| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Dep. of Crop and Soil Sciences, 3111 Plant Sciences Bldg., Univ. of Georgia, Athens, GA 30602-7272
* Corresponding author (dradclif{at}uga.edu)
Received for publication October 18, 2005.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
Abbreviations: CDE, convection dispersion equation • LSC, liquid scintillation counting • PVC, polyvinyl chloride • TOC, total organic carbon
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
Sources of environmental hormone contamination include both human and domestic animal waste. In Georgia, the leading state for poultry production in the United States, hormones originating from poultry litter (a mixture of chicken manure, spilled feed, feathers, and bedding materials) are of particular concern. In 2002, the state produced 1.29 billion broilers (Gallus gallus domesticus) and almost 2 billion kg of poultry litter (Georgia Agricultural Statistics Service, 2004). This litter, which contains the macronutrients N, P, and K, is normally land-applied as fertilizer; however, it also contains appreciable concentrations of estradiol and testosterone. The litter of 8-wk-old male and female broilers contains between 1.3 and 35 µg estradiol kg–1 and 0.3 to 55 µg testosterone kg–1 (Finlay-Moore et al., 2000; Jenkins et al., 2005).
The hormones present in land-applied poultry waste may contaminate surface or subsurface waters through field runoff and subsurface transport. Several studies have indicated land application of poultry litter increases concentrations of estradiol and testosterone in runoff (Nichols et al., 1997; Nichols et al., 1998; Finlay-Moore et al., 2000), providing evidence that runoff is a likely source of surface water contamination. The potential for estradiol and testosterone to be transported through the soil into groundwater, however, is not yet as well understood.
Relatively high sorption coefficients indicate both hormones are strongly sorbed to soils. Batch sorption isotherm experiments report both linear and nonlinear hormone sorption to soils and sediments. Linear sorption isotherm coefficients (Kd) range between 3.56 and 84.41 mL g–1 for estradiol and 4.57 and 27.3 mL g–1 for testosterone (Holthaus et al., 2002; Lee et al., 2003; Casey et al., 2005). The lowest Kd values for both ranges were for sorption to a soil which was 94% sand (Lee et al., 2003). Nonlinear Freundlich sorption coefficients (Kf) are between 10 and 6670 µg1 – n mLn g–1 for estradiol and 43.4 and 1200 mL g–1 µg1 – n for testosterone (Jürgens et al., 1999; Lee et al., 2003; Ying et al., 2003; Casey et al., 2003, 2004, 2005). High estradiol and testosterone soil sorption affinity indicates that the hormones are unlikely to experience significant soil transport; however, field studies suggest that high sorption affinity does not necessarily preclude hormone transport. Peterson et al. (2000) and Shore et al. (2004) have linked estradiol and testosterone in surface waters to the sub-soil transport of hormones through agricultural lands fertilized with domestic animal manure.
Several soil column transport experiments have investigated estradiol and testosterone soil transport. From these studies, it appears that transport of testosterone occurs more readily than estradiol (Casey et al., 2003, 2004; Das et al., 2004). However, based on current information, it is still difficult to assess the potential of either hormone to contaminate soil and groundwater under field conditions. Comparisons between studies are complicated by the use of different experimental methods. For example, studies differ in the number of pore volumes of hormone-free background solution applied to soil columns following the application of a pulse of hormone solution and the use of sand-soil mixtures (Casey et al., 2003, 2004, 2005; Das et al., 2004).
Hormone degradation and metabolite formation during transport further hampers determination of the potential for soil and groundwater contamination since the sorption characteristics of these metabolites differ from those of the parent compounds (Jürgens et al., 1999; Holthaus et al., 2002; Lee et al., 2003). Estradiol and testosterone are highly labile, and rapidly degrade under a variety of conditions. The half-life of estradiol under soil conditions similar to those of a temperate growing season is less than 0.5 d and the half-life of testosterone is between 0.35 and 0.88 d (Colucci et al., 2001; Lorenzen et al., 2005). Under anaerobic conditions, such as those present in saturated or aquatic environments, the reported range of half-lives is greater—between 0.2 and 9.7 d for estradiol and 0.3 and 7.3 d for testosterone (Jürgens et al., 1999; Lee et al., 2003; Ying et al., 2003). Estradiol transformation products include the metabolites estrone and estriol, whereas testosterone oxidizes to form androstenedione, androstanedione, and androstadienedione (Jürgens et al., 1999; Colucci et al., 2001; Casey et al., 2003, 2004, 2005; Das et al., 2004; Lorenzen et al., 2005; Jacobsen et al., 2005).
