ABSTRACT
Water treatment infrastructure facilities play an important role in ensuring drinking water safety. A survey of the drinking water treatment process was conducted in the urban peripheral areas of Beijing, China, and the impact of the main water treatment infrastructure facilities on microbial contamination was investigated. Sedimentation equipment, filtration facilities, and disinfection equipment were all significantly correlated with the concentration of heterotrophic plate counts in drinking water. The filtration facilities and disinfection equipment were also positively correlated with the concentrations of total coliform and Escherichia coli. The removal of microorganisms by different water disinfection methods gradually decreased in the order of ozone, chlorine, chlorine dioxide, and ultraviolet light. The effects of microbial contamination removal of different water pumping methods were as follows: direct water supply > pressure tank > secondary pressing pump station > water tower > high-level water tank, and the removal effects were 7.6, 7.4, 4.1, 3.6, and 1.7 times that of the self-flowing water supply. This study provides scientific support for the renovation and upgrading of microbial pollution reduction in drinking water in rural areas of the urban periphery.
HIGHLIGHTS
The drinking water treatment process and the main water treatment infrastructure facilities regarding their influence on microbial contamination were investigated.
The effect of microbial contamination removal using diverse water disinfection methods was discussed.
The relationship between the different water pumping methods and microbial contamination was also explored.
INTRODUCTION
The safety of drinking water in rural areas is an important indicator that reflects socioeconomic development and quality of life in the countryside (Nowicki et al. 2020). Numerous studies have demonstrated that microbial contamination is the primary factor that influences the quality of rural drinking water (Ashbolt 2015; Favere et al. 2021; WHO 2022). The main path of rural drinking water throughout the production technology is: water source – waterworks – user. Water treatment in waterworks has traditionally been a multistage process, such as coagulation–flocculation–settling (referred to as sedimentation), filtration, disinfection, and pumping (Thom et al. 2022). All these processes influence microbial contamination in drinking water.
Numerous studies have been conducted on the impact of drinking water treatment processes on microorganisms, primarily focusing on large waterworks in cities or small waterworks in remote rural areas (Yates 2019; Wysowska et al. 2021; Timmers et al. 2024). Nevertheless, the waterworks in the rural areas of the urban periphery are quite different from those in large-scale waterworks and remote rural areas. Compared to large-scale waterworks in cities, the scale of drinking water treatment in these areas is smaller, and the length of their water supply networks is relatively shorter. Furthermore, owing to the higher presence of high-rise buildings in urban peripheral areas, the scale of drinking water pumping in waterworks is larger than that in remote rural areas. Therefore, it is necessary to conduct research on the impact of the drinking water treatment process on microorganisms in rural waterworks of urban peripheral areas.
As the capital and national central city of China, Beijing has experienced rapid urbanization in recent years. In 2022, there were 68 urban public waterworks which supplied the majority of the urban population. In addition, there were 3,271 village waterworks and 472 decentralized water supply projects that provided water to residents in the rural areas surrounding the Beijing urban region (Beijing Municipal Water Authority 2023). Most urban public waterworks are large-scale waterworks. In addition to coagulation and sedimentation, filtration, and disinfection, advanced water treatment technologies, including ozone and activated carbon, have been incorporated into water purification processes in large-scale waterworks. Village waterworks and decentralized water supply projects are typically small-scale waterworks, and most of their water treatment processes are still conventional, including coagulation, sedimentation, filtration, and disinfection. Furthermore, some waterworks applied simply water treatment methods that only utilize a part of conventional water treatment processes, such as sedimentation and disinfection, filtration and disinfection, or only disinfection. A survey conducted by Ye et al. (2013) revealed that in rural areas surrounding the Beijing urban region, there were nearly 4,000 water wells, all of which were used for public water supply. Approximately 3.59 million rural inhabitants relied on these small centralized water supplies. Selecting the impact of water treatment processes on microorganisms in rural waterworks in Beijing urban peripheral areas as a pilot study has a certain reference significance for other urban peripheral areas and rural regions. To analyze the impact of the water treatment process on microorganisms, a survey of the drinking water treatment process was conducted in the peripheral urban areas of Beijing, China. Data on the entire water treatment process in rural areas were collected, and microbial indicators in drinking water were determined.
