Environ. Sci. Technol. 2010, 44, 2341–2347
Transport and Modeling of Estrogenic Hormones in a Dairy Farm Effluent through Undisturbed Soil Lysimeters LAURE D. STEINER,† V I N C E N T J . B I D W E L L , ‡ H O N G J . D I , * ,† KEITH C. CAMERON,† AND GRANT L. NORTHCOTT§ Centre for Soil and Environmental Quality, P.O. Box 84, Lincoln University, Lincoln 7647, Christchurch, New Zealand, Lincoln Ventures Ltd., P.O. Box 133, Lincoln 7640, Christchurch, New Zealand, and Plant and Food Research, Ruakura, Private Bag 3213, Hamilton, New Zealand
Received October 13, 2009. Revised manuscript received January 21, 2010. Accepted January 29, 2010.
The presence of endocrine-disrupting chemicals, including estrone (E1) and 17β-estradiol (E2), in surface waters has been associated with physiological dysfunction in a number of aquatic organisms. One source of surface and groundwater contamination with E1 and E2 is the land application of animal wastes. The processes involved in the transport of these hormones in the soil, when applied with animal wastes, are still unclear. Therefore, a field-transport experiment was carried out, where a dairy farm effluent spiked with E1 and E2 was applied on large (50 cm diameter and 70 cm depth) undisturbed soil lysimeters. The concentrations of E1 and E2 in the leachate were monitored over a 3-month period, during which irrigation was applied. The experimental data suggest that E1 and E2 were transported through preferential/macropore flow pathways. The data from the experiment also show that E1 and E2 are leached earlier than the inert tracer (bromide). This observation can be explained either by the presence of antecedent concentrations in the soil or by an enhanced transport of E1 and E2 through the soil. A statespace mixing-cell model was further developed in order to describe the transport of E1 and E2 by three transport processes in parallel. The inverse modeling of the leaching data did not support the hypothesis that antecedent concentrations of estrogens could be responsible for the observed breakthrough curves but confirmedthatestrogensweretransportedmainlyviapreferential/ macropore flow and also via an enhanced transport. The parameter values that characterized this enhanced transport strongly suggest that this enhanced transport is mediated by colloids. For the first time, the simultaneous transport of E1 and E2 was modeled under transient conditions, taking into account the advection-dispersion, preferential/macropore flow, and colloidalenhanced transport processes as well as E1 and E2 dissipation in the soil. These findings have major implications in terms of management practices to decrease E1 and E2 transport and water contamination.
* Corresponding author email:
[email protected]. † Lincoln University. ‡ Lincoln Ventures Ltd. § Plant and Food Research. 10.1021/es9031216
2010 American Chemical Society
Published on Web 02/18/2010
Introduction A number of endocrine-disrupting chemicals (EDCs) have been shown to affect the normal reproduction functions of aquatic organisms (1). Among the naturally occurring EDCs, two estrogens, estrone (E1) and 17β-estradiol (E2), are of major concern because they are found in the environment at concentrations that affect the normal endocrine function of many aquatic species (2). The threshold level of E2 and E1 above which endocrine-disrupting effects on aquatic organisms are observed, for a number of different end points, is in the range 1-100 ng L-1, depending on the species and life stage (1). Recognizing that estrogens are unlikely to occur in isolation, a conservative maximum predicted-no-effect concentration for total estrogens of 1 ng L-1 estradiol equivalents has been proposed in the U.K. (3). One major source of environmental water contamination with E1 and E2 is the application of animal wastes on land as fertilizer (4). In several studies, following agricultural waste application on land, concentrations of E1 and/or E2 in groundwater above 1 ng L-1 were observed (5). Kjær et al. (6) monitored E1 and E2 following the application of pig slurry using GC-MS/MS and measured, at 1 m depth, concentrations of E1 and E2 of 68.1 and 2.5 ng L-1, respectively. Similarly, following swine manure application, Thompson et al. (7) detected E2, using LC-MS, in monitoring wells at 2-3 m depth at concentrations ranging from 16 to 100 ng L-1. These results suggest that a fraction of the applied estrogen load is transported through the soil profile, despite the important sorption and degradation processes typically observed in packed soil columns (8). The deep leaching of estrogens in an undisturbed soil profile could be explained in some cases by the preferential transport of the estrogens through soil macropores (6, 9). Where preferential/macropore flow transport could not explain the presence of estrogens in the soil, it has been suggested that a colloidal-enhanced transport of the estrogens could take place (5, 7, 10, 11). Colloids are organic, inorganic, or biological particles of 1 nm to 1 µm in size. The enhanced transport of contaminants associated with colloids is explained by the colloid size-exclusion effect or the reduced sorption of the contaminants to the soil. A dairy farm effluent (DFE) is mainly composed of dairy cow excreta (ca. 10%), teat washings (ca. 4%), and wash water (ca. 86%) (12). Maximum natural E1 and E2 concentrations of 3123 and 331 ng L-1, respectively, have been measured in DFE in New Zealand (NZ). It is unknown whether the application of DFE on land, which is a common practice in NZ, is a threat to the groundwater quality and can, consequently, affect the quality of the connected surface waters and drinking water supplies. The objective of the present study is therefore to determine whether E1 and E2 in DFE applied on land are transported through the soil profile and to characterize the transport processes involved. DFE, spiked with additional E1 and E2 to ensure reliable detection at depth, was applied on undisturbed soil lysimeters under transient flow conditions, in order to measure E1 and E2 breakthrough in the leachate. The experimental data were used to test different hypotheses about estrogen transport processes through soil using a solute transport model.
