed in all samples, averaging 94.4% of the total Cr measured in Former Refinery soils. This result was not unexpected, as once Cr contacts soil, it is reduced to Cr(III), which is stable, insoluble, and immobile relative to Cr(VI) in soil conditions above pH 5 (Bartlett 1991; Losi et al. 1994).
Table Table 3.. Chromium (Cr) analytical results in soils (units in mg.kg−1 wet wt)
|Area||Sample ID||Total Cr||Cr(VI)||Cr(III)a||% Cr(VI)||% Cr(III)|
|Former cracking unit||FC1||33||<2b||32||3.0||97.0|
The extraction test we used to estimate relative bioavailability of Pb and Cr to the ecological receptors for the site was originally developed for estimating relative bioavailability of metals to humans. Extraction test parameters such as extraction time, pH, and temperature can affect bioavailability estimates (Ruby et al. 1993, 1996). For this investigation, simulated gastric fluids were maintained at 3 7 °C, a representative GI tract temperature in humans (Ruby et al. 1993), which is similar to mammals including 7 Peromyscus species (36.5 °C), gerbil (Meriones unquiculatus; 36.9 °C), tree shrew (Tupaia belangeri; 36–39 °C), hamster (Mesocricetus auratus; 35.8 °C), 2 rat species (Ratus ratus; R. norvegicus; 36.7–38.1 °C), 2 ground squirrel species (Spermophilus tridecemlineatus; S. richardsonii; 34.1–37.5 °C), ferret (Mustela putorius; 38 °C), fox (Alopex lagopus; 38 °C), porcupine (Erethizon dorsatum; 38 °C), and raccoon (Procyon lotor; 38 °C; McNab and Morrison 1963; Frappell et al. 1992; Swenson and Reece 1993; Aiello and Mays 1998; Refinetti 1999; Yoda et al. 2000). This temperature (37 °C) is at the lower end of the body temperature range for birds, which ranges from about 37 to 42 °C (McNab 1966; Wasser 1986). While 37 °C is considered a representative body temperature for humans, it is known that actual body temperatures for other mammals and for birds will vary on the basis of several factors, including individual metabolism rate and heat loss rate. On a mass-to-mass basis, birds have higher body temperatures than mammals because they have higher rates of metabolism and usually have lower rates of heat loss than mammals (McNab 1966). For example, the body temperature reported for the American robin is 43.2 °C (McNab 1966). The difference in body temperatures between mammals and birds is a source of uncertainty in the use of these results to evaluate bioavailability in the mammalian and avian receptors at the site. This source of uncertainty can be addressed by increasing the gastric fluid temperature to about 40 °C to represent bird body temperatures. Because increases in the temperature of the simulated gastric fluid increases the bioaccessibility measured, the bioaccessibility values reported here for robin may be low. A general rule of thumb is that kinetically controlled reaction rates double with each 10 °C increase in temperatures (Espenson 1981). Therefore, a 3 °C difference from the test conditions should only increase bioavailability by less than about 30%.
The mass of test material (0.4 g) and volume of test solution (40 mL) were based on examinations of stomachs and small intestines of rabbits (Ruby et al. 1993). Extraction times of 20, 40, and 60 min coincide with the emptying time of rabbit stomachs (Ruby et al. 1993). In adult humans, the GI tract is 80% empty after 60 min following ingestion of a meal, and a child's stomach empties in 54 to 68 min (Ruby et al. 1996).
The pH value of 2.5 used to estimate GI tract pH is the average during fasting (pH = 1.0–1.3) and fed conditions (pH = 2.8–4.1) in rabbit stomachs (Ruby et al. 1993). These values are in general agreement with the GI tract pH of the rat, which has a stomach fluid pH of 1.0 to 1.5 under fasting conditions but otherwise ranges from 2.6 to 5.1 (Freeman et al. 1992). The pH in the forestomach of rodents often ranges from 4.0 to 4.5 (Karasov and Hume 1997). Rats usually nibble intermittently but continuously during the day, and when food is ingested, it causes the stomach fluid pH to rise because of the buffering capacity of the food (Wixson and Davies 1993). Similarly, stomach pH in pigs increased to 5.1, 4.0, and 3.1 at 30, 60, and 90 min after ingesting food (Wixson and Davies 1993). In ruminants such as cattle, stomach pH usually ranges from 3.5 to 4.5 (Chaney et al. 1989). The gastric pH of 5 species of falconiforms on a mouse diet averaged 1.7 (range −1.3–1.8) 4 h before meals and averaged 2.7 (range −1.5–3.5) 2 h after meals (Duke et al. 1975). In the same study, the gastric pH for 2 owl species 4 h before mouse meals was 2.35, ranging from 2.2 to 2.5, and was 2.75 (range −2.7–2.8) 2 h after feeding. Duke et al. (1975) also reported gastric pH data for the turkey (Meleagris gallopavo) and domesticated hybrid duck (Anas platyrhynchos × Cairina moschata), which averaged 3.0 premeal and 2.3 postmeal and 2.1 premeal and 2.1 postmeal, respectively. The pH used (2.5) is similar to or lower than that observed in the species of concern and in other species where gastric pH data are available, especially under nonfasting conditions. Therefore, pH conditions for the species studied resulted in a conservative estimate of pH since metals generally are more bioavailable at lower pH values. Thus, the method likely resulted in an overestimation of bioavailability for many species, thereby providing a higher risk estimate.