MATERIALS AND METHODS
Study area
Data analysis
The evaluation of microbial indicators in drinking water was based on the Standard for Drinking Water Quality of China. There are three microbial indicators in this standard: HPC, total coliforms, and E. coli. The limits for total coliforms and E. coli are none detectable in 100 mL, respectively. The limit for HPC is less than 100 CFU·mL−1. Therefore, if either of these three items fails to meet the standard requirements, this implies that the microorganism does not conform to the standard. The microbial indicators that met the standard were displayed digitally for easy analysis, with 1 representing compliance with the standard and 2 indicating noncompliance. Likewise, the water treatment process and infrastructure facilities, such as sedimentation equipment, filtration facilities, filtration methods, disinfection equipment, disinfection methods, and pumping methods, were applied with classified variables for digitization. The classifications were as follows: sedimentation equipment (1 = sedimentation equipment, 2 = without sedimentation equipment), filtration facility (1 = filtration facility, 2 = without filtration facility), disinfection equipment (1 = disinfection equipment, 2 = without disinfection equipment or disinfection equipment but not in use), filtration methods (1 = integrated filter tank, 2 = filter bed), disinfection methods (1 = ozone, 2 = chlorine dioxide, 3 = chlorine, 4 = ultraviolet light, 5 = no disinfection), and pumping methods (1 = direct water supply, 2 = secondary pressing pump station, 3 = high-level water tank, 4 = water tower, 5 = pressure tank, and 6 = self-flowing water supply).
Spearman correlation analysis was used to explore the relationship between the different water treatment processes described earlier and the microbial concentrations of HPC, total coliforms, and E. coli. The general loglinear model analysis was adopted to analyze the effect of disinfection and pumping methods on microbial contamination.
RESULTS AND DISCUSSION
Information on the water treatment process in the survey areas
The microorganism qualification rate of all the surveyed water samples was 72%, among which HPC had a qualification rate of 73%, total coliform was 88%, and E. coli was 97%. These qualification rates were higher than those of rural drinking water in Beijing in 2013 (Ye et al. 2013) but lower than those of the Beijing urban area during the same period. The qualification rates were 100% within the Beijing urban area. The positive rates of total coliforms and E. coli were 12 and 3%, respectively. The total coliform positive rate was somewhat lower than that of the private and small public well water samples in Canada and the United States, whereas the E. coli positive rate was higher than that of these samples (Invik et al. 2017; Morris et al. 2022). This indicates that after years of rural water improvement projects, the drinking water quality in the rural areas of Beijing has significantly improved, but there is still a gap in the drinking water quality between the rural and urban areas.
Filtration methods . | Number of waterworks (%) . | Disinfection methods . | Number of waterworks (%) . | Pumping methods . | Number of waterworks (%) . |
---|---|---|---|---|---|
Integrated filter tank | 123 (57%) | Ozone | 74 (15%) | Direct water supply | 200 (39%) |
Filter bed | 91 (43%) | Chlorine dioxide | 60 (12%) | Secondary pressing pump station | 76 (15%) |
Chlorine | 53 (10%) | High-level water tank | 96 (19%) | ||
Ultraviolet light | 4 (1%) | Water tower | 16 (3%) | ||
No disinfection | 316 (62%) | Pressure tank | 29 (6%) | ||
Self-flowing water supply | 90 (18%) |
Filtration methods . | Number of waterworks (%) . | Disinfection methods . | Number of waterworks (%) . | Pumping methods . | Number of waterworks (%) . |
---|---|---|---|---|---|
Integrated filter tank | 123 (57%) | Ozone | 74 (15%) | Direct water supply | 200 (39%) |
Filter bed | 91 (43%) | Chlorine dioxide | 60 (12%) | Secondary pressing pump station | 76 (15%) |
Chlorine | 53 (10%) | High-level water tank | 96 (19%) | ||
Ultraviolet light | 4 (1%) | Water tower | 16 (3%) | ||
No disinfection | 316 (62%) | Pressure tank | 29 (6%) | ||
Self-flowing water supply | 90 (18%) |
Effects of sedimentation, filtration, disinfection equipment, and filtration methods on microorganisms
As shown in Table 2, the outcomes of the Spearman correlation analysis revealed that the treatment process exerted a considerable influence on the microbial contamination of drinking water, including sedimentation equipment, filtration facilities, filtration methods, and disinfection facilities. The results demonstrated that the sedimentation equipment, filtration facilities, and disinfection equipment were all significantly correlated with HPC (p = 0.002, 0.000, 0.000, and 0.009, respectively). Moreover, the filtration facilities and disinfection equipment were also positively correlated with the total coliforms (p = 0.001 and 0.000) and E. coli (p = 0.004 and 0.013, respectively), while the sedimentation equipment and filtration methods were not significantly correlated with total coliforms (p = 0.502 and 0.209) and E. coli (p = 0.445 and 0.099, respectively).