Materials and Methods Soil and DFE. The soil used for this experiment was a Templeton soil (Immature Pallic soil (13); Udic Haplustept (14)) from the Canterbury region of NZ where application of a DFE on the land is a common practice. Some major properties of the Templeton soil are reported in the Supporting Information. VOL. 44, NO. 7, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2341
Fresh DFE was collected from a local dairy farm less than 2 h before application on the lysimeters. The total nitrogen content of the DFE was 261 ( 3 mg L-1, and the total carbon content was 415 ( 0.7 mg L-1. The DFE had a pH of 7.8 ( 0.2 and a conductivity of 1462.3 ( 247.3 µS cm-1. The total dry matter content of the DFE was 2.0 ( 0.2 mg L-1, and the colloidal/dissolved dry matter, measured after filtration through a 1.2-µm-pore-size filter, was 0.8 ( 0.1 mg L-1. The DFE naturally contained low concentrations of E1 (66.2 ( 1.2 ng L-1) and E2 (5.0 ( 0.4 ng L-1). Details on the methods used to measure the reported DFE properties can be found in the Supporting Information. Field-Transport Experiment. Lysimeter Soil Columns. Six undisturbed columns of Templeton soil (50 cm in diameter and 70 cm depth) were collected from a site south of Christchurch on the Canterbury Plains (43°38′11′′ S; 172°26′18′′ E), according to a method developed by Cameron et al. (15), and installed in a field lysimeter facility at Lincoln University. Petroleum jelly was used to seal the gap between the soil column and the metallic casing around it in order to prevent edge-flow effects. The surface of each lysimeter was at the same level as the surrounding soil, and the pasture on each lysimeter was maintained at a height between 5 and 25 cm, before and during the experiment. Although urea and cow urine had been applied on the lysimeters 3 years before the start of the present experiment (16), no other treatment had been applied since then. Before the start of the present experiment, the soil moisture content in each lysimeter was brought to field capacity. Treatment Application. The Canterbury Regional Council permits land application of DFE provided the rate of application of nitrogen does not exceed 200 kg N ha-1 per year (17). Targeting the application of a third of the maximum yearly application, 5L of the effluent were applied on three of the lysimeters (L1-L3). Each 5 L of effluent was spiked with bromide (NaBr) to act as an inert tracer (232 ( 3 mg of Br L-1). The same volume of effluent was also spiked with 20 mL of a concentrated solution of E2 and E1 in ethanol for final concentrations of 4.50 ( 0.13 and 5.20 ( 0.16 mg L-1, respectively. After spiking, the effluent was stirred and poured by hand onto the soil surface to give as even a surface coverage as possible. The remaining lysimeters (S4-S6) were used as controls and received 5L of water spiked only with bromide at a concentration of 232 ( 3 mg of Br L-1. Irrigation. Irrigation was applied to reach a target of 25 mm of water every 3 days during the first month and of 25 mm week-1 during the second and third months, including the contribution of precipitation. Spray irrigation was applied simultaneously on all lysimeters using nozzles (Teeget FL5GC) placed 30 cm above the soil surface. To limit ponding effects, irrigation cycles of 30 s followed by a 5 min wait time were applied. The lysimeters were drained exclusively by gravity because they were collected from a site that normally has gravel at about 60 cm depth. The average evapotranspiration was measured on-site using weighing lysimeters. The total amount of water applied on the lysimeters over the period of the experiment (331 mm) was representative of the amount of rain that can be expected in Canterbury during a wet winter. More information on the irrigation pattern and the amount of evapotranspiration can be found in the Supporting Information. Leachate and Soil Sampling. The leachate samples were collected from the lysimeters L1-L3 in 10 L high-densitypolyethylene plastic vessels protected from the sunlight and sampled every 2-4 days during the first month and every 7 days during the following two months. At each of the 20 sampling events that took place during the 92 day experiment, the vessels were weighed to determine the volume of leachate collected and replaced by a new set of preweighed vessels. The volume of cumulated drainage collected for each 2342
