Sewage Markers as Determinants to Differentiate Origins of Emerging Organic Pollutants in an Urban Sri Lankan Water Drainage Network

to Differentiate Abstract: Urban sanitation is a major challenge during the rapid urbanization being experienced by developing countries, as a low sewerage infrastructure capacity and irregular onsite wastewater treatment raise the risk of surface water contamination. The application of speciﬁc sewage markers to characterize contaminant sources is therefore essential for managing urban sanitation issues. In this study, we investigated the concentrations of eight sewage markers (acetaminophen, caffeine, carbamazepine, cotinine, sulfamethoxazole, sulfapyridine, atenolol, and acesulfame) in various water sources within urban area of the Galle City, Sri Lanka. The total concentration of the eight markers was in the order of hospital discharge > sewage treatment plant (STP) inﬂuent > surface drainage system. Among the eight selected markers, acetaminophen was dominant in hospital discharge (70.2–123.6 µ g/L) while caffeine was the largest contributor to STP inﬂuent (16.2–68.7 µ g/L) and surface drainage (0.95–21.73 µ g/L). We then proposed and tested a set of criteria for evaluating the applicability of markers, including removal efﬁciency, concentration magnitude, excretion rate, and wastewater burden. The labile markers caffeine and acetaminophen were suitable for characterizing domestic gray and black wastewater, respectively. These results imply that the city’s drainage system receives both domestic graywater and human excretion, likely due to insufﬁcient on-site sanitation systems. The conservative marker carbamazepine was useful for tracking hospital residues over long distances; these results imply that hospital wastewater treatment was not working properly, accounting for pharmaceutical residues reaching surface water via a hidden discharges connected to the drainage system.


Introduction
According to the estimates and projections of the world urban and rural populations published by the United Nations, over half of the global population currently lives in urban areas. It is estimated that by 2050, the urban population will be 6.68 billion, compared to 3.09 billion in rural areas. The urbanization rate is two-to threefold more rapid in less developed regions than developed regions. The rate of change is projected to be fivefold in the near future [1].
In developing countries, rapid urbanization often exceeds urban infrastructure development, especially in terms of wastewater management [2]. Poor wastewater management and the outdated sanitation facilities continuously pose a potential pollution problem for the aquatic environment in the developing context [3][4][5]. The source of contaminants is difficult to identify because discharge pipes are often underground and disposed together with rainwater drainage. This raises the need for reliable markers to differentiate the pollutants' source in order to preserve healthy urban landscape and water quality and enhance environmental protection regulation.
Among the wastewater markers, the active compounds in pharmaceuticals have recently drawn attention [6]. Pharmaceuticals are ubiquitous in the aquatic environment in trace levels, and the main release pathway of pharmaceuticals into the environment is through human wastewater [7,8]. Ecotoxicology tests have demonstrated acute effects of high concentrations (g/L) of some compounds, such as antibiotics, diclofenac, ibuprofen, and carbamazepine, on fish, Daphnia, algae, and bacteria [7]. The potential human health hazards, however, was thought to be negligible [9][10][11]. Therefore, it is interesting to consider their occurrence as sewage markers due to various advantages, such as their high detection frequency, the availability of resistant markers, and especially the specificity to human fecal waste [6]. A number of application studies regarding pharmaceuticals in groundwater, surface water, and drinking water in developed countries have been well documented [7,[11][12][13]. However, the number of studies on pharmaceuticals in the water environment and their application in wastewater tracking in developing countries is still limited [14][15][16][17][18].
Noting that factors such as land use, population, and sanitation types are essential to selecting appropriate markers [6], additional data on pharmaceuticals in urban sanitation in the developing context are essential. In this study, we tested a set of criteria for identifying suitable chemical markers capable of distinguishing raw human excretion, graywater, and hospital contributions to the drainage system in Galle City, the ninth-largest urban area in Sri Lanka. Almost one-fifth of the Sri Lankan population is living in urban settings. Basic sanitation access in urban Sri Lanka has grown slightly, from 85% to 88%, over the period of 2000-2015 [19]. A 2016 Sri Lankan National Demographic and Health Survey reported that at that time 80% of the urban population relied on simple latrines, with only a negligible 5% equipped with a septic tank, and it remains unknown whether Sri Lankan urban on-site sanitation meets the requirements of UN sustainable development goals [20]. The sanitation conditions here are representative of other growing urban areas in low-income countries. The study results can provide a basis for improved wastewater management in Sri Lanka and elsewhere.