Indicator . | Correlation parameter . | Sedimentation equipment . | Filtration facility . | Disinfection equipment . | Filtration methods . |
---|---|---|---|---|---|
Heterotrophic plate counts | r | 0.139 | 0.189 | 0.225 | 0.179 |
p | 0.002 | 0.000 | 0.000 | 0.009 | |
Total coliform | r | −0.030 | 0.151 | 0.205 | 0.086 |
p | 0.502 | 0.001 | 0.000 | 0.209 | |
Escherichia coli | r | 0.034 | 0.013 | 0.110 | 0.113 |
p | 0.445 | 0.004 | 0.013 | 0.099 |
Indicator . | Correlation parameter . | Sedimentation equipment . | Filtration facility . | Disinfection equipment . | Filtration methods . |
---|---|---|---|---|---|
Heterotrophic plate counts | r | 0.139 | 0.189 | 0.225 | 0.179 |
p | 0.002 | 0.000 | 0.000 | 0.009 | |
Total coliform | r | −0.030 | 0.151 | 0.205 | 0.086 |
p | 0.502 | 0.001 | 0.000 | 0.209 | |
Escherichia coli | r | 0.034 | 0.013 | 0.110 | 0.113 |
p | 0.445 | 0.004 | 0.013 | 0.099 |
The findings indicated that sedimentation, filtration, and disinfection treatments all contribute to the reduction of HPC in drinking water, and the filter bed within the filtration methods is more effective in reducing HPC in water compared to the integrated filter tank. Filtration and disinfection treatments also had a significant impact on the removal of total coliforms and E. coli. These results are consistent with those of previous studies. Wysowska et al. (2021) revealed that initial coagulation in settling tanks and filtration through sand filters can significantly remove indicator microbes. Chowdhury et al. (2013) discovered that coagulation–flocculation–settling and filtration are capable of transforming pollutants like suspended matter, colloid, and some bacteria adsorbed on the surface of the water into precipitates and removing them via filtration. This study additionally revealed that there was no substantial variance in the removal of total coliforms and E. coli in drinking water, regardless of sedimentation. This is because conventional sedimentation removes only a small segment of total coliforms and E. coli, and the reproduction capability of total coliforms and E. coli is very strong (Oakley 2018). The effect of the removal of total coliforms and E. coli through the filtration method was not significant, which can be attributed to the fact that despite the filtration bed possessing a larger filtration area compared with the integrated filtration tank (de Souza et al. 2021), the positive rate of total coliforms and E. coli was lower, and the size of the filtration area did not impact its removal rate.
Relationship between water disinfection methods and microbial contamination
Effective disinfection of drinking water can control microbial contamination. There are numerous methods of disinfection in rural waterworks. In this study, a general loglinear model analysis was applied to systematically analyze the relationship between disinfection methods and microorganisms in drinking water to assess the influence of diverse disinfection methods on microorganisms. As shown in Table 3, the results indicate that the effect of different water disinfection methods is gradually reduced by ozone, chlorine, chlorine dioxide, and ultraviolet light. The sterilization effects of these disinfection methods on microorganisms are 5.8, 1.9, 1.6, and 0.3 times that of no disinfection, respectively. Compared with no disinfection, chlorine dioxide, and ultraviolet light disinfection showed no significant difference in the removal of microbial contamination.
