Evaluation of woodchips-amended biosand filter for nitrate and MS2 bacteriophage reduction

Household or point-of-use (POU) water treatments are suggested by the World Health Organization (WHO) as efficient and affordable tools for supplying safe drinking water (WHO 2012). Biosand filter (BSF) is one of the most promising POU systems. The ‘schmutzdecke,’ a biological layer that develops on top of the sand bed and is formed by a combination of microorganisms such as bacteria, fungi, algae, and other biological matter, is a distinctive characteristic of biosand filters. BSF can trap impurities and remove organic/inorganic compounds and several pathogens that cause diarrhea (Wang et al. 2014; Freitas et al. 2022). A recent study by Freitas et al. (2022) highlighted that improved water quality via the use of BSF has been achieved in terms of turbidity, metals, and pathogens reduction, with varying degrees of success. Freitas et al. (2022) summarized that the mean bacteria removal of BSFs in the field studies was reported to be ranged from 1.0 log to 2.9 log and 3.7 log for Escherichia coli and total coliform, respectively. As for viruses, BSF showed high removal rates (>5.0 log) of MS2 bacteriophage in groundwater (Freitas et al. 2022). However, the results of the nitrate removal rate in BSF have been varied, ranged from <5 to 53% (Freitas et al. 2022). Since dynamic nitrogen cycling (e.g., nitrification and denitrification) can occur within the filter media, some researches even pointed out that increased nitrate concentration was observed in BSF outflow, especially when influent waters present high nitrate concentration (Murphy et al. 2010; Pompei et al. 2017). However, groundwater is an important source of drinking water for rural populations, and nitrate is the most widespread contaminant in groundwater (Ward et al. 2018). There is a need to enhance the groundwater nitrate removal performance of BSF to provide safe drinking water. To our knowledge, only one research reported complete denitrification in a vinegar-amended anaerobic BSF (Snyder et al. 2016). In their study, vinegar was used to provide acetic acid as the electron to remove nitrate in the influent. However, repeated chemical addition is time and money consuming and could alter the taste of drinking water. This challenge requires adapted, low-cost, and sustainable solutions for BSF modification.

Denitrifying woodchip bioreactors have emerged as an engineering-based technology to reduce nitrate contamination from a wide range of waters, such as drainage water and urban stormwater (Christianson et al. 2017). The woodchips used in bioreactors support the growth of denitrifying bacteria, which can convert nitrate into inert nitrogen gas and has the potential to achieve up to 100% NO₃-N reduction. The ability of woodchip bioreactors to remove pathogens from wastewater has also been investigated, such as Escherichia coli and Salmonella (Soupir et al. 2018; Rambags et al. 2019). Some studies explored the potential of using woodchips as filling materials to construct column-based woodchip filters to treat wastewater, such as dairy-soiled water and agricultural wash water (Choudhury et al. 2016; Murnane et al. 2016). However, woodchip filters are specialized filtration systems primarily designed for nitrate removal, which may require periodic replacement of the woodchips, whereas BSF, despite its varying nitrate removal rate, can handle a broader range of contaminants and typically requires less maintenance. Therefore, the integration of the benefits of woodchip and biosand filters can improve water quality. However, there are potential chemical and microbiological risks associated with their use, especially related to the denitrification process to remove nitrate using woodchips. It is important to consider these risks to ensure the drinking water quality.

The primary objectives of this study were to design the woodchip-amended BSF (W-BSFs) and to investigate the functions of the woodchip on the enhancement of denitrification performance and MS2 virus removal of BSF. The microbial analysis was conducted to explore the removal mechanism and related microbiological results, and risks linked to water quality and health were discussed as well. The results of this study can contribute to a new understanding and development of POU water treatment systems that combine woodchip denitrification and BSF for nitrate and virus removal for drinking water treatment.

