Science of the Total Environment 470–471 (2014) 171–179
Contents lists available at ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Potential environmental implications of emerging organic contaminants in Taihu Lake, China: Comparison of two ecotoxicological assessment approaches Zhenhua Yan a, Xiaofan Yang a,b, Guanghua Lu a,⁎, Jianchao Liu a, Zhengxin Xie a, Donghai Wu a a b
Key Laboratory of Integrated Regulation and Resources Development of Shallow Lakes of Ministry of Education, College of Environment, Hohai University, Nanjing 210098, China College of Biological and Chemical Engineering, Anhui Polytechnic University, Wuhu 241000, China
H I G H L I G H T S • • • •
Most water areas were at significant risk for adverse effects based on HQ indexes. Compounds were ranked and hormones were regarded as the greatest hazards. The firstly applied EIBR in field suggested Zhushan Bays as the most stressful place. Both approaches, HQ and EIBR, are mutually consistent, with good positive linear correlation.
a r t i c l e
i n f o
Article history: Received 2 August 2013 Received in revised form 27 September 2013 Accepted 27 September 2013 Available online 15 October 2013 Editor: Damia Barcelo Keywords: Emerging organic contaminant Enhanced integrated biomarker response Hazard quotient Taihu Lake
a b s t r a c t In this study, the hazard quotient (HQ) and a novel enhanced integrated biomarker response (EIBR) were applied to indirectly/directly estimate the ecotoxicological risk of emerging organic contaminants in Taihu Lake. Nine out of sixteen target compounds were detected in most sampling points at comparable concentrations (1.58–206.95 ng/L). Simultaneously, changes in multi-biomarkers were measured in caged fish for 28 days. The 0HQ results preliminarily indicated that most water areas were at significant risk for adverse effects to aquatic organisms (HQ N 10). The prioritisation was then ranked and 17α-ethinylestradiol, diethylstilbestrol and 17β-estradiol were regarded as the greatest hazards. The EIBR, covering multi-biomarkers and their weighting, was applied to field study, and Zhushan Bay was suggested as the most stressful place, followed by Meiliang Bay. The HQ showed significant positive linear correlation with the EIBR (r = 0.848, P b 0.001), suggesting mutual consistency between the two approaches based on laboratory and field study in ecotoxicological risk assessment. © 2013 Elsevier B.V. All rights reserved.
1. Introduction The occurrence and implication of emerging organic contaminants in aquatic environments have recently become a matter of concern. These contaminants from municipal wastewater treatment plants, concentrated livestock farming operations, decentralised on-site treatment systems, manufacturing and hospital effluents, and urban and agricultural runoff enter the environment (Daughton and Ternes, 1999; Fenech et al., 2013; Kolpin et al., 2002; Liu and Wong, 2013; Swartz et al., 2006; Verlicchi et al., 2012) and have been detected in different aquatic environments, such as wastewater, surface water, ground water, and even drinking water throughout the world (Benotti et al., 2009; García-Galán et al., 2011; Stuart et al., 2012; Zhang et al., 2013a). The concentrations of these compounds are generally found in the ng/L–μg/L range in natural waters, and these low concentrations ⁎ Corresponding author. Tel.: +86 258 378 7894; fax: +86 258 378 7330. E-mail address: [email protected] (G. Lu). 0048-9697/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.scitotenv.2013.09.092
are thought to cause detrimental effects on both aquatic biota and human health due to their biological activities. For instance, Purdom et al. (1994) found extremely high plasma vitellogenin (Vtg) concentrations in caged fish living downstream from sewage treatment plants in the UK due to the presence of estrogenic chemicals in the effluents. Kidd et al. (2007) indicated that vanishingly low levels of 17α-ethynylestradiol (5–6 ng/L) in a Canadian lake led to the feminisation of male fathead minnows (Pimephales promelas), ranging from the inappropriate synthesis of Vtg to the presence of intersex individuals and, ultimately, the collapse of the population of the fish species. The mixtures of carbamazepine, fenofibric acid, propranolol, sulfamethoxazole and trimethoprim in the Douro River estuary also caused increases in Vtg, hepatic mass, cytoplasmic eosinophilia and cytochrome P450 1A immunoreactivity in male zebrafish liver (Madureira et al., 2012). The near extinction of Asian vultures following exposure to diclofenac is a key example where exposure to pharmaceuticals caused a population-level impact on non-target wildlife (Cuthbert et al., 2011).
172
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
Given that emerging organic contaminants are referred as “pseudopersistent” contaminants due to their wide production in daily life and continuous release into the aquatic environment (Daughton, 2002), there is an urgent need to conduct a comprehensive risk assessment to better understand and predict the negative consequences of these emerging contaminants. Diverse approaches have been developed to directly/indirectly assess the ecotoxicological risks of these compounds in the environment, such as the water quality index (Pesce and Wunderlin, 2000), the toxic unit concept (López-Doval et al., 2012), the integrative tissue pollutant index (Marigómez et al., 2013), the aquatic macroinvertebrate diversity indexes (Ginebreda et al., 2010) and the fish plasma model (Du et al., 2014; Huggett et al., 2003). The hazard quotient (HQ), a ratio between the predicted or measured environmental concentrations (PEC or MEC, respectively) and their non-observed effect or predicted non-effect concentrations (NOEC or PNEC, respectively) estimated from laboratory study, can provide a general characterisation of the chemical risks of contaminants according to the European Medicines Evaluation Agency (EMEA) guidelines on risk assessment, and ranking compounds of environmental concern for both regulatory and monitoring purposes in many investigations has proven beneficial (Cristale et al., 2013; García-Galán et al., 2011). It is a singlevalue estimate for screening-level risk assessment at early stages based on laboratory toxicity data. However, the HQ does not necessarily reflect the real ecosystem situation because many extra concurrent contaminations can be present, causing interactive or even synergistic effects on biota under certain environmental conditions. To overcome this limitation, a suite of biomarkers that reflect exposure to various types of contaminants and provide reliable indications of effects are increasingly employed as early warning signals in monitoring programmes for the ecotoxicological risk assessment of contaminants. Because contaminants are usually present as complex mixtures, there is no single biomarker that can provide a complete interpretation of the current environment (Cazenave et al., 2009). The integrated biomarker response (IBR) index, an early multivariate response method that summarises multiple biomarker responses into a single value, has been frequently used in environmental risk assessment in various studies (Kim et al., 2010; Li et al., 2011). However, the conventional IBR calculation suffers from two weak points: (1) the IBR result is strongly dependent on the arrangement of the biomarkers on the star plot; and (2) it always overlooks the weighting of each biomarker deviating from the normal healthy state (Liu et al., 2013; Sanchez et al., 2013). To avoid mistakes due to the above weak points, Liu et al. (2013) extended the environmental toxicity of perfluorinated chemicals in mussels (Perna viridis) using an enhanced integrated biomarker response (EIBR) approach and demonstrated that perfluorononanoic acid (PFNA) could be as equally potent as perfluorooctanesulfonate (PFOS) in terms of integrative toxicity because PFNA induced a higher toxic response at the cellular level with higher ecological significance than PFOS. Taihu Lake, the third largest freshwater lake in China, is located in the heart of Yangtze River Delta and has a surface area of 2338 km2 and a mean depth of 1.89 m. It serves as an indispensable water source for drinking water, agriculture and industry for a population of 20 million in several important cities in Eastern China, including Shanghai, Suzhou, and Wuxi. However, with the extraordinary economic development, population growth and urbanisation, increasing amounts of anthropogenic pollutants have been continuously discharged into the whole basin (Liu et al., 2009; Zhang et al., 2012). Taihu Lake is famous for its severe eutrophication, which has deteriorated the water quality and threatened the water supply. Additionally, the occurrence of emerging organic contaminants, such as polybrominated diphenyl ethers and synthetic musks, has been detected in the sediment and native fish from the lake (Yu et al., 2012; Zhang et al., 2013b). Our previous studies detected some pharmaceuticals and endocrine disrupting chemicals in different regions of the lake (Lu et al., 2011, 2013; Yan et al., 2012). However, almost all studies have focused on the biological effects based on the
conventional IBR approach but have overlooked biomarkers' weightings and thus could be biased and misleading. Moreover, previous studies failed to prioritise the health risks of those contaminants. In addition, public concern over the chemical and ecological quality of the lake requires a comprehensive chemical and biological assessment of the ecotoxicological risk in water bodies. Therefore, the goals of this study are to provide a more comprehensive method to assess the potential ecotoxicological risk of 16 emerging organic contaminants in Taihu Lake and to help distinguish the potential hazard factors. HQ was employed as an indicator to provide a general identification and indirect estimation of the chemical status in the lake based on laboratory study, and the EIBR approach was applied in the field study as an early warning signal to directly assess the ecological effects. Moreover, relationships between HQ and EIBR based on the laboratory and field studies were also discussed in different sampling points of Taihu Lake. 2. Materials and methods 2.1. Sampling points and water sampling According to the characteristics of the lake, the locations of eight sampling points in the northern section of Taihu Lake are illustrated in Fig. 1. These points were located in Gonghu Bay (P1, P2 and P3), Meiliang Bay (P4 and P5), Zhushan Bay (P6 and P7) and Lake Centre (P8). Water samples were collected twice from each sampling point during November 21–December 17, 2011 using a stainless steel bucket from 50cm below the water's surface and were immediately transferred to an amber glass bottle that was pre-rinsed with methanol. All sampling equipment was rinsed with the sample before sampling. The water samples were immediately transferred to the laboratory in an ice-packed container and stored at 4 °C for a maximum of 6 h before extraction. A total of 16 emerging organic contaminants were determined in this study based on the information found in the literature on their occurrence in surface water as well as their high usage and consumption in China (Luo et al., 2010; Zhao et al., 2009). These contaminants comprised five hormones (17α-ethinylestradiol, 17β-estradiol, diethylstilbestrol, estrone and estriol), four antibiotics (sulfamethoxazole, norfloxacin, ofloxacin and chloramphenicol), two anti-inflammatory analgesics (diclofenac and ibuprofen), two β-blockers (atenolol and propranolol), one antibacterial (triclosan), one stimulant (caffeine) and one plasticiser (bisphenol A). The purity and source of all analytical compounds are listed in the Supplementary material (Table S1). 2.2. Sample extraction and analysis After filtering through 0.45 μm glass fibre filters, the water samples (500 mL) were spiked with 100 ng of each surrogate standard and were passed through Oasis HLB solid phase extraction cartridges (200 mg, 6 mL, Waters, Milford, USA) that were pre-conditioned with 6 mL of methanol and 6 mL of ultrapure water. After further rinsing with 10 mL ultrapure water, the cartridges were dried under vacuum for at least 30 min and eluted twice with 5 mL of methanol. The internal standards (100 ng each) were added into the extracts before final evaporation to 1 mL under a gentle nitrogen stream. The final extract was stored in a 2mL amber glass vial at −20°C for further chemical analysis. Identification and quantification of the 16 target compounds were performed with a 1260 UHPLC instrument equipped with a 6460QQQ triple quadrupole mass spectrometer with electrospray ionisation (Agilent Technologies, Palo Alto, USA) operating in the multiple reactions monitoring (MRM) mode (Yan et al., 2012; Lu et al., 2013). Sulfamethoxazole, norfloxacin, ofloxacin, atenolol, propranolol and caffeine were measured in the positive ion mode, and the remaining 10 compounds were measured in the negative ion mode. The recovery, relative standard deviation and method detection limits for the different investigated compounds are summarised in the Supplementary material
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
173
Fig. 1. Location of the sampling points in Taihu Lake.
