Study area and sampling collection
Shanghai is one of the largest international metropolises in China and is a center of business, habitancy, farming, industry, and other activities. It covers an area of approximately 6340.5 km2 and is situated on the alluvial plain of the Yangtze River Delta with an average altitude of 4 m above sea level [16]. There are approximately 33,127 rivers and streams in Shanghai, with a total length and a total water surface area of approximately 24,915 km and 642.7 km2, respectively [17]. These rivers are interconnected, thus forming an urbanized river network. The implementation of the coastal development strategy in 1984 has had brought serious influences on the quality of the river network in Shanghai, and has even destroyed the original ecological balance [18].
River water samples from 53 sampling sites in the urbanized river network were collected from December 2018 to February 2019. The sampling sites were distributed equably across the entire city of Shanghai, spanning from 30° 40′ N to 31° 53′ N and from 120° 52′ E to 122° 12′ E, with 10 sites located in suburban towns, 6 sites in island areas, 11 sites in industrial areas, 15 sites in agricultural areas, and 11 sites in the inner city (Fig. 1). Additional file 1: Table S1 describes the detailed information on the sampling sites of diverse intensive land-use types. At each site, triplicates of river water samples were collected at depths of 5–15 cm from the middle of the river using a plexiglass sampler, stored in 1-L brown glass bottles, and immediately transported to a laboratory at 4 °C with preservatives (0.10 g/L ascorbic acid, 0.35 g/L ethylenediaminetetraacetic acid disodium salt, and 9.4 g/L potassium dihydrogen citrate). Then, the samples were analyzed within 24 h. In addition, the sampling sites were located using a global positioning system (GPS) device.
Analysis of OCPs and total suspended solids (TSS)
Pretreatment of surface water samples
A 1-L river water sample was directly filtered using 0.45-μm glass microfiber filters (GFF, ANPEL, China) that were baked in a muffle furnace at 400 °C for 4 h. OCPs in filtrate were extracted using solid phase extraction (SPE, ANPEL CG1824, China) with 6-mL cartridges filled with 500 mg C18 (ANPEL, China), and the SPE procedure was performed according to the method recommended by the United States Environmental Protection Agency (USEPA) [19]. First, the SPE column was activated into balance using 5 mL ethyl acetate, 10 mL methanol, and 10 mL ultra-pure water successively. Then, the loading water velocity was controlled at 10 mL/min. In the next step, which aimed to target the OCPs, the sample was eluted with 5 mL methanol and 5 mL dichloromethane (DCM) after extraction. The OCP eluent was dried using anhydrous sodium sulfate, which was also calcined for 4 h at 400 °C and then transferred with DCM and reduced to ~ 0.7 mL using an antiseptic 24-bit nitrogen blower (ANPEL DC24-Rt, China). Finally, the concentrated sample was fixed to 1 mL after adding ~ 0.3 mL methanol.
Pretreatment of the suspended particulate matter (SPM)
The suspended particulate matter (SPM) separated using GFF was filtered from 1 L of river water sample and allowed to dry naturally. After drying, the total suspended solid (TSS) content of the river water could be calculated according to the added weight of the GFF. The procedure for OCP extraction in the SPM was modified from the procedure used by Liu et al. In brief, the SPM in each sample was extracted using 10 mL methanol and 10 mL DCM in turn for 48 h. After filtered with a nylon syringe filter (ANPEL, China), the extractions were then reduced to ~ 0.7 mL using the antiseptic 24-bit nitrogen blower (ANPEL DC24-Rt, China). The concentrated sample was then purified using a self-packet anhydrous sodium sulfate column (2 g, baked for 4 h at 400 °C, and washed using DCM and n-hexane of twofold column volumes). After that, the sample was fixed to 1 mL using ~ 0.3 mL methanol for the gas chromatography–mass spectrophotometry (GC–MS) analysis.
OCP quantification and quality control
GC–MS (Thermo Fisher Trace DSQ II-MS, USA) in EI mode was used for the OCP quantification. The capillary column used was HP-5MS (30.0 m × 0.25 mm × 0.25 μm). The carrier gas was helium at a flow rate of 1.2 mL/min under the constant flow mode. The injector was set at 275 °C. The oven temperature was programmed as follows: initially at 70 °C (equilibrium time 1.5 min), increased to 200 °C at the rate of 10 °C/min, continually increased before reaching at 320 °C at the rate of 7 °C/min, and then held for 3 min.
