Abstract

Epikarst springs are commonly used for drinking water in karst mountainous areas, but they tend to bring health risks to residents because of their vulnerability. In this work, a modified slow sand filtration system (M-SSF) was established as a case study to purify and conserve the epikarst spring water. The outcomes indicate that the purification of M-SSF relies mainly on the adsorption and ion exchange of the filter medium (mixtures of heat-treated red clay and crushed limestone, MHRCCL) during the schmutzdecke juvenility, and on the schmutzdecke-formed food chain of pollutants → bacteria → protozoa after the schmutzdecke maturity. The closed water cellar lined with ceramic tiles could reduce the deterioration of epikarst spring water during storage. Via 16S rRNA sequencing, it was found that the high abundance of TM6_Dependentiae in purified epikarst spring water (PESW) suggested that the M-SSF system relies on the formation of a closed food chain to achieve effective water purification. The decrease of Pseudarcicella abundance in PESW indicated that M-SSF could effectively prevent the water quality from external influences represented by leeches. Besides, the 16S function prediction was used to qualitatively characterize microbial nitrogen metabolism, as well as organic matter degradation in water purification.

HIGHLIGHTS

  • The quality of epikarst spring water was significantly improved using the M-SSF.

  • TM6_Dependentiae indicated that the purification of M-SSF relies on the establishment of a closed food chain.

  • The M-SSF could prevent the purified epikarst spring water from being influenced by external pollution.

Graphical Abstract

Graphical Abstract
Graphical Abstract

INTRODUCTION

Karst mountainous areas are typically characterized by a large topographic relief, deep cutting depth, and rapid runoff of surface water and groundwater. Due to topographic and geological characteristics, karst mountainous areas tend to have extremely poor conditions for the construction of reservoirs, resulting in severe losses of water resources (Jiang et al. 2008; Wu et al. 2009; He et al. 2018). In this way, residents often experience water shortages in their daily lives, despite the moist climatic conditions of karst mountainous areas (Wang et al. 2005). As an important regulation and storage system, the epikarst zone can increase infiltration recharge and also delay the loss of water resources in a karst system after rainfall and, therefore, has significant functions for maintaining local water availability (Trček 2006; Hu et al. 2015; Champollion et al. 2018). Furthermore, epikarst springs usually have high outcropping positions and are easily exploited, as well as being widely distributed and suitable for meeting the dispersed domestic water demands, which can be valuable water supply sources for karst mountainous areas (Williams 2008).

In most cases, epikarst springs have shallow circulation depths (around 10 m). Their catchments typically have small areas and are covered by a thin soil layer with poor continuity, and a considerable portion of the bedrock is typically exposed (Li et al. 2007). Therefore, epikarst springs are very sensitive to environmental changes, such as rainfall and changes in land-use types, as demonstrated by a high vulnerability (Bakalowicz 2004; Pipan et al. 2018). Importantly, as cultivated land resources are generally scarce in karst mountainous areas, epikarst spring catchments are usually relatively fertile areas where human activities are concentrated, thus increasing the risk of pollution in epikarst spring water. The reverse succession of karst habitat is not conducive to inhibiting the migration of pollutants to the epikarst springs (Sun et al. 2019). The soil erosion also aggravates the rocky desertification of epikarst spring catchments, increasing the health risk of epikarst spring water (Zhao et al. 2018).

In addition, epikarst springs show notable seasonality so that water cellars are typically constructed for epikarst springs to store water (Ren et al. 2018). However, epikarst spring water usually directly enters these constructed water cellars without prior treatment, and the water quality further deteriorates during the storage process. Affected by agricultural and domestic pollution sources, water contamination primarily manifests as nitrate, turbidity, and microbial indicators exceed the guidelines for drinking water quality (WHO).

As one of the earliest water treatment processes, slow sand filtration (SSF) has aroused renewed interest in its application over the past three decades due to its lack of need for chemical additives, low energy consumption, and ease of operation and maintenance (Ellis & Aydin 1995; Weber-Shirk & Dick 1999; Devadhanam Joubert & Pillay 2008; Jenkins et al. 2011; Haig et al. 2014; Lautenschlager et al. 2014; Zipf et al. 2016; Oh et al. 2018). Water purification by SSF is primarily attributed to several microbially mediated processes, such as predation, adsorption, and bio-oxidation (Haig et al. 2011, 2015; Oh et al. 2018), which mainly occur in the schmutzdecke with active microbial aggregates on the SSF surface. Conventional SSF schmutzdecke requires 20–40 days to reach maturity (D'Alessio et al. 2015; Fish et al. 2017), while SSF with juvenile schmutzdecke could not achieve an ideal treatment effect. To reduce the cost of purification materials and overcome the sub-optimal performance during schmutzdecke juvenility, the mixtures of heat-treated red clay and crushed limestone (MHRCCL) modified SSF (M-SSF) in this study were used for purifying the epikarst spring water in karst mountainous areas. A closed water cellar lined with ceramic tiles was used to store the purified epikarst spring water (PESW), aiming at preventing PESW from deterioration during storage. The high-throughput sequencing was used to study the relationship between water purification and the bacterial community structure, and then, the purification mechanism of the M-SSF system of epikarst spring water were discussed.

