Multi-Endpoint Toxicity Tests and Effect-Targeting Risk Assessment of Surface Water and Pollution Sources in a Typical Rural Area in the Yellow River Basin, China

Abstract

Multi-endpoint toxicity tests were used to evaluate the acute toxicity, estrogenic activity, neurotoxicity, genotoxicity, and ecological risks of surface water and sewage from possible pollution sources in rural areas of the Yellow River (China). Toxicity testing results showed that the luminescence inhibition rates of acute toxicity ranged from not detected (ND) to 38%, the 17β-estradiol equivalent (E2-EQ) values of estrogenic activity ranged from 4.8 to 131.0 ng·L⁻¹, neurotoxicity was not detected, and the protein effect level index (PELI) values of genotoxicity ranged from 1 to 6.06. Neither acute toxicity nor genotoxicity were detected in the tributaries of the Yellow River (River 2) flowing through the investigated rural area. The distribution of high estrogenic activity sites was relatively scattered, but mainly located in the tributaries of River 2. Industrial, domestic, and livestock and poultry breeding sewage were all possible sources of toxicity, and the contribution of livestock and poultry to environmental estrogens in the surface water was significant. Furthermore, the potential effect-targeting risks of toxic substances in the surface water for aquatic organisms were assessed using the risk quotient method, by considering the toxic equivalent concentration. The results indicated that the risk of estrogenic activity was the main ecological risk in the surface water of this rural area. Except for the reservoir site, the other sampling sites showed a moderate to high estrogenic activity risk, especially in the tributaries of River 2.

1. Introduction

As suppliers of cereals and other natural resources, rural areas occupy an important position in various countries worldwide [1]. Today, with the continuous acceleration of urbanization, rural communities face many problems due to economic structural changes, product transformations, technological innovations, and so on [2]. Rural areas are mostly dominated by agriculture and animal husbandry. While pursuing rapid economic development, more and more chemicals, such as pesticides, herbicides, and antibiotics, are used in agriculture and livestock and poultry breeding [3,4]. As a result, a large number of these substances are discharged into rural areas, which seriously threatens human health. In addition, the development of rural industry and the improvement of people’s living standards also mean that emerging contaminants, such as perfluorinated compounds, microplastics, pharmaceuticals, and personal care products, enter the rural environment, causing huge environmental risks [5,6,7,8].

At present, all kinds of production activities require water, which also increases the risk of water pollution. Unlike in cities, water pollution in rural areas is more difficult to monitor and control, mainly due to the complex pollution sources. There are many types of pollution source, with an irregular distribution, which make it impossible for rural areas to detect certain pollutants purposefully and according to the potential pollution sources in the region, as is done in cities. In addition, due to gaps in some policies and a lack of awareness of environmental protection, the illegal discharge of pollutants in rural areas is serious, which makes it more difficult to correctly determine the pollution situation and source of pollution in water bodies [9]. Therefore, in order to effectively deal with the problem of water pollution in rural areas, it is necessary to conduct a comprehensive analysis of water bodies.

With the increase of diverse classes of pollutants in waterbodies, the environmental risks are also diverse [10], making water pollution assessment challenging but crucial [11]. With the increasing attention to health and the in-depth study of pollutants in the environment, toxicity indicators directly related to the health of organisms, such as acute toxicity, neurotoxicity, genotoxicity, and endocrine disrupting effects have become widely used [12,13,14]. Mixtures of pollutants may cause one or more toxic effects. Through the multi-endpoint toxicity analysis of water bodies, the comprehensive impact of complex mixtures of pollutants in water bodies can be evaluated. At present, multi-endpoint toxicity analysis has been widely used in urban rivers, wastewater treatment plants, industrial parks, and other sites, while there have been relatively few toxicity assessments and pollution analyses of waterbodies in rural areas, especially of surface water [15,16,17]. For the complex situation of water pollution in rural areas, a multi-endpoint toxicity test can reflect the overall problems of the water body and give insights about the existing pollution. Qualitative and quantitative analysis of the water body can be carried out by chemical analysis, to determine specific groups or pollutants. Moreover, using these multi-endpoint toxicity test results, the qualitative analysis of pollutants will be simpler and faster and more sensitive to potential toxicity.

