Occurrence of estrogenic chemicals in South Korean surface waters and municipal wastewaters

Abstract

Broad scale monitoring of estrogenic compounds was performed at 19 sampling points throughout the Yeongsan and Seomjin river basins and 5 wastewater treatment plants (WWTPs) adjacent to the Gwangju area, Korea, from December 2005 to August 2007. The concentrations of estrogenic compounds, including estrone (E1), 17β-estradiol (E2), 17α-ethynylestradiol (EE2), bisphenol-A, nonylphenol (NP) and 4-octylphenol (OP), in the samples was measured with gas chromatography/mass spectrometry (GC-MS). In addition, the estrogenic activities throughout the river were investigated using the E-screen assay. Of the six estrogenic chemicals, NP (114.6–336.1 ng L⁻¹) and EE2 (0.23–1.90 ng L⁻¹) were detected at the highest and lowest levels, respectively in both the river waters and the WWTP effluents. Bisphenol-A showed the largest concentration range, from 7.5 to 335 ng L⁻¹. The concentrations of E1, E2 and octylphenol ranges were 3.6–69.1, 1.2–10.7, and 2.2–16.9 ng L⁻¹, respectively. According to the calculated estradiol equivalent concentration (EEQ); however, no estrogenic contribution was observed due to the phenolic compounds in the river waters and effluents. E1 and E2 dominated in both the river water and effluent samples, with contributions to the calculated EEQ of over 79 and 77%, respectively. Conversely, EE2 was rarely detected in the river waters (21%) and effluents (0%). The largest contribution of EE2 to the calculated EEQ was 21% in the river water at S-7. The levels of E1, E2, and EE2 were remarkably decreased in the effluents, indicating that the 5 WWTPs did not contribute to the estrogenic effect of the receiving streams. Overall, the WWTPs did not contributed to the estrogenic activity of the receiving waters, but the livestock industry or wildlife may play an important role in the estrogenic contribution to river water.

---

Environmental impact

The estrogenic chemicals in natural water have been reported to cause reproductive disorders and adverse effects on the endocrine system in humans and other wildlife. This study is a rare investigation on the occurrence and distribution of estrogenic compounds performed in Korean river basins and wastewater treatment plants using a comparison of their concentrations via both biological and chemical approaches. The results revealed that WWTPs do not impact on the EEQ of the receiving waters, but rather the livestock industry or wild life may contribute to the estrogenicity in river water. Moreover, the possible under- or over-estimation of the estrogenic activity in an environmental matrix was discussed through critical data analyses via an E-screen assay and GC-MS.

Introduction

Endocrine disrupting chemicals mimic estrogens by binding to estrogen receptors in the human body, which can alter the homeostasis of the endocrine system. Failure to maintain homeostasis can lead to many deleterious effects, such as developmental problems, reproductive success, fecundity and longevity in aquatic species. Various natural and synthetic estrogens have been identified in laboratory animals, including phthalates, phenols, pharmaceuticals, pesticides and heavy metals. To estimate the adverse effects of these chemicals, extensive investigations have been performed to monitor or assess the status of estrogenic chemicals contamination in waste water treatment plant effluents, river waters, groundwater and drinking water.^1–4

The release of estrogenic chemicals mainly originates from wastewater treatment plant (WWTP) effluents, as they are excreted from humans and animals, or from industrial activities. In addition, the combined treatment process of human excretions with industrial and animal husbandry effluents at WWTPs has attracted attention due to the possible estrogenic activity of WWTPs effluents and their detrimental effects on humans and ecosystems.³ However, the environmental concentrations in rivers and surface waters not related to WWTP effluents has also revealed similar concentrations to those found in WWTP effluents.⁵ Considering the complicated nature of the discharge of estrogenic chemical contaminants, intensified investigations on surface waters related to drinking water, recreation and sports sources are required.

Many diverse methods have been adopted for the monitoring of environmental estrogens in river waters and WWTP effluents, including GC-MS/MS, LC/MS/MS, E-screen, yeast two-hybrid, ER-luciferase and enzyme linked immunosorbent assay (ELISA). Chemical analyses can give both quantitative and qualitative information on the chemicals in water samples. Conversely, biological approaches are able to reveal the real estrogenic activities of environmental samples. Biological methods, such as ELISA, are sometimes applied to estimate residual estrogen concentrations, but whole body or cell-based approaches normally assess the entire estrogenic activity in environmental samples. The environmental concentrations of natural estrogens, 17β-estradiol (E2) and estrone (E1), in river water have been reported to range from 0.1 to 100 ng L⁻¹, but concentrations higher than this range have been found in several studies; 592.7 ng L⁻¹ for E1 and 200 ng L⁻¹ for E2.^6,7 Several studies have reported E2 concentrations in river and surface waters; 0.15–3.6 ng L⁻¹ (Germany), 2.6–14.7 ng L⁻¹ (Japan), 1.4–32 ng L⁻¹ (France), <93 ng L⁻¹ (U.S.), 1.1–1.7 ng L⁻¹ (China) and <5.5 ng L⁻¹ (The Netherlands).^8,9 The 17β-estradiol equivalent concentration (EEQ), an alternative measure of the estrogenic activity, is commonly introduced to explain the potential hazard of whole sample adverse endocrine effects. The conversion of chemical concentrations to EEQ allows easier estimation or prediction of their effects than just knowing the concentration. Biologically measured EEQs are used mainly for E2, as well as for E1, and EE2. Pojana et al. [2004] observed a contribution of E1, E2 and EE2 to the total EEQ of 97% in a lagoon water.¹⁰ EEQ levels in river and surface waters have been observed to range from 0.5 to 1.1 ng-EEQ/L (Switzerland), 4.3 to 47 ng-EEQ/L (Italy) and 0.023 to 1.711 ng-EEQ/L (Portugal).^2,11,12

Theoretically, chemical measurements provide detailed information regarding each estrogenic chemical in surface waters, which are then converted into EEQ to predict the overall estrogenic activity. However, a number of chemicals that exert estrogenic activity, as well as their synergistic interaction, make precise calculations difficult. In addition, the various physicochemical characteristics of field samples can affect the biological sensitivity or tolerance of test species or cell lines, making correlation studies even more difficult.

In this study, chemical and biological estrogenic activities were measured using gas chromatography/mass spectrometry (GC-MS), as well as a human breast cancer cell line proliferation (E-screen) assay on river water samples collected at 19 points and on the influents and effluents of 5 WWTPs. Based on the results, the risk from estrogenic compounds in river water was evaluated by their conversion to EEQ values.