MATERIALS AND METHODS

Site description

The target epikarst spring is located in the east of the Xiaojiang River Basin and affiliated with Ading Village, Luxi County, Yunnan Province (Figure 1). The catchment of Ading Epikarst Spring (AES) is a typical mountainous area in the outer margin of the karst fault basin, with an altitude of 2,288 m, and has a subtropical plateau monsoon climate, with an annual average temperature of 13.0 °C and annual average rainfall of 1,200 mm. However, rainfall is extremely unevenly distributed in time, with rainfall between May and October accounting for about 85% of the total for the year. The formation of lithology in Ading Village is dominated by the limestone, dolomite, and mudstone of the Middle Triassic epoch. The burial depth of saturated water in Ading Village is more than 100 m, which is not suitable for well development and utilization (Figure 2). AES is developed in argillaceous dolomite, and its flow rate is small with obvious dynamic variations, which are approximately 0.15 L/s during the wet season and 0.08 L/s during the dry season. The soil type of AES catchment is carbonate red soil, and the soil layer is shallow with a thickness generally less than 0.3 m. The land-use types of AES catchment are villages and cultivated land, and the spring water quality is hence affected by pollution from agriculture, animal husbandry, and domestic sewage. The AES water is stored in cement-lined open water cellars, which are highly susceptible to external influences (including plant litter, animal excrement, and human pollution), and the water quality is easily deteriorated. The prior water quality test results of raw epikarst spring water (RESW) have shown that sensory indicators and microbial indicators in the water have partly exceeded the guidelines for drinking water quality (WHO) or standards for drinking Water Quality of China (GB 5749-2006) (Supplementary Table S1).

Figure 1

Karst geological conditions in the area where the AES is located. The karst geological schematic map includes part of the mountainous area and the river valley in the east of Xiaojiang basin. The black square in the topographic image shows the scope of the karst geological schematic map in Yunnan Province.

Figure 1

Karst geological conditions in the area where the AES is located. The karst geological schematic map includes part of the mountainous area and the river valley in the east of Xiaojiang basin. The black square in the topographic image shows the scope of the karst geological schematic map in Yunnan Province.

Figure 2

Schematic diagram of the karst geological profile in the area where the AES is located.

Figure 2

Schematic diagram of the karst geological profile in the area where the AES is located.

Establishment of the M-SSF system

The M-SSF system was established 1 m away from the AES (Supplementary Figure S1) during the dry season when the epikarst zone had the lowest water table (0.8 and 0.1 m during the wet season). It was lined with bricks and cement, leaving a groundwater outlet at the bottom, and ceramic tiles adhered to the inner wall. Previous studies have shown that the ceramic tile lining can decrease the hardness and total dissolved solids of cellar water, and can also reduce the attachment and growth of microorganisms compared with the cement lining (Zou et al. 2007). According to the investigation, residents' daily water consumption of AES was usually below 2 m3/day, hereby the size of the M-SSF system was designed (Figure 3). The slow sand filtration barrier (SSFB) of this system used a side colonization mode, which consisted of schmutzdecke colonization, adsorption layer, and supporting layer, and a stainless-steel cage and nylon filter cloth to be modularized for ease of cleaning and replacing. Both the supporting layer and the schmutzdecke colonization were filled with SiO2 particles (0.15–0.18 mm, analytical reagent) using for supporting and immobilizing the adsorption layer. The SiO2 particles were chemically stable and suitable for microbial colonization without affecting the water quality. The adsorption layer was comprised of the MHRCCL. Red clay was collected at a local uncontaminated woodland area (0.5–1 m) and was air-dried and milled into a size of 300-mesh, and then heat-treated at a temperature of 300 °C in a muffle furnace for 3 h under anaerobic conditions (Neytech, USA). Previous studies have shown that heat-treatment with 300 °C could sterilize both bacteria and fungi in soil effectively, and increase the specific surface area and cation exchange capacity and then enhance the adsorption capacity of clay (Duane & Wolf 1994; Guerrero et al. 2005; Chen et al. 2011). Crushed limestone was sifted within a narrow range of particle sizes (0.5–2 mm), elutriated, and then sterilized at a temperature of 300 °C for 3 h under anaerobic conditions (Neytech, USA). The grain size analyses and costs of red clay and crushed limestone are shown in Supplementary Table S2. Then, heat-treated red clay and crushed limestone were mixed thoroughly at a volume ratio of 4:7 and then compacted to a density of 1.75 g/cm3 aiming at making the adsorption layer with a reasonable permeability (≈0.2 m/h). In this condition, even under the maximum daily water consumption (2 m3/day, the head difference on both sides of SSFB is 0.83 m), the filtering velocity of SSFB in this study still meets the velocity limit range of SSF (<0.6 m/h), which could ensure the purifying effect (Zhu 2014).

Figure 3

Structural schematic diagram for the M-SSF system: (a) the profile schematic diagram of the M-SSF system and (b) the plane schematic diagram of the M-SSF system.

Figure 3

Structural schematic diagram for the M-SSF system: (a) the profile schematic diagram of the M-SSF system and (b) the plane schematic diagram of the M-SSF system.

Sample collection and testing

Based on the water quality test results of RESW both in the wet and dry seasons in 2017 (Supplementary Table S1), some indicators of RESW that had been found to pose potential health risks were monitored to research the purifying effect of the M-SSF system (Table 1). These indicators of RESW and PESW were monitored for 360 days from 26 May 2018. RESW and PESW samples (0.25 L for microbiological testing and 1.5 L for hydrochemical testing) were taken on days 2, 5, 10, 20, 40, 90, 180, and 360, and were stored in a cooler (4 °C) during transportation and then tested immediately (<24 h). Total coliforms, thermotolerant coliforms, and E. coli were enumerated using a multiple tube fermentation method, whereas heterotrophic plate counts (HPC) were assessed by a pour plate method. Additionally, the pH of these samples was measured using a SevenCompact pH meter (Mettler Toledo, Switzerland); turbidity and chromaticity were measured using a turbidity meter and a chromaticity meter, respectively (Hach, USA); nitrate was measured with an ion chromatograph (Dionex, USA); chemical oxygen demand (COD) and ammonia were measured with a spectrophotometer (Hach, USA). During the COD test, samples were digested (105 °C) in advance by a special reagent based on permanganate (Lianhua Technology Co., Ltd, China). Following digestion, samples were measured in a 30 mm cuvette with a light wavelength of 510 nm. During the measurement of ammonia, samples were pretreated with Nessler's reagent and then measured in a 10 mm cuvette at a light wavelength of 420 nm.