We selected a typical rural area in the lower reaches of the Yellow River in China. This area is mainly dominated by agriculture and livestock and poultry breeding, with industry in the development stage. Acute toxicity, estrogenic activity, neurotoxicity, and genotoxicity tests of surface water were conducted in this area, to analyze the multi-endpoint toxicity of the surface water. At the same time, a multi-endpoint toxicity analysis was carried out for many types of possible pollution sources in the local area. Combined with the toxicity of the surface water, the ecological risks for water bodies in the rural area were analyzed. This study can provide a reference for future pollution assessments and the prevention of surface water pollution in rural areas.

2. Material and Methods

2.1. Study Sites and Sample Collection

The study area is situated in a rural area on the east side of the lower reaches of the Yellow River, China. There are two main rivers in the area, which are formed by a levee breach when the Yellow River flows through. The whole area is a plain area, with high terrain in the southwest and low terrain in the northeast. Therefore, the two main rivers flow from west to east, forming many small tributaries and ditches (Figure 1). At present, River 1 serves as a water source river, to help supplement domestic water and drinking water in this rural area; while River 2 flows through the whole rural area and receives all kinds of water from regional production and domestic activities.

The sampling campaigns were carried out in September, the autumn rainy season. A total of 27 water samples were collected in the study area, including 21 surface water samples and 6 possible pollution source samples.

Table 1. Sampling sites in the selected study area.

Code	Sites	Types	Date
SW1	Yellow River diversion gate, headstream of River 1/2	Surface water	26 September
SW2-7	Main stream of River 2		26 September
SW8-15	Tributary of River 2		26 September
SW16-19	Tributary of River 2 adjacent to a goose farm		26 September
SW20	River 1		26 September
WSL	Reservoir on River 1	Surface water	26 September
WPS1	Chemical industrial park	Rainwater	27 September
WPS2	Planting farmland	Ditch water	27 September
WPS3	Duck farm	Fecal sewage	27 September
WPS4	Cattle farm		27 September
WPS5	Swine farm	Urine sewage	27 September
WPS6	sewage treatment plant	Effluent	27 September

displays the water sampling sites and sample types. Surface water samples were collected 0.5 m below the water surface with 1 L plexiglass water collector and then transferred to amber glass bottles. Each bottle was rinsed with the water sample in advance. Samples from pollution sources were collected directly in bottles. Then, 0.429 mL of concentrated hydrochloric acid (36%) was added to each 1 L water sample for acidification. Then the collected samples were transported to the laboratory in a low-temperature transport box and stored at 4 °C until analysis.

2.2. Sample Pretreatment for Toxicity Analysis

The samples were pretreated as described below, to pre-concentrate them for the toxicity assays, except for the acute toxicity test. Then, 500 mL of each sample was filtered through a 0.7 μm glass fiber membrane (Whatman) under vacuum. The filtered samples were processed by solid phase extraction (SPE) using Oasis^® HLB 6cc extraction cartridges (Waters). The HLB cartridges were conditioned with 5 mL methanol and equilibrated with 5 mL 5 mM HCl before use. The pre-filtered samples were flowed through a cartridge connected to a Visiprep DL (Supelco) under vacuum at a flow rate of 5–10 mL min⁻¹. Afterward, the cartridges were kept in vacuum aspiration until the residual liquid had drained. The cartridges were wrapped in aluminum foil and stored at −20 °C until elution.

The cartridges were successively eluted with 5 mL methanol and 5 mL mixed solution of n-hexane and acetone (1:1, v/v). The solution was added and allowed to stand for 5 min before elution. The flow rate of the whole elution process was maintained at 1–2 mL min⁻¹ or slower. Finally, the eluent was blown dry under gentle nitrogen and reconstituted with 1 mL methanol. The sample solutions were stored at −20 °C until analysis.

2.3. Acute Toxicity Analysis

An acute toxicity analysis was conducted on Centro LB 960 luminescence microplate readers using Vibrio fischeri and according to the test protocol ISO 11348-3 [18]. The samples were filtered through a 0.45 µm filter to remove particles and then adjusted to a pH value of 6.0–8.5 before analysis. After mixing 100 μL sample solutions with 100 μL bacterial suspensions, the decrease of bacterial luminescence at 30 min compared with the initial value was recorded. In addition, 2% NaCl gradient diluted 3,5-dichlorophenol (DCP) was used as positive control and 2% NaCl was used as negative control for each test. The relative luminescence inhibition rate (%) of each sample was calculated, to determine toxicity using the following formula:

$$\mathrm{Inhibition} \mathrm{rate}=\frac{I_0{ \times C}_{\mathrm{f}}{ - I}_{\mathrm{t}}}{I_0{ \times C}_{\mathrm{f}}} \times 100$$

