Waterborne viruses in urban groundwater environments

The proportion of global population using urban aquifers as drinking water sources increases every year and indeed the groundwater quality is not monitored adequately. Although norovirus has been identified as the first cause of groundwater-related outbreaks, the surveillance of waterborne viruses has been rather neglected. From ageing or disrupted sewer systems, occasional sewer discharges (e.g. combined sewer overflows, storm run-off), to poorly managed reclaimed water infiltration practices, multiple are the pathways that cause groundwater quality deterioration. This study revises the main viral contamination sources and the factors affecting viral contamination of groundwater bodies in terms of transport, inactivation, and survival of the viral particles. It also summarizes the methods used for those reporting the presence of human viruses in urban groundwaters. A total of 36 articles have been included in the method survey spanning a period of 24 years (1999– 2022). There is a need of systematic monitoring considering representative set of water-borne pathogens. The evaluation of the presence of human adenovirus seems a useful tool to predict the presence of other waterborne pathogens in groundwater. Large volume sampling methods, but also new passive sampling methodologies applied to groundwater, coupled to target massive sequencing approaches may elucidate the range of pathogens capable of contaminating urban groundwaters for further evaluation of risk.


Introduction
From the total earth's available fresh water a 98% corresponds to groundwater [1].The proportion of the global population using safely managed services increases every year but it is estimated that still 3.6 billion people (50% of the world population that uses groundwater sources) lacks safely managed services and consumes contaminated water [2].Groundwater plays an important role in the urban areas in terms of public health.Cities alter the quality and quantity of the local aquifer systems.Leakage from the water supply system and storm drainage, generates large amounts of recharge water but also leakage from sewage systems, individual septic tanks, direct injection of inadequately treated wastewater into aquifers or accidental spills may introduce large amounts of pathogens into the urban aquifers [3][4][5].Globally, the urban population using save groundwater sources increased from 57% in 2015 to 62% in 2020 [2].
Nevertheless, there is still space for improving sanitation management that will have a profound impact on the quality of groundwater, particularly when sewage contaminates drinkingwater sources.
A pathogen is the agent that causes disease to a host and a waterborne pathogen is the causative agent for a disease transmitted through water.The WHO Guidelines for drinking water quality [6] and the Global Waterborne Pathogen Project (GWPP) [7] have listed members of six viral families as the most important pathogens transmitted by a variety of means, including fecal-oral transmission routes.Noroviruses (NoV) and other enteric viruses, such as adenovirus (AdV), astrovirus (AstV), enterovirus (EV), hepatitis A virus (HAV) and rotavirus (RoV) are the major cause of waterborne diseases worldwide [7].Waterborne viral pathogens are detected in the environment because people suffering from viral infections excrete viral particles into wastewater containing feces, urine and desquamation cells.Even if wastewater is treated in a wastewater treatment facility, including secondary biological treatment, only 2 and 3 log of viruses are reduced which allows significant viral loads to be spread into the environment [8,9].
The wastewater virome is highly diverse and so the groundwater impacted with raw or treated urban sewage [10,11], being plant viruses and bacteriophages the most widely prevalent viruses in groundwater environments [12,13].A full picture of the groundwater virome can be obtained by applying untargeted viral metagenomics.When focusing on waterborne pathogens, and as observed in wastewater metagenomic analyses [10], Target Enrichment Sequencing (TES) and Amplicon Deep Sequencing (ADS) approaches, applied to groundwater used for irrigation, provide a more sensitive approach, allowing the detection of a diversity of viral species within a given family [14].
Around 30% of the disease outbreaks associated with drinking water, reported to the North American CDC's Waterborne Disease and Outbreak Surveillance System (WBDOSS), are related to untreated groundwater [15].Meanwhile, in the WHO European Region about 18% of investigated outbreaks are associated with sewage contamination of groundwater sources [16].The main contributing factor to disease outbreaks associated with untreated groundwater seems to be human sewage [17] and although the majority of outbreaks have an unidentified etiology, the US Waterborne Disease and Outbreak Surveillance System (WBDOSS) indicates hepatitis A virus (HAV) as the most commonly reported etiology for outbreaks associated with untreated groundwater [18].Findings of a meta-analysis study, considering occurrence of viral pathogens in public-supply groundwater systems, indicate the best indicators are E.coli and somatic coliphage, although both indicators may underestimate virus occurrence [19].Moreover, those correlations between viral pathogens and indicators are related to the study sample size and the number of pathogen-positive detections [20].Thus, assessing the hazard from viruses in those settings requires knowledge on quantitative virus concentrations as well as a complete understanding on how viral contamination occurs.