Table 1

Comparison of the representative water quality parameters of the PESW and the RESW on days 2, 5, 10, 20, 40, 90,180, and 360

SamplesHPCTotal coliformsThermotolerant coliformsE. coliVisible to the naked eyeChromaticityTurbiditypHCODNH3(N)
CFU/mLMPN/100 mL
°NTUmg/Lmg/Lmg/L
RESW Day 2 180 540 240 79 Russet precipitation 1.04 7.79 1.13 0.57 55.39 
PESW 110 ND <5 0.75 7.50 0.75 0.05 25.77 
RESW Day 5 640 350 170 49 Russet precipitation <5 1.02 7.74 1.07 0.48 50.52 
PESW 360 ND <5 0.81 7.52 1.02 0.19 30.69 
RESW Day 10 260 540 140 94 Russet precipitation 1.01 7.77 1.10 0.42 46.81 
PESW 420 ND <5 0.72 7.49 1.04 0.08 23.60 
RESW Day 20 2,500 1,600 1,600 1,600 Russet precipitation 1.17 7.79 1.14 0.51 53.58 
PESW 400 ND <5 0.69 7.50 0.90 0.07 21.21 
RESW Day 40 2,240 920 220 94 Suspended solids 1.13 7.67 1.63 0.59 62.35 
PESW 480 ND <5 0.58 7.52 0.88 0.03 23.34 
RESW Day 90 1,740 1,600 920 180 Suspended solids <5 1.11 7.64 1.60 0.57 59.39 
PESW 99 ND <5 0.66 7.53 1.10 0.02 23.87 
RESW Day 180 370 110 23 13 ND <5 0.67 7.78 1.22 0.32 44.20 
PESW 220 ND <5 0.54 7.65 1.05 0.04 24.61 
RESW Day 360 590 430 180 180 Russet precipitation <5 0.96 7.59 1.06 0.44 55.40 
PESW 380 ND <5 0.77 7.43 0.93 0.04 22.03 
SamplesHPCTotal coliformsThermotolerant coliformsE. coliVisible to the naked eyeChromaticityTurbiditypHCODNH3(N)
CFU/mLMPN/100 mL
°NTUmg/Lmg/Lmg/L
RESW Day 2 180 540 240 79 Russet precipitation 1.04 7.79 1.13 0.57 55.39 
PESW 110 ND <5 0.75 7.50 0.75 0.05 25.77 
RESW Day 5 640 350 170 49 Russet precipitation <5 1.02 7.74 1.07 0.48 50.52 
PESW 360 ND <5 0.81 7.52 1.02 0.19 30.69 
RESW Day 10 260 540 140 94 Russet precipitation 1.01 7.77 1.10 0.42 46.81 
PESW 420 ND <5 0.72 7.49 1.04 0.08 23.60 
RESW Day 20 2,500 1,600 1,600 1,600 Russet precipitation 1.17 7.79 1.14 0.51 53.58 
PESW 400 ND <5 0.69 7.50 0.90 0.07 21.21 
RESW Day 40 2,240 920 220 94 Suspended solids 1.13 7.67 1.63 0.59 62.35 
PESW 480 ND <5 0.58 7.52 0.88 0.03 23.34 
RESW Day 90 1,740 1,600 920 180 Suspended solids <5 1.11 7.64 1.60 0.57 59.39 
PESW 99 ND <5 0.66 7.53 1.10 0.02 23.87 
RESW Day 180 370 110 23 13 ND <5 0.67 7.78 1.22 0.32 44.20 
PESW 220 ND <5 0.54 7.65 1.05 0.04 24.61 
RESW Day 360 590 430 180 180 Russet precipitation <5 0.96 7.59 1.06 0.44 55.40 
PESW 380 ND <5 0.77 7.43 0.93 0.04 22.03 

ND, Not detected.

In addition, as schmutzdecke usually matures within days 20–40 (D'Alessio et al. 2015; Fish et al. 2017), samples of 16S rRNA gene sequencing were collected for the RESW and the PESW water samples on day 40 of M-SSF system operation. 5-L water samples were filtered using a 0.22-μm microporous membrane (47 mm diameter, Millipore, USA) to extract microorganisms in the water. The microorganism-enriched membrane was stored using dry ice during transportation (−78.5 °C) and sent for testing immediately. The V4–V5 hypervariable region of the bacterial 16S rRNA gene was amplified using the universal primer 515F (GTGC CAGCMGCCGCGG)/R907 (5CCGTCAATTCMTTTRAGTTT) (Mi et al. 2015), and the samples were sequenced using the Illumina MiSeq platform. The sequencing data could be found on the NCBI database (ID: PRJNA689665).