Previous transport experiments have also not taken into account the influence of preferential flow on hormone transport. To date, all studies have used repacked soil columns to eliminate both soil structure and soil macropores, the main source of physical nonequilibrium transport by rapid preferential flow in natural soils (Beven and Germann, 1982). By eliminating preferential flow, these studies were able to confine nonequilibrium transport to chemical nonequilibrium processes and simplify transport analysis. Transport experiments that use large undisturbed soil columns preserve many naturally occurring macropores and may better reflect transport through soils under field conditions. Similarly, differences in soil structure because of tillage regimes may affect the transport of hormones in soils. Conventional tillage destroys the structure of soil and disrupts macropores. This can reduce infiltration of water and may retard transport of hormones into the soil. Soils that are not tilled maintain a greater degree of soil structure and continuous macropores may allow greater hormone infiltration through preferential flow (Beven and Germann, 1982).
The objective of this study was to determine the soil sorption and transport characteristics for estradiol and testosterone in large undisturbed columns of conventionally tilled and no-till Cecil sandy loam. Cecil soils are the dominant soil series in the Georgia Piedmont region (Endale et al., 2002). Hormones were applied to soils at concentrations consistent with poultry litter application. An inverse model, HYDRUS-1D version 2.0 (
im
nek et al., 1998), estimated hormone transport parameters from the observed transport experiment data.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
Soil columns were extracted from the plots by pressing the sharpened edge of a steel cylinder into the soil with a tractor-mounted hydraulic soil probe. Seated inside the cylinder was a polyvinyl chloride (PVC) sleeve (15-cm i.d., 32-cm length). Once the cylinder was extracted from the plot, the sleeve containing the soil column was removed. Paper padded the ends of the column within the sleeve, which were sealed with a polyvinyl chloride cap. Padding and sealing the columns minimized further disturbance to the soil during transport to the laboratory. Columns were stored at 1.7°C immediately upon arrival at the laboratory.
Bulk soil samples for sorption experiments were extracted with augers (7.6-cm i.d., 10.2-cm length) from the 0- to 10- and 20- to 30-cm depths of both no-till and conventionally tilled plots. Tillage on conventionally tilled soils affected the first 15 cm of soil from the surface. The 0- to 10-cm soil depth represented the soil affected by tillage in conventionally tilled soils, whereas soil in the 20- to 30-cm depth would not have been tilled for either treatment. These soil samples were air-dried, ground, sieved (<2 mm), and stored at 20°C.
Table 1 gives values of soil characteristics for the treatments in this study. Bulk density was measured using rings (8.9-cm i.d. by 5.9-cm length) sampled at the center of the 0- to 10-cm depth and 20- to 30-cm depth in each plot.
|
The radiolabeled hormones 6,7-3H-estradiol and 4-14C-testosterone were used simultaneously during column transport experiments. The labels were differentiated during the transport experiment using a dual-label feature of a Beckman LS 5000TA liquid scintillation counter (LSC) (Beckman Coulter, Fullerton, CA). The dual-label feature was not necessary for sorption kinetics and sorption isotherm experiments because the two hormones were not combined in these experiments.
The environmental conditions of the experiments presented here were similar to those present in studies where hormone transformations have been reported (Jürgens et al., 1999; Lee et al., 2003; Ying et al., 2003). It is therefore likely that during the course of the sorption and transport experiments some hormones underwent transformation. The LSC used in this study is able to detect estradiol, testosterone, and their primary metabolites, since transformation does not affect the structures to which the labels are attached. However, LSC is unable to differentiate between the metabolites and their precursors. Although our methods could not distinguish between the parent compounds and their primary metabolites, we will refer to the radiolabeled compounds as estradiol and testosterone for the sake of simplicity.
Sorption Isotherms
Sorption kinetics and sorption isotherm batch experiments were performed by combining 1-g air-dried soil samples with 10 mL of estradiol or testosterone solution in 15-mL glass centrifuge tubes. Radiolabeled and unlabeled hormones were combined in a background solution of 0.01 M CaCl2 to create the desired concentration of estradiol or testosterone solution. The glass centrifuge tubes were sealed with Teflon-lined caps and shaken on a reciprocating shaker for prescribed time intervals. At the end of the time interval, the appropriate sample sets were removed and centrifuged for 10 min at 4000 x g. The supernatant was analyzed for radioactivity by LSC. The difference between initial and final solution radioactivity was attributed to sorption of the hormone. The percentage of hormone sorbed was then used to extrapolate the sorption of the unlabeled hormone.