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 7, 2010
lysimeter over time during the period of the experiment is provided in the Supporting Information. Triplicate soil samples were collected from the control lysimeters S4-S6 at two different soil depths, 5-10 cm (A horizon) and 30-35 cm (top of the B horizon), using a soil corer (2 cm diameter), and the sampling holes were backfilled after collection. Seven soil sampling events took place over the period of the experiment at days 1, 3, 6, 14, 30, 57, and 91 after treatment. The soil samples were passed through a 2-mm-mesh sieve and stored in plastic vials in the freezer at -18 °C until analysis. Extraction and Analysis. Estrogen Extraction from Leachate Samples. Leachate samples containing particulate materials were filtered through glass fiber filters through which colloidal material can pass (GF/C, 1.2 µm, Whatmann) using 0.5 cm filter aid (Hyflo SuperCel, Merck) (18). Leachate subsamples of 1.2 L, or smaller if not enough leachate was collected, were spiked with 60 ng of an internal standard (ISTD; 17βestradiol-2,4,16,16-d4, >98% atom D, C/D/N Isotopes, Canada) to account for any loss during sample manipulation and passed through solid-phase-extraction (SPE) cartridges (Superclean ENVI-18, 1 g, Supelco, Auckland, NZ) in order to extract E1 and E2. The SPE cartridges were then dried under full vacuum for 10 min and stored in their original packaging in the freezer at -18 °C until elution. After the cartridges were thawed for 15 min, the estrogens were eluted with acetone under gravity. After the elution procedure was repeated a second time, a full vacuum was applied to the cartridges to remove any residual acetone. The samples were stored in the freezer (-18 °C) before cleanup and analysis. Estrogen Extraction from Soil Samples. Estrogens were extracted using a sonication and shaking method. In brief, 6 g of sieved moist soil was extracted with 10 mL of an isopropyl alcohol/water (1:1) mixture for 10 min in a sonication bath (25 °C, Sonorex Digital 10P) followed by 30 min on a flat back shaker (235 rpm, IKA KS501), and the extraction was repeated twice. The sample extract was passed through a SPE cartridge (OASIS 500 mg) and eluted with dichloromethane/diethyl ether (80:20) and 2% methanol. Estrogen Analysis. Soil and leachate sample extracts were cleaned through silica gel and gel permeation chromatography according to a method adapted from Sarmah et al. (18). E1 and E2 in the sample extract were derivatized by acylation with trifluoroacetic anhydride. Analysis of the derivatized sample was carried out on an Agilent 6890N gas chromatograph coupled to an Agilent 5975A inert XL mass spectrometer and a CTC autosampler, according to a method described by Sarmah and Northcott (19). The method limit of quantification (MLQ) was defined for the leachate and soil matrices by a signal-to-noise ratio of 10. The MLQ for E1 and E2 was 1.0 ng L-1 for leachate samples and 0.1 ng g-1 for soil samples. The relative standard error of the mean calculated for all samples in which a fixed concentration of ISTD (E2-d4) was added before sample extraction was 3.0%. The ISTD accounted for losses occurring during all of the sample manipulation steps, and the 3.0% error includes all variations introduced by sample manipulation from extraction to analysis. Bromide Analysis. At each leachate sampling event, 20 mL leachate subsamples were collected and filtered through a 0.2 µm membrane filter before analysis for bromide by ion-exchange chromatography (Dionex DX-120, Dionex Corp.). The ions were separated in a Dionex AS9-SC column with a weak carbonate/sodium bicarbonate mobile phase and detected by conductivity.