The groundwater was obtained from a natural aquifer underneath the Nathan M. Newmark Civil Engineering Laboratory, University of Illinois at Urbana-Champaign. The concentrations of nitrate in the groundwater were below the detection limit of the measurement for nitrate of 0.02 mg/L. This water source was extensively used by previous studies, and the total organic carbon (TOC) was found to be around 2.35 mg/L (Bradley et al. 2011). The KNO₃ solution was used to adjust the groundwater nitrate concentration to 50 mg-NO₃/L during the experiment period. This nitrate level is higher than that mentioned by the United States Environmental Protection Agency (USEPA) standards (USEPA 2011) and needed to be treated since for short-term (acute) exposure to nitrate for bottle-fed infants. The value should not exceed 50 mg-NO₃/L (Murphy et al. 2010). Ward et al. (2018) estimated that 2% of public-supply wells and 6% of private wells exceeded the maximum contaminant level for nitrate (which is approximately equivalent to 50 mg/L-NO₃) in public drinking water supplies in the United States. In many developing countries where biosand filters are used, such as rural areas in India, nitrate concentrations in groundwater are much higher, ranging from 45.7 to 66.6 mg/L-NO₃. The situation is even more severe in The Gaza Strip, where some areas have reported extremely high nitrate levels, reaching concentrations of up to 500 mg/L-NO₃ (Ward et al. 2018). MS2 bacteriophage (ATCC 15597-B1) was selected as a surrogate virus due to its similarities in size and morphology to human enteric viruses. The MS2 was propagated in the lab following the method outlined in our previous study (Wang et al. 2014). MS2 was propagated with E. coli (ATCC 15597) in the tryptic soy broth liquid medium. After 24 h of incubation, the cell debris was removed by centrifugation (30,000 rpm at 20 °C) followed by filtration through a 0.22 μm filter (Millipore Sigma SCGPS01RE SteritopTM). The purified MS2 stock was divided into 5 mL volumes and stored at −80 °C. To protect the phage particles during freezing and thawing, a cryoprotectant is often added to the phage suspension, such as glycerol or dimethyl sulfoxide (AHPA 2002). A total of 250 μL of MS2 solution (with 4.57 × 10⁶ PFU/mL) was inoculated into 1 L influent solution with 50 mg/L nitrate to prepare the groundwater as the influents. The concentration of MS2 solution was further confirmed by the double-layer agar method with E. coli Famp (ATCC 15597) as a host (USEPA 2001). Dilutions containing 30–300 plaques were employed to determine the PFU per milliliter.

A series of laboratory experiments were conducted to evaluate the performance of W-BSF in nitrate and MS2 removal under groundwater containing nitrate and MS2 at room temperature (25 °C). The experiments were continuously operated over 273 days (39 weeks). A ripening period was not adopted due to a new type of BSF. The average filtration flow rates were 0.040 m³/day for Filter A and 0.035 m³/day for Filter B throughout the duration of the experiments. The low filtration rate in our current study is to ensure enough time for denitrification to occur in woodchips. The filters were charged daily with 1 L of influent feedwater, and around 0.8 L effluents was produced. Influent water, consisting of groundwater with a nitrate level of 50 mg/L, was spiked with MS2 (4.57 × 10⁶ PFU/ml, a commonly used targeted MS2 concentration) (pulse injection) during the microbial test days (once a week). The effluent samples were collected weekly for 39 weeks at the outlet after every microbial test day with a retention time of 24 h of feeding with an influent containing nitrate and MS2 bacteriophage. The flow rate of the two filters was measured weekly. The plug flow condition was verified using a tracer test with 10 L of groundwater containing 0.1 M NaCl and was determined based on the shape of the tracer concentration breakthrough curve. Since our current study focuses on nitrate removal, we did not perform the particle counts to monitor breakthrough.