(Table S1). Further details of the chemical analyses are provided in the Supplementary material.
An in situ biomonitoring with mature male goldfish was conducted at each sampling point in November 2011 to provide actual biological status information in Taihu Lake, as described in our previous study (Wang et al., 2011). Briefly, healthy fish were acclimatised for at least two weeks in the laboratory and were transferred to each sampling point in Taihu Lake. The fish were caged in non-toxic columnar polyethylene mesh cages (diameter 60 cm and height 1.50 m) that were fixed by three wooden pegs and completely suspended in the water approximately 20 cm below the water surface. Control fish (C) were kept in four 60 L aquariums containing 30 L dechlorinated municipal water under constant aeration in our laboratory and were fed daily. Four fish were sampled at each point following exposure for 7 d, 14 d, 21 d and 28 d after which they were immediately transported to the laboratory. After anaesthetisation with 100 mg/L MS-222, blood, liver and gonad samples were collected from the fish and were treated and stored for biomarker assays, as described in detail in the Supplementary material.
concentration was quantified using an ELISA kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) following the manufacturer's instructions. Gonadal DNA damage was measured by the comet assay and was calculated by tail length × DNA % in the tail/100, as previously described by Roy et al. (2003). Ethoxyresorufin-O-deethylase (EROD) activity was quantified using a microplate reader at 572 nm (Molecular Device VersaMax, Sunnyvale, USA) following the description of Lu et al. (2009). Glutathione S-transferase (GST) activity was determined using 1-chloro-2,4-dinitrobenzene as a substrate at 340 nm (Frasco and Guilhermino, 2002). Glutathione peroxidase (GPx) activity was quantified at 340 nm by monitoring the oxidation of NADPH to NADP (Berntssen et al., 2003). Superoxide dismutase (SOD) activity was measured by the inhibition of the auto-oxidation of pyrogallol as described by Marklund and Marklund (1974). Catalase (CAT) activity was determined by the method of ammonium molybdate (Góth, 1991). The VTG values and enzymatic activities were all normalised to the total protein of each sample to control the potential differences between treatments and replicates. Protein concentrations were determined using the method developed by Bradford (1976) with bovine serum albumin as the standard. Further details of the biomarker assays are provided in the Supplementary material.
2.4. Biomarker assays
2.5. Environmental implications
Serum Vtg concentrations were determined by a competitive enzyme-linked immunosorbent assay (ELISA) with purified goldfish Vtg as the standard (Rempel et al., 2006). The serum 17β-estradiol
2.5.1. Hazard quotient indexes The HQ model was developed in an effort to quantify the actual potential ecotoxicological risk of specific species exposures to certain
2.3. In situ biomonitoring
174
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
contaminants in the surrounding natural environment. In the present study, the hazard quotient hqij of compound i in sampling point j was estimated as follows: hqij ¼
MECij PNECi
ð1Þ
where MECij and PNECi are the “measured environmental concentrations” and “predicted non-effect concentrations”, respectively. The PNEC values used in this study were obtained from the literature according to the EMEA guidelines (European Commission, 2003). If more than one value was available, then the lowest value was used. Meanwhile, the total hazard quotient for each point j was calculated as the sum of hqij for the mixture components, assuming that compounds acted additively at environmentally relevant concentrations (Zhang et al., 2010): HQj ¼
X
ð2Þ
hqij:
i
A commonly used risk screening benchmark for ecotoxicological effects from contaminated water was then applied: HQ b 1.0, no significant risk; 1.0 ≤ HQ b 10, a small potential risk for adverse effects; 10 ≤ HQ b 100, significant potential for adverse effects; and HQ ≥ 100, potential adverse effects should be expected (Cristale et al., 2013; Sánchez-Avila et al., 2012). The relative contribution weight of the different compounds to the total hazard quotient at each sampling point was then estimated using the following expression: hqijð%Þ ¼
hqij 100: HQi
ð3Þ
2.5.2. Enhanced integrated biomarker response values To address the biomarkers as a whole, an enhanced integrated biomarker response (EIBR), developed by Beliaeff and Burgeot (2002) and later modified by Liu et al. (2013), was calculated to evaluate the toxicity at different sampling points. Briefly, the biomarker response data for each point were standardised according to the Eq. (4): Yi ¼ ðXi−mÞ=s
ð4Þ
where Yi is the standardised value of the biomarker, Xi is the mean value of a biomarker at each point, and m and s are the mean value and the standard deviation of a biomarker considering all sampling points, respectively. Using standardised data, Zi was then calculated as Zi = Yi or Zi = –Yi in the case of a biomarker responding to contamination by induction or inhibition, respectively, and the minimum value (mini) for each biomarker at all points was calculated from the standardised response value. The score of each biomarker response (Si) was calculated as: Si ¼ Zi þ j minij:
ð5Þ
Finally, to obtain an integrated multi-biomarker response, the EIBR value was calculated as the sum of the weighting of the biomarker and the biomarker score as follows: EIBR ¼
n X i¼1
n X
Si Wi=
Wi
ð6Þ
i¼1
where Wi is the weighting of each biomarker i, and molecular (Vtg, EROD, GST, GPx, SOD and CAT) and cellular (17β-estradiol and DNA damage) biomarkers are weighted as 1 and 2, respectively, because it is assumed that an alteration at the cellular level would have a greater impact on the health of the organisms than changes at the molecular level (Liu et al., 2013).
2.6. Statistical analysis The results for each biomarker are presented as the mean±standard deviation. All data were tested for normality and equal variance using Shapiro–Wilk's and Levene's tests, respectively. A one-way analysis of variance (ANOVA) was used to analyse the differences between treatments within each experiment, and Dunnett's t test was used to identify significantly different treatments with significance set at Pb0.05. All statistical analyses were performed with SPSS 13.0. 3. Results and discussion 3.1. Concentrations of emerging organic contaminants A summary of the concentrations of emerging organic contaminants is shown in Table 1. Seven out of sixteen target compounds (17β-estradiol, diethylstilbestrol bisphenol A, sulfamethoxazole, caffeine, norfloxacin and ofloxacin) were detected at all sampling points, estrone and 17αethinylestradiol were detected at 75% of sampling points, and the remaining compounds were not detected at any points in the lake. Overall, bisphenol A, sulfamethoxazole, caffeine and norfloxacin were more abundant than the other compounds in this region, with average concentrations of 49.8, 43.2, 47.0 and 31.9 ng/L, respectively. The high concentrations of sulfamethoxazole and norfloxacin in water might be ascribed to their low removal in conventional wastewater treatment plants (WWTPs) (Lin et al., 2009; Watkinson et al., 2007). Although the elimination efficiencies of caffeine and bisphenol A in conventional WWTPs are more than 90% (Nakada et al., 2006; Sui et al., 2010), their massive consumption and usage in daily life could result in their high abundance in waters (Carballa et al., 2005; Sim et al., 2010). 3.2. Biological effects No mortality or deformities occurred in fish at any points during the periods of in situ exposure. The changes in serum Vtg, 17β-estradiol, gonadal DNA damage, liver EROD, GST, GPx, SOD and CAT activities are shown in Table 2. In the controls, these biomarkers did not significantly change during the experimental periods, and no Vtg expression was observed. Compared with the controls, serum Vtg in the samples increased significantly in a time dependent manner at all points, as did the serum 17β-estradiol levels and gonadal DNA damage in most cases. In previous studies, these biomarkers were mainly used to indicate the health status of endocrine and reproductive function and to estimate the presence of estrogens in waters (Lu et al., 2010). The enhanced serum Vtg and 17β-estradiol levels and DNA damage in the present study could be attributed to the prevalence of estrogens in these regions, especially in Meiliang Bay and Zhushan Bay. In addition, the presence of caffeine
Table 1 Individual and total concentrations of nine detected emerging organic contaminants at different sampling points (ng/L). Compounds
Sampling points P1
P2
Estrone 3.37 2.96 17β-Estradiol 7.21 3.01 17α-Ethinylestradiol 2.28 1.64 Diethylstilbestrol 34.6 9.05 Bisphenol A 39.7 44.9 Sulfamethoxazole 65.2 85.4 Caffeine 20.4 24.4 Norfloxacin 39.4 35.0 Ofloxacin 13.1 11.7 Total 225 218 a
P3
P4
P5
P6
3.06 1.58 3.61 2.76 3.51 8.03 17.28 9.51 a 1.75 1.81 4.00 n.d. 3.16 15.4 15.0 16.9 19.1 207 35.6 14.8 75.1 34.7 28.0 21.0 23.3 39.5 89.8 78.1 50.4 24.8 24.5 25.2 10.5 9.74 9.61 13.0 188 343 225 185
n.d. means below the method detection limit.