Viral fate and contamination sources
Waterborne viruses can occur in groundwater where there is a poor soil cover, where the soil cover has been breached by the human action or when the soil has been affected by natural disasters (earthquakes, floodings).Land application of sewage effluents and sludges as well as leachates from sewerage systems or septic tanks are the main sources of waterborne virus contamination of urban groundwaters.Sewer leakage is a chronic problem for many cities worldwide due to aquifer recharge under rapid urbanization, ageing of infrastructures, insufficient maintenance and rehabilitation and in many cases poor construction materials [5,21,22].There is an extensive literature assessing both infiltration and exfiltration of urban sewer systems [23].To this intend, flow or solute transport models as well as nitrogen isotope observations have been used to calculate that in some areas groundwater receives more than 500 liters of sewage a day [24].In most of the developing world, the objective of good sanitation is yet unresolved and on-site sanitation is still the most widespread mode of sanitation available [25].The two primary contaminants from these systems are pathogens (mainly bacteria and viruses) and nitrate.
Direct sewer discharges after an earthquake have also a great impact on groundwater supplies, but few earthquake-related studies have the focus on the microbial quality of groundwater.After the earthquakes in New Zealand-2010 [26], Nepal-2015 [27], Korea-2016 [28] and Japan-2016 [29] the microbial quality of groundwater changed, as well as the number of bacteria species and their abundance in the studied groundwater ecosystems.Metagenomic studies have been limited to 16S, 18S, and 23S rRNA genes and waterborne viruses have been shortly considered.Bacterial pathogens remain viable and attached to the sediments after earthquake sewage discharges [30] and it is assumable that waterborne viruses will do so as they have extended survival rates.Thus, viruses have the potential to reach these aquifers, perhaps at a longer travel time and in reduced concentrations posing health risks to citizens after long periods of time.
Stormwater and reclaimed water infiltration practices to promote groundwater recharge, reduce runoff during intensive rain events [31].However, waterborne pathogens present in stormwater have the potential to affect surrounding areas when infiltrated.Although bacterial contamination of groundwater after infiltration has been reported [32] few studies have examined the efficacy of infiltration practices for pathogenic organisms.In 2021 the Minnesota Department of Health published the first study charactering viruses in stormwater before and after reusing it to recharge the aquifer [33].The study monitored 5 waterborne viruses (AdV, EV, HAV, NoV and RoV) from 5 sampling sites across the stormwater infiltration system.RoV was detected throughout the study period in 3 out of 5 sampling sites indicating that groundwater may be susceptible to contamination by viruses when in proximity to stormwater infiltration systems, especially during periods of substantial recharge [33].With the prediction of more intense rain events due to climate change, sewer overflows may increase in the urban environments.Widespread evidences of sewage contamination following rainfall have been found in literature being directly related to the level of urbanization and population density [8,34].