Experimental

Chemicals and materials

The standard chemicals; nonylphenol, 4-octylphenol, bisphenol-A, 17β-estradiol, 17α-ethynylestradiol and estrone, were obtained from Aldrich (St. Lousi, Mo, USA). For trace analyses and derivatization, at or below the part per billion levels, dichloromethane, hexane and acetone (Burdick & Jackson; Muskegon, MI, USA) were used with solid phase extraction (SPE) cartridges. Internal and surrogate standards, bisphenol-Ad₁₆ and pyrened₁₀, were purchased from Aldrich (USA). Derivatization reagent, N-methyl-N-(trimethylsilyl) trifluoroacetamide (MSTFA), was obtained from Sigma Aldrich (St. Loise, MO, USA). Sep Pak^®C18 cartridges (Waters) were employed for chemical extraction from the water samples. For cleanup, laboratory manufactured 3 layered columns were prepared, with Na₂SO₄ (Sigma Aldrich, USA), Florisil (Supelco, USA) and Na₂SO₄ at a ratio of 1 : 2 : 2.

For the E-screen assay, stock solutions of NP, 4-OP, BPA, E1, E2 and EE2 were prepared in acetone. The MCF 7 BUS cell line was supplied by Dr Soto (Tufts University School of Medicine, Boston, MA, USA). Dulbecco's Modified Eagle's Medium (DMEM), fetal bovine serum (FBS), trypsin and 0.4% sulforhodamine B (SRB) dying solution were purchased from Gibco-BRL life technologies (Basel, Switzerland). DMEM and phenol red-free DMEM, supplemented with 10% FBS treated with 5% charcoal (Sigma, USA) and 0.5% dextran (Sigma, USA) were used for the bioassay.

Site description and sampling

Water samples were collected from the Yeongsan and Seomjin Rivers, which are located in South Korea and used as resources for agriculture, livestock farming, industry, as well as drinking water. The basins of the Yeongsan and Seomjin Rivers occupy 3,371.4 and 4,896.5 km² and extend 136 and 212.3 km, respectively. There is a dam at the end of the Yeongsan River estuary, with a storage capacity of 250 million m³, while the Seomjin River flows directly to sea. The flow rates of the two rivers fluctuate greatly throughout the year, especially with the seasons. Most of the rain is localized during the monsoon season between June and August; therefore, the rivers have an abundance of water during this season. The largest supply of nonpoint source pollutants to the rivers also occurs during this season. Conversely, during the dry season, the influxes of most pollutants to the rivers are caused by point source discharges. Samples were collected from 19 sites along the rivers (e.g., upstream and downstream of the WWTPs) and from 5 WWTPs (A–E). These WWTPs receive industrial, manure and municipal wastewaters.

Most of the WWTPs use activated sludge treatment methods, but two use rotating biological contactor (RBC) methods, with a capacity of about 10,000 m³ day⁻¹. Of the 5 WWTPs, the largest (WWTP-B) is located in Gwangju province, with discharges of effluent into the Yeongsan River of 0.6 million m³ day⁻¹. WWTP-A is located upstream of WWTP-B and discharges 8,000 m³ day⁻¹ of effluent to a sampling point between Y-1 and Y-2. WWTP-C is located at a tributary of the Yeongsan River and discharges effluent to a sampling point between Y-4 and Y-5 (11,000 m³ day⁻¹). WWTP-D is on the main stream of the Yeongsan River, is regarded as a positive source of pollutants to Y-8 and discharges 22,500 m³ day⁻¹ of effluents. WWTP-E is on the mainstream of the Seomjin River and discharges 2,600 m³ day⁻¹ of effluents to a sampling point between S-3 and S-4.

Grab samples were collected on six occasions for river water, between June 2006 and August 2007, and eight times for WWTPs effluent, between December 2005 and August 2007. The sampling points are shown in Fig. 1. Samples were collected using a 1.0 L stainless steel bucket and stored in 4 L amber glass bottles with polytetrafluoroethaene (PTFE) caps. Samples were acidified on site with hydrochloric acid, to prevent bacterial activities (pH 2.5 ∼ 3.0), and stored in an ice box. Samples were refrigerated at 4 °C and extracted for GC-MS analysis and E-screen assay within 3 days of sampling. After extraction, the solvents were reconstituted for the bioassay. All glassware was rinsed in a stepwise order of surfactant, ethanol, acetone and deionized water.

Fig. 1 The sampling sites along the Yeongsan and Seomjin River basins (), as well as the locations of the wastewater treatment plants () in South Korea.

Gas chromatography and mass spectrometry analysis

Samples were loaded onto pre-conditioned Waters Sep-Pak^®Vac 6cc C18 cartridges, at a flow rate of 10 ml min⁻¹, followed by deionized water to wash the cartridges. Cartridges were dried by blowing with dry nitrogen for 30 min and then eluted using dichloromethane. Cartridge conditioning was carried out by passing 5 ml of methanol and then 5 ml of deionized water through the cartridges. Prior to sample loading, the suspended particles were removed by filtration through a pre-rinsed glass fiber filter (GF/C, 0.45μm, Whatman^®). The retentates were sonicated in methanol to recover the estrogenic compounds absorbed onto the suspended particles, which were then added to the dichloromethane eluent of the C18 cartridge. The eluent mixture from both the retentate and cartridge was dehydrated with anhydrous sodium sulfate and concentrated to 500 μl, with 400 μl used for the E-screen assay and 100 μl for the GC-MS analysis. The 100 μl samples were evaporated to dryness under a nitrogen stream at 40 °C, followed by derivatization with MSTFA in 2 ml of acetone for 15 min and then evaporated to 500 μl. To purify the derivatized samples, a cleaning column packed with Na₂SO₄, florisil and Na₂SO₄ was employed at a ratio of 1 : 2 : 1, in grams. Column conditioning was carried out by passing 15 ml of hexane through the column before sample loading. Columns were then eluted with 15 ml of hexane, then evaporated to dryness. The pellet on the bottom was resuspended in 100 μl of acetone and transferred to a 2 ml amber vial containing 10 μl of the internal standard; bisphenol-Ad₁₆. The vial was derivatized by adding 90 μl of MSTFA for 15 min. Finally, a 200 μl sample was used for the GC-MS analysis.