Parameter . | Estimate . | Std. error . | Z . | Sig. . | 95% Confidence interval . | |
---|---|---|---|---|---|---|
Lower bound . | Upper bound . | |||||
Constant | 4.732 | 0.094 | 50.411 | 0.000 | 4.548 | 4.916 |
[microorganism = 1] | 0.584 | 0.117 | 4.984 | 0.000 | 0.354 | 0.813 |
[microorganism = 2] | 0a | |||||
[disinfection method = 1] | −2.860 | 0.403 | −7.091 | 0.000 | −3.650 | −2.070 |
[disinfection method = 2] | −1.991 | 0.271 | −7.352 | 0.000 | −2.522 | −1.460 |
[disinfection method = 3] | −2.206 | 0.298 | −7.403 | 0.000 | −2.790 | −1.622 |
[disinfection method = 4] | −3.816 | 0.639 | −5.969 | 0.000 | −5.068 | −2.563 |
[disinfection method = 5] | 0a | |||||
[microorganism = 1] * [disinfection method = 1] | 1.771 | 0.427 | 4.150 | 0.000 | 0.935 | 2.608 |
[microorganism = 1] * [disinfection method = 2] | 0.493 | 0.317 | 1.557 | 0.119 | −0.127 | 1.113 |
[microorganism = 1] * [disinfection method = 3] | 0.616 | 0.343 | 1.795 | 0.053 | −0.057 | 1.289 |
[microorganism = 1] * [disinfection method = 4] | −0.584 | 0.902 | −0.647 | 0.517 | −2.351 | 1.184 |
[microorganism = 1] * [disinfection method = 5] | 0a | |||||
[microorganism = 2] * [disinfection method = 1] | 0a | |||||
[microorganism = 2] * [disinfection method = 2] | 0a | |||||
[microorganism = 2] * [disinfection method = 3] | 0a | |||||
[microorganism = 2] * [disinfection method = 4] | 0a | |||||
[microorganism = 2] * [disinfection method = 5] | 0a |
Parameter . | Estimate . | Std. error . | Z . | Sig. . | 95% Confidence interval . | |
---|---|---|---|---|---|---|
Lower bound . | Upper bound . | |||||
Constant | 4.732 | 0.094 | 50.411 | 0.000 | 4.548 | 4.916 |
[microorganism = 1] | 0.584 | 0.117 | 4.984 | 0.000 | 0.354 | 0.813 |
[microorganism = 2] | 0a | |||||
[disinfection method = 1] | −2.860 | 0.403 | −7.091 | 0.000 | −3.650 | −2.070 |
[disinfection method = 2] | −1.991 | 0.271 | −7.352 | 0.000 | −2.522 | −1.460 |
[disinfection method = 3] | −2.206 | 0.298 | −7.403 | 0.000 | −2.790 | −1.622 |
[disinfection method = 4] | −3.816 | 0.639 | −5.969 | 0.000 | −5.068 | −2.563 |
[disinfection method = 5] | 0a | |||||
[microorganism = 1] * [disinfection method = 1] | 1.771 | 0.427 | 4.150 | 0.000 | 0.935 | 2.608 |
[microorganism = 1] * [disinfection method = 2] | 0.493 | 0.317 | 1.557 | 0.119 | −0.127 | 1.113 |
[microorganism = 1] * [disinfection method = 3] | 0.616 | 0.343 | 1.795 | 0.053 | −0.057 | 1.289 |
[microorganism = 1] * [disinfection method = 4] | −0.584 | 0.902 | −0.647 | 0.517 | −2.351 | 1.184 |
[microorganism = 1] * [disinfection method = 5] | 0a | |||||
[microorganism = 2] * [disinfection method = 1] | 0a | |||||
[microorganism = 2] * [disinfection method = 2] | 0a | |||||
[microorganism = 2] * [disinfection method = 3] | 0a | |||||
[microorganism = 2] * [disinfection method = 4] | 0a | |||||
[microorganism = 2] * [disinfection method = 5] | 0a |
aThis parameter is set to zero because it is redundant.
Model: Poisson.
Design: microorganism + disinfection method + microorganism * disinfection method.
Chlorine compounds are common chemicals often used as oxidizing agents for drinking water disinfection and water treatment. The popularity of chlorine is not only due to its lower cost but also due to its higher oxidizing potential, which provides a minimum level of residual chlorine throughout the distribution system and protects against microbial recontamination with a remarkable residual effect (Drogui & Daghrir 2015). This also explains the significant positive correlation (p = 0.053) between microbial removal with and without chlorine disinfection in this study. The ozone disinfection capacity was 5.8 times that of the nondisinfection. This is because ozone has the strongest disinfectant effect because it has the strongest oxidizing power among several disinfectants, and it can remove some chlorine-resistant bacteria from drinking water (Ding et al. 2019). The chlorine dioxide disinfection capacity was 1.9 times that of nondisinfection. This further confirms that chlorine dioxide is a strong oxidant, microbicide, and preservative (Jefri et al. 2022). Chlorine dioxide decomposes easily in water, making it difficult to optimize the operation control of the generator, often resulting in the failure of continuous disinfection and regrowth of microorganisms. It has been reported that UV light can effectively remove indicator organisms, but this disinfectant cannot guarantee complete protection from pathogens at specified doses, and microbial growth after disinfection is a major limitation of UV light (Shekhawat et al. 2021). Microbial regrowth caused chlorine dioxide and UV light disinfection to have no significant effect on microbial inactivation in this study (p = 0.119 and 0.517, respectively).