Hormone sorption kinetics were measured on soils from the 0- to 10-cm and 20- to 30-cm depths of one no-till and one conventionally tilled plot. This was done by combining soil samples from the two depths with solution containing 0.50 µg mL–1 estradiol or 0.50 µg mL–1 testosterone as described in the Batch Experiments section. To create the 0.50 µg mL–1 estradiol solution, 0.374 KBq of 6,7-3H-estradiol and unlabeled estradiol were combined in a background solution of 0.01 M CaCl2. To create the 0.50 µg mL–1 testosterone solution 0.148 KBq of 4-14C-testosterone and unlabeled testosterone were combined in a background solution of 0.01 M CaCl2. Samples were shaken for 2, 24, 48, 72, 120, or 144 h at 20°C. Three replicate samples were shaken from each hormone treatment and soil depth at each time interval. At the end of a time interval, the appropriate sample sets were removed, centrifuged, and the resulting supernatant analyzed for radioactivity as previously described. Kinetic sorption data was then analyzed to determine the time to equilibrium sorption for estradiol and testosterone and this became the shaking time for sorption isotherm samples.
Batch equilibrium experiments were used to determine estradiol and testosterone sorption characteristics for soils from the three conventionally tilled and three no-till plots. Separate estradiol and testosterone sorption isotherms were plotted for soils from the 0- to 10-, and 20- to 30- cm depths of each plot (24 total isotherms). Estradiol and testosterone solutions were prepared as previously described. Initial hormone concentrations of 0.001, 0.010, 0.100, 0.500, and 1.00 µg mL–1 were used to create sorption isotherms. These concentrations were chosen because they were within the range of estradiol found in poultry litter applied to pasture (Finlay-Moore et al., 2000; Shore et al., 1995). Treatments were performed in triplicate. Previously completed kinetic sorption experiments indicated that maximum sorption occurred within 72 h of shaking, therefore samples were shaken on a reciprocating shaker for 72 h at 20°C. Following shaking, the samples were centrifuged for 10 min at 4000 x g. The radioactivity of the resulting supernatant was determined by LSC as previously described.
Batch equilibrium experiments were also performed to determine the sorption affinity of Cl– to soils from the 0- to 10-cm and 20- to 30-cm depths of one conventionally tilled and one no-till plot. Appropriate amounts of CaCl2 were combined with water to create initial Cl– solution concentrations of 0.005, 0.03, 0.37, 0.59, and 0.70 mg mL–1. As with the hormone batch equilibrium experiments, 1 g of air-dried soil was combined with 10 mL of solution in a 15-mL glass centrifuge tube and capped. Each treatment was performed in triplicate. Tubes were shaken on a reciprocating shaker for 2 h. The samples were centrifuged for 10 min at 4000 x g. The Cl– concentration remaining in solution was determined using a digital chloridometer (Labconco, Kansas City, MO).
Batch equilibrium sorption data were described by the linear sorption equation
 | [1] |
 | [2] |
The linear sorption coefficient was also expressed as the organic carbon partitioning coefficient Koc = Kd/foc, where foc is the fraction of soil organic carbon [total organic carbon (TOC) x 100–1] (Soil Science Society of America, 1996).
The Freundlich adsorption parameters (Kf and n) were analyzed using a split-split plot statistical design where tillage treatment was the whole plot and depth and hormone were the split-plot "treatments" (Gomez and Gomez, 1984). The appropriate error terms were used to test for main effects and interactions. This model was fit to the results using the SAS software package (SAS Institute, 1999).
Column Transport
Transport experiments were performed using six soil columns, one column from each of the three no-till plots and three conventionally tilled plots. In preparation for column transport experiments, each soil column (15-cm diam., 32-cm length) was removed from its PVC sleeve and placed upright inside a 16.5-cm diameter PVC cap. A single, central drain was drilled into each cap and a brass nipple attached to the exterior. A 3- by 3-cm square of cheesecloth and 2 cm of acid-washed sand covered the drain opening inside the cap. Several thin layers of saran resin dissolved in acetone were applied to the sides of the column to prevent wax penetration into the soil. A 2-cm coat of paraffin wax was applied to the sides of the column to prevent solute sideflow and provide structural stability to the soil. A PVC collar (15-cm diameter) was placed around the top of the column to prevent wax from spreading over the soil surface and to allow ponding of pulse and background solutions on the top of the column during transport experiments.