The GC-MS analysis was performed on a Shimadzu GC-17A gas chromatograph coupled to a Shimadzu QP-5050 mass spectrometer (Shimadzu, Kyoto, Japan). Samples were separated using a Restek®, XTI-5 fused silica capillary column (RESTEK Corp., Bellefonte, PA, USA), 30 m in length, with a 0.25 μm film thickness and 0.25 mm internal diameter. A 3 μl sample was injected in the splitless mode, with a sampling time of 2 min and He (99.9999%) as the carrier gas at a flow rate of 1.0 ml min⁻¹. The oven temperature was maintained at 100 °C for 2 min, then ramped to 300 °C at a rate of 10 °C min⁻¹, and held at 300 °C for 10 min. The temperatures of the injection port and interface were 280 and 290 °C, respectively. The mass spectra were obtained under electron impact ionization, with an ionization voltage of 70 eV and ion source temperature of 280 °C. Target compounds were quantitatively determined in the selective ion mode (SIM).

Table 1 shows the retention times, limits of quantification, recoveries, relative standard deviations and m/z values of the six estrogenic compounds. The peaks for the compounds from the GC were eluted in the order: NP, OP, BPA, E1, E2 and EE2. The average recoveries of the target estrogens ranged from 69.4 to 132.8%, with relative standard deviation between 2.62 and 9.55%. The limits of quantification were lowest for E1, E2 and OP and highest for EE2, with concentrations ranging from 0.5 to 10.0 ng L⁻¹.

E-screen assay

400 μl of the samples in dichloromethane were completely dried under a stream of nitrogen and resuspended in 100 μl of DMSO. The extracts were then diluted to the desired concentration in phenol red-free DMEM, supplemented with 10% FBS treated with 5% charcoal and 0.5% dextran. Human breast cancer cells (MCF-7 BUS) were routinely maintained and grown in Dulbecco's modified Eagle medium, supplemented with 10% fetal bovine serum and antibiotic-antimycotic, under a 5% CO₂ and 95% air atmosphere and saturating humidity conditions at 37 °C. During the experiments, DMEM, without phenol-red, but supplemented with 10% FBS treated with 5% charcoal and 0.5% dextran, was used to increase the sensitivity of the E-screen assay, as explained by Lee et al. [2008].¹³ Before the experiments, MCF-7 BUS cells were acclimatized in phenol red–free DMEM, containing FBS treated with 5% charcoal and 0.5% dextran, for 24 h. At the start of each experiment, MCF-7 BUS cells were trypsinized and seeded into 24 well plates, at 10⁴cells/well, in 400 μl of the above-described medium. The cells were allowed to attach for 24 h. The medium was then replaced with the above-mentioned medium, containing varying concentrations of the test compounds, with the cells then cultured for 6 days.

Table 1 Analytical parameters for the target compounds, RT; retention time, RSD; relative standard deviation, LOD; limit of detection

Parameters	RT	LOD/ng L^−1	LOQ/ng L^−1	RSD (%)	Recovery (%)	m/z
Nonylphenol	12.88–13.60	1.0	3.3	2.98	115.1	193, 221, 207, 235, 295
p-octylphenol	13.78	0.1	0.3	2.62	69.4	179, 278
Bisphenol-A	17.86	0.5	1.7	4.52	132.8	357, 372, 386
Estrone	21.75	0.1	0.3	9.55	93.7	218, 257, 342
17β-estradiol	22.03	0.1	0.3	3.95	128.7	285, 416
17α-ethynylestradiol	23.02	1.0	3.3	3.83	95.9	196, 232, 425
Pyrene-d10	14.19	92.5	305.3	16.77	121.4	212, 106, 211

---

The growth of cells was determined using the sulforhordamine B (SRB) assay. The test medium was removed, with 300 μl of 10% trichloroacetic acid (Merck, Darmstadt, Germany) added to each well to fix the cells onto the well plate. After 40 min, the well plates were washed five times with tap water and then dried. The cells were dyed with 0.4% SRB solution for 1 h, with 1% acetic acid used to remove the excess SRB solution from the wells. Four-hundred microlitres of 10 mM Tris solution (pH 10.5) were added to each well, and the absorbance measured at 560 nm using an enzyme-linked immunosorbent assay reader (uQuant; Bio-Tek Instruments, Winooski, VT, USA). The details of the SRB assay have been described previously by Körner et al. (2001).¹⁴ In general, cell growth is affected by intrinsic and extrinsic factors, such as cell passage number and difference in the cell conditions within each of the well plates. To minimize these variations, the raw absorbance data were normalized using the following equation.¹⁵

where Abs_test is the absorbance of the test compound in a well, Abs_nc the absorbance of the negative control well and Abs_pc the absorbance of the positive control well.

Results and discussion

Phenolic compounds in river water

GC-MS analyses were performed on the samples, which were collected on six occasions, from the 19 sampling points along the Yeongsan and Seomjin River basin. A total of 114 samples were collected and analyzed for nonylphenol (NP), 4-octylphenol (OP), bisphenol-A (BPA), estrone (E1), 17β-estradiol (E2) and 17α-ethynylestradiol (EE2). Table 2 shows the results of the chemical analyses of the river water samples. The results are expressed as the mean analyte concentrations, with standard deviations given in parentheses. The target analytes were present in measurable concentrations up to several hundred ng L⁻¹. Of the 6 chemicals, NP, with 211 theoretical isomers, exhibited the highest concentrations, ranging from 114.63 to 336.14 ng L⁻¹. Due to its isomers, NP had a wide range of retention times on the chromatogram (12.88–13.60 min). OP, a degradation product of 4-octylphenol polyethoxylate, showed a mean concentration more than 30 times lower (2.24–16.78 ng L⁻¹) than NP, which can be explained by the relatively high usage of nonylphenol polyethoxylate (80% in production) in detergent production compared to octylphenol polyethoxylate (20%).¹⁶ The main use of OP is in the production of phenol/formaldehyde resins (98% of commercially used OP compare to 2% for octylphenol polyethoxylate); however, their concentrations that reach the environment is unclear.¹⁷ Therefore, most of the OP in natural water can be attributed to an octylphenol polyethoxylate origin. The OP concentrations observed in this study were relatively low compared to those found in other studies on river samples. For example, the environmental concentrations of OP reported in river waters ranged from 70 to 4,200 ng L⁻¹ in Portugal, 1.0 to 2,470 ng L⁻¹ in China, and 13 ng L⁻¹ in EU countries.^4,18,19 The reason for the large differences in the OP concentrations between this and other studies is unclear, but suggests the latent possibility of OP contamination in the Yeongsan and Seomjin River basins through changes in the patterns of its use or for some other reason. However, considering the low estrogenic activities compare to E2, NP and OP, they are thought to rarely contribute to the estrogenic activity. The occurrences of NP and OP in river waters were revealed to have a weak relationship via a Pearson correlation (r = 0.37). The levels of NP in spring and summer differed greatly from those in fall and winter. Likewise, OP showed a significantly higher concentration in summer (p < 0.05; Student's t-test). Bisphenol-A was detected at lower concentrations than NP, but higher than OP, and had its highest concentration at Y-3 (335.56 ng L⁻¹). This sampling point was located downstream of the largest WWTP (B), which discharges 600,000 m³ day⁻¹ of effluent, indicating an adverse effect of the effluent from B on the receiving stream (Y-3). The mean concentration of BPA (60.5 ng L⁻¹) was three times lower than that of NP (208.4 ng L⁻¹), but was observed to have a 9 times higher concentration than OP, with a mean concentration of 6.96 ng L⁻¹. At most observation points, the BPA concentrations ranged up to several tens of ng L⁻¹. As a conclusion, these three phenolic compounds were not detected at high enough concentrations so as to cause estrogenic activity.