Data Analysis and Modeling Model Description. A model was constructed, in an Excel spreadsheet, to test the assumption that bromide and estrogens can be simultaneously subjected to transport via
FIGURE 1. Schematic representation of the model with up to three transport processes running in parallel: advection-dispersion (AD), colloidal-enhanced transport (Col), and preferential/macropore flow (macro). See the text and Supporting Information for a detailed description of each parameter. preferential/macropore flow (macro), advection-dispersion (AD), or colloidal-enhanced transport (Col) processes. In this model, these three parallel transport processes do not interact with each other and the transporting water flux associated with each transport process is allocated according to the fractions FAD, FCol, and Fmacro (Figure 1). For each transport process, solute transport through the mobile phase is described by an AD mechanism that is simulated by the mixing-cell model of Bidwell (20). This model is a finite-volume numerical model of solute transport comprising a series of cells through which the transporting water flows. Within each mixing cell, solute is completely mixed, can interact with nontransporting phases, and can undergo degradation. Dispersion is simulated by the number n of mixing cells in the series, such that dispersivity λ ) L/2n [L], for a transport depth L [L]. The dynamics of macropore flow, and to a lesser extent the dynamics of colloidalenhanced transport, are simulated by setting n to a large value so that the resulting value of dispersivity is very small. The mixing-cell model has cumulative drainage flux, rather than time, as the index variable, so that transient flux inputs of varying magnitude are represented by events at uneven intervals of cumulative flux. Dispersion processes are determined by the incremental flux of each event, whereas degradation processes are calculated from the time between occurrences of flux inputs. Solute inputs to the model are generated by partitioning the solute mass M [M] applied to the land surface into an immobile mass fraction, resident at the land surface, and an initial mobile resident solute mass fraction (Fm). Solute flux input cin [M L-3] is generated by leaching the immobile mass at a fractional rate a [L-1 cumulative flux] into the mobile phase. The initial resident solute concentration cr(j ) 0) [M L-1] is distributed into the mobile phase of the mixing cells according to a depth-dependent exponential distribution
exp(-2rλ/d), for cell r, where d [L] is a characteristic depth parameter. To simulate macropore flow transport, the whole solute mass applied is assumed to be mobile. For the Col process, the solute mass MCol represents the mass of estrogens associated with colloids. Reversible, linear, equilibrium sorption of estrogens to the immobile phase is simulated in the mixing-cell model by use of the retardation factor R. The R factor can be regarded as the ratio between the total resident solute mass and the dissolved mass on a soil volume basis. In the present model, the R factor is used as a multiplier of the mobile transport fraction θm, so that each cell volume becomes 2θmR. To simulate the colloidal-enhanced transport, estrogens are assumed to be sorbed to colloids and the RCol factor represents the attachment of colloids to the soil-immobile solid phase. It is also assumed that the solutes transported by macropore flow do not sorb to the soil. Degradation of E2 and E1, either free or associated with colloids, is described by the following time-dependent processes, assuming that E1 is the daughter product of E2: dCE2 ) -k2CE2 dt dCE1 ) -k1CE1 + k2CE2 dt where C [M L-3] and k [T-1] are the concentration and degradation rate parameters, respectively. It is also assumed that the resident time of the solute transported by macropore flow is too short for degradation to occur. The degradation of estrogen during the first 2 days of the experiment within the collecting vessel by DFE bacteria transported by preferential/macropore flow is taken into account by means of a first-order process with rate parameter κ [T-1] for the first leaching event only. VOL. 44, NO. 7, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2343
TABLE 1. E1, E2, and Bromide Mass Recoveries (%) in the Leachate over the Period of the Experimenta % recovery
bromide E1 E2
L1
L2
L3
S4
S5
S6
85 0.05 0.06
70 7.6 0.3
68 13 0.7
75 n/a n/a
86 n/a n/a
55 n/a n/a
a The background E1 and E2 concentrations were not measured in the leachate samples of the control lysimeters S4-S6 (n/a ) not applicable).
Filtration of colloids is simulated as a first-order process, which depends on transport distances associated with flux events. In the model, the filtration rate σ [L-1] is applied to colloidal movement between mixing cells according to transport distances determined by cell dimension 2λ. The complete computational mathematics of the model are provided in the Supporting Information. Fitting Procedure and Data Analysis. The experimental data were used to estimate the parameter values by inverse modeling. The bromide data were first used to adjust the λ, θm, Fm, a, d, FAD, and Fmacro parameters in the AD and macro processes in each lysimeter, using constraints or the results from previous experiments when available. Those parameters were unchanged during estimation of the R, k, and κ parameters for estrogen transport via AD or macro processes. However, the transport parameters used to simulate the colloidal transport of estrogens were adjusted independently because bromide is a solute tracer and not a colloidal tracer. The adjusted parameter values were considered acceptable if they were in accordance with the physical reality that they described. An R2 value was calculated to evaluate the quality of the fit of the model to the data, and the relative residuals F(yi,fi) were calculated to determine where the model mismatched the data (Supporting Information).