To prevent entrapped air from interfering with solute flow, each column was purged with gaseous CO2 for 24 h before saturating the column. Once purged, the column was slowly saturated from the bottom with a 0.01 M CaNO3 background solution. A mariotte bottle was used to pond 3 cm of 0.01 M CaNO3 on the surface of the column and the flow rate was monitored until it had reached steady state. Once steady state was achieved, the flow of background solution was stopped and the 3-cm head was allowed to infiltrate the soil column. Just before complete infiltration, the head was restored with pulse solution from a second mariotte bottle. The pulse consisted of 1.1 L (approximately 0.52 pore volume) of 0.533 µg estradiol (17 700 Bq 6,7-3H-estradiol and 0.525 µg unlabeled estradiol) and 1.21 µg 4-14C-testosterone (9430 Bq) in a 0.01 M CaCl2 solution. These amounts were chosen to reflect the amount of estradiol and testosterone present in poultry litter when applied to pasture at a rate of 5 Mg ha–1 (Shore et al., 1995). After the pulse had entered the column, the 0.01 M CaNO3 background solution was restored and 21 L (approximately 10 pore volumes) of background solution were allowed to flow through the column before halting flow and disconnecting the column from the mariotte bottle and fraction collector. A 10-cm long, three-pronged time domain reflectometry (TDR) waveguide (Dynamx, Houston, TX) was inserted into the surface of the saturated column and volumetric water content (
) determined with a cable tester (Tektronix, Beaverton, OR) before draining the remaining solution from the column.
Hormone Distribution
Once the column water drained to field capacity, four soil cores were removed from the column with a push probe (2 cm i.d. by 32 cm). Care was taken to avoid sampling near holes created by the three-pronged TDR waveguide and to ensure that minimum compaction occurred within the soil cores. Visual inspection of soil cores indicated a maximum compaction of approximately 1 cm for any single core. Soil samples were divided into 1-cm increments by depth, air-dried, and crushed. Soil from all four cores were combined according to depth and thoroughly mixed. A 2-g soil subsample from each 1-cm increment was mixed with 10 mg of cellulose (Sigmacell 100, Sigma-Aldrich) and oxidized using a biological oxidizer (OX-500, R.J. Harvey Instruments, Hillsdale, NJ). During oxidation, all 3H and 14C present in the soil were released and captured in two liquid scintillation cocktails (R.J. Harvey Instruments, Hillsdale, NJ). Cocktails were analyzed for radioactivity by LSC as previously described.
Column Effluent Analysis
Radiolabeled estradiol and testosterone concentrations in column effluent indicated hormone breakthrough. A 2-mL sample was taken from each 67-mL effluent fraction, combined with 18 mL of Scintiverse BD (Fisher Scientific, Pittsburgh, PA), and analyzed for radioactivity by LSC. Total hormone present in each fraction of effluent was determined by multiplying the mass of radiolabeled hormone found in 1 mL of each fraction collector tube by 67 mL, the average volume of effluent estimated present in each tube. The breakthrough of the pulse solution was also traced by detecting Cl– present in the effluent.
Differences between hormone breakthroughs for the two tillage treatments were analyzed for statistical significance using a paired t test to account for differences between the two hormones across tillage conditions. Paired t tests also determined significance of hormone recovery for individual hormones across tilling conditions. Paired t tests were performed using SAS version 8.02 (SAS Institute, 1999).
Solute Transport Models
Hormone transport was modeled with the HYDRUS-1D software package (Version 2.0; imnek et al., 1998). Three conceptual models were considered to describe the movement of solute in the column transport experiment. The low sorption affinity of the Cl– tracer made it highly unlikely to experience significant chemical nonequilibrium processes. Any observed nonequilibrium transport characteristics were instead attributed to physical processes such as macropore or preferential flow. A two-domain, dual-porosity solute transport model with linear sorption described the physical nonequilibrium transport of the Cl– tracer (van Genuchten and Wagenet, 1989; Brusseau and Rao, 1990). This model assumes that the liquid phase present in the column can be divided into a mobile fraction, m (cm3 cm–3), and an immobile fraction, im (cm3 cm–3). Solute exchange between the two domains is modeled as a first-order process with 
as the mass transfer coefficient (imnek et al., 1998). The fraction of sorption sites in contact with mobile water is f. The overall convection dispersion equation (CDE) is:
 | [3] |
b is bulk density (g cm–3), 
is dispersivity (cm), v is mean pore water velocity (cm min–1), and x is distance (cm) (imnek et al., 1998). The inverse capability of HYDRUS-1D was used to estimate the transport parameters , f, im, and from a fit of the numerical solution to Eq. [3] to the observed Cl– breakthrough curve data. To limit the number of variables modeled and increase the uniqueness of the solution, Kd was fixed to the value obtained from the sorption experiments.
The HYDRUS-1D model can consider either physical or chemical nonequilibrium, but not both. Therefore, the hormones were modeled first with a chemical nonequilibrium model based on a two-site sorption model (Nkedi-Kizza et al., 1984). This model divides sorption into Type-1 and Type-2 sites where f is the fraction of Type-1 sites. On Type-1 sites, sorption is assumed to be instantaneous and described by the Freundlich equation (Casey et al., 2003). On the remaining (Type-2) sites, sorption is considered to be a first-order kinetic process where is the first-order kinetic sorption coefficient (min–1). Transport is described using the following form of the CDE:
 | [4] |
from a fit of the numerical solutions to Eq. [4] to the observed hormone breakthrough curve data. We assumed the distribution of pore water velocities in a given column would be the same for Cl– as for the hormones. Thus, values of were fixed to those previously determined for a column by the physical nonequilibrium model of the Cl– breakthrough curves. Although we measured soil hormone concentrations in the soil at the end of the experiment, we did not include these in the objective function when fitting transport parameters. This was because we thought the soil measurements underestimated the mass of hormone in the soil (discussed below in the results section).