Pre-Survey
The center of Galle City (Galle Four Gravet Divisional Secretariat) has the highest population density in the Southern Province (4212 per km 2 ) [20], but the city has no centralized sewage treatment plants. We initially investigated the sanitation situation here using a questionnaire survey of 31 random local households during October 2019 (Supplementary Materials) to assess how toilet/laundry/bathroom/kitchen wastewater was collected and disposed of on-site; options included septic systems, pit latrines, and direct discharge (open defecation).

Sample Collection
We collected water samples at sites representative of sewage treatment plant (STP) influent and effluent, hospital discharge, and surface water. Samples were collected in 50 mL polypropylene bottles (duplicated at each sampling site), acidified to pH 2 using 1 M hydrochloric acid, and kept frozen before transportation to Tokyo, Japan. Three sampling sites were considered.
The Hikkaduwa STP (6 • 8 54.85 N, 80 • 6 34.53 E, not shown on the map) is the only municipal wastewater treatment facility in the Southern Province. This STP receives and treats domestic wastewater of Hikkaduwa town, located 20 km northwest from the study site. It uses 2 facultative ponds (6500 m 3 and 6300 m 3 ) and a polishing pond (7000 m 3 ) to achieve a retention time of 3-4 days before discharging treated water into a natural lake.
Influent samples were collected after the sand-screening stage, while effluent samples were collected from the discharge mouth of the polishing pond.
Karapitiya Teaching Hospital is the largest national hospital in the Southern Sri Lanka with 600 beds. Its on-site sewage treatment facility uses an activated sludge process followed by chlorination and discharges the treated water into the municipal drainage network. Hospital-treated discharge was collected at the effluent mouth (6 • 3 54.34 N, 80 • 13 30.06 E).
The Galle municipal drainage network consists of the Moragoda, Moda, and Kepu canals, which mainly function to mitigate floods caused by excess storm water during the long rainy season from October to March. However, these canals also receive wastewater from residential areas as well as treated effluent from hospital wastewater. Grab samples were collected from the middle of these channels using a rope and a stainless-steel bucket.

Chemicals and Standards
The target chemical markers included the artificial sweetener acesulfame (ACS), the stimulants caffeine (CAF), the tobacco metabolite cotinine (COT), over-the-counter acetaminophen (ACT), and four prescribed pharmaceuticals (atenolol, ATE; carbamazepine, CBZ; sulfamethoxazole, SFM; and sulfapyridine, SFP). This list was selected based on three criteria: (i) detection in screening test results from the Hikkaduwa STP (Table S1), (ii) a literature review of ubiquitous chemical markers in surface water, and (iii) an adaptation of current analysis methodology. High-purity grade (>98%) pharmaceutical standards and internal standards were purchased from Sigma-Aldrich (Darmstadt, Germany). Further details of the target compounds are provided in the SI. Stock solutions were prepared in MeOH and stored in a refrigerator at −20 • C. Mobile phase solutions (LC/MS grade), water, and formic acid were obtained from Wako (Tokyo, Japan).

LC-MS/MS Analysis
Prior to analysis, samples were thawed and filtered by a single-use 0.2 µm syringe filter with a regenerated cellulose membrane (Minisart RC 4, Sartorius™, Germany). Filtered samples were transferred into a 1.5 mL glass amber vial and spiked with internal standards. Samples were then ready to be injected into an LC-MS/MS. The total time from sample collection to analysis was 4 days. Instrumental analysis was carried out using a DIONEX UltiMate 3000 (Thermo Scientific, Waltham, MA, USA) coupled to a triple quadrupole mass spectrometer (MS/MS) (ABSciex 5500 QTRAP (Sciex, Concord, Ontario, Canada). Chromatographic separation was achieved on a Kinetex Biphenyl column (2.6 µm, 100 Å, LC Column 50 mm × 2.1 mm, Phenomenex), following a published method [21]. A 10point calibration standard curve ranging from 0.1 to 100 µg/L was used for all substances except carbamazepine (0.01-10 µg/L). A blank sample was added to the HPLC run after each 10 samples. Details of method validation are included in the SI.