The GI system in birds shares with mammals the same fundamental components for digestion of food: A tubular intestine with proximal and/or distal fermentation chambers (Vispo and Karasov 1996; Karasov and Hume 1997). Minimal food (e.g., starch and fiber) digestion has been reported in the crops of some birds, with pH being as low as 4.5 (Ziswiler and Farner 1972; Vispo and Karasov 1996). The pH of gastric fluid secreted by the proventriculus has been reported to range from 0.2 to 1.2, while the pH in the ventriculus (or gizzard) ranges from 0.7 to 2.8 (Ziswiler and Farner 1972).
The amount of time that food travels through the digestive system from mouth to anus is termed the mean retention time. A summary of mean retention times for various species of birds indicates times ranging from approximately 45 to 90 min for nectar-, fruit-, or insect-eating birds; from approximately 60 to 190 min for leaf- and twig-eating birds; and from approximately 170 to 390 min for seed-eating birds (Karasov 1990; Levey and Karasov 1992, 1994; Karasov and McWilliams 2005). For the American robin, which is a seasonal frugivore, there is a difference in gut retention times depending on the food type. The mean retention time for fruit in the GI tract of the robin was 48 min, while for crickets it was 65 min (Levey and Karasov 1992). Although stomach emptying times or intestinal transit times were not reported, the actual times for these processes are expected to be less than the mean retention times.
Mean retention times in mammals are longer than in birds, with a general increasing trend in retention time with increasing body mass of the species. Mean retention times summarized by Stevens and Hume (1998) range from 3.4 h (vole) to 48 h (pig), with the actual retention time in the stomach to be less than this amount. For example, the 3 h required to completely empty a rabbit stomach (Ruby et al. 1993) is considerably less than the mean retention time of 27 to 39 h reported for rabbits for particles and fluids, respectively (Stevens and Hume 1998). The rabbit gut retention time of about 3 h is longer than retention times for other species, so bioavailability was likely overestimated. However, Ruby et al. (1993) calibrated their in vitro study to the rabbit model and subsequently found that these results were similar to later swine- and primate-based studies (e.g., Ruby et al. 1996).
Larger particle sizes would result in a smaller surface-to-mass ratio. As the rate of mineral dissolution is proportional to the particle size, the small particle size results in enhanced Pb dissolution (Davis et al. 1993). Larger particles would therefore dissolve less Pb. Since Pb concentrations were generally higher in the <250-μm fraction than in the bulk soil, the inclusion of data only from the <250-μm soil fraction in the extraction test likely increased Pb dissolution estimates in our study.
These data suggest that the assumptions used in the extraction test provide a reasonably accurate predictor of relative bioavailability and generate a reasonable estimate for the eastern cottontail while providing a conservative estimate for the other 3 species of concern because of the conservative body temperature, pH, and mean gut retention time values. Regulatory approval was granted by the Texas Commission on Environmental Quality (TCEQ), which considered this approach appropriate for assessing Pb and Cr risks at the site.
Until further research is done, however, extrapolation of in vitro bioavailability determinations to other species should be done with caution. While the current procedure mimics mammalian body temperatures of about 3 7 °C, it is less than avian body temperatures of 38 to 42 °C. Differences in gastric pH need to be examined for both bird and mammal species. The gastric pH of 2.5 may need to be lowered to account for gastric pH in some species of birds, such as falconiforms, and under fasting conditions. The sampling time of up to 60 min may not be sufficient to mimic gastric emptying time for some bird species. In ruminants, dietary intake 1st passes through the reducing environment of bacterial fermentation in the rumen, where complexation of metals can occur, changing the speciation and potential absorption in the upper GI tract.