Water samples were filtered (0.45 μm filters) and analyzed for nitrate concentration (Continuous flow Technicon Autoanalyzer II). The nitrate concentration was determined using the USEPA-approved Brucine Colorimetric Method (USEPA 1971), based on a color reaction between nitrate ions and brucine (an organic compound), which produces a reddish-brown color. The intensity of the color is directly proportional to the nitrate concentration in the sample. The MS2 was quantified with the plaque-forming unit (PFU) assay with the double-layer agar method (Wang et al. 2014). The double-layer technique is a valuable tool to differentiate between different types of microorganisms based on their colony characteristics and to quantify their abundance in the original sample, including a bottom layer of solid agar (usually a nutrient agar) and a top layer of a different agar medium containing the sample to be tested. After quantifying viruses and measuring nitrate, the one-way and two-way analysis of variance (ANOVA) tests were conducted to determine the factors (the proportion of woodchip and sand) affecting MS2 and nitrate removal. 16S rRNA sequencing was conducted to examine the microbial diversity across depths in the woodchip-amended filters (Jang 2020). The sand samples were collected from the top of the sand layer, ports 2 and 3 from Filters A and B at 38 weeks. DNeasy PowerSoil Kit (QIAGEN, Hilden, Germany) was used to extract DNA from the sand samples. The extracted DNA was amplified for 16S rRNA genes by the forward and reverse primer for archaea and bacteria using the Illumina platform at the Roy J. Carver Biotechnology Center at the University of Illinois at Urbana − Champaign. For the bacterial community analysis, V4_515F_New and V4_806R_New primers were used as forward and reverse primers, respectively. The Fluidigm constructed library was quantitated by the quantitative polymerase chain reaction and sequenced on one MiSeq flowcell for 251 cycles from each end of the fragments using a MiSeq 500-cycle sequencing kit version 2. Fastq files were generated and demultiplexed with the bcl2fastq v2.17.1.14 Conversion Software (Illumina). The water control and negative control were included in this study to check the contamination during DNA extraction. The water control was the elution buffer used at the final step in DNA extraction without sand samples. The negative control was the elution buffer, which examined all DNA extraction steps. Qiime 2 (https://qiime2.org/) was used to generate a tree for phylogenetic diversity analysis and taxonomic analysis. The demultiplexing sequencing was the first step to know which barcode sequence is associated with each sample. The metadata file was modified based on this study condition, and the barcode sequence was defined in the file. The sequence quality control and feature table construction were generated with DADA2. The DADA2 method was a pipeline for detecting and correcting Illumina amplicon sequence data. After the quality filtering step, the table and rep-seq files were available as the outputs. The table file provided information on how many sequences are associated with each sample and with each feature, histograms of those distributions, and some related summary statistics. In addition, the rep-seqs file provided a mapping of feature IDs to sequences and links to easily BLAST each sequence against the NCBI database. The last step was generating a tree for phylogenetic diversity and taxonomic analyses.

The average MS2 removal increased over time ranging from 1.32-log10 to 2.85-log10 for Filter A and from 1.31-log10 and 2.46-log10 for Filter B. After 22 weeks, the average MS2 removal at the outlet was 2.26-log10 ± 0.49 and 2.14-log10 ± 0.41 for Filters A and B, respectively. After 30 weeks, the variance (standard deviation) in MS2 removal by the whole filter decreased to ± 0.14 for Filter A and ± 0.17 for Filter B. In our study, Filter A and Filter B could remove the bacteriophage MS2 about 1.91-log10 (98.8%) by Filter A and 1.88-log10 (98.7%) by Filter B for 39 weeks. There was no significant difference in MS2 removal at the outlet between Filters A and B (p > 0.05), which indicated that the addition of woodchips into the BSF to construct the W-BSF would not affect virus removal performance compared with reported BSF (Bradley et al. 2011; Elliott et al. 2011). This result also indicated that the filters reached steady-state conditions around 30 weeks. However, since the woodchip-amended BSFs also presented generally good performance before 30 weeks, further investigation is required to establish a formal and reliable ripening period for woodchip-amended BSFs. This finding demonstrated that a single woodchip-amended filter could exhibit fairly good MS2 removal performance over a long-term period. MS2, while not pathogenic itself, is a reliable indicator of fecal contamination, which can harbor harmful waterborne pathogens. Our results suggested that the W-BSF decreased the levels of MS2 in the effluent, suggesting the potential positive role of W-BSF in reducing the outbreaks of gastrointestinal illnesses and viral infections within communities. This correlation underscores the practical implications of W-BSF to minimize the health risks associated with fecal contamination.