Removal Efficiency (RE) in the STP
Following previous research suggesting the use of the chemical marker attenuation rate during treatment in the STP to determine persistence [13][14][15], the removal efficiency was estimated using influent and effluent data from the Hikkaduwa STP: where C inf and C eff are the concentrations of specific compounds in the raw and treated wastewater, respectively. Compounds with RE > 90% were determined to be labile markers, while conservative markers were defined as compounds with RE < 25%.

Concentration Magnitude (CM) of Target Compounds in Raw Wastewater
As a reliable wastewater chemical indicator should be significantly higher than the analytical sensitivity [15,16], we defined concentrations 50 times higher than the instrument quantification limit (IQL) in raw wastewater as one screening criterion: where CM ww is the magnitude of the concentration of a specific compound and C ww is the average concentration in the raw wastewater source (either STP or hospital discharge). Details of IQL determination are given in the SI.

Excretion Rate
The target compounds represent either daily consumption (acesulfame, caffeine, and cotinine) or oral medication (acetaminophen, atenolol, carbamazepine, sulfamethoxazole, and sulfapyridine). Once consumed, some compounds are transformed during metabolism while others remain unchanged, before all are excreted along with urine and feces. As it is essential to consider this factor as input information when considering the origin of human wastewater, we collected excretion rates from previous studies and summarized these in the Results and Discussion section.

Wastewater Burden (WB) in Surface Water
The term WB was adapted from a previous study [15] to estimate the wastewater load to surface water using a conservative indicator: where C sur is the concentration of a conservative indicator in the surface water at each sampling site and C ww is the average concentration of a conservative indicator in the effluent wastewater.

Study Area Sanitation
Our survey data showed that pits are dominantly used for on-site sanitation in Galle City (~80%, Table 1), consistent with national data from the 2016 Demographic Health Survey [20]. However, the total lack of sewerage in Galle City is well behind the national rate of 12.3%. This gap is compensated for by a higher percentage of septic tank use (19.2%). Smaller living spaces in urban areas mean less space to cover and safely abandon full latrines and construct new ones or to replace these with more advanced sanitation facilities [5]. Wastewater from kitchens, bathrooms, and laundry areas (i.e., graywater) has traditionally been given lower priority than excretory effluent in sanitation management [21] In rural areas, each house has a dedicated space where discharged graywater can infiltrate into the soil. In contrast, urban graywater is often discharged into drainage networks; for example, two-thirds of the households surveyed reported that their graywater is directly discharged into the city's canals (Supplementary Materials). To distinguish the potential pharmaceutical pathways in black water and graywater, questions about disposal practices were included in the survey, with a majority of respondents selecting "disposed with solid waste" or "burned with solid waste." In general, domestic graywater is unlikely to include pharmaceuticals.