P7
P8
n.d.a 8.00 3.00 6.41 6.63 20.7 52.4 33.1 11.1 141
n.d.a 4.36 n.d. 3.56 30.8 15.6 48.3 22.7 10.2 136
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
175
Table 2 Multi-biomarker responses in fish at eight sampling points after exposure for 7 d (A), 14 d (B), 21 d (C) and 28 d (D). Asterisks indicate values that are significantly higher than the controls (P b 0.05).a Points
Vtg
E2
TM
EROD
GST
GPx
SOD
CAT
A C P1 P2 P3 P4 P5 P6 P7 P8
bLOD b 87.8 ± 11.9 83.4 ± 7.60 90.0 ± 4.17 103 ± 7.89 101 ± 4.11 97.6 ± 8.70 84.2 ± 5.40 51.0 ± 1.31
309 ± 11.3 351 ± 12.3 322 ± 8.43 324 ± 15.4 383 ± 12.3* 387 ± 14.1* 391 ± 12.2* 322 ± 15.9 297 ± 18.9
0.573 ± 0.142 1.33 ± 0.464 0.919 ± 0.281 1.00 ± 0.343 3.12 ± 0.580* 2.06 ± 0.905* 2.27 ± 0.778* 0.904 ± 0.230 0.756 ± 0.0879
12.0 ± 1.90 15.5 ± 2.67 12.5 ± 0.691 11.7 ± 0.754 11.9 ± 2.20 12.5 ± 2.69 15.2 ± 1.96 15.2 ± 2.06 12.0 ± 2.73
35.5 ± 7.27 46.8 ± 8.42 47.3 ± 9.95 43.8 ± 9.91 43.6 ± 5.31 45.9 ± 11.3 51.0 ± 7.48* 55.2 ± 3.39* 36.5 ± 4.56
71.2 ± 15.1 50.8 ± 9.48* 56.0 ± 6.21 61.2 ± 7.46 60.1 ± 10.9 50.1 ± 4.25* 47.2 ± 9.05* 47.1 ± 5.45* 64.1 ± 9.51
6.79 ± 1.18 10.09 ± 2.01 10.6 ± 3.07* 10.5 ± 1.97* 9.33 ± 0.401 11.9 ± 2.36* 11.9 ± 1.84* 12.3 ± 1.12* 7.74 ± 2.46
438 ± 56.9 326 ± 67.8* 338 ± 67.1 354 ± 56.1 286 ± 76.0* 361 ± 30.8 309 ± 43.9 248 ± 21.6* 402 ± 84.7
B C P1 P2 P3 P4 P5 P6 P7 P8
bLOD b 119 ± 12.7 121 ± 9.04 122 ± 2.92 150 ± 10.9 135 ± 10.7 135 ± 8.56 94.5 ± 4.93 50.1 ± 5.07
308 ± 15.3 424 ± 11.8* 363 ± 11.2 361 ± 13.3 430 ± 12.9* 411 ± 9.78* 408 ± 12.0* 398 ± 15.0* 333 ± 6.06
0.669 ± 0.167 1.92 ± 0.833* 1.91 ± 1.28* 1.85 ± 0.256* 3.63 ± 1.19* 3.79 ± 1.67* 4.10 ± 1.22* 1.69 ± 0.387* 1.12 ± 0.411*
11.5 ± 1.52 15.7 ± 4.86 13.7 ± 2.21 12.7 ± 2.77 12.6 ± 1.08 13.5 ± 2.45 16.9 ± 3.92* 16.6 ± 2.73* 12.1 ± 2.66
36.5 ± 5.26 51.0 ± 7.93* 50.1 ± 4.58* 45.0 ± 10.0 45.1 ± 4.73 47.2 ± 10.1 56.2 ± 3.38* 57.1 ± 8.69* 37.8 ± 5.71
73.1 ± 12.9 49.7 ± 5.61* 56.8 ± 9.73* 60.3 ± 4.92 59.4 ± 10.2 50.4 ± 10.5* 42.3 ± 4.56* 45.0 ± 5.65* 64.3 ± 7.95
7.15 ± 1.59 10.5 ± 1.80* 11.1 ± 1.45* 11.1 ± 1.02* 9.51 ± 1.03 12.5 ± 2.61* 12.4 ± 2.21* 13.0 ± 1.07* 7.81 ± 2.00
433 ± 47.5 354 ± 56.5* 353 ± 48.2* 358 ± 9.61* 297 ± 19.7* 354 ± 61.5* 313 ± 12.4* 246 ± 40.8* 407 ± 43.0
C C P1 P2 P3 P4 P5 P6 P7 P8
bLOD b 141 ± 3.17 152 ± 8.61 140 ± 9.07 173 ± 5.92 204 ± 8.23 199 ± 10.6 105 ± 8.82 62.2 ± 8.21
313 ± 3.22 510 ± 16.2* 414 ± 29.7* 438 ± 8.66* 549 ± 5.84* 502 ± 12.3* 505 ± 11.0* 411 ± 5.00* 407 ± 9.70*
0.642 ± 0.189* 2.17 ± 0.0909* 2.06 ± 0.927* 1.97 ± 0.125* 4.17 ± 0.622* 3.26 ± 0.591* 3.43 ± 0.156* 1.46 ± 0.204* 1.04 ± 0.28*
12.4 ± 2.29 15.9 ± 3.88 14.3 ± 1.25 12.7 ± 2.79 12.1 ± 1.67 13.9 ± 3.76 17.1 ± 1.83* 16.8 ± 4.10* 12.0 ± 2.17
37.6 ± 5.10 51.2 ± 6.89 51.8 ± 6.96 45.8 ± 8.68 44.9 ± 17.4 48.3 ± 7.53 57.0 ± 7.34* 60.5 ± 9.07* 40.1 ± 6.48
72.6 ± 5.80 48.2 ± 2.75* 52.7 ± 10.3* 58.3 ± 11.7* 56.8 ± 5.49* 48.8 ± 6.06* 41.7 ± 5.84* 46.0 ± 13.2* 65.0 ± 2.27
6.67 ± 1.38 11.1 ± 1.83* 11.6 ± 1.16* 11.0 ± 0.978* 9.85 ± 1.84 12.9 ± 0.78* 13.4 ± 2.52* 14.3 ± 3.13* 8.49 ± 2.00
465 ± 47.0 363 ± 86.7* 344 ± 90.5* 346 ± 50.0* 305 ± 52.9* 339 ± 55.9* 322 ± 22.7* 240 ± 27.8* 384 ± 31.9
D C P1 P2 P3 P4 P5 P6 P7 P8
bLOD b 158 ± 15.8 151 ± 8.06 142 ± 8.07 191 ± 4.73 209 ± 2.68 204 ± 8.94 106 ± 5.79 69.1 ± 5.19
315 ± 5.73 506 ± 14.5* 423 ± 14.7* 435 ± 10.6* 552 ± 10.4* 530 ± 7.55* 520 ± 9.97* 346 ± 15.3 367 ± 15.8
0.611 ± 0.178 2.11 ± 0.733* 2.16 ± 1.17* 2.09 ± 0.984* 5.20 ± 0.745* 3.76 ± 1.18* 4.06 ± 0.368* 1.70 ± 0.219* 1.49 ± 0.275*
11.9 ± 3.08 15.3 ± 2.63 13.6 ± 1.00 12.2 ± 0.668 12.3 ± 3.10 13.5 ± 2.21 16.7 ± 2.74* 16.3 ± 3.70 12.1 ± 3.05
37.7 ± 5.83 48.3 ± 3.18 48.3 ± 4.66 45.6 ± 3.28 45.3 ± 8.09 46.1 ± 4.92 53.1 ± 11.1* 55.6 ± 4.74* 38.2 ± 5.65
70.1 ± 11.1 45.3 ± 11.0* 51.6 ± 7.91* 57.9 ± 1.37 54.1 ± 11.3* 47.1 ± 3.00* 42.1 ± 7.67* 43.5 ± 3.70* 64.9 ± 9.85
7.03 ± 1.33 10.4 ± 0.98* 10.8 ± 1.22* 10.3 ± 2.21* 9.39 ± 1.38 12.7 ± 2.08* 12.5 ± 2.39* 14.2 ± 1.18* 8.09 ± 0.987
465 ± 25.3 331 ± 80.5* 369 ± 38.6* 345 ± 72.8* 280 ± 21.9* 345 ± 66.7* 312 ± 65.6* 258 ± 66.0* 423 ± 38.1
a Vtg = Serum vitellogenin concentration (ng/mg); E2 = Serum 17β-estradiol concentration (pg/mL); TM = Comet tail moment; EROD = Ethoxyresorufin-O-deethylase activity (pmol/mg/min); GST = Glutathione S-transferase activity (nmol/mg/min); GPx = Glutathione peroxidase activity (nmol/mg/min); SOD = Superoxide dismutase activity (U/mg); CAT = catalyse activity (μmol/mg/min). b bLOD: below the detection limit.