Relationship between water pumping methods and microbial contamination
Because of the variations in the distance from the waterwork to the residential house, height of the building, and water pressure of the water distribution network, the pumping methods will also be different. The general loglinear model analysis was also used to systematically analyze the relationship between pumping methods and microorganisms in rural drinking water. As shown in Table 4, the results revealed that the effect of microbial contamination control on different water pumping methods gradually decreased in the order of direct water supply, pressure tank, secondary pressing pump station, water tower, and high-level water tank. The microbial contamination control effects of these pumping methods are 7.6, 7.4, 4.1, 3.6, and 1.7 times that of the self-flowing water supply.
Parameter . | Estimate . | Std. error . | Z . | Sig. . | 95% Confidence interval . | |
---|---|---|---|---|---|---|
Lower bound . | Upper bound . | |||||
Constant | 3.942 | 0.139 | 28.286 | 0.000 | 3.668 | 4.215 |
[microorganism = 1] | −0.265 | 0.212 | −1.254 | 0.210 | −0.680 | 0.149 |
[microorganism = 2] | 0a | |||||
[pumping method = 1] | −0.557 | 0.231 | −2.413 | 0.016 | −1.010 | −0.105 |
[pumping method = 2] | −1.024 | 0.271 | −3.777 | 0.000 | −1.555 | −0.493 |
[pumping method = 3] | −0.192 | 0.207 | −0.927 | 0.354 | −0.598 | 0.214 |
[pumping method = 4] | −2.438 | 0.492 | −4.959 | 0.000 | −3.401 | −1.474 |
[pumping method = 5] | −2.438 | 0.492 | −4.959 | 0.000 | −3.401 | −1.474 |
[pumping method = 6] | 0a | |||||
[microorganism = 1] * [pumping method = 1] | 2.025 | 0.291 | 6.969 | 0.000 | 1.456 | 2.595 |
[microorganism = 1] * [pumping method = 2] | 1.417 | 0.340 | 4.161 | 0.000 | 0.749 | 2.084 |
[microorganism = 1] * [pumping method = 3] | 0.514 | 0.294 | 1.746 | 0.081 | −0.063 | 1.091 |
[microorganism = 1] * [pumping method = 4] | 1.287 | 0.589 | 2.185 | 0.029 | 0.132 | 2.441 |
[microorganism = 1] * [pumping method = 5] | 2.000 | 0.553 | 3.614 | 0.000 | 0.915 | 3.084 |
[microorganism = 1] * [pumping method = 6] | 0a | |||||
[microorganism = 2] * [pumping method = 1] | 0a | |||||
[microorganism = 2] * [pumping method = 2] | 0a | |||||
[microorganism = 2] * [pumping method = 3] | 0a | |||||
[microorganism = 2] * [pumping method = 4] | 0a | |||||
[microorganism = 2] * [pumping method = 5] | 0a | |||||
[microorganism = 2] * [pumping method = 6] | 0a |
Parameter . | Estimate . | Std. error . | Z . | Sig. . | 95% Confidence interval . | |
---|---|---|---|---|---|---|
Lower bound . | Upper bound . | |||||
Constant | 3.942 | 0.139 | 28.286 | 0.000 | 3.668 | 4.215 |
[microorganism = 1] | −0.265 | 0.212 | −1.254 | 0.210 | −0.680 | 0.149 |
[microorganism = 2] | 0a | |||||
[pumping method = 1] | −0.557 | 0.231 | −2.413 | 0.016 | −1.010 | −0.105 |
[pumping method = 2] | −1.024 | 0.271 | −3.777 | 0.000 | −1.555 | −0.493 |
[pumping method = 3] | −0.192 | 0.207 | −0.927 | 0.354 | −0.598 | 0.214 |
[pumping method = 4] | −2.438 | 0.492 | −4.959 | 0.000 | −3.401 | −1.474 |
[pumping method = 5] | −2.438 | 0.492 | −4.959 | 0.000 | −3.401 | −1.474 |
[pumping method = 6] | 0a | |||||
[microorganism = 1] * [pumping method = 1] | 2.025 | 0.291 | 6.969 | 0.000 | 1.456 | 2.595 |
[microorganism = 1] * [pumping method = 2] | 1.417 | 0.340 | 4.161 | 0.000 | 0.749 | 2.084 |
[microorganism = 1] * [pumping method = 3] | 0.514 | 0.294 | 1.746 | 0.081 | −0.063 | 1.091 |
[microorganism = 1] * [pumping method = 4] | 1.287 | 0.589 | 2.185 | 0.029 | 0.132 | 2.441 |
[microorganism = 1] * [pumping method = 5] | 2.000 | 0.553 | 3.614 | 0.000 | 0.915 | 3.084 |
[microorganism = 1] * [pumping method = 6] | 0a | |||||
[microorganism = 2] * [pumping method = 1] | 0a | |||||
[microorganism = 2] * [pumping method = 2] | 0a | |||||