(1)

where I₀ is the initial luminescence intensity of Vibrio fischeri, I_t is the luminescence intensity of Vibrio fischeri after 30 min of being mixed with the water samples, and C_f is the correction factor for the contact time of 30 min;

$$C_{\mathrm{f}}=\frac{C_{\mathrm{t}}}{C_0}$$

(2)

where C₀ is the initial luminescence intensity of negative control sample, and C_t is the luminescence intensity of Vibrio fischeri after 30 min of being mixed with the negative control sample. The acute toxicity test is valid when the C_f is between 0.6 and 1.8.

2.4. Estrogenic Activity Analysis

The estrogenic activity of the water sample after SPE could be determined using the yeast estrogen screening (YES) method [19]. Yeast strains stored at −80 °C were added to SD/-Trp-Leu medium and cultured with shaking at 30 °C for 24–30 h. Then, 180 μL of the bacterial suspensions was added to a 96 well plate until an absorbance level of 0.7–1.0 at 600 nm (OD600) was achieved. Concentrated samples were made into a series of twofold dilutions with 0.2 M phosphate-buffered saline (PBS), and 20 μL of diluted samples were added to a 96 well plate. 17β-estradiol (E2) was prepared as a positive control in a series of concentration gradients in methanol for each plate. The OD600 was measured after the mixing of the bacterial suspensions, and samples was incubated at 30 °C, 800 rpm for 4 h. Next, 120 μL of test-buffer (100 mL PBS, 270 μL β-Mercaptoethanol and 3.33 mL SDS 0.1%) and 20 μL of chloroform were added sequentially and pre-incubated for 10 min. The chromogenic reaction was started by adding 40 μL of 8 mg mL⁻¹ chlorophenolred-β-D-galactopyranoside (ONPG) and terminated by adding 100 μL of 1 M Na₂CO₃ after 1 h. The supernatant was separated by centrifugation at 2500 rpm for 2 min, and the absorbance was measured at 420 nm. Combined with the OD600 value, the β-galactosidase activity (U) was expressed using the following formula [20]:

$$U=\frac{{OD}_{420\mathrm{s}}{ - OD}_{420\mathrm{b}}}{{t \times V \times OD}_{600}} \times D$$

(3)

where t is the time of chromogenic reaction (min); V is the volume of the final test (mL); D is the dilution ratio between the final test and adding test-buffer; OD_420s and OD_420b are the absorbance of samples and blank at 420 nm, respectively; and OD₆₀₀ is the absorbance of samples at 600 nm.

Dose–response curves for E2 and samples were plotted as the concentration and relative enrichment factor (enrichment factor multiplied by dilution factor) versus β-galactosidase activity, respectively. The comprehensive estrogenic activity of the water samples was measured by dividing the EC50 of the two curves and expressed as the 17β-estradiol equivalent (E2-EQ).

2.5. Neurotoxicity Analysis

Acetylcholinesterase (AChE) activity was used to indicate neurotoxicity and measured according to a modified Ellman method [21,22]. The enriched samples were added to 96 well plates and diluted step by step with 0.05 M PBS. Then, 10 μL of N-bromosuccinimide and 20 μL of the mixture of PBS and 4 g L⁻¹ ascorbic acid (1:1, v/v) were added. Afterward, 10 μL of 800 U L⁻¹ AChE solution was added and mixed on a plate shaker for 10 min; 140 μL of the test solution (2 mL S-Acetylthiocholine iodide 78 mM, 2 mL 5,5′-dithion-bis-2-nitrbenzoic acid 7.8 mM, and 10 mL PBS) were added; and the absorbance at 420 nm was immediately measured for 4 min (30 s intervals). The inhibition rate of enzyme activity was calculated from the enzyme velocity and fitted to the dose–response curve. Serial concentrations of parathion were used as a positive control, and the toxicity was expressed as parathion equivalent (PT-EQ) through the IC50 value.