Table 2 Mean (± standard deviation) concentrations of the estrogenic compounds in the Yeongsan (Y) and Seomjin (S) River basins; results of six analyses from June 2006 to August 2007 (n = 6), nd; non-detected

Site	Mean concentration/ng L^−1
	nonylphenol	octylphenol	bisphenol-A	estrone	17β-estradiol	17α-ethynylestradiol
Y-1	248.89 (145.95)	10.67 (13.81)	21.51 (28.24)	5.10 (6.22)	6.12 (6.83)	nd
Y-2	155.20 (132.62)	12.89 (30.92)	29.19 (36.58)	24.11 (23.84)	7.27 (8.05)	1.08 (2.65)
Y-3	321.00 (177.20)	5.17 (10.49)	335.56 (277.52)	69.05 (142.81)	7.98 (8.50)	nd
Y-4	157.88 (154.93)	3.66 (5.37)	13.95 (10.51)	24.98 (25.64)	5.46 (5.97)	1.60 (3.93)
Y-5	209.56 (81.94)	5.93 (9.65)	12.09 (13.61)	19.76 (20.50)	7.86 (13.58)	1.76 (4.32)
Y-6	133.53 (149.48)	7.12 (10.98)	10.05 (7.68)	17.68 (22.65)	5.08 (5.67)	0.87 (2.14)
Y-7	278.41 (162.64)	7.54 (12.80)	172.39 (75.42)	15.13 (10.20)	9.41 (13.29)	0.23 (0.56)
Y-8	186.18 (80.30)	4.25 (5.41)	86.26 (95.03)	13.10 (20.87)	7.94 (11.74)	nd
Y-9	336.14 (297.98)	10.25 (16.50)	104.29 (88.28)	18.19 (22.02)	6.05 (6.45)	nd
Y-10	180.77 (133.96)	3.39 (6.43)	100.49 (93.62)	15.89 (19.77)	1.83 (2.73)	0.38 (0.93)
Y-11	228.89 (145.39)	5.45 (8.31)	96.02 (113.84)	34.04 (30.22)	10.70 (11.01)	nd
Y-12	238.12 (259.00)	16.78 (36.28)	49.50 (50.16)	29.99 (29.04)	4.20 (5.78)	nd
S-1	296.68 (353.19)	6.12 (14.99)	7.54 (6.28)	3.63 (4.48)	1.15 (2. 82)	nd
S-2	189.89 (251.36)	3.22 (5.44)	11.98 (12.07)	16.13 (17.25)	2.94 (5.07)	nd
S-3	258.95 (64.62)	13.54 (28.66)	13.42 (18.06)	11.70 (18.25)	7.39 (4.52)	nd
S-4	159.92 (113.69)	2.24 (4.08)	18.62 (17.58)	22.63 (21.50)	6.84 (8.33)	nd
S-5	114.63 (39.15)	3.20 (4.66)	32.35 (38.27)	19.50 (19.59)	8.42 (7.63)	nd
S-6	126.01 (60.78)	5.08 (6.43)	9.90 (4.41)	16.45 (16.55)	7.51 (5.28)	nd
S-7	139.80 (52.30)	5.73 (5.63)	24.46 (29.41)	15.02 (18.53)	3.31 (5.67)	1.90 (4.66)

---

Natural and synthetic estrogens in river water

The natural and synthetic (EE2) estrogens investigated in this study were E1, E2 and EE2. EE2, one of the most commonly used medications for the control of fertility, was detected at a very low average environmental concentration at 6 sampling points in both the Yeongsan and Seomjin River basins (0.23–1.76 ng L⁻¹). The highest mean EE2 concentration was observed at Y-5 (a downstream of C treatment plant) in the Yeongsan River basin. In a comparison of the basins, the Yeongsan River (50%) overwhelmed the Seomjin River (14%) in its detection frequency. EE2 was only observed at one sampling point in the Seomjin River basin. However, there was no significant difference in the observed concentrations between the two basins. The remaining 50 (the Yeongsan River) and 86% (the Seomjin River) of the sampling points had levels below the detection limits. Interestingly, the EE2 concentration at Y-3 point, downstream of WWTP-B (capacity; 600,000 m³ day⁻¹), was below the detection limit. Considering WWTP effluents are the main source of EE2 (0.2–7.5 ng L⁻¹ in the Netherlands, 0.3–0.5 ng L⁻¹ in Germany, and >7.0 ng L⁻¹ in the UK), the WWTP-B system probably favors the control (degradation, transformation, and sorption) of EE2 (an activated sludge system with hydraulic retention time of 10 h).^8,20 In support of the statement above, the EE2 concentration at WWTP-B decreased from 16.3 ng L⁻¹ (influent) to below the detection limit (effluent).