The internal standard method recommended by the USEPA (2012) was adopted in this study to quantify the concentrations of OCPs. The internal standards are pure compounds that are isotopically labeled analogs of selected method analytes that are added to each sample prior to extraction in a known amount in order to measure the relative responses of the OCPs, and also determine the recoveries in the real samples [19]. Deuterated polycyclic aromatic hydrocarbons (PAHs) of 100 ng/μL (d10-acenaphthene, d10-phenanththrene, d12-chrysene) were used as the internal standard solution (Sigma, USA) in this study [19].
The linear correlation coefficients (R2) of the standard calibration curves for target OCPs ranged from 0.9906 to 0.9993, with a relatively high repeatability. Procedural blanks and field blanks were analyzed routinely to monitor the procedural performance and interference. One blank and one standard were conducted every ten samples. The limits of detection (LOD) of all the OCPs ranged between 0.24 and 1.83 ng/L with a signal-to-noise ratio (S/N) of three. In addition, the limits of quantification (LOQ) ranged from 0.80 to 6.14 ng/L, with an S/N of 10. All the method analyses were under the LOD in the blank samples. The recoveries from this method were 62.5%–126.3% during the surface water phase and 60.8%–114.8% during the SPM phase. Values for OCPs lower than the LOQ were reported as half of the LOQ, while values lower than the LOD were substituted with zero prior to the statistical analysis.
The standard OCPs included etridiazole, chloroneb, propachlor, α-HCH, β-HCH, γ-HCH, δ-HCH, HCB, chlorothalonil, heptachlor, aldrin, dacthal (DCPA), heptachlor epoxide, trans-chlordane, cis-chlordane, endosufan I, endosufan II, trans-nonachlor, 4,4′-DDE, 4,4′-DDD, 4,4′-DDT, dieldrin, endrin, chlorobenzilate, endosufan sulfate, and methoxychlor, and the internal standards were purchased from Sigma. All of the solvents (e.g., ethyl acetate, methanol, and DCM) used for the sample processing and analysis were high-performance liquid chromatography (HPLC) grade and obtained from ANPEL.
Statistical analysis
The partition coefficient of the OCPs between the SPM and surface water (Kp, L/g) was calculated as follows [18]:
$$ K_{p} = C_{p} /C_{d} $$
(1)
where Cp is the OCPs concentration in the SPM, ng/g; and Cd the OCPs concentration in the surface water, ng/L.
ArcGIS version 10.2 (ESRI, USA) was used for the spatial distribution analysis of the 26 OCP concentrations and their ecological risks. The graphs were drawn using the interpolation method, and the inverse distance weight (IDW) method was adopted in the interpolation model.
Ecological risk assessment
The ecological risk of OCPs to aquatic organisms was assessed using the risk quotient (RQ) method, which was performed by the calculation of the RQ for the detected OCPs [20].
$$ {\text{RQ}} = {\text{EEC/PNEC}} $$
(2)
where RQ is the risk quotient; EEC the measured concentration of each OCP, mg/L; and PNEC the predicted no-effect concentration for a particular OCP, mg/L.
The PNEC was calculated as follows [20]:
$$ {\text{PNEC}} = {\text{HC}}_{ 5} /{\text{AF}} $$
(3)
where HC5 is the OCP concentration at which 5% of species are exposed to chronic hazards, mg/L; and AF an assessment factor, which considers the uncertainty between laboratory toxicology tests and the real environment. In this study, AF = 100 [4].
HC5 was calculated using species sensitivity distribution (SSD) method [21]. The acute toxicity data, such as the median lethal concentration (LC50) or median effective concentration (EC50) of fish (freshwater), daphnia (freshwater), green alga (freshwater), earthworm (freshwater), fish (seawater), and shrimp (seawater) for 26 OCPs, were obtained from ECOSAR version 2.0 (USEPA, USA) as far as possible (Additional file 1: Table S2). Due to the scarcity of chronic toxicity data, no observed effect concentration (NOEC) or lowest observed effect concentration (LOEC) of fish (freshwater), daphnia (freshwater), and green alga (freshwater) for 26 OCPs were obtained using ECOSAR version 2.0 (USEPA, USA) (Additional file 1: Table S3). The acute toxicity data of each species for the same OCP monomer were arranged in order from small to large, and the corresponding cumulative probability (P) was calculated as follows:
$$ P = n/\left( {N + 1} \right) $$
(4)
where n is the ordinal number. The minimum ordinal number was set to 1, and the maximum ordinal number was set to N.