2.6. Genotoxicity Analysis

CHK1 acts as a checkpoint in response to any DNA general damage [23,24]. We selected Saccharomyces cerevisiae (ATCC 201388) containing GFP-tagged CHK1 to detect neurotoxicity. Yeast strains were grown in clear-bottom black 384 well plates (Costa) with SD medium at 30 °C for 4–6 h, until reaching early exponential growth (OD600 about 0.2–0.4). Then, 10 μL of concentrated samples (diluted from 500× to 100× and 10× with medium) and controls (medium only) were added to each well. Plates were placed in a micro-plate reader to measure the absorbance (OD600 for cell growth) and GFP signal (excitation 485 nm, emission 525 nm for protein expression) every 10 min for 24 h.

The protein expression profiling data of yeast were processed as described in previous studies [25,26,27,28]. OD and GFP raw data were corrected using the background OD and GFP signal of the medium control. The protein expression P for each measurement was normalized by the cell number, as P = (GFP_corrected/OD_corrected). The alteration in protein expression of CHK1 at each time point due to chemical exposure was represented as I = P_experiment/P_control. The accumulative protein expression change over the exposure period was represented by the protein effect level index (PELI).

$$\mathrm{PELI}=\frac{{{\displaystyle \int }}_{t=0}^{t}I \mathrm{d}t}{\mathrm{exposure} \mathrm{time}}$$

(4)

where t is the exposure time. Genotoxicity was represented by PELI and genotoxicity positive was defined as having a PELI value greater than 1.5. [25,26].

2.7. Ecological Risk Assessment

The potential risks of toxic substances in the surface water to aquatic organisms were assessed using a risk quotient (RQ) method. The RQ values were calculated using the following formula:

$$\mathrm{RQ}= \frac{\mathrm{MEC}}{\mathrm{PNEC}}$$

(5)

where MEC is the measured environmental concentration, and PNEC is the predicted no effect concentration in water. In this study, the equivalent concentration of quality control substance corresponding to the different toxicities was used as the MEC value; that is, E2-EQ was used for estrogenic activity and PT-EQ was used for neurotoxicity. According to European Commission Technical Guidance Document, the PNEC value is usually obtained by dividing the toxicity data of the most sensitive species by the assessment factor (AF), as shown in the following equation:

$$\mathrm{PNEC}=\frac{\mathrm{E}\left(\mathrm{L}\right)\mathrm{C}50}{\mathrm{AF}} \mathrm{for} \mathrm{acute} \mathrm{toxicity}$$

(6)

$$\mathrm{PNEC}=\frac{\mathrm{NOEC}\left(\mathrm{LOEC}\right)}{\mathrm{AF}} \mathrm{for} \mathrm{chronic} \mathrm{toxicity}$$

(7)

where EC50 or LC50 was the half maximum effect concentration or the median lethal concentration in acute toxicity, and the AF value was set as 1000; NOEC or LOEC was the no observed effect concentration or lowest observed effect concentration in chronic toxicity, with AF values of 10, 50, or 100 [29].

The PNEC values for aquatic organisms of different compounds are shown in

Table 2. Aquatic toxicity data and PNEC values of the target compounds.

Compound	Taxonomic Group	Toxicity	NOEC/LOEC or EC50/LC50 (ng L−1)	AF	PNEC (ng L−1)	Reference
17β-Estradiol	Algae	Acute	2,480,000	1000	2480	[30]
	Invertebrate	Chronic	1000	100	10	[31]
	Fish	Acute	2000	1000	2	[32]
Parathion	Invertebrate	Acute	2500	1000	2.5	[33]
	Fish	Acute	1,500,000	1000	1500	[33]

. To avoid underestimating the potential ecological risks, the lowest PNEC was selected for calculation.

3. Results and Discussion

3.1. Acute Toxicity Assessment with Vibrio fischeri

Water samples from the study sites were evaluated for acute toxicity using the Vibrio fischeri luminescence inhibition rate, and samples with a luminescence inhibition rate of more than 10% were identified as having acute toxicity [18]. Acute toxicity was determined in some surface water samples and pollution source samples. As shown in Figure 2, no acute toxicity was detected in River 1, where the water supply reservoir was located, nor was acute toxicity detected in the main stream of River 2. In the surface water samples collected from the tributary of River 2, except for the goose farm, acute toxicity was only detected in SW10 and SW13, and the inhibition rate was 10% and 38%, respectively. SW10 was collected downstream of the chemical industrial park on the tributary of River 2, and acute toxicity was detected in the rainwater from the surface runoff of the chemical industrial park, with an inhibition rate of 26%. Due to a deficient rainwater collection system, the downstream rivers of the chemical industrial park were likely polluted by rainwater runoff [34], which was a possible reason for the acute toxicity of SW10. Among the surface water samples, SW13 had the strongest acute toxicity, with an inhibition rate of 38%. The tributary where the sampling site was located often received domestic sewage and garbage from the surrounding large villages, as well as rainwater carrying livestock and poultry fecal matter. These were the possible reasons for the acute toxicity of the site.