E1 and E2 were detected at all sampling points, and showed higher concentrations than EE2 in the river water samples. Also, these natural estrogens have no relationship with EE2 in their occurrence patterns or concentrations. Because of its high stability and estrogenicity, EE2 is normally categorized as an important estrogen in estimations of its effects on humans and aquatic ecosystems. In this investigation; however, E1 and E2 triggered greater concerns than EE2 because of their high concentrations and frequencies of detection. The E1 concentrations ranged from 3.63–69.05 ng L⁻¹, which were relatively high compared to the results for river water samples studied in France, the Netherlands, Italy and Germany, which were found at levels of 0.8–8.0 ng L⁻¹, 0.1–3.4 ng L⁻¹, 1.5 ng L⁻¹ and 0.1–4.1 ng L⁻¹, respectively, but similar to the results of studies in China (22.4–66.2 for Dan-Shui, 4.29–49.8 for Beitang and 5.00–55.3 for the Dagu River).²¹ The E2 concentrations in the Yeongsan and Seomjin River basins ranged from 1.15 to 10.70 ng L⁻¹, but those of E1 and E2 in the river water samples were about 50.0 and 15.0 times higher, respectively, than that of EE2. In a comparison between E1 and E2, the former showed a 3.3 times higher concentration than the latter in the river water samples. Similar results were observed by Lei et al. [2009], who reported the presence of E1 at two-fold higher concentrations than E2 in river waters in China, which was attributed to the easy degradation characteristic of E2 into E1 under natural water conditions.²¹ When discharged, E2 transforms into E1 within a very short time. For example, Ternes et al. [2002] reported that 95% of E2 in sludge was degraded into E1 within 3 h.²² Also, Jürgens et al. [1999] reported that the microorganisms in river water transformed E2 to E1, with a half life of 0.2 to 9 days at 20 °C, depending on the number of bacteria in the river water.²³ Therefore, without other sources, E2 might be barely detectable in the receiving stream. However, due to the presence of the conjugated form (a slow releasing reservoir of E2), E2 could be detected in the receiving stream after the time full degradation of the E2 in the effluent would have been expected. Although E2 was present at very low concentrations, at most several tens ng L⁻¹, it can readily diffuse through the cell membrane and ultimately up or down regulate many genes.

The E1 concentration in the effluent from WWTP-B was as low as 8.13 ng L⁻¹, but the highest E1 concentration was observed at Y-3 (downstream of WWTP-B). Also, a significantly higher concentration of E1 was observed at Y-3 compared to Y-2 (upstream of Y-3 and WWTP-B), indicating a slight contribution to the E1 concentration of Y-2 from Y-3 (p > 0.05; Student's t-test). For example, a sharp increase in the E1 concentration was observed in August 2007 at Y-3 (360 ng L⁻¹) compared to Y-2 (below detection limit). Therefore, WWTP-B might be a possible source of E1 to the downstream area, not only by direct discharge of E1, but also due to the degradation of E2 to E1 in the Yeongsan River basin. Although E1 is the degradation product of E2, there was only a weak (medium positive relationship by Pearson correlation) relationship between the E1 and E2 concentrations in all of the river samples (r = 0.3461). The levels of three estrogenic compounds slightly varied with the season.

Estrogenic compounds in WWTP influents and effluents

The concentrations of NP at the 5 WWTPs were significantly different from those in the river water samples, but BPA and OP showed similar concentration variations. As shown in Fig. 2a, NP showed the highest concentration among the six target chemicals in both the influents and effluents to and from the WWTPs. The highest mean concentration of NP of 982 ng L⁻¹ was observed in the influent to WWTP-B, but it showed good degradation, with a mean of 79.4%, via the treatment process. BPA exhibited a lower mean degradation efficiency than NP, at 73.3%. In the case of OP, 100% degradation was observed at WWPTs-A and C, but only 55.2% at WWTP-E in the Seomjin River basin.

Fig. 2 Mean concentrations (± standard deviation, n = 8) of phenolic compounds [a] and E1, E2 and EE2 [b] in the WWTP influents and effluents, Inf: influent, Eff: effluent.

The results for the natural and synthetic estrogens are given in Fig. 2b. As depicted in this figure, E1 showed the highest concentration among the three estrogens, and EE2 the lowest. E1 peaked at WWTP-D, but E2 and EE2 were highest at WWTP-B. The EE2 concentration in the influent to WWTP-B was about 9 times higher than that at Y-5, which shows the highest concentration of all the river samples. However, it was degraded to below the detection limit (100%) after treatment. A much lower degradation efficiency was observed in the effluent from an activated sludge process at the WWTPs by Baronti et al. [2000], who reported degradation of 85% for 15 samples.²⁴ The high degradation efficiency found in this study was due to the difficulty in analyzing the complex effluent samples below the ng L⁻¹ level. The limit of detection for EE2 was 1.0 ng L⁻¹; therefore, below this concentration, the peaks on the chromatograms were not counted, but considered as non-detected. E1 showed the highest degradation efficiency among the three estrogens (87.8%), and was fairly higher than the 61% found by Baronti et al. [2000]. WWTP effluents are normally considered a major source of endocrine disruptors.²⁴ Contrary to previous results, our five target WWTPs were considered lesser sources of endocrine disruptors. E1 and E2 were significantly high in the river water samples than in the effluents from the WWTPs. Therefore, nonpoint sources may prevail as the primary inputs of endocrine disruptors to river waters in our study area.

Estrogenic activity in river water

According to previous research, phenolic compounds yield very low E1, E2 and EE2 estrogenic activities.¹⁴ In the case of OP, the estrogenic activity was about 10⁴ times lower than that of E2. The mean concentrations of OP and E2 were 6.96 and 6.19 ng L⁻¹, respectively, and did not show any significant differences in their concentrations (p > 0.05), indicating a slight contribution of OP to the total estrogenic activity in the river water samples. NP also showed a negligible contribution to the total estrogenic activity. This low contribution pattern was well explained by Pojana et al. [2003] in their lagoon water study.¹¹ They observed that estradiol and ethynylestradiol contributed to over 97% of the total estrogenic activity in their lagoon water samples. To estimate the estrogenic activities in river water, the EEQ values were calculated from the chemical concentrations, and then compared with the biologically measured EEQ (Table 3). In brief, the concentrations obtained from the GC-MS analyses were converted to E2 equivalent concentrations based on the dose response curves derived from the E-screen assay. According to the E-screen assay, the EEQ of E2 was highest, followed by EE2 and lastly E1 (data not given). NP, OP and BPA showed very low estrogenic activities compared to E1, E2 and EE2; therefore, their contributions to the calculated EEQ were negligible in the river water samples, which corresponded well with the results of Körner et al. [2001], who showed very low estrogenic contributions of the phenolic compounds compared to E1, E2 and EE2 in both river and effluent samples.¹⁴ From Table 3, the phenolic compounds did not contribute at all to the calculated EEQ at any of the sampling points. The contributions to the calculated EEQ were mainly derived from E1 and E2, and partly from EE2. Despite the high E1 concentration in the river water samples, E2 contributed to the estrogenic activity more than E1, as a result of the estrogenic potency of E2 being 10 times higher than that of E1 (data not given). As much as a 40% contribution of E1 to the mean calculated EEQ was observed at two sampling points in the Yeongsan River basin, with an average of 24.0%. Slightly lower E1 contributions (20.3%) were found in the Seomjin River basin, but the contributions from E2 were the same as for the Yeongsan River basin.