The urban groundwater quality has been specially studied in sub-Saharan Africa [21], where groundwater is the most resilient source of drinking water and indeed groundwater quality is not measured adequately.Reports of fecal contamination from eleven African countries included only one data on virus, which linked AdV and RoV contamination to pit latrine proximity [35].From 1948 to 2013, 649 epidemiological studies worldwide reported groundwater outbreaks.An extensive revision of all the published data, identified 17 pathogens as responsible groundwater-related outbreaks [36].NoV were responsible for most outbreaks followed by Campylobacter, Shigella, HAV and Giardia.Five viruses among the 17 identified pathogens were listed: NoV (39 outbreaks), HAV (21 outbreaks), RoV (2 outbreaks), AdV (1 outbreak), EV (1 outbreak), but the incidence of human enteric viruses in urban environments has been rather neglected in groundwater quality studies.
The methods used to study the virus fate in groundwater were reviewed using boolean literature searches (AND and OR) in PubMed, Google Scholar, and Web of Science.The selected search terms were groundwater, aquifer, borehole, spring, urban, cities, virus.The final algorithm as implemented in PubMed was, as example, as follows: (("groundwater" OR "aquifer" OR "borehole" OR "spring") AND ("urban" OR "city") AND ("virus")).The tool connectedpapers.comwas used to fill in the gaps as it helped elucidating similar papers in the field based on co-citation and bibliographic coupling.The literature was screened to extract information on mean sample volume analyzed, concentration method, virus analyzed and incidence.Articles providing poor data on viral occurrence or method description were excluded as well as nonenglish literarure, studies performed under controlled conditions or using spiked viruses and focused on bacteriophages or microbial fecal indicators.Articles strongly focused on agricultural areas were finally excluded from this review to put the eye on the occurrence of human waterborne viruses.A total of 36 studies were identified, reviewed, and are summarized in Table 1 and more extensively at the supplementary information (S1 Fig).Human AdV, NoV (genogroups I and II) or EV were included in 100% of the analysis performed by the reviewed literature (Table 1) and detected in 88, 72 and 89% of the studies in which they were measured.In 44% of the revised studies, all three viruses were included.Other waterborne pathogens like RoV, Hepatitis A and E virus (HAV and HEV) were both less frequently analyzed (64, 47 and 14% of the studies respectively) and detected (65,59, 40% of the literature in which they were included respectively) (S1 Fig) .Interestingly about one of every four studies included the analysis of human polyomavirus (PyV) and detected JCPyV in 67% of the reports.This human pathogen, shed in urine by one third of healthy adults asymptomatically and highly persistent in the environment, is a good indicator of human-specific fecal contamination [69].
Half of the published studies reporting the presence of viruses in groundwater wells supplying drinking water have been focused in North America where specific groundwater supply types have specific rules implemented [70][71][72].In the United States and Canada groundwater systems are categorized into one of three supply types: large-scale systems (supplying cities with more than 1000 inhabitants), community systems (supplying less than 1000 people), and largely untreated private domestic wells (serving single households).Although large urban systems, which are disinfected, are more likely to be adequately maintained and appropriately treated, 15% of the sources are positive for enteric pathogen, with no differences according to groundwater system type [73].