The logarithmic value of the OCP’s acute toxicity data for each species was used as the abscissa, and the corresponding cumulative probability was used as the ordinate. Seven types of SSD curve models were selected in this study, namely the Boltzmann, Logistic, Gaussian, Lorentzian, Exponential, Logarithm, and Log-normal models. Origin version 8.0 (OriginLab, USA) was used for drawing and model fitting. The model suitable for deducing the OCP concentration at which 5% of species are exposed to acute hazards (AHC5) was determined according to goodness of fit, and the AHC5 values were obtained (Additional file 1: Table S4).
The ratios between the acute toxicity data (EC50 or LC50) and the chronic toxicity data (NOEC or LOEC) of fish (freshwater), daphnia (freshwater), and green alga (freshwater) were determined, and then the geometric average value (FACR) of the three ratios was obtained. Therefore, the HC5 was eventually calculated as follows [21] (Additional file 1: Table S4):
$$ {\text{HC}}_{ 5} = {\text{AHC}}_{ 5} / {\text{FACR}}. $$
(5)
According to different evaluation criteria and management purposes, the classification of ecological risks was different [4, 6, 22, 23]. Overall, an RQ value greater than 1 indicates that the concentrations of OCPs in the environment are higher than the predicted no-effect concentrations. Subsequently, in the prior literature, ecological risks could be divided into five levels based on the RQ value: very low risk (RQ < 0.01), low risk (0.01 ≤ RQ < 0.1), moderate risk (0.1 ≤ RQ < 1), high risk (1 ≤ RQ < 10), and very high risk (RQ ≥ 10) [23].
Health risk assessment
The river network of Shanghai is primarily used for landscape water, and in this exposure scenario, the health risks of OCPs from respiratory intake are nearly negligible [4, 24]. Therefore, in this study, the carcinogenic and non-carcinogenic risks of OCPs from dermal contact and mistaken oral intake were focused on [25]. The citizens of Shanghai were divided into eight population groups according to age and gender: children (4–10 years), adolescents (11–17 years), adults (18–60 years), and seniors (61–70 years) for both males and females [26]. The exposure dose for each population group caused by dermal contact was calculated as follows [27]:
$$ {\text{EED}} = C_{w} \times {\text{SA}} \times {\text{PC}} \times {\text{ET}} \times {\text{EF}} \times {\text{CF}} \times {\text{ED}}/\left( {{\text{BW}} \times {\text{AT}}} \right) $$
(6)
where EED is the estimated exposure dose per unit body weight per day of OCPs, mg/(kg·day); Cw the concentration of OCPs in river waters, mg/L; SA the skin contact area, cm2; PC the dermal permeation constant of each OCP, cm/h; ET the daily exposure time, h/day (Additional file 1: Table S5); EF the annual exposure frequency, d/a (Additional file 1: Table S5); CF the conversion factor, (0.001 L/cm3); ED the exposure duration, a (Additional file 1: Table S6); BW the body weight, kg (Additional file 1: Table S6); and AT the average contact time, days (Additional file 1: Table S5).
The skin contact area (SA) was calculated as follows [27]:
$$ {\text{SA}} = 239 \times H^{0.417} \times {\text{BW}}^{0.517} \times {\text{SER}} $$
(7)
where H is the body height, cm (Additional file 1: Table S6); and SER the skin exposure ratio (Additional file 1: Table S6).
The dermal permeation constant of each OCP (PC) was calculated as follows [28]:
$$ {\text{PC}} = 10^{{\left( { - 0.280 + 0.66{ \log }\,K_{ow} - 0.0056{\text{MW}}} \right)}} $$
(8)
where Kow is the octanol–water partition coefficient of each OCP (Additional file 1: Table S7); and MW the molecular weight of each OCP (Additional file 1: Table S7).