might also induce estrogenic activity in males, e.g., Vtg production (Li et al., 2012). Significant EROD induction and GST induction were only observed at points 6 and 7 during most exposure periods, and obvious increases in GST activity were also observed at 7 d at P1–2. To date, the biotransformation of emerging contaminants, particularly pharmaceutical residues, in fish and other aquatic organisms has not been well understood (Connors et al., 2013; Gomez et al., 2010; Smith et al., 2010). The markedly induced EROD and GST activities in Zhushan Bay might suggest the presence of high levels of other organic contaminations in this region. Antioxidant enzymes in sampled fish showed spatial variability, with distinct GPx and CAT inhibition at P1–7 during most exposure periods and significant activation of SOD at most study points, with the exception of P4 and P8, compared to the controls. The significant increases in the SOD activities suggest an increased production of superoxide anions and H2O2 in fish. In general, these powerful and potentially harmful oxidising agents are metabolised by CAT and GPx into harmless agents, including H2O and O2 (Box et al., 2007). However, in the current study, the GPx and CAT activities were significantly inhibited in most cases, which suggested an incomplete defence chain against reactive oxygen species that could result in oxidative damage in organisms (Huang et al., 2007).
3.3. Environmental implications 3.3.1. Hazard quotient indexes Chemical analysis of the emerging organic contaminants in the aquatic environment is of limited use unless these data are related to their potential ecological effects via a risk assessment approach. To estimate the potential adverse effects of contaminants detected in this monitoring, an ecotoxicological risk evaluation was performed on non-target organisms. Toxicologically relevant concentrations for each detected compound used in the HQ calculation are exhibited in the Supplementary material (Table S2). Among the contaminants, no significant risk (HQ b 1) was observed for estrone, bisphenol A, norfloxacin or ofloxacin for most points, with average indexes varying from 0.307 to 0.984. For 17β-estradiol, sulfamethoxazole and caffeine, high HQ indexes, ranging from 1.60 to 3.81, indicated their low significant risk. Additionally, the presence of 17α-ethinylestradiol and diethylstilbestrol showed significant potential for adverse effects with the highest HQ indexes (18.1 and 13.0, respectively). These results suggest that the HQ index of each compound is mainly based on its toxicological potency and detected concentration in the water body. A sum of the HQ of each detected compound was performed for each point and is shown in Fig. 2. This provided an intuitive view of the total
176
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
Fig. 2. The total hazard quotient at different sampling points in Taihu Lake.
Sampling points
ecotoxicological risk at all points. The total HQ indexes in whole regions ranged from 9.75 to 67.37, whereas P6 showed the highest HQ index relative to the other points and presented a significant risk. In addition to P6, significant potential for adverse effects was also observed at points 1–5 and 7. The lowest HQ index of 9.75 was observed at P8, suggesting low significant risk for adverse effects in this region. In summary, points near the inlet of the inflowing rivers from the cities of Suzhou, Wuxi, Changzhou and Yixing have much higher risks than those in the centre of the lake. Previous studies have demonstrated that pollution of organic pollutants at lakeside and at the inlet of inflow rivers was more serious compared with the other points and showed a relatively high toxicity due to the human activities along the lakes and rivers (Zhang et al., 2012; Cristale et al., 2013). Moreover, the water diversion from the Yangtze River through the Wangyu River may also bring contaminants into the lake (Wang et al., 2011). Because hormones presented HQ indexes approximately 10 to 100 times higher than the other detected compounds, it is evident that 17α-ethinylestradiol, diethylstilbestrol and 17β-estradiol are the greatest hazards, and the potential for adverse effects in the lake is attributed almost exclusively to them (see Fig. 3). Additionally, the presence of sulfamethoxazole and caffeine accounted for 6.13% and 6.15% of the estimated total risk on average. In particular, at P3 and P8, the contribution ratio of sulfamethoxazole and caffeine to the total HQ reached 25.8% and 18.4%, respectively. Hence, to protect the ecological health of Taihu Lake, the prioritisation of the key control objectives of the contaminants should be chosen as 17α-ethinylestradiol, diethylstilbestrol, 17β-estradiol, sulfamethoxazole and caffeine. These results accord with the previous ranking investigation of Murray et al. (2010), who suggested 17α-ethinylestradiol and 17β-estradiol as the highest priority pollutants for regulation of pharmaceuticals in surface water. Sulfamethoxazole was also highlighted as a priority pharmaceutical in
P8
Estrone
P7
17β-Estradiol
P6
17α-Ethinylestradiol
P5
Diethylstilbestrol Bisphenol A
P4
Sulfamethoxazole
P3
Caffeine
P2
Norfloxacin
P1 0%
Ofloxacin
20%
40%
60%
80%
100%
Contribution ratio Fig. 3. Percent contribution of different detected compounds to the total hazard quotient indexes.
the water environment of China based on pharmaceutical consumption, removal and potential ecological effects by Sui et al. (2012). However, it should be noted that the current results from a singleseason study may not reflect the true hazards of the lake. The present study was conducted in the winter season, which means that seasonrelated consumption patterns and environmental conditions could lead to differences when comparing to other seasons. Lacey et al. (2012) found that the low water temperature (b10°C) during the winter season could dramatically decrease the biodegradation efficiencies of biodegradable contaminants in WWTPs and result in high concentrations in effluents. In addition, the weakened natural degradation processes (e.g., photodegradation) and increased human consumption of pharmaceuticals in colder periods owing to low temperature should be taken into account for the maximum levels of pharmaceuticals observed in the Llobregat River during this period (Osorio et al., 2012). All of these influencing factors could result in higher ecotoxicological risks in the lake environment in the winter season. 3.3.2. Enhanced integrated biomarker response values Although the HQ approach can give a preliminary determination of the possible effects on aquatic organisms, the detection of only a few classes of contaminants may not be sufficient to estimate the real effects of a complex mixture at actual environmental concentrations for an accurate ecotoxicological risk assessment. Moreover, the possible synergic interactions between contaminations with different modes of action and the bioaccumulation and bio-magnification in organisms may also lead to an underestimation of the risk in water. Instead, the application of multiple biomarkers could avoid the above defects and improve the differentiation between healthy and stressed organisms (Brooks et al., 2011), despite the fact that it is difficult to be understood by nonspecialists. The integrated biomarker response (IBR), which scores and summarises the responses of a set of biomarkers in a single value, provides an intuitive interpretation of the relative toxicity of point contaminations to the public. This approach has been widely used in many field and laboratory studies to assess the variation of environmental quality (Kim et al., 2010; Li et al., 2011). However, the conventional IBR system is being challenged when the employed biomarkers cover multibiological organisation levels. In this study, an enhanced integrated biomarker response (EIBR) that considered the significance of each biomarker at different levels was applied. Because the impacts of bioendpoint alterations on the health status of organisms normally increased through molecular, cellular, physiological and individual avenues, the weighting of each biomarker was appointed according to this hierarchy as 1 and 2 (Liu et al., 2013). A similar weighting has been applied in the pollution monitoring programme under the Europe Water Framework Directive (Hagger et al., 2009). Without weighting, the IBR results could be biased and even misleading. The EIBR values were calculated by integrating the eight biomarkers and their weighting, and the results are represented as a star plot (see Fig. 4). In general, the EIBR values indicated a large range of spatial variation of ecological effects at different points. Relatively high EIBR values were observed at sampling points 4–7, located in the Zhushan and Meiliang Bays, as well as P1, situated at the inlet of the Wangyu River, with the highest value at P6. P2 and P3, located in the Gong Bay presented low EIBR values during the periods of in situ exposure, and P8, situated in the centre of the lake showed the lowest EIBR value, which was only 0.21-fold that at P6. Given that the EIBR is a visualised indicator of ecotoxicological risk, the current results suggest that Zhushan and Meiliang Bays are the most stressful places for organisms, especially at P6 near the end of Zhushan Bay. In addition to anthropogenic influences, the almost enclosed areas of these two bays could increase the difficulty of the exchange of water with Gong Bay and the centre of the lake, especially at the end of the bays, which may further result in the accumulation of contaminants in these areas. For P1, the re-suspension processes of
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
P1
14d
P2
2
21d 28d
1 P7
P3
0
P6
P4
2
HQ indexes
P8
3
7d
3
177
y= 0.0245x+0.7511 r = 0.848 P
Contents lists available at ScienceDirect
Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv
Potential environmental implications of emerging organic contaminants in Taihu Lake, China: Comparison of two ecotoxicological assessment approaches Zhenhua Yan a, Xiaofan Yang a,b, Guanghua Lu a,⁎, Jianchao Liu a, Zhengxin Xie a, Donghai Wu a a b
Key Laboratory of Integrated Regulation and Resources Development of Shallow Lakes of Ministry of Education, College of Environment, Hohai University, Nanjing 210098, China College of Biological and Chemical Engineering, Anhui Polytechnic University, Wuhu 241000, China
H I G H L I G H T S • • • •
Most water areas were at significant risk for adverse effects based on HQ indexes. Compounds were ranked and hormones were regarded as the greatest hazards. The firstly applied EIBR in field suggested Zhushan Bays as the most stressful place. Both approaches, HQ and EIBR, are mutually consistent, with good positive linear correlation.