Factors affecting viral fate in groundwater bodies
Viral fate in groundwater will be affected by both physicochemical properties of soil and water and the diverse factors affecting the viral particle.Those factors will mainly rely on four processes: filtration, adsorption, survival, and die-off/inactivation.

Virus filtration and adsorption to aquifer sediments
Filtration results when the pathogens are too large to fit through the soil or aquifer pores.The extent of filtration depends on the type of soil and rocks through which groundwater flows.For example, silts are more effective at trapping viruses than sands (Fig 1).Physical removal by pores is less effective for viruses than other pathogens because of the small size of viruses.The pore size of the aquifer soil influences the travel time, which can be important to consider because viruses lose their infectivity with time in the groundwater [74].Aquifer types are diverse, and it is difficult to stablish a direct relationship between soil materials and presence of waterborne viruses.Wallender and coworkers, reviewed and reported that 26.2% of waterborne outbreaks were associated with karst limestone bedrock, because bedrock fracturing, unconsolidated bedrock materials and thin subsoil layers may provide pathways for contamination transport [17].Nevertheless, the association between geology and viral contamination is not clear as different studies present diverse number of monitored wells in each aquifer type and period of time [63].
Adsorption occurs when the microorganisms become attached to particles and removes them from the water or at least delays their transport.Virus adsorption onto sediment grains is considered the primary removal mechanism in groundwater, with a complex dependence on the chemistry of the sediment and water.For example, wells located in sand and gravel aquifers are more likely to be virus contaminated than those in other aquifer types [38].Considering both filtration and adsorption factors viruses can move large distances in those scenarios, reaching depths of 67 m and horizontal migration of 408 m in glacial till and 1600 m in fractured limestone [40,75].
Together with sediment surface charges, viral surface properties affect the degree of virus attachment during transport [76,77].The isoelectric point (IEP) of viral surface is used to predict the electrostatic interactions of viruses because when pH is above the IEP, the viral particles are negatively charged and will be adsorbing to positively charged soils [78].Groundwater pH typically ranges from 6 to 8,5 and IEP of most of the waterborne viruses is lower [78][79][80][81][82] (Fig 2).Rainfall water, typically at pH7, can decrease the pH of groundwater and enhance virus transport [83].The same effect will occur with the organic contamination from surface runoff or sewage leachate, that by introducing nitrates (NO3-) will make pH of water to decrease.Within the saturated zone of the aquifer, it is expected the presence of divalent ions or low groundwater pH favors adhesion [84].It has been calculated that adhesion/aggregation of virus to soil particles are responsible for most of the viral loss in groundwater and it is important to consider that this phenomenon is reversible.If water physicochemical conditions change viral particles may be released.In opposite to divalent ions and pH, organic matter in the medium may compete with viruses for adsorption on surfaces [84].
Traditionally, field injection studies have used bacteriophages and there is a lack in literature of studies using viruses infecting human [85].Nevertheless, in lab-scale studies and using model enveloped viruses (bacteriophage ϕ6), aggregation of sediment with virus occurred regardless of mineral type [86].Large numbers of virions remained viable post-aggregation indicating that small-sized aggregates, which may travel more readily through porous media, may pose an infection risk.Hydrogeologic conditions of fractured bedrock aquifers may for example create wide capture zones and short groundwater travel times, leading contamination but at very low concentrations [58].

Virus survival and inactivation
Although viruses cannot replicate outside their hosts, waterborne viruses can survive in groundwater environments for long periods and remain infective in water.One of the first studies facing the occurrence of viruses in groundwater included cell culture assays to determine the presence of infectious viruses [38].This work concluded that in 21 of the 448 samples analyzed viruses were infective.Nevertheless, researchers generally use spiked viruses to produce artificially contaminated water and study viral survival which may be not representative of the real threats in the groundwater environments [87].
The survival of viruses in groundwater was first studied in 1982, when EV demonstrated to persist longer times (24h) than bacteria [88].One of the first studies assessing specific virus persistence in groundwater revealed that temperature was the only variable significantly correlated with the decay rates of two different EV (poliovirus 1 and echovirus 1) [89].Few years later temperature, dissolved oxygen concentration and the presence of microorganisms were also listed as factors influencing EV survival [90].The presence of groundwater microorganisms was then described as the most influential factor affecting the decay of viruses, while temperature, soil moisture or desiccation, adsorption to soil oxygen and nutrient levels indirectly influence virus decay by influencing the activity of the groundwater bacteria [91,92].Viruses typically persist longer in moist environments compared to dry conditions but also soil type seems to protect differently the viral particles due to the absorption strength.Many authors have stated that clay soils increase virus survival rates compared to sandy soils [93].A linear correlation is stablished between temperature and the viral decay, indicating a clear relationship between virus survival and temperature [87,94].
Some studies have adapted the methodology and used genome quantification assays to increase sensitivity and reproducibility in the survival study results [74,95].Other authors have used culturable surrogates (Murine norovirus and Tulane virus) to assess the persistence of nonculturable viruses like the human norovirus [96,97].Human norovirus surrogates remained infectious in a cell culture assay up to 100 days of incubation and the presence of intact viral capsids enclosing the genomes (evaluated using a RNase assay coupled with RT-qPCR) was present up to 100 days after inoculation [97].
When considering genome detection, long-term genome persistence of NoV and AdV has been observed.Several studies using spiked AdV showed that its DNA genome can remain detectable 1343 days when stored in dark conditions at 4˚C [59], during 385-day study at 4˚C [74] or during a 672-day study at 12˚C [95].Also NoV RNA genome was detectable after 1277 days [59] and even 3 years [98].Along with an increasing awareness about groundwater contamination, more information about the main factors affecting virus survival in groundwater appears in literature.