Table 3 Mean (± standard deviation, n = 6) measured and calculated estradiol equivalent concentration (EEQ) of the river water samples collected between June 2006 and August 2007, and the percentage contribution of each chemical analyte to the calculated EEQ. NP; nonylphenol, OP; octylphenol, BPA; bisphenol-A, E1; estrone, E2; 17β-estradiol, EE2; 17α-ethynylestradiol, Y; Yeongsan River, S; Seomjin River

Site	Calculated EEQ (ng-EEQ/L)	Measured EEQ (ng-EEQ/L)	Contribution to the calculated EEQ (%)
			NP	OP	BPA	E1	E2	EE2
Y-1	6.62 (6.66)	0.38 (0.23)	0.0	0.0	0.0	6.6	93.4	0.0
Y-2	10.40 (9.90)	4.29 (6.76)	0.0	0.0	0.0	24.1	72.0	3.9
Y-3	15.32 (22.95)	2.40 (4.09)	0.0	0.0	0.0	47.7	52.3	0.0
Y-4	8.76 (6.72)	0.72 (0.68)	0.0	0.0	0.0	28.0	60.7	11.3
Y-5	10.63 (14.65)	0.80 (0.63)	0.0	0.0	0.0	18.1	71.4	10.5
Y-6	7.49 (6.50)	1.00 (0.81)	0.0	0.0	0.0	25.9	74.1	0.0
Y-7	11.06 (13.56)	5.45 (5.38)	0.0	0.0	0.0	13.9	86.1	0.0
Y-8	9.24 (11.78)	6.27 (8.43)	0.0	0.0	0.0	14.1	85.9	0.0
Y-9	7.90 (7.51)	3.57 (3.99)	0.0	0.0	0.0	23.2	76.8	0.0
Y-10	3.68 (3.94)	1.99 (2.77)	0.0	0.0	0.0	46.5	53.5	0.0
Y-11	14.28 (11.52)	2.91 (3.05)	0.0	0.0	0.0	24.4	75.6	0.0
Y-12	7.30 (7.69)	3.12 (3.29)	0.0	0.0	0.0	41.9	58.1	0.0
S-1	1.49 (2.68)	1.03 (0.64)	0.0	0.0	0.0	0.0	100.0	0.0
S-2	4.57 (4.62)	3.51 (4.47)	0.0	0.0	0.0	35.5	64.5	0.0
S-3	8.55 (5.46)	2.10 (3.03)	0.0	0.0	0.0	13.6	86.4	0.0
S-4	9.13 (9.19)	0.86 (0.55)	0.0	0.0	0.0	25.0	75.0	0.0
S-5	10.40 (6.58)	1.85 (3.14)	0.0	0.0	0.0	18.9	81.1	0.0
S-6	9.18 (4.45)	1.89 (2.39)	0.0	0.0	0.0	18.0	82.0	0.0
S-7	5.59 (9.22)	3.05 (3.39)	0.0	0.0	0.0	24.7	54.3	21.0

---

The means of the EEQ values measured by the E-screen assay in the Yeongsan and Seomjin River basins ranged from 0.38 to 4.29 ng-EEQ/L and 0.86 to 3.51 ng-EEQ/L, respectively. The highest estrogenic contamination was expected at Y-3 due to WWTP-B (the largest treatment plant in the study area) being located upstream of this sampling point. However, Y-3 revealed 2.40 ng-EEQ/L in the E-screen assay, which was slightly higher than that measured in the effluent, which had a value of 2.04 ng-EEQ/L (Table 4). Interestingly, the EEQ at Y-2 (upstream of Y-3 and WWTP-B) revealed the highest value among the three points, at 4.29 ng-EEQ/L. The negligible influence of the EEQ on the receiving stream was also reported by Bacci et al. [2009], where effluents with 3-fold higher EEQ values did not increase the estrogenic activity in the receiving stream, probably due to MCF-7 cell proliferation being prohibited by the presence of anti-estrogenic chemicals in the effluents or river water.⁶

Table 4 Mean (± standard deviation, n = 8) measured and calculated EEQs in the WWTP influents and effluents, and the removal rates of the estrogenic activity in terms of each EEQs between December 2005 and August 2007^a

WWTPs	Calculated EEQ					Measured EEQ
	Influent		Effluent		Removal (%)	Influent		Effluent		Removal (%)
	mean	std	mean	std		mean	std	mean	std
a Mean of 8 samples for each WWTP, Pearson; Pearson correlation, std; standard deviation.
A	27.59	19.29	3.71	5.51	86.5	21.92	12.11	0.38	0.21	98.3
B	13.52	16.46	2.65	4.90	80.4	30.27	10.79	2.04	2.73	93.3
C	10.79	12.85	2.61	5.64	75.8	17.65	6.74	1.65	1.48	90.7
D	15.07	16.80	2.01	5.27	86.7	21.64	10.31	0.91	0.67	95.8
E	14.67	13.81	2.76	3.72	81.2	19.18	9.85	3.84	3.74	80.0
mean	16.33	6.51	2.75	0.61	82.1	22.13	4.88	1.76	1.33	91.6
Pearson	r = 0.04 for influent, r = −0.21 for effluent, r = 0.49 for removal rate

---

At some investigation points along the Yeongsan River (Y-4, Y-5 and Y-6), the measured EEQ values were less than 1.00 ng-EEQ/L. Such low estrogenic activities were also observed by Oh et al. [2000] in 4 major rivers in Korea.²⁵ They divided the rivers into upstream and downstream of potential sources, and then performed E-screen assays. The highest concentrations they found were 7.43 ng-EEQ/L upstream and 2.05 ng-EEQ/L downstream, but reported very low estrogenic activities, which were as low as 0.01 ng-EEQ/L. From the monitoring that has been performed on river waters in Korea, certain river points have been found to be free from estrogenic contamination, presumably due to the low input of estrogenic chemicals (a long distance from the source) or the high self-depuration system.