a r t i c l e
i n f o
Article history: Received 2 August 2013 Received in revised form 27 September 2013 Accepted 27 September 2013 Available online 15 October 2013 Editor: Damia Barcelo Keywords: Emerging organic contaminant Enhanced integrated biomarker response Hazard quotient Taihu Lake
a b s t r a c t In this study, the hazard quotient (HQ) and a novel enhanced integrated biomarker response (EIBR) were applied to indirectly/directly estimate the ecotoxicological risk of emerging organic contaminants in Taihu Lake. Nine out of sixteen target compounds were detected in most sampling points at comparable concentrations (1.58–206.95 ng/L). Simultaneously, changes in multi-biomarkers were measured in caged fish for 28 days. The 0HQ results preliminarily indicated that most water areas were at significant risk for adverse effects to aquatic organisms (HQ N 10). The prioritisation was then ranked and 17α-ethinylestradiol, diethylstilbestrol and 17β-estradiol were regarded as the greatest hazards. The EIBR, covering multi-biomarkers and their weighting, was applied to field study, and Zhushan Bay was suggested as the most stressful place, followed by Meiliang Bay. The HQ showed significant positive linear correlation with the EIBR (r = 0.848, P b 0.001), suggesting mutual consistency between the two approaches based on laboratory and field study in ecotoxicological risk assessment. © 2013 Elsevier B.V. All rights reserved.
1. Introduction The occurrence and implication of emerging organic contaminants in aquatic environments have recently become a matter of concern. These contaminants from municipal wastewater treatment plants, concentrated livestock farming operations, decentralised on-site treatment systems, manufacturing and hospital effluents, and urban and agricultural runoff enter the environment (Daughton and Ternes, 1999; Fenech et al., 2013; Kolpin et al., 2002; Liu and Wong, 2013; Swartz et al., 2006; Verlicchi et al., 2012) and have been detected in different aquatic environments, such as wastewater, surface water, ground water, and even drinking water throughout the world (Benotti et al., 2009; García-Galán et al., 2011; Stuart et al., 2012; Zhang et al., 2013a). The concentrations of these compounds are generally found in the ng/L–μg/L range in natural waters, and these low concentrations ⁎ Corresponding author. Tel.: +86 258 378 7894; fax: +86 258 378 7330. E-mail address: [email protected] (G. Lu). 0048-9697/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.scitotenv.2013.09.092
are thought to cause detrimental effects on both aquatic biota and human health due to their biological activities. For instance, Purdom et al. (1994) found extremely high plasma vitellogenin (Vtg) concentrations in caged fish living downstream from sewage treatment plants in the UK due to the presence of estrogenic chemicals in the effluents. Kidd et al. (2007) indicated that vanishingly low levels of 17α-ethynylestradiol (5–6 ng/L) in a Canadian lake led to the feminisation of male fathead minnows (Pimephales promelas), ranging from the inappropriate synthesis of Vtg to the presence of intersex individuals and, ultimately, the collapse of the population of the fish species. The mixtures of carbamazepine, fenofibric acid, propranolol, sulfamethoxazole and trimethoprim in the Douro River estuary also caused increases in Vtg, hepatic mass, cytoplasmic eosinophilia and cytochrome P450 1A immunoreactivity in male zebrafish liver (Madureira et al., 2012). The near extinction of Asian vultures following exposure to diclofenac is a key example where exposure to pharmaceuticals caused a population-level impact on non-target wildlife (Cuthbert et al., 2011).
172
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
Given that emerging organic contaminants are referred as “pseudopersistent” contaminants due to their wide production in daily life and continuous release into the aquatic environment (Daughton, 2002), there is an urgent need to conduct a comprehensive risk assessment to better understand and predict the negative consequences of these emerging contaminants. Diverse approaches have been developed to directly/indirectly assess the ecotoxicological risks of these compounds in the environment, such as the water quality index (Pesce and Wunderlin, 2000), the toxic unit concept (López-Doval et al., 2012), the integrative tissue pollutant index (Marigómez et al., 2013), the aquatic macroinvertebrate diversity indexes (Ginebreda et al., 2010) and the fish plasma model (Du et al., 2014; Huggett et al., 2003). The hazard quotient (HQ), a ratio between the predicted or measured environmental concentrations (PEC or MEC, respectively) and their non-observed effect or predicted non-effect concentrations (NOEC or PNEC, respectively) estimated from laboratory study, can provide a general characterisation of the chemical risks of contaminants according to the European Medicines Evaluation Agency (EMEA) guidelines on risk assessment, and ranking compounds of environmental concern for both regulatory and monitoring purposes in many investigations has proven beneficial (Cristale et al., 2013; García-Galán et al., 2011). It is a singlevalue estimate for screening-level risk assessment at early stages based on laboratory toxicity data. However, the HQ does not necessarily reflect the real ecosystem situation because many extra concurrent contaminations can be present, causing interactive or even synergistic effects on biota under certain environmental conditions. To overcome this limitation, a suite of biomarkers that reflect exposure to various types of contaminants and provide reliable indications of effects are increasingly employed as early warning signals in monitoring programmes for the ecotoxicological risk assessment of contaminants. Because contaminants are usually present as complex mixtures, there is no single biomarker that can provide a complete interpretation of the current environment (Cazenave et al., 2009). The integrated biomarker response (IBR) index, an early multivariate response method that summarises multiple biomarker responses into a single value, has been frequently used in environmental risk assessment in various studies (Kim et al., 2010; Li et al., 2011). However, the conventional IBR calculation suffers from two weak points: (1) the IBR result is strongly dependent on the arrangement of the biomarkers on the star plot; and (2) it always overlooks the weighting of each biomarker deviating from the normal healthy state (Liu et al., 2013; Sanchez et al., 2013). To avoid mistakes due to the above weak points, Liu et al. (2013) extended the environmental toxicity of perfluorinated chemicals in mussels (Perna viridis) using an enhanced integrated biomarker response (EIBR) approach and demonstrated that perfluorononanoic acid (PFNA) could be as equally potent as perfluorooctanesulfonate (PFOS) in terms of integrative toxicity because PFNA induced a higher toxic response at the cellular level with higher ecological significance than PFOS. Taihu Lake, the third largest freshwater lake in China, is located in the heart of Yangtze River Delta and has a surface area of 2338 km2 and a mean depth of 1.89 m. It serves as an indispensable water source for drinking water, agriculture and industry for a population of 20 million in several important cities in Eastern China, including Shanghai, Suzhou, and Wuxi. However, with the extraordinary economic development, population growth and urbanisation, increasing amounts of anthropogenic pollutants have been continuously discharged into the whole basin (Liu et al., 2009; Zhang et al., 2012). Taihu Lake is famous for its severe eutrophication, which has deteriorated the water quality and threatened the water supply. Additionally, the occurrence of emerging organic contaminants, such as polybrominated diphenyl ethers and synthetic musks, has been detected in the sediment and native fish from the lake (Yu et al., 2012; Zhang et al., 2013b). Our previous studies detected some pharmaceuticals and endocrine disrupting chemicals in different regions of the lake (Lu et al., 2011, 2013; Yan et al., 2012). However, almost all studies have focused on the biological effects based on the
conventional IBR approach but have overlooked biomarkers' weightings and thus could be biased and misleading. Moreover, previous studies failed to prioritise the health risks of those contaminants. In addition, public concern over the chemical and ecological quality of the lake requires a comprehensive chemical and biological assessment of the ecotoxicological risk in water bodies. Therefore, the goals of this study are to provide a more comprehensive method to assess the potential ecotoxicological risk of 16 emerging organic contaminants in Taihu Lake and to help distinguish the potential hazard factors. HQ was employed as an indicator to provide a general identification and indirect estimation of the chemical status in the lake based on laboratory study, and the EIBR approach was applied in the field study as an early warning signal to directly assess the ecological effects. Moreover, relationships between HQ and EIBR based on the laboratory and field studies were also discussed in different sampling points of Taihu Lake. 2. Materials and methods 2.1. Sampling points and water sampling According to the characteristics of the lake, the locations of eight sampling points in the northern section of Taihu Lake are illustrated in Fig. 1. These points were located in Gonghu Bay (P1, P2 and P3), Meiliang Bay (P4 and P5), Zhushan Bay (P6 and P7) and Lake Centre (P8). Water samples were collected twice from each sampling point during November 21–December 17, 2011 using a stainless steel bucket from 50cm below the water's surface and were immediately transferred to an amber glass bottle that was pre-rinsed with methanol. All sampling equipment was rinsed with the sample before sampling. The water samples were immediately transferred to the laboratory in an ice-packed container and stored at 4 °C for a maximum of 6 h before extraction. A total of 16 emerging organic contaminants were determined in this study based on the information found in the literature on their occurrence in surface water as well as their high usage and consumption in China (Luo et al., 2010; Zhao et al., 2009). These contaminants comprised five hormones (17α-ethinylestradiol, 17β-estradiol, diethylstilbestrol, estrone and estriol), four antibiotics (sulfamethoxazole, norfloxacin, ofloxacin and chloramphenicol), two anti-inflammatory analgesics (diclofenac and ibuprofen), two β-blockers (atenolol and propranolol), one antibacterial (triclosan), one stimulant (caffeine) and one plasticiser (bisphenol A). The purity and source of all analytical compounds are listed in the Supplementary material (Table S1). 2.2. Sample extraction and analysis After filtering through 0.45 μm glass fibre filters, the water samples (500 mL) were spiked with 100 ng of each surrogate standard and were passed through Oasis HLB solid phase extraction cartridges (200 mg, 6 mL, Waters, Milford, USA) that were pre-conditioned with 6 mL of methanol and 6 mL of ultrapure water. After further rinsing with 10 mL ultrapure water, the cartridges were dried under vacuum for at least 30 min and eluted twice with 5 mL of methanol. The internal standards (100 ng each) were added into the extracts before final evaporation to 1 mL under a gentle nitrogen stream. The final extract was stored in a 2mL amber glass vial at −20°C for further chemical analysis. Identification and quantification of the 16 target compounds were performed with a 1260 UHPLC instrument equipped with a 6460QQQ triple quadrupole mass spectrometer with electrospray ionisation (Agilent Technologies, Palo Alto, USA) operating in the multiple reactions monitoring (MRM) mode (Yan et al., 2012; Lu et al., 2013). Sulfamethoxazole, norfloxacin, ofloxacin, atenolol, propranolol and caffeine were measured in the positive ion mode, and the remaining 10 compounds were measured in the negative ion mode. The recovery, relative standard deviation and method detection limits for the different investigated compounds are summarised in the Supplementary material
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
173
Fig. 1. Location of the sampling points in Taihu Lake.