Data analysis

Bacterial sequences were clustered into operational taxonomic units (OTUs) with a maximum distance of 3% by the UPARSE pipeline (Edgar 2013). Using the SILVA database (silva128/16S_bacteria), sequences were phylogenetically assigned to phylum, class, and other taxonomic levels by MOTHUR. The relative abundance of a given phylogenetic group was then calculated. Based on these data, the circos diagram was made by R. Based on PICRUSt, the 16S function prediction for the RESW and the PESW are as follows. The effect of the number of copies of the 16S marker gene in the genome of the species was removed, then each OTU was compared with the clusters of orthologous groups (COG) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) to obtain the COG family information or the KEGG orthology information corresponding to the OTU. Then, the COG and pathway abundances were calculated.

RESULTS AND DISCUSSION

M-SSF system effects

The M-SSF system operation started at the beginning of the wet season, at this time, the water table in the epikarst zone was low. With an increase of precipitation, the water table of the epikarst zone was raised by a large margin in the rainy season. The recharge of epikarst water by precipitation also aggravated the pollution of RESW (Table 1), which showed increases in turbidity, COD, HPC, coliforms, and so on. Because of the land-use types of the AES catchment dominated by residential and cultivated land, the farmyard manure and household waste had been accumulated in the soil during the dry season. With the rainy season beginning, the rainfall mobilized these accumulated pollutants in the soil to the epikarst water, which suggested that the thin soil layer in the epikarst zone could not protect the epikarst spring water from pollution effectively. Further analysis showed that HPC, coliforms, and turbidity of the RESW reached their peak on day 20, while ammonia, nitrate, and COD reached their peak on day 40 (Table 1). These results agree with other studies in which particulates arrive prior to solutes on account of the exclusion process, which occurs in fissure flows of epikarst zones (Pronk et al. 2009; Savio et al. 2019; Bandy et al. 2020). Moreover, the pH of the RESW decreased slightly, which could be attributed to the dilution effect of rainfall (He & Li 2005).

In contrast to the RESW, the water quality of the PESW was improved significantly (Table 1). Generally, the conventional SSF with juvenile schmutzdecke during the initial stage could not achieve the optimal performance. However, due to using MHRCCL as the filtration medium, the M-SSF system in this study could ensure efficient purification of epikarst water with juvenile schmutzdecke (Figure 4). The M-SSF system mainly relied on the adsorption, interception, and ion exchange of MHRCCL for water purification before its schmutzdecke matured. Although coliforms in PESW were not removed completely on days 5 and 10, coliforms of PESW were significantly eliminated compared to RESW by the M-SSF system before its schmutzdecke matured. With superior specific surface area and cation exchange capacity, solutes in PESW such as nitrate and ammonia were largely purified. The M-SSF system of AES exhibited acceptable purifying effects during the adsorption dominating stage. By day 20, the schmutzdecke was approaching maturity, and the main mechanism of water purification in the M-SSF system shifted from adsorption dominating to microbial in schmutzdecke dominating. The M-SSF system relied on microbially mediated mechanisms (e.g., predation and bio-oxidation) in the schmutzdecke to achieve effective water purification (Bauer et al. 2011; Lee & Oki 2013; Haig et al. 2014; Oh et al. 2018). The purifying effect on AES water was further improved, as can be seen by the complete removal in coliforms (Table 1), suggesting that the M-SSF system achieved optimal performance during the schmutzdecke maturity. Over time, the biofilm got thicker and the permeability of M-SSF system decreased. It could be observed that the waterhead of both sides of SSFB changed from 5 cm (day 2) to 12 cm (day 180) and then to 21 cm (day 360). The increase of bilateral waterhead might mitigate the effect of the decreasing permeability of SSFB, so that the M-SSF system would maintain relatively stable filter flows for as long as possible. Compared with the top colonization mode of the SSFs in our previous study (Zhao et al. 2019), the side colonization mode tended to maintain a longer optimal performance time without easily clogging which could be attributed to the inconsistency direction of filtration and particle precipitation direction.

Figure 4

Comparison between a conventional SSF and the M-SSF system: (a) the operational period of the conventional SSF and the performance of each stage and (b) the operational period of M-SSF and the performance of each stage.

Figure 4

Comparison between a conventional SSF and the M-SSF system: (a) the operational period of the conventional SSF and the performance of each stage and (b) the operational period of M-SSF and the performance of each stage.

Furthermore, the water quality indicators of the RESW were worse in the wet season than the dry season (Supplementary Figure S2), especially HPC and coliforms in summer. Besides rainfall mobilization of accumulated pollutants in the soil, the RESW was also subject to bacteria breeding and external contaminations such as pollution from animals and humans due to its open storage mode (Ahmed et al. 2011; WHO 2011; Zhang et al. 2017). To address this problem, the storage form of water cellar in the M-SSF system was improved. By adopting a closed water storage mode using ceramic tile lining, the external contaminations and bacteria breeding were effectively reduced. The monitoring results showed that water quality indicators of PESW were stable throughout 360 days (Table 1), indicating that the M-SSF system had the potential to prevent the water quality of the PESW from deterioration during the storage process.