The physical nonequilibrium model (Eq. [3]) was also applied to the hormone data, assuming nonlinear sorption (Eq. [2]). In this case HYDRUS-1D estimated f, im, and . As in the previously described chemical nonequilibrium analysis, values of were assumed to be the same as the values generated by the physical nonequilibrium analysis of the Cl– breakthrough curves.
Since the most likely transport scenario for hormones was a combination of physical and chemical nonequilibrium, a third, quasi-combined model was also applied to the hormone data. This model consisted of the chemical nonequilibrium model and an "effective dispersivity" to account for physical nonequilibrium. Thus for the quasi-combined model, HYDRUS-1D estimated , f, and for the observed hormone data. Brusseau and Rao (1990) refer to the use of the conventional CDE with an effective dispersivity as an alternative to physical and chemical nonequilibrium models. Paired t tests were used to test the estimated parameters for significance across tillage and hormone treatments with the SAS software package (SAS Institute, 1999).
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
In conventionally tilled soil, estradiol sorption occurred most rapidly between 0 and 2 h for both the 0- to 10-cm and 20- to 30-cm depth soils (Fig. 1a). Estradiol concentrations reached sorption equilibrium in soil of both depths in between 48 and 72 h. This pattern of rapid sorption, followed by slower, more gradual sorption, and eventually reaching equilibrium between 5 and 240 h was similar to those of previous studies (Jürgens et al., 1999; Holthaus et al., 2002; Bowman et al., 2002; Lee et al., 2003; Casey et al., 2003, 2005; Yu et al., 2004). Sorption to the soil from the 0- to 10-cm depth appeared to decrease slightly after 144 h of shaking. Because the experiment ceased at 144 h it was difficult to determine whether the decrease was due to a pattern or anomalies in the data. An actual decrease would indicate desorption of hormone from the soil similar to the desorption reported by Lai et al. (2000). They also reported sorption that occurred more rapidly than this study with estradiol sorption peaking within 1 h and desorption occurring within 5 h. The reason for the difference between sorption times is unknown. It is possible that the apparent desorption of estradiol in this experiment is actually the result of metabolite formation over time. Other studies have found that metabolites possess sorption characteristics that differ from parent compounds and decrease soil sorption (Lee et al., 2003; Casey et al., 2005). However, since there was only a slight decrease in our study, either the sorption characteristics of the metabolite were similar to the parent hormone, or only a small amount of metabolite was formed.
|
The testosterone kinetic sorption pattern was similar to estradiol for the 0- to 10-cm depth, but not the 20- to 30-cm depth (Fig. 1b). Like estradiol, sorption to 0- to 10-cm depth soil occurred most rapidly between 0 and 2 h and appeared to reach equilibrium between 48 and 72 h. Testosterone sorption to the 20- to 30-cm-depth soil reached equilibrium within 2 h of shaking. Such rapid sorption was similar to the testosterone sorption observed by Casey et al. (2004). In their experiment maximum testosterone sorption occurred within 5 h of shaking and was followed by testosterone desorption over the next 45 h. In contrast to the results of that experiment, no desorption was evident in our study. Like estradiol, the greatest testosterone sorption was to the 0- to 10-cm soil depth for all time intervals, with a maximum sorption of 3.22 µg g–1. Testosterone sorption for the 20- to 30-cm soil depth attained a maximum of 2.34 µg g–1.
Sorption Isotherms
A comparison of Freundlich nonlinear sorption isotherms with linear sorption isotherm for goodness of fit indicated that Freundlich nonlinear sorption isotherms best described estradiol and testosterone sorption data (Table 2). Nonlinear regression of estradiol and testosterone sorption experiment data using Eq. [2] resulted in mean r2 values of 0.99, whereas mean r2 values for linear isotherms (Eq. [1]) were 0.92 for estradiol and 0.96 for testosterone (Table 2). These results concur with several studies that have reported nonlinear hormone sorption (Lai et al., 2000; Casey et al., 2003; Lee et al., 2003; Ying et al., 2003; Yu et al., 2004).
|
|
|
The Kf values for both hormones decreased with depth, possibly due to the reduction of soil TOC with depth. There was a significant tillage-by-depth interaction in that Kf values decreased more sharply with depth in no-till than in conventionally tilled soil (p < 0.05). This is likely due to a sharper TOC decrease with depth for the no-till soil (57% decrease between 0- to 10- and 20- to 30-cm soil) vs. the conventionally tilled soil (50% decrease with depth). Crop litter left undisturbed on the surface of no-till soils concentrates TOC in the top 0- to 10-cm whereas conventional tillage incorporates more organic matter to deeper depths of the conventionally tilled soil, reducing the difference between surface and subsurface TOC.