The data obtained from the in vitro and speciation studies were used to refine risk assessment calculations for site receptors and also used as input into risk management. As such, protective concentration levels (PCLs) were calculated for site soils according to the TCEQ ecological risk assessment guidance (TCEQ 2001). Based on the premise that a hazard quotient (HQ) of 1.0 is protective of a receptor exposed to a specific chemical, comparative PCLs were calculated. This PCL calculation method is based on the site-specific exposure dose and is a ratio of the target risk (in this case “1”) and the calculated risk (i.e., the HQ) multiplied by the chemical concentration. In this case, we used the 95% upper confidence limit (UCL) on the mean concentration. The 95% UCL soil concentration was divided by the HQ to yield the comparative PCL as follows:
Likewise, PCLs based on the no-observable-adverse-effect level (NOAEL) and the lowest-observable-adverse-effect level (LOAEL; from Finley et al. 1976; Heinz and Haseltine 1981; Osborn et al. 1983; Zakrzewska 1988; Elbetieha and Al-Hamood 1997) were calculated from NOAEL- and LOAEL-based HQs as follows:
In these PCL calculations, the chemical concentration represented by the estimated dose was attributed to the soil because soil was the abiotic medium contributing most of the dose for terrestrial receptors at the site. For example, soil had 16% and food items (i.e., vegetation and insects) 84% of the total estimated ingested Pb for the American robin (data not shown). Soil concentrations and wildlife toxicity thresholds used in calculating HQs were on a wet-weight basis. The arithmetic average of the NOAEL- and LOAEL-PCL for each receptor equals the comparative PCL for that receptor. The lowest of the comparative PCLs for each chemical becomes the final ecological PCL for that chemical (Table 4).
The final ecological PCL for both Pb and Cr was indicative of the soil concentration that should be protective of the most sensitive receptor (i.e., the most sensitive of the robin, quail, mouse, or rabbit). Thus, PCLs can be used as cleanup levels protective of the most sensitive ecological receptor exposed to soils.
Table Table 4.. Protective concentration levels (PCLs) used to manage lead (Pb) and chromium (Cr) risk to ecological receptors. All units in mg.kg−1 wet weight
|Receptor||Comparative PCL for Cr(VI)||Comparative PCL for Cr(III)||Comparative PCL at 100% Pb bioavailability||Comparative PCL at 23.9% Pb bioavailability|
The data obtained from the extraction study were used to generate the PCL cleanup levels for Pb. Cleanup levels generated using site-specific bioavailability and speciation data were compared to cleanup levels based on the assumption of 100% bioavailability (Table 4). Considerable increases in Pb cleanup levels were achieved for all receptors, as the Pb PCL increased from 51.5 to 78.3 mg.kg−1. Similar results were obtained when site-specific speciation data for Cr were used. The comparative PCL for Cr nearly doubled from 58.6 to 112 mg.kg−1. For Pb, the change in bioavailability from 100% to 23.9% was the only contributor to the PCL change and resulted in a relatively consistent PCL increase across receptors examined. For Cr, however, the primary factor influencing the PCL change was toxicity reference values (TRVs; data not shown). The TRVs were originally based on studies where Cr(VI) species, such as potassium dichromate, were used; these were replaced with studies using Cr(III) species, such as chromium chloride. For mammals, TRVs based on studies using Cr(III) species were hundreds of times greater than TRVs based on Cr(VI) species; bird TRVs based on studies using Cr(III) were within 6-fold of TRVs derived using Cr(VI) species. This accounts for the nonlinearity of the increases in PCL values. For this same reason, the receptor for which estimated risks was greatest for Cr changed from the white-footed mouse to the American robin (Table 4).
These data provided evidence that site-specific bioavailability and speciation need to be considered when performing ecological risk assessments on soils contaminated with Pb and Cr. The costs associated with conducting the bioavailability and speciation studies were de minimus relative to the reduction in remediation costs realized at the site. The resulting PCLs were used to manage and segregate soils, allowing for more cost-effective ecological risk management at the site. While the 1.5- to 2-fold increase in Pb and Cr PCLs achieved here resulted in substantial reductions in soil remediation costs at the site, we realize that given the conditions and limitations of the data, achieving a less than 2-fold increase in PCLs may be less meaningful at other sites where Pb and Cr are of concern in soils. Whether such increases in PCLs are worth pursuing at other sites depends on site-specific conditions and risk management goals.