Estrogenic activity in WWTP influents and effluents

The mean measured EEQ values were considerably higher in the WWTP influents than in the river water. The highest EEQ value found in the influents was 30.27 ng-EEQ/L at WWTP-B, but with an average of 22.13 ng-EEQ/L for the 5 WWTPs (Table 4). The mean measured EEQ value in the influent was sharply decreased after the final treatment, with an average overall removal rate of 91.6%, and showed a maximum at WWTP-A (98.2%) and a minimum at WWTP-E (80.0%). The average removal rate was somewhat decreased to 82.1% in terms of the calculated EEQ (Table 4), which was substantially the same as the mean removal rates for E1, E2 and EE2. To estimate the contribution of the WWTPs to the receiving stream, the EEQ values measured in the effluents were compared with those both up- and downstream. The mean measured EEQ at Y-2 sharply increased from 0.38 to 4.29 ng-EEQ/L after the WWTP-A discharge point, but there was no indication that WWTP A contributed to the estrogenic activity downstream (p = 0.14). The other 4 WWTPs contributed to slight increases or big decreases in the measured EEQ values found in the receiving streams, indicating little or no influence on the estrogenic activity due to WWTP effluents. For example, the EEQ at S-4, which receives effluent from WWTP-E (3.84 ng-EEQ/L), was dramatically decreased from 2.10 to 0.86 ng-EEQ/L. Similarly, Schilirò et al. [2009] reported a decrease in the EEQ value of an effluent receiving stream, where the EEQ concentrations were decreased by 50% in the receiving stream.³ They concluded that WWTPs do not impact on the EEQ of the recipient river, which was also adopted in this study. The well accepted suggestions for this type of result is due to the dilution and self-depuration effects of the receiving river. Moreover, an anti-estrogenic mechanism can explain the decreased EEQ found in this study, suggesting the need for a parallel investigation of both the estrogenic and anti-estrogenic effects (biological approach) or concentrations (chemical approach) in the effluent.²⁶

Correlation between GC-MS analysis and E-screen assay

Table 3 shows the mean measured (E-screen) and calculated (calculated from chemical concentration) EEQ values for the river samples collected on six occasions. The means of the calculated EEQ values were higher than those measured at all the sampling points. The difference was the largest at Y-1 by a factor of 17.4 (calculated EEQ is higher than measured EEQ by a factor of 17.4), and ranged from 1.3 to 17.4, with a mean and standard deviation of 5.3 and 4.7, respectively. This corresponded well with other studies, which showed a low correlation between the measured and calculated EEQ values in sea water and WWTP effluent. Beck et al. [2006] observed a significant difference in a comparison between measured (by yeast estrogen screen) and calculated EEQ values, which they explained was probably due to the presence of anti-estrogens or unknown factors.²⁷ Hashimoto et al. [2007] chemical analyses revealed higher estrogenic activities than a bioassay on a sea water extract from Suruga Bay, Japan.²⁸ According to Körner et al. [2001], when the calculated and measured EEQ values were compared, the concentrations could only explain 13 to 95% of the measured EEQ, probably due to the effect of a mixture of estrogenic and anti-estrogenic compounds in the matrix, as well as the complexity of the matrix during extraction.¹⁴ The antagonistic mechanism can effect the estrogenic receptor binding sites, which might cause an underestimation of the biologically measured EEQ. However, the E-screen assay may not always underestimate the EEQ due to the effects of the matrix or anti-estrogens in river water samples. Conflicting results were observed by Bicchi et al. [2009] in their comparative study between the E-screen assay and GC-MS analyses of WWTP effluents.⁶ They reported that the calculated EEQ explained only 33.8% of the measured EEQ value, which was due to the presence of undetected EDCs in their chemical analyses. However, they observed a high correlation between the calculated and measured EEQ values using the Spearman correlation (r = 0.824). This differed greatly from our results, which showed no relationship between the measured and calculated EEQ values (r² = 0.01).

When all the individual data were considered (total 114 samples for river water), the calculated EEQ showed relatively higher activity than that measured (E-screen) in 75.4% of samples. This result shows that the calculated EEQ is generally higher than that measured, but in some case (here, in 24.6% case), the presence of undetected or non-targeted chemicals in river water can increase the estrogenic activity in the E-screen assay. Low concentrations, around or below the detection limit, are likely to decrease the precision of both the E-screen assay and GC-MS analysis. The detection limit in the GC-MS analysis can be affected by the matrix conditions. In other word, under certain matrix conditions, interference can reduce the overall efficiency of the extraction, clean up and derivatization. The measured EEQ can also be influenced by synergistic or antagonistic effects of chemicals in a mixture. Because of these problems, the precisions of E-screen assay and GC-MS analysis are likely to be low under certain matrix conditions or with various chemical compositions.

Using the Pearson correlation between the calculated and measured EEQ values in WWTPs influents and effluents, no relationship was observed (r = 0.04 for influent and −0.21 for effluent). According to the GC-MS analysis, the contribution of E2 to the total estrogenic activity was the most significant, with a mean of 76% in both the influents and effluents. The EEQ values due to E2 in effluents are of importance because E2 is a natural hormone from both human and livestock sources, which can cause the most severe estrogenic impact on the receiving river. Also, the presence of E2 is provocative because the microorganisms in river water will transform E2 to E1 and; therefore, this could affect the estrogenic activity, even after their breakdown. The removal rate for the calculated EEQ was 9.48% higher than that measured, but both methods showed good removal efficiencies of over 82% as a result of wastewater treatment. When individual chemicals were considered (E1, E2 and EE2), the contribution of E1 was dramatically decreased after the treatment process, from 25.2 to 8.5%. Conflicting results were reported by Bicchi et al. [2009] with respect to WWTP effluents.⁶ They reported a greater contribution of E1 to the total estrogenic activity, and suggested the biotransformation of E2 into E1 due to the treatment process. In this investigation; however, E2 accounted for 91.5% of the total estrogenic activity in the effluents, implying the probable simultaneous degradation of E1 and E2 during the treatment process, but the slow replenishment of E2 due to the decomposition of conjugated E2. Overall, the degradation patterns were similar for both the calculated and measured EEQ values, but no significant relationship was observed, regardless of treatment.