Results and Discussion Bromide and Estrogen Recoveries. The bromide, E1, and E2 recoveries in the leachate from a soil depth of 70 cm in lysimeters L1-L3 are shown in Table 1. The lower recovery of estrogens in comparison to the bromide recovery is attributed to the rapid biodegradation and the important sorption of estrogens in the soil. In most cases, E1 recoveries were larger than E2 recoveries probably because E2 is transformed into E1 in the soil. Bromide and Estrogen Breakthrough Curves. In lysimeters L2 and L3, the early bromide peak concentrations within the first 2 days, which correspond to ca. 0.1 pore volume (PV) of cumulative drainage, strongly suggest the presence of soil macropores created by biological activity or physicochemical processes (21), which are responsible for the preferential nonequilibrium transport of solute in soil (22) (Figure 2d,g). In lysimeter L1, however, there was no sign of the presence of macropores, which therefore suggests that bromide was transported only via an AD process (Figure 2a). The observed concentrations of E1 and E2 in the leachate reported over the cumulated drainage (or cumulated volume of leachate) are shown in the Supporting Information. The presence of preferential/macropore flow in Templeton soil and its impact on nutrient leaching has already been observed (23, 24). In lysimeters L2 and L3, early peaks of E1 and E2 appeared in the leachate simultaneously with the early bromide peak, within 2 days (or ca. 0.1 PV), demonstrating that estrogens were also transported via macro pathways (Figure 2e,f,h,i). As a result of the important sorption of estrogens to the soil, it is expected that estrogen breakthrough would be 2344
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 44, NO. 7, 2010
delayed over the inert tracer (bromide) breakthrough. However, in all lysimeters, the E1 and E2 peak concentrations in the leachate (after day 2) appeared before the corresponding bromide peak concentration (Figure 2). Three hypotheses were formulated to explain the earlier breakthrough of estrogens over bromide. The first hypothesis is that estrogens are transported only by the AD and macro processes. In the second hypothesis, the presence of antecedent estrogens within the soil profile is assumed, while considering that the deeper the presence of antecedent estrogens, the faster they would reach the bottom of the lysimeter following irrigation. The third hypothesis is that an estrogen-enhanced transport occurs in the soil, which could be a colloidal-enhanced transport, especially in the presence of DFE. Inverse Modeling of Bromide and Estrogen Transport. The predicted bromide, E1, and E2 breakthrough curves at 70 cm soil depth in lysimeters L1-L3 are shown in Figure 2, together with the observed data, and the values of the model parameters obtained by the inverse modeling are shown in Table 2. Bromide. The parameter θm represents the fraction of water participating in solute transport. In the macro transport process, the whole water content is considered to be mobile and to participate to the transport of E1 and E2. Therefore, θm,macro represents the depth- and time-averaged total water content of the soil. The θm,macro value obtained by inverse modeling (0.31) is closed to the theoretical volumetric soil moisture at the field capacity for Templeton soil (as reported in the Supporting Information). In the AD transport process, only a fraction of the total water content is participating in solute transport and, therefore, θm,AD is smaller than θm,macro. The degree of saturation of the soil, as well as the soil structure and texture, have an impact on the value of θm,AD. In order to simulate the AD transport of bromide in lysimeters L2 and L3, which showed signs of the presence of preferential flow, a larger characteristic depth (d) than that in lysimeter L1 was necessary. A large d value suggests that solute was rapidly transported along the soil profile by preferential flow and diffused into the soil matrix, because of a concentration gradient, further down the soil profile, where AD transport took place, as suggested by Ghodrati et al. (22). The lack of sensitivity of the mass-transfer rate parameter (a) made its precise adjustment difficult. Therefore, a similar transfer rate of 5 m-1, determined by Bidwell (25) to describe bromide leaching in undisturbed Templeton soil lysimeters, was used to predict bromide leaching by AD in all lysimeters in the present model. The longitudinal dispersivity (λ) is dependent on the pore properties of the soil, with a larger pore-size distribution leading to a larger dispersivity (26). Therefore, the λ value for the macro process, which accounts for transport via large pores only, is smaller than that for the AD