Sorption isotherm experiments for Cl– resulted in Kd values ranging from 0.3 to 0.5 mL g–1 for soils from the 0- to 10-cm and 20- to 30-cm depths of conventionally tilled and no-till plots. Overall sorption of Cl– to soils was deemed to be low enough for Cl– to be used as a nonreactive tracer during transport experiments (although we included sorption of Cl– when we modeled transport with HYDRUS-1D).
Although r2 values for hormone Kd were lower than those of Kf, they depicted the data with sufficient accuracy to determine the Koc for tillage and depth treatments (Table 2). The range of log Koc values for all estradiol at both depths was between 3.54 and 3.90 (mean log Koc = 3.72), within the range of previously reported values (Lai et al., 2000; Holthaus et al., 2002; Lee et al., 2003; Ying et al., 2003; Yu et al., 2004; Casey et al., 2005). The testosterone log Koc range of between 3.39 and 3.70 (mean log Koc = 3.52) also corresponded well with previously reported log Koc values (Lee et al., 2003).
Hormone Recovery
Breakthrough of detectable concentrations of 6, 7-3H-estradiol and 4-14C-testosterone radioactivity occurred in all soil columns. Overall, slightly more testosterone mass broke through soil columns than estradiol. However, this difference was not statistically significant (p = 0.09). Such a trend was expected given the lower sorption affinity of testosterone. This result also agrees with previous hormone transport studies that indicate testosterone is the more likely of the two hormones to experience significant subsurface transport (Casey et al., 2003, 2004; Das et al., 2004). Of the hormones initially present in the pulse, an average of 27% of the estradiol (SE = 7.1) and 42% of the testosterone (SE = 8.3) leached from the six columns. Greater breakthrough of estradiol occurred for conventionally tilled columns than no-till columns. However, no other tillage effects were observed during the transport experiment for either hormone treatment.
Oxidation of soil column samples recovered an average of 29% of the estradiol (SE = 5.1) and 17% of the testosterone (SE = 1.2). The larger percentage of initial estradiol recovered in soil was expected given the higher estradiol Kf values reported by the batch adsorption isotherms. The majority of the soil-bound estradiol and testosterone (
50%) sorbed to the top 10 cm of soil with the remainder fairly evenly distributed throughout the column (Fig. 2). The pattern is not surprising, give2 the greater sorption affinity both hormones display for soils from the 0- to 10-cm depth. This distribution pattern corresponded well with previously reported column experiments that also found the majority of soil-bound hormone within the top 10 cm (Casey et al., 2003, 2004). Furthermore, an earlier study at this site found that preferential flow begins below the 10-cm depth of tillage (Gupte et al., 1996), so more sorption in the 0- to 10-cm depth is expected.
|
Because land-applied manure contains both estradiol and testosterone, the simultaneous application of both hormones in the transport experiment should provide a more accurate depiction of transport under field conditions than studies in which each hormone is applied separately (Casey et al., 2003, 2004, 2005; Das et al., 2004). The results of the sorption experiments, however, may not present a completely accurate portrait of the sorption observed in either the transport experiments or the environment. Competitive sorption between estradiol and testosterone may influence the sorption characteristics of one or both hormones, and as degradation occurs the formation of metabolites may contribute to the effect of competitive sorption. Yu et al. (2004) reported competitive sorption between estradiol and estrone reduced the sorption coefficients (Koc) of both hormones. Our study was unable to determine what effect competitive sorption might have had on hormone transport because the sorption experiments were performed on each hormone separately rather than simultaneously and detection of metabolites was not feasible.
Twelve soil cores were initially extracted from the three conventionally tilled and three no-till plots (two columns from each plot). These undisturbed soils displayed highly variable steady effluent flux (Jw) rates. The Jw values of three conventionally tilled and one no-till soil column were too low for transport experiments to be completed within a reasonable amount of time. These columns were discarded. The three remaining conventionally tilled soil columns represented each of the three conventionally tilled plots. To maintain experimental consistency, three no-till columns were chosen to represent each of the three no-till plots in the experiment. It became evident during the transport experiment that the Jw values of the selected conventionally tilled columns (average Jw = 0.38 cm min–1) were similar to those of the no-till soil columns (average Jw = 0.51 cm min–1). By selecting columns with higher flow rates, we may have unintentionally biased the selection in favor of conventionally tilled columns with more structure. In addition, testing only six soil columns, as opposed to the original twelve, reduced the statistical power of comparisons between tillage.