Weak acute toxicity was also observed in the samples WPS3 and WPS5 collected from livestock and poultry farms, with inhibition rates of 17% and 16%, respectively. As livestock and poultry breeding sewage, feed additives, antibiotics, and their metabolites were the main factors causing acute toxicity [35]. As shown in Figure 2, the acute toxicity was detected in sample SW17 collected from the river adjacent to the goose farm, with an inhibition rate of 25%. The geese excrement during the period of geese stocking along the river would directly enter the river, and the wastes produced during the period of greenhouse breeding might also leak into the river. This suggests that large-scale livestock and poultry stocking or unregulated breeding may cause toxic substances to leak into the water body, thus generating water toxicity. However, this acute toxicity might be diluted by the river water or eliminated by the water body self-purification, because samples SW18 and SW19 collected downstream of the goose farm showed no obvious inhibition of luminescent bacteria.

3.2. Estrogenic Toxicity Assessment by YES

The collected water samples were tested for estrogenic effects using a YES assay, and the test results were expressed as the E2-EQ value. As shown in Figure 3, the E2-EQ value of sample SW20 collected from River 1 was 18.8 ng L⁻¹, which might have been related to the farmland and livestock and poultry breeding enterprises distributed on both sides of the river. No estrogenic activity was detected in sample WSL collected from the water supply reservoir on River 1. Referring to the latest Japanese drinking water quality standard, which stipulated that the concentration limit of 17β-estradiol was 80 ng L⁻¹, the water supply source in this area met the requirements on estrogen indicators of drinking water. The E2-EQ value of the seven samples from the main stream of River 2 was 6.6~25.2 ng L⁻¹, and the E2-EQ value of the eight samples from the tributaries was 6.6~39.6 ng L⁻¹. The estrogenic activity of the tributaries of River 2 was higher than that of the main stream. To analyze the estrogenic activity correlation between the main stream and the tributaries, further material analysis was required. Compared with the estrogenic activity in the lower Yellow River Basin reported by other studies, the activity in this study was relatively high; the highest estrogen activity was 0.21 ng L⁻¹ in the lower reaches of Tongguan of the Yellow River in rainy season [36]; while the estrogen activity in Zhengzhou basin of the Yellow River was 0.72~1.19 ng L⁻¹ [37]. However, it should be noted that the sampling sites reported above were located in the main stream of the Yellow River and not in rural areas, and there have been few reports on environmental surface water in the rural basin of the Yellow River.

In addition to farmland samples, different degrees of estrogenic activity were observed in samples collected from possible pollution sources. Strong estrogenic activities were observed in the livestock and poultry fecal sewage samples WPS3 and WPS4, with E2-EQ values of 74.1 and 32.9 ng L⁻¹, respectively. As the samples were filtered and pretreated, most of the estrogenic substances adsorbed by solids were removed, and untreated fecal sewage discharged into rivers might cause more serious pollution [38]. The strongest estrogenic activity was observed in swine urine sewage WPS5, with an E2-EQ value of 7265.2 ng L⁻¹. Previous studies have shown that human or animal urine exhibits a higher estrogenic activity than feces and that swine and poultry only excrete estrogens through urine, so compared to fecal sewage, urine sewage might contribute more environmental estrogens to rivers [39,40]. Estrogenic activity was also observed in the chemical park sample WPS1, with an E2-EQ value of 44.4 ng L⁻¹. However, compared with urban areas, the industrial enterprises in the rural area were relatively undeveloped, and were very concentrated in the study area, so the contribution of industrial production to the estrogenic activity in the surface water was low. The number of livestock and poultry breeding enterprises in the study area was extremely high, the locations were scattered, and their waste or sewage could enter surface water in many ways, such as runoff or seepage, accompanied by high concentrations of environmental estrogens [41,42,43]. The samples collected adjacent to the goose farm proved this to a certain extent. The E2-EQ value of sample SW16 collected upstream of the goose farm was only 4.7 ng L⁻¹, while the E2-EQ value of samples SW17 and SW18 collected from the farm and downstream were 131.0 ng L⁻¹ and 77.5 ng L⁻¹, respectively. The geese in the farm were free-ranging by the river, during which their excreta would enter the riverside land or directly into the river, and it was also found that the sewage in the breeding shed was irregularly discharged into the river through pipelines during sampling; all these factors led to an increase in the estrogenic activity of the river water, which also affected the downstream waters. The E2-EQ value of sample SW19 collected from further downstream was 4.9 ng L⁻¹, indicating that the self-purification and dilution of the water effectively reduced the content of environmental estrogens in the water body.