(Table S1). Further details of the chemical analyses are provided in the Supplementary material.
An in situ biomonitoring with mature male goldfish was conducted at each sampling point in November 2011 to provide actual biological status information in Taihu Lake, as described in our previous study (Wang et al., 2011). Briefly, healthy fish were acclimatised for at least two weeks in the laboratory and were transferred to each sampling point in Taihu Lake. The fish were caged in non-toxic columnar polyethylene mesh cages (diameter 60 cm and height 1.50 m) that were fixed by three wooden pegs and completely suspended in the water approximately 20 cm below the water surface. Control fish (C) were kept in four 60 L aquariums containing 30 L dechlorinated municipal water under constant aeration in our laboratory and were fed daily. Four fish were sampled at each point following exposure for 7 d, 14 d, 21 d and 28 d after which they were immediately transported to the laboratory. After anaesthetisation with 100 mg/L MS-222, blood, liver and gonad samples were collected from the fish and were treated and stored for biomarker assays, as described in detail in the Supplementary material.
concentration was quantified using an ELISA kit (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) following the manufacturer's instructions. Gonadal DNA damage was measured by the comet assay and was calculated by tail length × DNA % in the tail/100, as previously described by Roy et al. (2003). Ethoxyresorufin-O-deethylase (EROD) activity was quantified using a microplate reader at 572 nm (Molecular Device VersaMax, Sunnyvale, USA) following the description of Lu et al. (2009). Glutathione S-transferase (GST) activity was determined using 1-chloro-2,4-dinitrobenzene as a substrate at 340 nm (Frasco and Guilhermino, 2002). Glutathione peroxidase (GPx) activity was quantified at 340 nm by monitoring the oxidation of NADPH to NADP (Berntssen et al., 2003). Superoxide dismutase (SOD) activity was measured by the inhibition of the auto-oxidation of pyrogallol as described by Marklund and Marklund (1974). Catalase (CAT) activity was determined by the method of ammonium molybdate (Góth, 1991). The VTG values and enzymatic activities were all normalised to the total protein of each sample to control the potential differences between treatments and replicates. Protein concentrations were determined using the method developed by Bradford (1976) with bovine serum albumin as the standard. Further details of the biomarker assays are provided in the Supplementary material.
2.4. Biomarker assays
2.5. Environmental implications
Serum Vtg concentrations were determined by a competitive enzyme-linked immunosorbent assay (ELISA) with purified goldfish Vtg as the standard (Rempel et al., 2006). The serum 17β-estradiol
2.5.1. Hazard quotient indexes The HQ model was developed in an effort to quantify the actual potential ecotoxicological risk of specific species exposures to certain
2.3. In situ biomonitoring
174
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
contaminants in the surrounding natural environment. In the present study, the hazard quotient hqij of compound i in sampling point j was estimated as follows: hqij ¼
MECij PNECi
ð1Þ
where MECij and PNECi are the “measured environmental concentrations” and “predicted non-effect concentrations”, respectively. The PNEC values used in this study were obtained from the literature according to the EMEA guidelines (European Commission, 2003). If more than one value was available, then the lowest value was used. Meanwhile, the total hazard quotient for each point j was calculated as the sum of hqij for the mixture components, assuming that compounds acted additively at environmentally relevant concentrations (Zhang et al., 2010): HQj ¼
X
ð2Þ
hqij:
i
A commonly used risk screening benchmark for ecotoxicological effects from contaminated water was then applied: HQ b 1.0, no significant risk; 1.0 ≤ HQ b 10, a small potential risk for adverse effects; 10 ≤ HQ b 100, significant potential for adverse effects; and HQ ≥ 100, potential adverse effects should be expected (Cristale et al., 2013; Sánchez-Avila et al., 2012). The relative contribution weight of the different compounds to the total hazard quotient at each sampling point was then estimated using the following expression: hqijð%Þ ¼
hqij 100: HQi
ð3Þ
2.5.2. Enhanced integrated biomarker response values To address the biomarkers as a whole, an enhanced integrated biomarker response (EIBR), developed by Beliaeff and Burgeot (2002) and later modified by Liu et al. (2013), was calculated to evaluate the toxicity at different sampling points. Briefly, the biomarker response data for each point were standardised according to the Eq. (4): Yi ¼ ðXi−mÞ=s
ð4Þ
where Yi is the standardised value of the biomarker, Xi is the mean value of a biomarker at each point, and m and s are the mean value and the standard deviation of a biomarker considering all sampling points, respectively. Using standardised data, Zi was then calculated as Zi = Yi or Zi = –Yi in the case of a biomarker responding to contamination by induction or inhibition, respectively, and the minimum value (mini) for each biomarker at all points was calculated from the standardised response value. The score of each biomarker response (Si) was calculated as: Si ¼ Zi þ j minij:
ð5Þ
Finally, to obtain an integrated multi-biomarker response, the EIBR value was calculated as the sum of the weighting of the biomarker and the biomarker score as follows: EIBR ¼
n X i¼1
n X
Si Wi=
Wi
ð6Þ
i¼1
where Wi is the weighting of each biomarker i, and molecular (Vtg, EROD, GST, GPx, SOD and CAT) and cellular (17β-estradiol and DNA damage) biomarkers are weighted as 1 and 2, respectively, because it is assumed that an alteration at the cellular level would have a greater impact on the health of the organisms than changes at the molecular level (Liu et al., 2013).
2.6. Statistical analysis The results for each biomarker are presented as the mean±standard deviation. All data were tested for normality and equal variance using Shapiro–Wilk's and Levene's tests, respectively. A one-way analysis of variance (ANOVA) was used to analyse the differences between treatments within each experiment, and Dunnett's t test was used to identify significantly different treatments with significance set at Pb0.05. All statistical analyses were performed with SPSS 13.0. 3. Results and discussion 3.1. Concentrations of emerging organic contaminants A summary of the concentrations of emerging organic contaminants is shown in Table 1. Seven out of sixteen target compounds (17β-estradiol, diethylstilbestrol bisphenol A, sulfamethoxazole, caffeine, norfloxacin and ofloxacin) were detected at all sampling points, estrone and 17αethinylestradiol were detected at 75% of sampling points, and the remaining compounds were not detected at any points in the lake. Overall, bisphenol A, sulfamethoxazole, caffeine and norfloxacin were more abundant than the other compounds in this region, with average concentrations of 49.8, 43.2, 47.0 and 31.9 ng/L, respectively. The high concentrations of sulfamethoxazole and norfloxacin in water might be ascribed to their low removal in conventional wastewater treatment plants (WWTPs) (Lin et al., 2009; Watkinson et al., 2007). Although the elimination efficiencies of caffeine and bisphenol A in conventional WWTPs are more than 90% (Nakada et al., 2006; Sui et al., 2010), their massive consumption and usage in daily life could result in their high abundance in waters (Carballa et al., 2005; Sim et al., 2010). 3.2. Biological effects No mortality or deformities occurred in fish at any points during the periods of in situ exposure. The changes in serum Vtg, 17β-estradiol, gonadal DNA damage, liver EROD, GST, GPx, SOD and CAT activities are shown in Table 2. In the controls, these biomarkers did not significantly change during the experimental periods, and no Vtg expression was observed. Compared with the controls, serum Vtg in the samples increased significantly in a time dependent manner at all points, as did the serum 17β-estradiol levels and gonadal DNA damage in most cases. In previous studies, these biomarkers were mainly used to indicate the health status of endocrine and reproductive function and to estimate the presence of estrogens in waters (Lu et al., 2010). The enhanced serum Vtg and 17β-estradiol levels and DNA damage in the present study could be attributed to the prevalence of estrogens in these regions, especially in Meiliang Bay and Zhushan Bay. In addition, the presence of caffeine
Table 1 Individual and total concentrations of nine detected emerging organic contaminants at different sampling points (ng/L). Compounds
Sampling points P1
P2
Estrone 3.37 2.96 17β-Estradiol 7.21 3.01 17α-Ethinylestradiol 2.28 1.64 Diethylstilbestrol 34.6 9.05 Bisphenol A 39.7 44.9 Sulfamethoxazole 65.2 85.4 Caffeine 20.4 24.4 Norfloxacin 39.4 35.0 Ofloxacin 13.1 11.7 Total 225 218 a
P3
P4
P5
P6
3.06 1.58 3.61 2.76 3.51 8.03 17.28 9.51 a 1.75 1.81 4.00 n.d. 3.16 15.4 15.0 16.9 19.1 207 35.6 14.8 75.1 34.7 28.0 21.0 23.3 39.5 89.8 78.1 50.4 24.8 24.5 25.2 10.5 9.74 9.61 13.0 188 343 225 185
n.d. means below the method detection limit.