In our current study, we only reported two types of woodchip-amended biosand filters and investigated the impact of the proportion of sand and woodchip on nitrate removal. Further research is necessary to provide a comprehensive evaluation to determine if woodchip-amended biosand filters can be improved to meet all drinking water quality criteria (such as pH, DOC, and dissolved oxygen). For example, the information on the profiles of dissolved oxygen is important for considering whether aeration at the end of the treatment train for supplementing dissolved oxygen in the treated water is required, although a study by Young-Rojanschi & Madramootoo (2014) pointed out that dissolved oxygen levels at 5 and 10 cm of media depth in intermittent filters reached an average of 0 mg/L by 24 h of residence time on day 60 of the experiment. In addition, if organic matter derived from the woodchip-amended is available for denitrification, then the deterioration of the treated water quality by the organic matter remaining after the treatment (e.g., increases in turbidity or TOC concentration) could be of potential chemical risks. Therefore, monitoring the water chemical quality in the W-BSF is essential for future research. In addition, since nitrate removal by denitrification relies on biological processes, future studies should also investigate the microbiological risks of proposed W-BSF by evaluating the microbial levels in the effluent. On the basis of the aforementioned consideration, we believe an Integrated Microbial-Chemical Risk Assessment should be established through long-term study to identify and minimize both microbial and chemical risk. From an engineering application perspective, the optimization of the filtration rate to meet technological efficiency and its applicability in the field is needed. Our future research, based on current nitrate removal results, will further investigate these limitations since they are critical for the potential of woodchip-amended BSF to scale-up implementation in households.

In this study, we investigated the potential for nitrate and virus removals by amending the BSF with woodchip materials for drinking water treatment. The addition of woodchips into BSF to build W-BSF can present good performance in terms of nitrate and MS2 removal. For the influent nitrate concentration of 50 mg/L, Filter A and Filter B could reduce nitrate up to 40 and 36% by the whole filter, respectively, to reach the EPA standard. Our results showed that nitrate reduction increased with the increase in woodchip proportion. Microbial community analysis indicated that the nitrate removal by the woodchip-amended biosand filters could be attributed to the denitrifying bacteria, such as the family Streptomycetaces, the genera Pseudomonas and Bacillus, and relative abundances of the phylum Nitrospirae. Specifically, the phylum Nitrospirae present in the filter sample is known to perform nitrification or nitrite oxidation. Overall, the relative abundance of Nitrospirae in Filter A (2.75%) is lower compared with Nitrospirae in Filter B (3.92%), with was also consistent with the nitrate removal performance of Filter A (40%) and Filter B (36%), showing that a lower nitrate removal was achieved by Filter B. The woodchip-amended BSF could remove the bacteriophage MS2 about 1.91-log10 (98.8%) by Filter A and 1.88-log10 (98.7%) by Filter B for 39 weeks. The difference in sand–woodchip proportion did not have a significant impact on the MS2 reduction, demonstrating that a single W-BSF can maintain its MS2 removal performance to a fairly good level over a long-term period. Overall, this study indicated the potential of W-BSFs in achieving waterborne pathogen and nitrate removal to reduce chemical and microbiological contaminants in the drinking water systems, contributing directly to public health and safe drinking water for communities.

This work was partially supported by the National Institute of Food and Agriculture, U.S. Department of Agriculture, Hatch project (No. ILLU-741-337).

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

The authors declare there is no conflict.