[microorganism = 2] * [pumping method = 3] | 0a | |||||
[microorganism = 2] * [pumping method = 4] | 0a | |||||
[microorganism = 2] * [pumping method = 5] | 0a | |||||
[microorganism = 2] * [pumping method = 6] | 0a |
aThis parameter is set to zero because it is redundant.
Model: Poisson.
Design: constant + microorganism + pumping method + microorganism * pumping method.
The water from the self-flowing water supply remains for several hours, along with residual chlorine attenuation, temperature increase, and microbial growth. The direct water supply generally employs an inverter constant-pressure water supply or nonnegative-pressure water supply equipment (Jiang et al. 2011). A well-pressure tank serves as a water storage reservoir that aids in maintaining water pressure at a constant level (USEPA 2006). The significant control effects on microbial contamination of the direct water supply and pressure tank result in their operation in a completely sealed manner, ensuring that the water quality is not subject to secondary pollution. The ground elevations of pipeline networks in rural areas vary significantly. The early solution for high-rise water supply was to install a water tower or high-level water tank on the top of the building, with fill pumps at the bottom of the building, forming a simple gravity down feed arrangement (Ramana et al. 2015). The establishment of secondary pressure pump stations in the water network is another approach to guarantee a more uniform water supply pressure in all sections of the pipeline network in rural areas (Cheung et al. 2013). During the pumping processes of the secondary pressure pump station, water tower, and high-level water tank, chlorine and other disinfectants are generally added to ensure minimum residual disinfectant content in the water supply system and prevent secondary pollution caused by microbial growth.
CONCLUSIONS
The treatment process exerts a significant influence on microbial contamination of drinking water in rural areas on the urban periphery. Sedimentation, filtration, and disinfection treatments all contributed to the reduction of HPC in drinking water. Both filtration and disinfection treatments also had a significant impact on the removal of total coliforms and E. coli. The effects of different water disinfection methods gradually decreased in the order of ozone, chlorine, chlorine dioxide, and ultraviolet light. Compared with self-flowing water supply, the pumping methods such as direct water supply, pressure tank, secondary pressing pump station, water tower, and high-level water tank ensure the supply of high-rise water in rural areas on the urban periphery, ensure the residual level of disinfectant in the water supply system to protect against microbial contaminants, and limit the growth of microbial within the distribution system. However, improper operation, maintenance, and inadequate handling of sedimentation equipment, filtration facilities, and disinfection equipment can also diminish the efficiency of microbial agent removal. Therefore, selecting the most suitable water treatment method based on the characteristics of the local rural areas, and regular cleaning and routine maintenance of water treatment equipment will also contribute to the prevention and control of microbial pollution in drinking water in rural areas on the urban periphery.
ACKNOWLEDGEMENT
This study was supported by grants from the Major Science and Technology Project of Water Pollution Control and Management in China (no. 2018ZX07502001). The authors are grateful to all the participants and the local staff of the study areas for their participation in this study.
DATA AVAILABILITY STATEMENT
All relevant data are included in the paper or its Supplementary Information.
CONFLICT OF INTEREST
The authors declare there is no conflict.