P7
P8
n.d.a 8.00 3.00 6.41 6.63 20.7 52.4 33.1 11.1 141
n.d.a 4.36 n.d. 3.56 30.8 15.6 48.3 22.7 10.2 136
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
175
Table 2 Multi-biomarker responses in fish at eight sampling points after exposure for 7 d (A), 14 d (B), 21 d (C) and 28 d (D). Asterisks indicate values that are significantly higher than the controls (P b 0.05).a Points
Vtg
E2
TM
EROD
GST
GPx
SOD
CAT
A C P1 P2 P3 P4 P5 P6 P7 P8
bLOD b 87.8 ± 11.9 83.4 ± 7.60 90.0 ± 4.17 103 ± 7.89 101 ± 4.11 97.6 ± 8.70 84.2 ± 5.40 51.0 ± 1.31
309 ± 11.3 351 ± 12.3 322 ± 8.43 324 ± 15.4 383 ± 12.3* 387 ± 14.1* 391 ± 12.2* 322 ± 15.9 297 ± 18.9
0.573 ± 0.142 1.33 ± 0.464 0.919 ± 0.281 1.00 ± 0.343 3.12 ± 0.580* 2.06 ± 0.905* 2.27 ± 0.778* 0.904 ± 0.230 0.756 ± 0.0879
12.0 ± 1.90 15.5 ± 2.67 12.5 ± 0.691 11.7 ± 0.754 11.9 ± 2.20 12.5 ± 2.69 15.2 ± 1.96 15.2 ± 2.06 12.0 ± 2.73
35.5 ± 7.27 46.8 ± 8.42 47.3 ± 9.95 43.8 ± 9.91 43.6 ± 5.31 45.9 ± 11.3 51.0 ± 7.48* 55.2 ± 3.39* 36.5 ± 4.56
71.2 ± 15.1 50.8 ± 9.48* 56.0 ± 6.21 61.2 ± 7.46 60.1 ± 10.9 50.1 ± 4.25* 47.2 ± 9.05* 47.1 ± 5.45* 64.1 ± 9.51
6.79 ± 1.18 10.09 ± 2.01 10.6 ± 3.07* 10.5 ± 1.97* 9.33 ± 0.401 11.9 ± 2.36* 11.9 ± 1.84* 12.3 ± 1.12* 7.74 ± 2.46
438 ± 56.9 326 ± 67.8* 338 ± 67.1 354 ± 56.1 286 ± 76.0* 361 ± 30.8 309 ± 43.9 248 ± 21.6* 402 ± 84.7
B C P1 P2 P3 P4 P5 P6 P7 P8
bLOD b 119 ± 12.7 121 ± 9.04 122 ± 2.92 150 ± 10.9 135 ± 10.7 135 ± 8.56 94.5 ± 4.93 50.1 ± 5.07
308 ± 15.3 424 ± 11.8* 363 ± 11.2 361 ± 13.3 430 ± 12.9* 411 ± 9.78* 408 ± 12.0* 398 ± 15.0* 333 ± 6.06
0.669 ± 0.167 1.92 ± 0.833* 1.91 ± 1.28* 1.85 ± 0.256* 3.63 ± 1.19* 3.79 ± 1.67* 4.10 ± 1.22* 1.69 ± 0.387* 1.12 ± 0.411*
11.5 ± 1.52 15.7 ± 4.86 13.7 ± 2.21 12.7 ± 2.77 12.6 ± 1.08 13.5 ± 2.45 16.9 ± 3.92* 16.6 ± 2.73* 12.1 ± 2.66
36.5 ± 5.26 51.0 ± 7.93* 50.1 ± 4.58* 45.0 ± 10.0 45.1 ± 4.73 47.2 ± 10.1 56.2 ± 3.38* 57.1 ± 8.69* 37.8 ± 5.71
73.1 ± 12.9 49.7 ± 5.61* 56.8 ± 9.73* 60.3 ± 4.92 59.4 ± 10.2 50.4 ± 10.5* 42.3 ± 4.56* 45.0 ± 5.65* 64.3 ± 7.95
7.15 ± 1.59 10.5 ± 1.80* 11.1 ± 1.45* 11.1 ± 1.02* 9.51 ± 1.03 12.5 ± 2.61* 12.4 ± 2.21* 13.0 ± 1.07* 7.81 ± 2.00
433 ± 47.5 354 ± 56.5* 353 ± 48.2* 358 ± 9.61* 297 ± 19.7* 354 ± 61.5* 313 ± 12.4* 246 ± 40.8* 407 ± 43.0
C C P1 P2 P3 P4 P5 P6 P7 P8
bLOD b 141 ± 3.17 152 ± 8.61 140 ± 9.07 173 ± 5.92 204 ± 8.23 199 ± 10.6 105 ± 8.82 62.2 ± 8.21
313 ± 3.22 510 ± 16.2* 414 ± 29.7* 438 ± 8.66* 549 ± 5.84* 502 ± 12.3* 505 ± 11.0* 411 ± 5.00* 407 ± 9.70*
0.642 ± 0.189* 2.17 ± 0.0909* 2.06 ± 0.927* 1.97 ± 0.125* 4.17 ± 0.622* 3.26 ± 0.591* 3.43 ± 0.156* 1.46 ± 0.204* 1.04 ± 0.28*
12.4 ± 2.29 15.9 ± 3.88 14.3 ± 1.25 12.7 ± 2.79 12.1 ± 1.67 13.9 ± 3.76 17.1 ± 1.83* 16.8 ± 4.10* 12.0 ± 2.17
37.6 ± 5.10 51.2 ± 6.89 51.8 ± 6.96 45.8 ± 8.68 44.9 ± 17.4 48.3 ± 7.53 57.0 ± 7.34* 60.5 ± 9.07* 40.1 ± 6.48
72.6 ± 5.80 48.2 ± 2.75* 52.7 ± 10.3* 58.3 ± 11.7* 56.8 ± 5.49* 48.8 ± 6.06* 41.7 ± 5.84* 46.0 ± 13.2* 65.0 ± 2.27
6.67 ± 1.38 11.1 ± 1.83* 11.6 ± 1.16* 11.0 ± 0.978* 9.85 ± 1.84 12.9 ± 0.78* 13.4 ± 2.52* 14.3 ± 3.13* 8.49 ± 2.00
465 ± 47.0 363 ± 86.7* 344 ± 90.5* 346 ± 50.0* 305 ± 52.9* 339 ± 55.9* 322 ± 22.7* 240 ± 27.8* 384 ± 31.9
D C P1 P2 P3 P4 P5 P6 P7 P8
bLOD b 158 ± 15.8 151 ± 8.06 142 ± 8.07 191 ± 4.73 209 ± 2.68 204 ± 8.94 106 ± 5.79 69.1 ± 5.19
315 ± 5.73 506 ± 14.5* 423 ± 14.7* 435 ± 10.6* 552 ± 10.4* 530 ± 7.55* 520 ± 9.97* 346 ± 15.3 367 ± 15.8
0.611 ± 0.178 2.11 ± 0.733* 2.16 ± 1.17* 2.09 ± 0.984* 5.20 ± 0.745* 3.76 ± 1.18* 4.06 ± 0.368* 1.70 ± 0.219* 1.49 ± 0.275*
11.9 ± 3.08 15.3 ± 2.63 13.6 ± 1.00 12.2 ± 0.668 12.3 ± 3.10 13.5 ± 2.21 16.7 ± 2.74* 16.3 ± 3.70 12.1 ± 3.05
37.7 ± 5.83 48.3 ± 3.18 48.3 ± 4.66 45.6 ± 3.28 45.3 ± 8.09 46.1 ± 4.92 53.1 ± 11.1* 55.6 ± 4.74* 38.2 ± 5.65
70.1 ± 11.1 45.3 ± 11.0* 51.6 ± 7.91* 57.9 ± 1.37 54.1 ± 11.3* 47.1 ± 3.00* 42.1 ± 7.67* 43.5 ± 3.70* 64.9 ± 9.85
7.03 ± 1.33 10.4 ± 0.98* 10.8 ± 1.22* 10.3 ± 2.21* 9.39 ± 1.38 12.7 ± 2.08* 12.5 ± 2.39* 14.2 ± 1.18* 8.09 ± 0.987
465 ± 25.3 331 ± 80.5* 369 ± 38.6* 345 ± 72.8* 280 ± 21.9* 345 ± 66.7* 312 ± 65.6* 258 ± 66.0* 423 ± 38.1
a Vtg = Serum vitellogenin concentration (ng/mg); E2 = Serum 17β-estradiol concentration (pg/mL); TM = Comet tail moment; EROD = Ethoxyresorufin-O-deethylase activity (pmol/mg/min); GST = Glutathione S-transferase activity (nmol/mg/min); GPx = Glutathione peroxidase activity (nmol/mg/min); SOD = Superoxide dismutase activity (U/mg); CAT = catalyse activity (μmol/mg/min). b bLOD: below the detection limit.