Bacterial community structures of the PESW and the RESW

According to the sequencing results of 16S rRNA, the test samples all had high coverage (>99%), indicating that the test results had good representativeness. The PESW and the RESW had different degrees of variation in species taxonomy, diversity, and evenness. In terms of species taxonomy, the number of all taxonomic levels in the PESW and the RESW did not seem to have significant changes (Table 2), but there were substantial differences in species diversity and evenness between PESW and RESW. By using the Shannon diversity index as an example, the RESW (<3) was markedly smaller than the PESW (>3). According to previous research results, a Shannon diversity index of <3 indicates that β-diversity of water might be polluted in a high probability, and a Shannon diversity index of >3 indicates a high probability of β-unpolluted water (Yang et al. 2017). The more diverse bacterial communities, the more enriched their environmental functions are, and a higher bacterial community diversity also means enhanced stability of environmental functions. The coexistence of multiple species can provide a greater guarantee of comprehensive environmental functions; that is, when some species lose a certain environmental function under ecological pressure, other species can fill this gap of the environmental function (Wittebolle et al. 2009; Lautenschlager et al. 2014; Saifullah & Purnomo 2015; Fish et al. 2017). Similarly, the greater the bacterial community evenness, the better its tolerance and stability in environmental stresses. When the bacterial community evenness is low; that is, when the communities are controlled by only a few species, the resistance to disturbance can only occur when the predominant species are tolerant of disturbance (Wittebolle et al. 2009; Haig et al. 2015). In comparison, when the bacterial community evenness is high, so is the probability of the presence of species that are resistant to water quality disturbance.

Table 2

Amounts of the different taxonomic levels, diversity indices, and evenness indices of the bacterial communities in the PESW and the RESW

SamplesTaxonomic levels
Diversity indices
Evenness indices
Coverage
PhylumClassOrderFamily GenusSpeciesOTUShannonSimpsonAceChao 1HeipShannonevenSimpsoneven
RESW 22 37 72 121 207 275 343 2.09 0.33 466.95 479.12 0.02 0.36 0.01 99.60% 
PESW 22 35 73 121 208 271 341 3.94 0.04 400.71 410.16 0.15 0.68 0.08 99.75% 
SamplesTaxonomic levels
Diversity indices
Evenness indices
Coverage
PhylumClassOrderFamily GenusSpeciesOTUShannonSimpsonAceChao 1HeipShannonevenSimpsoneven
RESW 22 37 72 121 207 275 343 2.09 0.33 466.95 479.12 0.02 0.36 0.01 99.60% 
PESW 22 35 73 121 208 271 341 3.94 0.04 400.71 410.16 0.15 0.68 0.08 99.75% 

The bacterial community structure of RESW and PESW at the phylum level was analyzed. The results show that in the RESW, Proteobacteria, Bacteroidetes, and Actinobacteria were the predominant phyla; however, Proteobacteria, Bacteroidetes, TM6_Dependentiae, and Actinobacteria were the predominant phyla in the PESW (Figure 5). As the denitrifier mostly exists in Proteobacteria, the nitrate decreasing in PESW (Table 1) may be related to the abundance of Proteobacteria increasing. Also, the schmutzdecke of SSF is rich in the biological population, including bacteria, protozoa, and various microbial secretions. These microbes form a virtuous food chain, and TM6_Dependentiae in PESW seems to be an indicator that bacterium confirms the formation of a food chain in SSF. Previous studies have found that TM6_Dependentiae is a type of bacterium that primarily parasitizes protozoa (e.g., heterotrophic flagellates) and lives a highly host-related lifestyle. Their genomes are small (1.0–1.5 Mb) and they lack complete biosynthetic pathways for various essential cellular building blocks, including amino acids, lipids, and nucleotides (Yun Kit Yeoh et al. 2016; Deeg et al. 2018). Furthermore, the Chromulinavorax destructans generated by TM6_Dependentiae can lyse heterotrophic protists and thus TM6_Dependentiae enter the water. Therefore, the high abundance of TM6_Dependentiae in PESW indicated that there was a high abundance of heterotrophic protists in the M-SSF (Deeg et al. 2018). As predators at the top of the SSF food chain, the heterotrophic protists were too large to pass through the SSF; however, the lysis of heterotrophic protists engaged themselves as a carbon source in the food chain and re-participated in the material cycle of the SSF. The high abundance of TM6_Dependentiae in PESW found in this study confirms that its host heterotrophic protists were widely present in the schmutzdecke as predators. Moreover, TM6_Dependentiae, an indicator bacterium, could provide evidence that the formation of a complete food chain is the main mechanism for water purification by the mature SSF.

Figure 5

Bacterial community structure analysis on the phylum level: (a) Pieplot of RESW and (b) Pieplot of PESW.

Figure 5

Bacterial community structure analysis on the phylum level: (a) Pieplot of RESW and (b) Pieplot of PESW.

Bacterial community structure analysis on the genus level (Figure 6) shows that the genera of RESW primarily included Acidovorax (54.24%), Pseudarcicella (17.24%), Acinetobacter (6.41%), and Rhodobacteraceae (3.93%). In particular, the abundance of Pseudarcicella in the RESW was 460 times that in the PESW. Pseudarcicella primarily parasitizes in the skin of leeches (Kampfer et al. 2012), and its existence could serve as a sign that RESW is affected by external animal pollution. By contrast, Pseudarcicella rarely existed in the PESW, which suggested that the M-SSF system could cut off the external pollution effectively and prevent epikarst water from deteriorating during storage. The bacterial genera of the PESW primarily included Hydrogenophaga (13.38%), Herbaspirillum (11.31%), Novosphingobium (8.95%), Porphyrobacter (6.23%), Methylotenera (5.33%), Altererythrobacter (4.69%), and Silanimonas (3.71%). Among these bacterial genera, Novosphingobium can degrade aromatic compounds (Liu et al. 2005), and Methylotenera are obligate methylamine-degrading bacteria (Kalyuzhnaya et al. 2006). Some of these bacterial genera are related to the degradation of organic matter, as well as the removal of total nitrogen, and are primarily those that remain after water purification using the M-SSF.