Methods for the surveillance of viruses in urban groundwaters
Most of the large surveillance studies (with more than 100 samples) took place in North America [4,5,37,39,42,43,54,58,63,65,67].Only one study addressing frequency and patterns of viral contamination was focused in West Africa [35] and any of the reviewed reports and research papers from Europe or Asia included more than 40 samples (Table 1).Electropositive filtration was the primary viral concentration method used by 64% ( 23 Potential associations between study sample size, sample volume and the percentage of virus detection (positive samples/samples analyzed) were analyzed using Pearson's correlation coefficient tests.A weak and negative Pearson correlation coefficient (r = -0.29)was observed between the percentage of virus incidence and the number of samples analyzed.It needs to be pointed out that the studies included in this review were performed under different scenarios and include: studies performed during an outbreak investigation [41,59], which usually consider few samples but find high percentages of incidence (>50%) and studies aiming to draw an accurate snapshot of presence of viruses in specific groundwater systems [4,38,63,65].Whatever is the aim, and as happens in clinical studies, one can draw a precise and accurate conclusion only with an appropriate sample size.A small number of samples may not give a result that enables to detect a difference between groundwater wells (e.g.aquifer types, drinking water systems).Nevertheless, a very large sample size, which may give a clearer overview of the fate of viruses, will involve higher amounts of time and resources which cannot be covered by each interested country.
Electropositive filtration, which does not typically require preconditioning of the sample, has been the gold standard method used as a primary concentration step over the first 20 years (1999-2018) of waterborne virus research in groundwater (Fig 3).Since 2019, ultrafiltration devices (e.g.Rexeed 25S and Leoceed 21H), previously used for virus concentration [99] and commonly used in clinical settings for hemodialysis, have been employed for sampling larger volumes of water for viruses.Viral recoveries from hollow-fiber ultrafilters are higher than those from electropositive filters and have the added benefit of being a multi-pathogen concentrator [99].
Regardless of the concentration methodology, the sample volume analyzed has decreased over the years among the studies examined (Fig 4).Most of the studies have been published over the second half of the studied period being HAdV detected incidence the highest compared to the incidence of other viruses analyzed.As stated in section 1, more than 80% of the literature analyzed and detected NoV as the major source of the outbreaks although the HAdV occurrence was higher in all the studies where both were analyzed.This confirms that the evaluation of the presence of HAdV could be a useful tool to predict the presence of other waterborne pathogens in groundwater.
There is little existing literature on groundwater virome.The relatively low abundance of viral particles within groundwater systems requires cumbersome library preparation protocols to enrich viral RNA and DNA.Despite, pre-amplification protocols, studies reporting the diverse viral communities describe Myoviridae, Siphoviridae and Podoviridae as the three most dominant families in groundwater [13,100,101].All three families are mainly comprised of tailed dsDNA bacteriophages that infect bacteria and archaea.In the recent years, three massive sequencing approaches (High Throughput Sequencing (HTP), Target Enrichment Sequencing (TES) and Amplicon Deep Sequencing (ADS)) have been tailored to study viruses infecting human and other vertebrates [11,12,14,68,102].To our knowledge only one study was able to detect waterborne viruses, HAdV 41 and HEV, using a SISPA pre-amplification and a HTP sequencing approach [12].Avian adenoviruses and sequences related to human and bovine alpha-and gamma-herpesvirus were found in other studies mapping double stranded DNA virus from sub-Saharan urban settlements [11].While targeted approaches like TES, that uses biotinylated oligonucleotide probes designed to bind coding sequences of specific viral taxa known to infect vertebrates, can provide a snapshot of unexpected viral pathogens and potentially zoonotic pathogens from large volumes of water [14,68], ADS may provide higher sensitivity for the identification of viral types or variants over specific outbreak investigation.