Column Transport Analysis
During the column transport experiment, peak Cl– concentrations all occurred within 12 to 103 min, which corresponded to between 0.55 to 0.85 pore volumes (Fig. 3). Peak tracer concentrations occurring at less than one pore volume clearly indicated that preferential flow influenced Cl– transport for all columns. In addition, asymmetrical Cl– breakthrough curves ending in long tails confirmed the presence of physical nonequilibrium processes in all columns. A physical nonequilibrium transport model described the transport of the Cl– very well (Table 5). This model produced large values for the relative immobile water content in the columns (average imob –1 = 0.62). Such values indicated that the soil matrix was relatively immobile compared to a smaller fraction of the soil (macropores) where most of the transport occurred. The values for dispersivity () were large, but less than the column length. This is consistent with the interpretation that dispersivity is a measure of the average ped diameter or distance solute travels in a pore before mixing with solute from another pore (Leij and Dane, 1989).
|
|
Hormone peak concentrations occurred after the Cl– peak in only one (no-till) soil column. Initial estradiol and testosterone breakthrough occurred simultaneously (65 min, 0.49 pore volumes), but lagged behind Cl– breakthrough (NT 1 in Fig. 3). Estradiol concentrations reached their peak almost 1 h after Cl– (159 min, 1.40 pore volumes) whereas testosterone concentrations did not peak until 193 min (1.73 pore volumes) had elapsed. These peak concentrations represented much more rapid movement through the soil than would be predicted from the sorption isotherms. Using the average values for Kd from Table 2 for estradiol (16.0 mL g–1) and testosterone (8.7 mL g–1), the average b from Table 1 (1.65 g cm–3), a water content of 0.33 cm3 cm–3, and assuming linear adsorption (n = 1), the retardation coefficient (R = 1 + b Kd/) was 83 for estradiol and 47 for testosterone. The retardation coefficient is an estimate of the number of pore volumes at which one expects the peak concentration for a narrow pulse input of a sorbed solute to occur (Jury et al., 1991). Clearly, peak concentrations occurred much earlier in our undisturbed columns.
Colloid transport can cause rapid movement of hormones (Dizer et al., 2002), but we used dilute solutions of CaNO3 and CaCl2 (0.01M) to prevent clay dispersion and saw no evidence (such as turbidity in out-flow samples) of clay dispersion. We also used stock solutions instead of poultry manure as our source of hormones so water-soluble organic matter was not present as a potential colloid.
Hormone Transport Modeling
In this study, the HYDRUS-1D model of physical nonequilibrium transport did not provide a satisfactory fit to the hormone data (Table 5). The average r2 values (0.64 for estradiol and 0.79 for testosterone) were much less than the value for Cl– (0.99). The fitted parameters indicated a greater degree of nonequilibrium was required to fit the hormone breakthrough curves compared to the Cl– breakthrough curves—the fraction of adsorption sites in the mobile region (f) and the mass transfer coefficient between mobile and immobile regions () decreased, and the immobile water content (im) either increased (estradiol) or stayed the same (testosterone).
The HYDRUS-1D model of chemical nonequilibrium transport did not provide an adequate fit to the hormone data either (Table 5). The average r2 values for fit of the predicted breakthrough curves for estradiol (r2 = 0.52) and testosterone (r2 = 0.68) were relatively poor compared to the Cl– data (r2 = 0.99). The model predictions either missed the peak or failed to describe the long tail. There were no significant differences between estimated transport parameters for estradiol or testosterone, or between tillage treatments for either hormone. Estimates of the fraction of Type-1 sites (f) for estradiol were well within the range of f values reported by Casey et al. (2003). However, mass transfer coefficient () values were all at least 100 times less than previously reported values, and indicated a much slower rate of sorption on Type-2 kinetic sorption sites. Testosterone breakthrough curve parameters were similar to estradiol parameters with values also showing slow sorption of testosterone to Type-2 sites.
Only the quasi-combined model of physical and chemical nonequilibrium provided a satisfactory fit to the hormone data with r2 approaching that of the Cl– data (Table 5 and Fig. 3). The estimates of effective dispersivity generated by this model were much larger than those found in the Cl– data and much longer than the column length. Large effective dispersivity estimates indicated a nonequilibrium process in addition to that of chemical nonequilibrium. We believe this second process was physical nonequilibrium as the Cl– data clearly showed that physical nonequilibrium (due to preferential flow through macropores) occurred in these undisturbed soil columns. The presence of chemical nonequilibrium is supported by the data on kinetic sorption. The comparative success of the quasi-combined model in describing data, as compared to either the physical or chemical nonequilibrium models, is further evidence that both physical and chemical nonequilibrium processes were important to hormone transport.