Conclusions

The main contributors to the mean calculated EEQ of the river water samples were E1, E2 and EE2, but E2 accounted for most of this, implying that E2 plays a major role in the estrogenic activity in the Yeongsan and Seomjin River basins. Although other xenoestrogens, including NP, OP and BPA, were detected at high concentrations compared to natural hormones, they rarely contributed to the estrogenic activity (0.0%). The levels of the natural estrogens that contributed most to the EEQ varied slightly according to the season.

The influents to WWTPs contain significantly high levels of estrogenic chemicals compared to river water and effluents. After treatment; however, the concentrations and measured EEQ values were greatly decreased, showing similar estrogenic activity to that found in river water using the measured EEQ (p = 0.37), but significantly lower estrogenic activity via the calculated EEQ (p < 0.05), suggesting no significant effect of WWTP effluents on the receiving rivers in this study. Therefore, although WWTP effluents are generally regarded as a main source for the supply of estrogens to river water, unidentified sources of estrogens, originating from, for example, the livestock industry or wildlife, may also play important roles in the estrogenic activity of river water.

Moreover, the present study has demonstrated that bioassays and chemical analyses of estrogens, when applied to environmental matrices, are unstable due to complexities of the matrix and the effects of chemical mixtures, which can cause synergistic or antagonistic effects on the biological endpoints. Therefore, when the estrogenic activity in environmental samples is sought, combined methods, such as GC-MS analysis and the E-screen assay, will be required to obtain greater reliability. With respect to anti-estrogens, the introduction of negative effect criteria, such as the Reverse Estradiol EQuivalent (R-EEQ) concentration, should be considered. It is also desirable to consider whether conjugates of estrogens or unknown chemicals, which may increase or decrease the estrogenicity, are present.

Acknowledgements

This research was supported by grants from the Yeongsan River Environment Research Center.

References

1. S. D. Kim, J. Cho, I. S. Kim, B. J. Vanderford and S. A. Snyder, Water Res., 2007, 41, 1013.
2. R. A. Rudel, S. J. Melly, P. W. Geno, G. Sun and J. G. Brody, Environ. Sci. Technol., 1998, 32, 861.
3. T. Schilirò, C. Pignata, R. Rovere, E. Fea and G. Gilli, Chemosphere, 2009, 75, 335.
4. L. Quirós, R. Céspedes, S. Lacorte Bruguera, P. Viana, D. Raldúa, D. Barceló and B. Piña, Environ. Toxicol. Chem., 2005, 24, 389.
5. C. G. Campbell, S. E. Borglin, F. B. Green, A. Grayson, E. Wozei and W. T. Stringfellow, Chemosphere, 2006, 65, 1265.
6. C. Bicchi, T. Schiliròa, C. Pignata, E. Fea, C. Cordero, F. Canale and G. Gilli, Sci. Total Environ., 2009, 407, 1842.
7. D. W. Kolpin, E. T. Furlong, M. T. Meyer, S. D. Zaugg, L. B. Barber and H. T. Buxton, Environ. Sci. Technol., 2009, 36, 1202.
8. A. C. Belfroid, A. Van der Horsta, A. D. Vethaak, A. J. Schäfer, G. B. Rijs, J. Wegener and W. P. Cofino, Sci. Total Environ., 1999, 225, 101.
9. J. Gong, Y. Ran, D. Chen, Y. Yang and X. Ma, Environ. Monit. Assess., 2009, 156, 199.
10. G. Pojana, A. Bonfà, F. Busetti, A. Collarin and A. Marcomini, Environ. Toxicol. Chem., 2004, 23, 1874.
11. G. Pojana, A. Critto, C. Micheletti, C. Carlon, F. Busetti and A. Marcomini, Chimia, 2003, 57, 542.
12. E. L. M. Vermeirssen, M. J. F. Suter and P. Burkhardt-Holm, Environ. Toxicol. Chem., 2006, 25, 2413.
13. J. Lee, B. C. Lee, J. S. Ra, J. Cho, I. S. Kim, N. I. Chang, H. K. Kim and S. D. Kim, Chemosphere, 2008, 71, 1582.
14. W. Körner, P. Spengler, U. Bolz, W. Schuller, V. Hanf and J. W. Metzger, Environ. Toxicol. Chem., 2001, 20, 2142.
15. N. Rajapakse, E. Silva, M. Scholze and A. Kortenkamp, Environ. Sci. Technol., 2004, 38, 6343.
16. R. White, S. Jobling, S. A. Hoare, J. P. Sumpter and M. G. Parker, Endocrinology, 1994, 135, 175.
17. OSPAR Commission, ISBN 1-904426-19-0, 2003, pp.7–8.
18. R. Loos, B. M. Gawlik, G. Locoro, E. Rimaviciute, S. Contini and G. Bidoglio, Environ. Pollut., 2009, 157, 561.
19. J. L. Zhao, G. G. Ying, L. Wang, J. F. Yang, X. B. Yang, L. H. Yang and X. Li, Sci. Total Environ., 2009, 407, 962.
20. B. C. Lee, J. Lee, H. Y. Kim, C. N. Duong, J. S. Ra, N. I. Chang, H. K. Kim and S. D. Kim, Korean Society of Environmental Engineers, 2007, 29, 1145.
21. B. Lei, S. Huang, Y. Zhou, D. Wang and Z. Wang, Chemosphere, 2009, 76, 36.
22. M. D. Jürgens, K. I. E. Holthaus, A. C. Johnson, J. L. Smith, M. Hetheridge and R. J. Williams, Environ. Toxicol. Chem., 2002, 21, 480.
23. T. A. Ternes, P. Kreckel and J. Mueller, Sci. Total Environ., 1999, 225, 91.
24. C. Baronti, R. Curini, G. D'ascenzo, A. Di Corcia, A. Gentili and R. Samperi, Environ. Sci. Technol., 2000, 34, 5059.
25. S. M. Oh, S. Y. Choung, Y. Y. Sheen and K. H. Chung, Sci. Total Environ., 2000, 263, 161.
26. Q. Sun, S. Deng, J. Huang, G. Shen and Y. Gang, Environ. Toxicol. Pharmacol., 2008, 25, 20.
27. I. C. Beck, R. Bruhn and J. Gandrass, Chemosphere, 2006, 63, 1870.
28. S. Hashimoto, Y. Ueda, R. Kurihara and F. Shiraishi, Environ. Toxicol. Chem., 2007, 26, 279.

---