Conclusions
Waterborne viruses enter urban aquifers through multiple sources and pathways, mainly from sewer leakage (a.k.a.exfiltration) but also combined sewer overflows, urban runoff and WWTP effluent recharge areas.Although groundwater sources are considered resilient sources of drinking water, viruses (and specially norovirus) are reported as the main cause of gastrointestinal outbreaks linked to consumption of contaminated groundwater.Viral filtration, adsorption, survival, and inactivation within the aquifer are well known processes.Nevertheless, systematic studies considering a representative set of waterborne pathogens are needed in each scenario, considering that high thermal stability and long survival periods, may not mimic that of pathogens that are present due to recent contamination.There is also a limited systematic monitoring data for waterborne virus in urban groundwater with only 36 references available.Thus, viral risk assessment studies [54,67] related to groundwater consumption are commonly performed during viral outbreaks resulting in poorly defined risks because of the limited temporal resolution.Large surveillance studies have been focused within North America, where there are specific legislations and funds to study groundwater, whereas most of the urban population using groundwater as a source is in Africa and Asia.Important knowledge gaps for a sustainable management of the groundwater sources may need large volume sample methods coupled to target massive sequencing approaches.Understanding the range of pathogens capable of contaminating groundwater is important for assessing risks.Passive sampling, as an alternative to grab sampling for the monitoring of viruses in diverse water environments [103], may be the key to popularize microbial quality monitoring, enabling larger sample sizes and/or higher sampling frequencies.Since the Urban Waste Water Treatment legislations do not require disinfection prior to discharge and uncontrolled discharges of sewage will continue to occur, sustainable and biobased solutions are the only way of reducing viral pathogens in the receiving groundwater bodies.

Fig 3 .
Fig 3. Solar chart and heatmap summarizing the combination of primary and secondary concentration methodologies used by the reviewed studies.Year of publication (from 1999 to 2022) and bibliography reference numbers are indicated in each solar slice, representing each reviewed research study.Heatmap cells contain the number of articles using a specific combination of methodologies.Colors spanned from purple (frequently used combination) to yellow (combination used in one research paper).https://doi.org/10.1371/journal.pwat.0000168.g003

Fig 4 .
Fig 4. Percentage of virus detection according to sample volume analyzed in each of the studies included in this review.Concentric circles represent waterborne viruses (Human adenovirus (HAdV), Enterovirus (EV), Norovirus (NoV), Rotavirus (RoV) and Hepatitis A and E viruses (HAV and HEV)) and one human fecal indicator (HPyV (Human polyomavirus)) analysed and detected in each of the revised literaure.https://doi.org/10.1371/journal.pwat.0000168.g004

Table 1 . Studies investigating waterborne virus contamination in urban groundwaters, percentage of incidence and viruses analyzed in each study.
Purple indicates that the viral DNA or RNA has been analyzed and detected and grey analyzed but not detected in any of the tested samples.AdV: adenovirus, PyV: polyomavirus, EV: enterovirus, NoV: norovirus, RoV: rotavirus, HAV and HEV: hepatitis A and E virus. https://doi.org/10.1371/journal.pwat.0000168.t001