The exposure dose for each population group through mistaken oral intake was calculated as follows [27]:
$$ {\text{EED}} = {\text{Cw}} \times {\text{IR}} \times {\text{EF}} \times {\text{ED}}/\left( {{\text{BW}} \times {\text{AT}}} \right) $$
(9)
where IR is the oral intake rate, L/day (Additional file 1: Table S6).
The incremental lifetime cancer risk of population groups caused by a single OCP through a single exposure pathway in the river waters of Shanghai was calculated based on Eq. 10 [27].
$$ {\text{ILCR}}_{\text{i}} = {\text{EED}}_{\text{i}} \times {\text{SF}}_{\text{i}} $$
(10)
where ILCRi is the incremental lifetime cancer risk caused by the ith OCP; EEDi is the estimated exposure dose per unit body weight per day of the ith OCP, mg/(kg·day); and SFi is the cancer slope factor for the specific exposure pathway of the ith OCP, kg·day/mg (Additional file 1: Table S7).
The incremental lifetime cancer risk of population groups caused by 26 OCPs through dermal contact or mistaken oral intake from the river waters of Shanghai was calculated based on Eq. 11 [27].
$$ {\text{ILCR}} = \mathop \sum \limits_{i}^{n} {\text{ILCR}}_{i} $$
(11)
where ILCR is the total incremental lifetime cancer risk caused by 26 OCPs.
The loss of life expectancy was calculated using Eq. 12 [29].
$$ {\text{LLE}} = 62 \times {\text{ILCR}}/10^{ - 5} $$
(12)
where LLE is the total loss of life expectancy caused by 26 OCPs, min.
The non-carcinogenic risk of population groups caused by a single OCP through a single exposure pathway from the river waters of Shanghai was calculated according to Eq. 13 [27].
$$ {\text{HI}}_{i} = {\text{EED}}_{i} /{\text{RfD}}_{i} $$
(13)
where HIi is the non-carcinogenic risk hazard index caused by the ith OCP; and RfDi is the non-carcinogenic reference dose for the specific exposure pathway of the ith OCP, mg/(kg·day) (Additional file 1: Table S7).
The non-carcinogenic risk of population groups caused by 26 OCPs through dermal contact or mistaken oral intake from the river waters of Shanghai was calculated according to Eq. 14 [27].
$$ {\text{HI}} = \mathop \sum \limits_{i}^{n} {\text{HI}}_{i} $$
(14)
where HI is the total non-carcinogenic risk hazard index caused by 26 OCPs.
Due to the possible uncertainties in the body weights and heights at different age stages and sexes and with different probabilities of OCP concentrations in river waters, the probabilistic evaluation of health risks was conducted using a Monte Carlo simulation with 5000 runs [26]. Random body weights, body heights, and concentrations of OCPs were generated for each run based on the normal, normal, and lognormal distributions, respectively. The mean and standard deviation (SD) for generating the random body weights and heights that aligned the corresponding distribution for each age stage and each sex were obtained from the China Health Statistics Yearbook (Additional file 1: Table S6) [30]. In addition, the mean and SD for generating the random OCP concentrations for each intensive land-use type were calculated from the 53 sampling sites (Additional file 1: Table S8). The probabilistic estimation approximates the probabilistic health risks from OCPs in the river network of Shanghai [31].
Studies have shown that when the incremental lifetime cancer risk (ILCR) is higher than 1.0 × 10−6, the carcinogenic risk cannot be ignored, and an ILCR above 1.0 × 10−5 indicates a “not negligible” risk [32]. The USEPA regards an ILCR in the range of 1.0 × 10−7 to 1.0 × 10−4 as an acceptable carcinogenic risk value [33]. Some other studies have classified the ILCR into five levels: values smaller than 1.0 × 10−6 indicate very low risk; values between 1.0 × 10−6 and 1.0 × 10−4 indicate low risk; values ranging from 1.0 × 10−4 to 1.0 × 10−3 indicate moderate risk; values between 1.0 × 10−3 and 1.0 × 10−1 represent high risk; and values greater than 1.0 × 10−1 represent very high risk [24]. In this study, a five-level evaluation method was adopted to evaluate the carcinogenic risk of OCPs in the river waters of Shanghai. In addition, if the hazard index (HI) is less than 1, the non-carcinogenic risk could be considered negligible to local residents, which can be used as a criterion for a non-carcinogenic risk assessment [27].