In addition, the E2-EQ value of 7.9 ng L⁻¹ was detected in the effluent sample WPS6 of the sewage treatment plant, indicating that the sewage still contained some environmental estrogens after treatment, which was also reported in previous studies [44]. Although the situation of other sewage treatment plants in the study area might be different, this was still one of the possible factors that might have caused more environmental estrogens to enter the surface water, especially the sewage with high levels of environmental estrogens.

3.3. Neurotoxicity Assessment

Neurotoxicity was assessed using AChE activity, and the results were expressed as PT-EQ value. Among all the samples collected in River 1 and River 2, only sample WPS1 collected from the chemical industrial park was observed with neurotoxicity, with a PT-EQ value of 4.0 μg L⁻¹, as shown in Figure 4. AChE can be inhibited by many substances, such as organophosphorus pesticides, organochlorine pesticides, heavy metals, bisphenol-based compounds, etc. [45,46,47]. For rural areas, the pesticides and herbicides used in agricultural production activities are the main sources of neurotoxic substances. However, according to the test results, no neurotoxicity was detected in sample WPS2 collected from a farmland ditch and the surface water sample SW16 collected near farmland. The pesticides and herbicides used in local agricultural activities were mainly emamectin benzoate, bifenthrin, acetamiprid, and propisochlor, which would not produce neurotoxicity within the specified range [48,49,50]. There were many factories in the chemical industrial park, such as a fertilizer factory and wig factory, whose wastes such as heavy metals or dyes would cause inhibition of AChE, which might explain the neurotoxicity detected in WPS1; however, no neurotoxicity was detected in samples SW9 and SW10 collected downstream of the chemical industrial park, indicating that the sewage or rainwater of the chemical industrial park did not cause neurotoxicity in the surface water.

3.4. Genotoxicity Assessment

The genotoxicity of the samples was determined using the cell cycle checkpoint kinase CHK1 as a biomarker. The occurrence of genotoxicity was reflected by the up-regulation of CHK1, and the determined result was expressed as the PELI. As shown in Figure 5, no genotoxicity was detected in the samples collected from River 1 and the main stream of River 2. Genotoxicity was detected in five of the eight samples collected from the tributaries of River 2, SW11–SW15, with PELI ranging from 1.56 to 6.06, all of which were downstream of River 2. Compared with the middle reaches flowing through relatively developed counties, the downstream basin was composed of several large villages. More extensive pollution sources and the relatively basic sewage collection and treatment facilities made the surface water more prone to toxicity.

For samples collected from possible pollution sources, only the rainwater sample WPS1 collected from the chemical industrial park had genotoxicity, with a PELI of 5.40; the surface water samples SW9 and SW10 collected downstream of the park had no detected genotoxicity, indicating that the sewage and rainwater in the park did not cause genotoxicity risks to the relevant rivers. Some herbicides and insecticides used in agricultural activities can cause genotoxicity, so the impact of farmland planting requires a follow-up investigation. In addition, livestock and poultry breeding was an important source of hormone substances, and some also showed genotoxicity. Both endogenous and exogenous hormones can show genotoxicity, manifested in causing DNA damage, which has been confirmed in previous studies [51,52,53,54]. Although the livestock and poultry sewage contained high concentrations of hormone substances, no genotoxicity was detected in the collected sewage samples.