The quasi-combined model predictions of sorbed hormones in the soil at the end of the experiment are compared to the measured mass in Fig. 2a and 2b. The model predictions show roughly the same pattern as the measured data (high near the surface and decreasing with depth), but the predicted masses are higher than those observed. These results support our view that the soil sample data underestimated the actual hormone mass due to problems in sampling such large columns.
Previous column transport studies have focused on exploring the effect of chemical nonequilibrium transport alone on hormone transport by using repacked soil columns to intentionally eliminate physical nonequilibrium. The presence of physical nonequilibrium in this study, however, appeared to increase the relative concentration of hormone transported through soil. Casey et al. (2003, 2004) repacked soil into columns and recovered, on average, only 5% of the initial estradiol mass and 21.4% of the initial testosterone mass present in the pulse solution, significantly less than the 27% average initial estradiol mass and 42% initial testosterone mass recovered here. Some column studies have recovered a greater percentage of their initial hormone mass in effluent than our study. However, these studies used much smaller columns and followed pulse solutions with between 66 and 100 pore volumes of background solution (Das et al., 2004; Casey et al., 2005).
The addition of physical nonequilibrium in this experiment may also have facilitated high peak concentrations of initial hormone mass to leach from the columns. Based on detectable hormone radioactivity in the effluent, peak concentrations of initial estradiol mass recovered in effluent ranged from 0.006 to 0.123 µg L–1, within the range reported to cause endocrine disruption in fish (Desbrow et al., 1998; Routledge et al., 1998). Peak concentrations of initial testosterone mass ranged from 0.031 to 0.378 µg L–1.
Study Limitations
While our study was able to observe the effects of preferential flow on hormone transport, several factors limit the ability of this study to accurately predict the transport of hormones under field conditions. Time constraints necessitated selection of columns with higher water flux, possibly obscuring tillage effects on hormone transport by introducing an unintentional experimental bias toward columns with more macropores. The hormone sorption affinity of the experimental apparatus was not determined. While the amounts of hormone used in both the sorption and transport experiments represented realistic amounts applied to soil in poultry litter, the experiments used only pure estradiol and testosterone and did not consider the effect that poultry litter may have had on hormone transport. Dizer et al. (2002) found evidence indicating water-soluble organic matter in animal manure facilitates transport of estradiol and other endocrine disruptors into soil.
Finally, the experimental design of the sorption and transport experiments did not incorporate analysis for the possible development of hormone metabolites. At the start of the experiments in early 2002 the current wealth of literature regarding the extent and rate of hormone degradation did not exist. Given our resource constraints, combined with a lack of relevant literature, we deemed the presence of the radiolabeled 3H and 14C tracer in both experiments sufficient to indicate the presence of estradiol and testosterone, respectively. More recent reports of rapid degradation under a wide variety of conditions have indicated a high likelihood that at least a portion of the detected radiolabels were attached to metabolites rather than the parent compounds. Hormones degraded into metabolites still display estrogenic and androgenic properties, albeit to a lesser extent, thus keeping our results relevant. The presence of a large percentage of the initial pulse radioactivity in effluent indicates the possibility that macropore flow may facilitate the transport of either parent hormone or metabolite into groundwater where continuing bioactivity may threaten both humans and wildlife.
|
SUMMARY AND CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONSUMMARY AND CONCLUSIONSREFERENCES |
|---|
imnek, and G.L. Larsen. 2004. Fate and transport of testosterone in agricultural soils. Environ. Sci. Technol. 38:790–798.[Medline]imnek, J. Lee, G.L. Larsen, and H. Hakk. 2005. Sorption, mobility, and transformation of estrogenic hormones in natural soil. J. Environ. Qual. 34:1372–1379.imnek. 2003. Fate and transport of 17ß-estradiol in soil-water systems. Environ. Sci. Technol. 27:2400–2409.[CrossRef]imnek, J., M. ejna, and M.Th. van Genuchten. 1998. The HYDRUS-1D software package for simulating the one-dimensional movement of water, heat, and multiple solutes in variably-saturated media. Version 2.0. USDA–ARS Salinity Laboratory, Riverside, CA.
) through columns of conventionally tilled (CT) and no-till (NT) Cecil sandy loam. Each graph possesses two y-axes. The left y axis represents the relative concentration of hormones (C C0–1). The right y axis represents the relative concentration of Cl– (C C0–1). Solid lines indicate the fitted model generated by HYDRUS-1D. A physical nonequilibrium model was fit to the Cl– data and a quasi-combined physical and chemical nonequilibrium model was fit to the hormone data.