Figure 6

Circos diagram on the genus level of the PESW and the RESW. u_f, unclassified family; n_f, no rank family; n_p, no rank phylum.

Figure 6

Circos diagram on the genus level of the PESW and the RESW. u_f, unclassified family; n_f, no rank family; n_p, no rank phylum.

Based on the results from the 16S function prediction, the COG associated with ammonia, nitrate, and nitrite were selected for analysis. It is seen from Figure 7(a) that there was a higher relative abundance of ammonia-, nitrate-, and nitrite-related COG in the RESW than in the PESW (the definitions of COG are shown in Supplementary Table S3), suggesting a higher relative abundance of microorganisms associated with the metabolism of these nitrogen compounds in the RESW. The higher abundance of nitrogen metabolism microorganisms in the RESW suggests a greater selective pressure for nitrogen compounds to form in the RESW than in the PESW (Zelaya et al. 2019), which is consistent with the results of the water quality testing. Similarly, the RESW had a higher relative abundance of most of the metabolic pathways associated with the degradation of organic pollutants (Figure 7(b)), especially aromatic compounds (Supplementary Table S4), than the PESW, which is consistent with previous studies (Bai et al. 2013). This also implies that the RESW was more intensely affected by organic pollution, which is also consistent with the COD, as an indicator of dissolved organic pollution in these water quality tests. The results of the 16S function prediction further confirmed the effect of the M-SSF system and qualitatively characterized the biological process of water purification. However, to better characterize the environmental functions of microorganisms present in epikarst spring waters, further research such as metagenomic sequencing or macro-transcriptome sequencing should be done. In addition, further research on genes encoding enzymes involved in nitrogen metabolism (e.g., ammonia monooxygenases, nitrite reductases, and nitrous oxide reductases) in the schmutzdecke might be helpful to understand the purification mechanism of nitrate and ammonia in SSF (Braker et al. 1998; Moore et al. 2011; Zhou et al. 2011; Yang et al. 2019).

Figure 7

Stacked bar charts of relative abundance in the PESW and the RESW based on 16S functional prediction. (a) Chart of COG related to ammonium, nitrate, and nitrite metabolism; the description of each COG is shown in Supplementary Table S1; (b) chart of pathways associated with organic pollutant degradation; the description of each pathway is shown in Supplementary Table S2.

Figure 7

Stacked bar charts of relative abundance in the PESW and the RESW based on 16S functional prediction. (a) Chart of COG related to ammonium, nitrate, and nitrite metabolism; the description of each COG is shown in Supplementary Table S1; (b) chart of pathways associated with organic pollutant degradation; the description of each pathway is shown in Supplementary Table S2.

CONCLUSION

The results presented here highlight the high level of sensitivity epikarst spring water experience with regards to pollution from human activity, and outline an effective, simple, and affordable SSF system based on MHRCCL for treatment of these epikarst spring water sources upon which local communities are dependent. This paper makes good strides to characterize the microbial community that establishes itself in the SSF and shows a meaningful relationship between bacterial and protozoan communities, which provides new evidence for SSF to build a closed food chain responsible for the improvement in the safety of drinking water. A more in-depth study of how the communities are established could yield a better understanding of how the critical functions are delivered.

ACKNOWLEDGEMENTS

This research was supported by the (1) Natural Science Foundation of Hebei Province (C2020504001), (2) National Key R&D Program of China (2016YFC0502502), (3) China Geological Survey (DD20190356 and DD20189262), (4) Chinese Academy of Geological Sciences (YKWF201628), and (5) National Natural Science Foundation of China (No. 41272301).

DATA AVAILABILITY STATEMENT

All relevant data are included in the paper or its Supplementary Information.