process, which accounts for a wider range of the soil pore sizes. The R factor was set to 1, assuming no sorption of the inert tracer to the solid phase. The large Fm values obtained for all of the lysimeters suggest that most of the applied solute mass was initially partitioned into the mobile water. Estrogens. Following the application of 5 L of DFE on a soil column that contains macropores, it is expected that a portion of the DFE microbes are leached through the macropores directly in the collecting vessel, where they might degrade E1 and E2 only for the period of time before the first sampling (i.e., 2 days). E1 and E2 degradation rates in the collecting vessel, κ1 and κ2, were found by inverse modeling to be slightly different between lysimeters L2 and L3, even though the ratio of κ2 to κ1 is similar. In lysimeter L2, the extent of macro pathways was more important, and therefore
FIGURE 2. Bromide (4), E2 (b), and E1 (O) concentrations in the leachate at 70 cm depth over time measured in the field (symbols) and predicted by the model (full line) in lysimeters L1 (a-c), L2 (d-f), and L3 (g-i). more bacteria might have been cotransported, leading to a faster degradation. The kAD and RAD factors needed to be specifically determined for E1 and E2. Using an inverse-modeling approach and applying no constraint on kAD and RAD at all, no combination of these two parameters giving a satisfactory description of the observed data could be found for any of the lysimeters. Therefore, the first hypothesis suggesting that the estrogen breakthrough could be explained by the transport of estrogen via the AD and macro processes only was rejected. The RAD factor for E1 and E2 was, nonetheless, set within a range of R values reported in the literature by authors modeling estrogen transport in soil (9, 27, 28). These R values ranged from 42 to 249 for E2 and from 14 to 137 for E1 and were either directly reported by the authors or estimated from reported Kd values, using the relation R ) 1 + FbKd/θ (29). For further modeling exercises, the kAD values were estimated as being equal to the average degradation rates measured in a separate incubation experiment (data not shown) carried out under temperature and moisture conditions similar to those of the present experiment. Those degradation rates, derived from a simple exponential model, were of 1 and 2 day-1 for E1 and E2, respectively, assuming that estrogen degradation occurs in both mobile and immobile water. Using these kAD and RAD, reported in Table 2, the model predicted that no estrogens are transported by the AD transport process to a depth of 70 cm.
To test the second hypothesis involving antecedent concentrations of estrogens in the soil, estrogen concentrations measured in the soil of the control lysimeters, reported in the Supporting Information, were used. The input concentration cin was set to zero, and the initial concentration cr,j)0 was used to simulate antecedent concentrations along the soil profile. The estimation of cr,j)0 in each cell along the soil profile was based on the highest E1 + E2 background concentrations measured in lysimeters S4-S6. Those background concentrations in the soil were 5.9 ng g-1 in the 5-10 cm layer and 2.3 ng g-1 in the 30-35 cm layer. Between those two points of reference, the initial concentrations cr,j)0 were assumed to be linearly distributed along the soil profile. Otherwise, keeping the same parameter values estimated previously, the model predicted no breakthrough of either E2 or E1 at 70 cm depth in lysimeter L1, even under this worst-case scenario. Similar exercises for the other lysimeters also resulted in the conclusion that antecedent concentrations could not explain the observed breakthrough curves. The third hypothesis to explain the earlier breakthrough of estrogens over bromide involves the colloidal-enhanced transport of estrogens in the soil, which assumes that estrogens are bound to colloids and that colloids are transported in the soil. Mineral, organic, or biological colloids are present in the soil as well as in the DFE. The sorption of hydrophobic chemicals like estrogens to colloids is believed to be rapid (30, 31) and to involve hydrophobic (32) and VOL. 44, NO. 7, 2010 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2345
TABLE 2. Values of the Model Parameters Obtained by Inverse Modeling That Describe Bromide, E1, and E2 Transport in the Leachate in Lysimeters L1-L3 by macro, AD, and Col Processes solute-process
θm
λ [m]
Fm
R
Br-AD E2-AD E1-AD E2-Col E1-Col
0.265 0.265 0.265 0.056 0.046
0.058 0.058 0.058 0.012 0.012
0.85 0.85 0.85 0.18 0.15
1.0 42 14 1.5 1.4
Br-AD E2-AD E1-AD E2-Col E1-Col Br-macro E2-macro E1-macro
0.280 0.280 0.280 0.042 0.042 0.310 0.310 0.310
0.035 0.035 0.035 0.012 0.012