The discharge of domestic sewage was also a possible source of genotoxicity. The sewage collection and treatment facilities in the area are not perfect, resulting in untreated domestic sewage being directly discharged into the river, and there was also an accumulation of garbage. In previous reports, domestic sewage showed the same genotoxicity as mixed sewage containing industrial sewage [55], and the liquid produced by domestic wastes can also cause serious genotoxicity [56]. In addition, only one marker, CHK1, was selected in this study. CHK1 is activated for DNA repair after DNA damage, which could reflect general DNA damage [24]. Its upregulation indicated that genotoxicity existed, but there are many DNA damage and repair pathways, and judgment by CHK1 alone might lead to false negatives [57]. Therefore, the genotoxicity test method and sample test results used in this paper were not sufficient to make a complete assessment and judgment of the surface water and pollution sources in the study area, and follow-up research is needed.

3.5. Risk Assessment

The ecological risks for estrogen activity and neurotoxicity, as well as the degree of toxicity and acute toxicity were assessed for all 27 samples. According to the calculated RQ value, the environmental risks were divided into four levels: no risk (0 ≤ RQ < 0.1), low risk (0.1 ≤ RQ < 1), medium risk (1 ≤ RQ < 10), and high risk (RQ ≥ 10) [58]. According to the inhibition of luminescent bacteria, acute toxicity was divided into four levels: no toxicity (0 ≤ inhibition rate < 10%), low toxicity (10% ≤ inhibition rate < 20%), medium toxicity (20% ≤ inhibition rate < 50%), and high toxicity (50% ≤ inhibition rate ≤ 100%). The RQ values and inhibition rate of water samples in the rural area are shown in Figure 6.

Among the surface water samples, the sample from River 1 had a medium risk for aquatic organisms from estrogen activity, but the sample collected from the water source reservoir had no risk. The ecological risk to aquatic organisms of River 2 was mainly from estrogenic activity. Three of the seven samples collected from main stream were high risk, with RQ values ranging from 11.1 to 12.6, and the rest were medium risk. Six of the twelve samples collected from the tributary were high risk, with RQ values ranging from 11.1 to 65.5, and the rest were medium risk. Compared with the main stream, the tributary had more high risk points and higher risks, which might be related to the small water volume and slow flow rate, making the concentration of toxic substances relatively higher and more easily enriched. According to previous reports, the reproduction and development of aquatic organisms were affected by low concentrations of estrogens. For example, a decrease in egg production was observed in zebrafish exposed to 10 ng L⁻¹ ethinylestradiol [59], and female grayling exposed to 1 ng L⁻¹ 17β-estradiol were shown to ovulate earlier [60]. In this paper, the main risk to surface water in the study area was found to be estrogen risk, through the assessment of ecological risk for the most sensitive aquatic organisms. In addition, both industrial rainwater and livestock sewage showed a high risk of estrogenic activity, indicating that both industrial production and livestock and poultry breeding were important sources of environmental estrogen in the surface water. There were many livestock and poultry breeding enterprises in the study area, being widely spaced in the River 1 and River 2 basins, especially on the tributaries. Surface water samples adjacent to livestock and poultry breeding sources, such as SW13, SW17, and SW18, all showed a high risk, indicating that the contribution of livestock and poultry breeding to environmental estrogen in the surface water might be significant. In view of the high risk of estrogenic activity in surface water in rural areas, appropriate measures should be taken to reduce environmental risks, such as strengthening the supervision of wastewater treatment and sewage discharge in production activities, reducing the occurrence of illegal discharge, and reducing environmental estrogen substances used in production activities, especially livestock and poultry breeding.

4. Conclusions

In this study, a multi-endpoint toxicity analysis was carried out for environmental water bodies and sewage from possible pollution sources in rural areas of the lower Yellow River basin. Of the two rivers studied, River 1, which was the water source river, had no obvious toxicity risk; while River 2, which flowed through most of the region, presented different degrees of acute toxicity, estrogenic activity, and genotoxicity risks. The estrogenic activity of River 2 was generally high, with E2-EQ values ranging from 4.8 to 131.0 ng·L⁻¹, and both acute toxicity and genotoxicity occurred in the tributary of River 2. Untreated industrial sewage was considered to be one of the sources of acute toxic substances, domestic sewage was inferred to have contribution to the acute toxicity and genotoxicity, and livestock and poultry breeding sewage was found to be the source of acute toxicity and estrogenic activity, especially for the environmental estrogens in surface water. Through a risk assessment, most samples showed high risk of estrogenic activity, and the risk in tributaries was higher than that of the main stream. This study provides a reference for subsequent research on water bodies in rural areas.