REFERENCES

REFERENCES
Ahmed
W.
,
Gardner
T.
&
Toze
S.
2011
Microbiological quality of roof-harvested rainwater and health risks: a review
.
Journal of Environment Quality
40
(
1
),
13
21
.
Bakalowicz
M.
2004
The epikarst, the skin of karst
.
Karst Waters Institute Special Publication
9
,
16
22
.
Bandy
A. M.
,
Cook
K.
,
Fryar
A. E.
&
Zhu
J.
2020
Differential transport of Escherichia coli isolates compared to abiotic tracers in a karst aquifer
.
Ground Water
58
(
1
),
70
78
.
Bauer
R.
,
Dizer
H.
,
Graeber
I.
,
Rosenwinkel
K. H.
&
Lopez-Pila
J. M.
2011
Removal of bacterial fecal indicators, coliphages and enteric adenoviruses from waters with high fecal pollution by slow sand filtration
.
Water Research
45
(
2
),
439
452
.
Champollion
C.
,
Deville
S.
,
Chéry
J.
,
Doerflinger
E.
,
Le Moigne
N.
,
Bayer
R.
,
Vernant
P.
&
Mazzilli
N.
2018
Estimating epikarst water storage by time-lapse surface-to-depth gravity measurements
.
Hydrology and Earth System Sciences
22
(
7
),
3825
3839
.
Chen
H.
,
Zhao
J.
,
Zhong
A.
&
Jin
Y.
2011
Removal capacity and adsorption mechanism of heat-treated palygorskite clay for methylene blue
.
Chemical Engineering Journal
174
(
1
),
143
150
.
D'Alessio
M.
,
Yoneyama
B.
,
Kirs
M.
,
Kisand
V.
&
Ray
C.
2015
Pharmaceutically active compounds: their removal during slow sand filtration and their impact on slow sand filtration bacterial removal
.
Science of the Total Environment
524
,
124
135
.
Deeg
C. M.
,
Emma George
M. M. Z.
,
Husnik
F.
,
Keeling
P. J.
&
Suttle
C. A.
2018
Chromulinavorax destructans, a pathogenic TM6 bacterium with an unusual replication strategy targeting protist mitochondrion
.
BioRxiv
379
388
. https://doi.org/10.1101/379388
Devadhanam Joubert
E.
&
Pillay
B.
2008
Visualisation of the microbial colonisation of a slow sand filter using an environmental scanning electron microscope
.
Electronic Journal of Biotechnology
11
(
2
),
119
125
.
Duane
C.
&
Wolf
H. D. S.
1994
Methods of soil analysis: part 2 – microbiological and biochemical properties
.
SSSA Book Series
5
,
41
51
.
Ellis
K. V.
&
Aydin
M. E.
1995
Penetration of solids and biological activity into slow sand filters
.
Water Research
29
(
5
),
1333
1341
.
Guerrero
C.
,
Mataix-Solera
J.
,
Gómez
I.
,
García-Orenes
F.
&
Jordán
M. M.
2005
Microbial recolonization and chemical changes in a soil heated at different temperatures
.
International Journal of Wildland Fire
14
(
4
),
385
400
.
Haig
S. J.
,
Collins
G.
,
Davies
R. L.
,
Dorea
C. C.
&
Quince
C.
2011
Biological aspects of slow sand filtration: past, present and future
.
Water Science and Technology: Water Supply
11
(
4
),
468
472
.
Haig
S. J.
,
Quince
C.
,
Davies
R. L.
,
Dorea
C. C.
&
Collins
G.
2014
Replicating the microbial community and water quality performance of full-scale slow sand filters in laboratory-scale filters
.
Water Research
61
,
141
151
.
Haig
S.
,
Quince
J.
,
Davies
C.
,
Dorea
R. L.
&
Collins
C. C. G.
2015
The relationship between microbial community evenness and function in slow sand filters
.
MBio
6
(
5
),
e00729
15
.
He
X.
&
Li
Q.
2005
Hydrochemical variations of the epikarst zone its environmental effects: an example spring of Landiantang at Nongla, Mashan
.
Journal of Guangxi Normal University
23
(
2
),
103
106
(in Chinese)
.
He
Z.
,
Liang
H.
,
Yang
Z.
,
Huang
F.
&
Zeng
X.
2018
Water system characteristics of Karst river basins in South China and their driving mechanisms of hydrological drought
.
Natural Hazards
92
(
2
),
1155
1178
.
Kalyuzhnaya
M. G.
,
Bowerman
S.
,
Lara
J. C.
,
Lidstrom
M. E.
&
Chistoserdova
L.
2006
Methylotenera mobilis gen. nov., sp. nov., an obligately methylamine-utilizing bacterium within the family Methylophilaceae
.
International Journal of Systematic and Evolutionary Microbiology
56
(
Pt 12
),
2819
2823
.
Kampfer
P.
,
Busse
H. J.
,
Longaric
I.
,
Rossello-Mora
R.
,
Galatis
H.
&
Lodders
N.
2012
Pseudarcicella hirudinis gen. nov., sp. nov., isolated from the skin of the medical leech Hirudo medicinalis
.
International Journal of Systematic and Evolutionary Microbiology
62
(
Pt 9
),
2247
2251
.
Lautenschlager
K.
,
Hwang
C.
,
Ling
F.
,
Liu
W. T.
,
Boon
N.
,
Köster
O.
,
Egli
T.
&
Hammes
F.
2014
Abundance and composition of indigenous bacterial communities in a multi-step biofiltration-based drinking water treatment plant
.
Water Research
62
,
40
52
.
Liu
Z. P.
,
Wang
B. J.
,
Liu
Y. H.
&
Liu
S. J.
2005
Novosphingobium taihuense sp. nov., a novel aromatic-compound-degrading bacterium isolated from Taihu Lake, China
.
International Journal of Systematic and Evolutionary Microbiology
55
(
Pt 3
),
1229
1232
.
Mi
Z.
,
Dai
Y.
,
Xie
S.
,
Chen
C.
&
Zhang
X.
2015
Impact of disinfection on drinking water biofilm bacterial community
.
Journal of Environmental Sciences
37
,
200
205
.
Moore
T. A.
,
Xing
Y.
,
Lazenby
B.
,
Lynch
M. D.
,
Schiff
S.
,
Robertson
W. D.
,
Timlin
R.
,