might also induce estrogenic activity in males, e.g., Vtg production (Li et al., 2012). Significant EROD induction and GST induction were only observed at points 6 and 7 during most exposure periods, and obvious increases in GST activity were also observed at 7 d at P1–2. To date, the biotransformation of emerging contaminants, particularly pharmaceutical residues, in fish and other aquatic organisms has not been well understood (Connors et al., 2013; Gomez et al., 2010; Smith et al., 2010). The markedly induced EROD and GST activities in Zhushan Bay might suggest the presence of high levels of other organic contaminations in this region. Antioxidant enzymes in sampled fish showed spatial variability, with distinct GPx and CAT inhibition at P1–7 during most exposure periods and significant activation of SOD at most study points, with the exception of P4 and P8, compared to the controls. The significant increases in the SOD activities suggest an increased production of superoxide anions and H2O2 in fish. In general, these powerful and potentially harmful oxidising agents are metabolised by CAT and GPx into harmless agents, including H2O and O2 (Box et al., 2007). However, in the current study, the GPx and CAT activities were significantly inhibited in most cases, which suggested an incomplete defence chain against reactive oxygen species that could result in oxidative damage in organisms (Huang et al., 2007).
3.3. Environmental implications 3.3.1. Hazard quotient indexes Chemical analysis of the emerging organic contaminants in the aquatic environment is of limited use unless these data are related to their potential ecological effects via a risk assessment approach. To estimate the potential adverse effects of contaminants detected in this monitoring, an ecotoxicological risk evaluation was performed on non-target organisms. Toxicologically relevant concentrations for each detected compound used in the HQ calculation are exhibited in the Supplementary material (Table S2). Among the contaminants, no significant risk (HQ b 1) was observed for estrone, bisphenol A, norfloxacin or ofloxacin for most points, with average indexes varying from 0.307 to 0.984. For 17β-estradiol, sulfamethoxazole and caffeine, high HQ indexes, ranging from 1.60 to 3.81, indicated their low significant risk. Additionally, the presence of 17α-ethinylestradiol and diethylstilbestrol showed significant potential for adverse effects with the highest HQ indexes (18.1 and 13.0, respectively). These results suggest that the HQ index of each compound is mainly based on its toxicological potency and detected concentration in the water body. A sum of the HQ of each detected compound was performed for each point and is shown in Fig. 2. This provided an intuitive view of the total
176
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
Fig. 2. The total hazard quotient at different sampling points in Taihu Lake.
Sampling points
ecotoxicological risk at all points. The total HQ indexes in whole regions ranged from 9.75 to 67.37, whereas P6 showed the highest HQ index relative to the other points and presented a significant risk. In addition to P6, significant potential for adverse effects was also observed at points 1–5 and 7. The lowest HQ index of 9.75 was observed at P8, suggesting low significant risk for adverse effects in this region. In summary, points near the inlet of the inflowing rivers from the cities of Suzhou, Wuxi, Changzhou and Yixing have much higher risks than those in the centre of the lake. Previous studies have demonstrated that pollution of organic pollutants at lakeside and at the inlet of inflow rivers was more serious compared with the other points and showed a relatively high toxicity due to the human activities along the lakes and rivers (Zhang et al., 2012; Cristale et al., 2013). Moreover, the water diversion from the Yangtze River through the Wangyu River may also bring contaminants into the lake (Wang et al., 2011). Because hormones presented HQ indexes approximately 10 to 100 times higher than the other detected compounds, it is evident that 17α-ethinylestradiol, diethylstilbestrol and 17β-estradiol are the greatest hazards, and the potential for adverse effects in the lake is attributed almost exclusively to them (see Fig. 3). Additionally, the presence of sulfamethoxazole and caffeine accounted for 6.13% and 6.15% of the estimated total risk on average. In particular, at P3 and P8, the contribution ratio of sulfamethoxazole and caffeine to the total HQ reached 25.8% and 18.4%, respectively. Hence, to protect the ecological health of Taihu Lake, the prioritisation of the key control objectives of the contaminants should be chosen as 17α-ethinylestradiol, diethylstilbestrol, 17β-estradiol, sulfamethoxazole and caffeine. These results accord with the previous ranking investigation of Murray et al. (2010), who suggested 17α-ethinylestradiol and 17β-estradiol as the highest priority pollutants for regulation of pharmaceuticals in surface water. Sulfamethoxazole was also highlighted as a priority pharmaceutical in
P8
Estrone
P7
17β-Estradiol
P6
17α-Ethinylestradiol
P5
Diethylstilbestrol Bisphenol A
P4
Sulfamethoxazole
P3
Caffeine
P2
Norfloxacin
P1 0%
Ofloxacin
20%
40%
60%
80%
100%
Contribution ratio Fig. 3. Percent contribution of different detected compounds to the total hazard quotient indexes.
the water environment of China based on pharmaceutical consumption, removal and potential ecological effects by Sui et al. (2012). However, it should be noted that the current results from a singleseason study may not reflect the true hazards of the lake. The present study was conducted in the winter season, which means that seasonrelated consumption patterns and environmental conditions could lead to differences when comparing to other seasons. Lacey et al. (2012) found that the low water temperature (b10°C) during the winter season could dramatically decrease the biodegradation efficiencies of biodegradable contaminants in WWTPs and result in high concentrations in effluents. In addition, the weakened natural degradation processes (e.g., photodegradation) and increased human consumption of pharmaceuticals in colder periods owing to low temperature should be taken into account for the maximum levels of pharmaceuticals observed in the Llobregat River during this period (Osorio et al., 2012). All of these influencing factors could result in higher ecotoxicological risks in the lake environment in the winter season. 3.3.2. Enhanced integrated biomarker response values Although the HQ approach can give a preliminary determination of the possible effects on aquatic organisms, the detection of only a few classes of contaminants may not be sufficient to estimate the real effects of a complex mixture at actual environmental concentrations for an accurate ecotoxicological risk assessment. Moreover, the possible synergic interactions between contaminations with different modes of action and the bioaccumulation and bio-magnification in organisms may also lead to an underestimation of the risk in water. Instead, the application of multiple biomarkers could avoid the above defects and improve the differentiation between healthy and stressed organisms (Brooks et al., 2011), despite the fact that it is difficult to be understood by nonspecialists. The integrated biomarker response (IBR), which scores and summarises the responses of a set of biomarkers in a single value, provides an intuitive interpretation of the relative toxicity of point contaminations to the public. This approach has been widely used in many field and laboratory studies to assess the variation of environmental quality (Kim et al., 2010; Li et al., 2011). However, the conventional IBR system is being challenged when the employed biomarkers cover multibiological organisation levels. In this study, an enhanced integrated biomarker response (EIBR) that considered the significance of each biomarker at different levels was applied. Because the impacts of bioendpoint alterations on the health status of organisms normally increased through molecular, cellular, physiological and individual avenues, the weighting of each biomarker was appointed according to this hierarchy as 1 and 2 (Liu et al., 2013). A similar weighting has been applied in the pollution monitoring programme under the Europe Water Framework Directive (Hagger et al., 2009). Without weighting, the IBR results could be biased and even misleading. The EIBR values were calculated by integrating the eight biomarkers and their weighting, and the results are represented as a star plot (see Fig. 4). In general, the EIBR values indicated a large range of spatial variation of ecological effects at different points. Relatively high EIBR values were observed at sampling points 4–7, located in the Zhushan and Meiliang Bays, as well as P1, situated at the inlet of the Wangyu River, with the highest value at P6. P2 and P3, located in the Gong Bay presented low EIBR values during the periods of in situ exposure, and P8, situated in the centre of the lake showed the lowest EIBR value, which was only 0.21-fold that at P6. Given that the EIBR is a visualised indicator of ecotoxicological risk, the current results suggest that Zhushan and Meiliang Bays are the most stressful places for organisms, especially at P6 near the end of Zhushan Bay. In addition to anthropogenic influences, the almost enclosed areas of these two bays could increase the difficulty of the exchange of water with Gong Bay and the centre of the lake, especially at the end of the bays, which may further result in the accumulation of contaminants in these areas. For P1, the re-suspension processes of
Z. Yan et al. / Science of the Total Environment 470–471 (2014) 171–179
P1
14d
P2
2
21d 28d
1 P7
P3
0
P6
P4
2
HQ indexes
P8
3
7d
3
177
y= 0.0245x+0.7511 r = 0.848 P