Lanza
S.
,
Ryan
M. C.
,
Aravena
R.
,
Fortin
D.
,
Clark
I. D.
&
Neufeld
J. D.
2011
Prevalence of anaerobic ammonium-oxidizing bacteria in contaminated groundwater
.
Environmental Science & Technology
45
(
17
),
7217
7225
.
Pipan
T.
,
Culver
D. C.
,
Papi
F.
&
Kozel
P.
2018
Partitioning diversity in subterranean invertebrates: the epikarst fauna of Slovenia
.
PLoS One
13
(
5
),
e0195991
.
Pronk
M.
,
Goldscheider
N.
,
Zopfi
J.
&
Zwahlen
F.
2009
Percolation and particle transport in the unsaturated zone of a karst aquifer
.
Ground Water
47
(
3
),
361
369
.
Ren
R.
,
Yang
C.
&
Kuang
Y.
2018
Study on exploitation model of epikarst spring karst water Wumengshan area
.
Ground Water
40
(
2
),
24
26
(in Chinese)
.
Saifullah
D. H.
&
Purnomo
B. H.
2015
Water quality of Situ Cibanten based on Shannon-Weaver diversity index value
.
Jurnal Perikanan dan Kelautan
5
(
1
),
1
4
.
Savio
D.
,
Stadler
P.
,
Reischer
G. H.
,
Demeter
K.
,
Linke
R. B.
,
Blaschke
A. P.
,
Mach
R. L.
,
Kirschner
A. K. T.
,
Stadler
H.
&
Farnleitner
A. H.
2019
Spring water of an alpine karst aquifer is dominated by a taxonomically stable but discharge-responsive bacterial community
.
Frontiers in Microbiology
10
,
28
.
Wang
Y.
,
Yuan
D.
&
Yang
S.
2005
Effective exploitation model of karst water in Xiaojiang Basin, Luxi County, Yunnan
.
Carsologica Sinica
24
(
4
),
305
311
(in Chinese)
.
Weber-Shirk
M. L.
&
Dick
R. I.
1999
Bacterivory by a chrysophyte in slow sand filters
.
Water Research
33
(
3
),
631
638
.
WHO
2011
Guidelines for Drinking-Water Quality
, 4th edn.
World Health Organization
,
Geneva, Switzerland
.
Williams
P. W.
2008
The role of the epikarst in karst and cave hydrogeology: a review
.
International Journal of Speleology
37
(
1
),
1
10
.
Wittebolle
L.
,
Marzorati
M.
,
Clement
L.
,
Balloi
A.
,
Daffonchio
D.
,
Heylen
K.
,
Vos
P. D.
,
Verstraete
W.
&
Boon
N.
2009
Initial community evenness favours functionality under selective stress
.
Nature
458
(
7238
),
623
626
.
Wu
P.
,
Tang
C.
,
Zhu
L.
,
Liu
C.
,
Cha
X.
&
Tao
X.
2009
Hydrogeochemical characteristics of surface water and groundwater in the karst basin, southwest China
.
Hydrological Processes
23
(
14
),
2012
2022
.
Yang
H.
,
Zhang
G.
,
Yang
X.
,
Wu
F.
,
Zhao
W.
,
Zhang
H.
&
Zhang
X.
2017
Microbial community diversity and differences in cellar water of typical rainwater harvesting area
.
Environmental Science
38
(
11
),
4733
4747
(in Chinese)
.
Yang
X. C.
,
Han
Z. Z.
,
Ruan
X. Y.
,
Chai
J.
,
Jiang
S. W.
&
Zheng
R.
2019
Composting swine carcasses with nitrogen transformation microbial strains: succession of microbial community and nitrogen functional genes
.
Science of the Total Environment
688
,
555
566
.
Yun Kit Yeoh
Y. S.
,
Parks
D. H.
&
Hugenholtz
P.
2016
Comparative genomics of candidate phylum TM6 suggests that parasitism is widespread and ancestral in this lineage
.
Molecular Biology and Evolution
33
(
4
),
915
927
.
Zelaya
A. J.
,
Parker
A. E.
,
Bailey
K. L.
,
Zhang
P.
,
Van Nostrand
J.
,
Ning
D.
,
Elias
D. A.
,
Zhou
J.
,
Hazen
T. C.
,
Arkin
A. P.
&
Fields
M. W.
2019
High spatiotemporal variability of bacterial diversity over short time scales with unique hydrochemical associations within a shallow aquifer
.
Water Research
164
,
114917
.
Zhang
G.
,
Zhao
L.
,
Wu
F.
,
Zhang
M.
,
Yue
Z.
,
Wang
Q.
,
Yang
S.
,
Wang
J.
&
Gong
S.
2017
Study on rough filtration plus slow sand filtration techniques for cellar water of villages and towns in northwest China
.
Advances in Engineering Research
143
,
154
166
.
Zhao
Z.
,
Shen
Y.
,
Shan
Z.
,
Yu
Y.
&
Zhao
G.
2018
Infiltration patterns and ecological function of outcrop runoff in epikarst areas of Southern China
.
Vadose Zone Journal
17
(
1
),
1
10
.
Zhao
Y.
,
Wang
X.
,
Liu
C.
,
Wang
S.
,
Wang
X.
,
Hou
H.
,
Wang
J.
&
Li
H.
2019
Purification of harvested rainwater using slow sand filters with low-cost materials: bacterial community structure and purifying effect
.
Science of the Total Environment
674
,
344
354
.
Zhou
Z. F.
,
Zheng
Y. M.
,
Shen
J. P.
,
Zhang
L. M.
&
He
J. Z.
2011
Response of denitrification genes nirS, nirK, and nosZ to irrigation water quality in a Chinese agricultural soil
.
Environmental Science and Pollution Research International
18
(
9
),
1644
1652
.
Zhu
S.
2014
The Experimental Study of Modified Filter Media for Cellar Water Treatment Using Biology Slow Filtering Techniques
.
Lanzhou Jiaotong University
,
Lanzhou, China
(in Chinese)
.
Zipf
M. S.
,
Pinheiro
I. G.
&
Conegero
M. G.
2016
Simplified greywater treatment systems: slow filters of sand and slate waste followed by granular activated carbon
.
Journal of Environmental Management
176
,
119
127
.
Zou
S.
,
Li
Z.
,
Chen
H.
,
Liang
B.
&
Xia
R.
2007
A preliminary discussion on the pollution of dispersed water supply in southwest Karst Area
.
Resources and Environment in the Yangtze Basin
16
(
2
),
240
244
(in Chinese)
.