Occurrence and Removal Efficiency of Targeted Compounds in Raw Wastewater
The concentrations of target compounds in raw wastewater from the Hikkaduwa STP and Karapitiya Hospital were comparable with those frequently detected in previous studies in developed countries, confirming the appropriateness of the selected list (Table 2). ACT and CAF accounted for the vast majority of the total concentrations in both STP influent and hospital discharge. In the former, the contribution of CAF and ACT equally made up 45% of the total concentration, while ACT accounted for more than 80% of the latter. The proportion of ACS, though a negligible 1% in hospital discharge, made up nearly a tenth of that in urban raw wastewater (Figure 1). Despite their high concentrations in raw wastewater, up to 99% of CAF and ACT was removed by the STP (Figure 2), indicating their labile characteristic; this was consistent with other studies where CAF and ACT were present in only minute or undetected concentrations in effluent [26,27]. The other compounds were all categorized as conservative according to their RE values. There was no significant difference between the concentrations of ACS, ATE, COT, and SFP in the STP influent and effluent, contradicting a previous report in which the REs of ACS, ATE, and COT varied from 50% to 80%. CBZ is known for its high recalcitrance to biological oxidation, as shown by its negative RE in many studies on pharmaceutical removal in STPs [28][29][30][31]. In our study, the mean concentration of CBZ in the effluent was double that in the influent, reflecting its conservative characteristic. Our SFM also had a negative RE, other than one instance where the influent was higher than the effluent (0.1 µg/L and 0.4 µg/L). Other study has reported moderate to high REs for SFM in STPs [24]. Negative REs can be explained by multiple factors, such as the conversion from conjugated metabolite parent compounds during the WWTP process or the retardation of persistent compounds onto sludge, which is released later in the effluent [27]. This could also be explained by the fact that we did not take hydraulic retention time into consideration (i.e., STP influent and effluent samples were collected on the same day) [24].
The hospital discharge samples showed that the labile compounds ACT and CAF still occurred in elevated concentrations in hospital wastewater (Table 3), although the two compounds were categorized as labile. Wiest et al. [22] documented the ready elimination of ACT in hospital wastewater; although influent ACT levels were~10 times higher than in this study (813.5 µg/L), it was reduced to a minute concentration (0.04 µg/L in the discharge). This indicates that the current treatment process used by the Karapitiya Hospital was not sufficient to remove pharmaceutical compounds. The complicated composition of hospital discharge suggests the need for longer-term monitoring study.

Selecting Wastewater Indicators to Differentiate Pollutant Sources
The concentrations of ACT, CAF, and ACS in urban sewage were significantly higher than those in IQL (>50 times), while ACT, CAF, and CBZ exceeded the criteria for hospital wastewater. Hence, these four were selected as wastewater indicators. The stimulant CAF and the artificial sweetener ACS are included in everyday beverages, such as tea, coffee, and soft drinks, so both black water and graywater may contain these compounds. However, the fact that CAF is readily degradable and entirely metabolized in the human body suggests that black water is not a major source and that it is more likely to be a graywater indicator. ACS, however, is not metabolized and is excreted unchanged through the kidneys. The fact that ACS can be in both graywater and human excretion makes it impossible to differentiate black water sources in surface water. However, due to the persistence of ACS during wastewater treatment, it can be used to generalize the impacts of domestic wastewater on surface water. ACT and CBZ are active pharmaceutical ingredients. The over-the-counter mild pain reliever ACT was marked as labile due to its efficient elimination in STPs (>99%), while the prescribed CBZ is known for its great attenuation. As expected, the CBZ level was significantly lower than that of ACT in urban wastewater owing to its limited use. In terms of the excretion rate, both compounds are extensively transformed during metabolism. Our findings therefore suggest the following applications of these compounds for identifying pollutant sources in Galle City.   Figure 3 presents the total concentration of each marker by sampling site. Notably, elevated CAF concentrations were detected in the middle of the Moda canal (MODA3) and Moragoda canal (MOR3 and MOR4), which can be explained by the higher-density housing in those areas. In addition, the dilution effect of seawater might have lowered the graywater flow in estuary sampling locations. However, two sampling sites in the Kepu canal had equally high total concentrations. ACT and CAF were used to evaluate the impact of urban wastewater as both are ubiquitous in the drainage system within a comparable range (ACT 0.57-4.70 µg/L and CAF 0.95-21.73 µg/L) (Figure 4a). CAF (representing graywater discharge) was dominant at all sampling points except for MOR5. Based on end-use data [40] and the graywater direct discharge rate of 80% obtained in this study, we estimated that~700 m 3 of untreated graywater is disposed of daily into the city's drainage system. We therefore predicted that graywater from domestic discharge is dominant in the drainage system, as demonstrated by the high CAF concentrations. In contrast, the lower ACT levels represent the impact of human excretion on the drainage system, where the most relevant source of black water is likely the insufficient RE of current on-site sanitation methods. This was supported by the trace occurrence of other pharmaceutical compounds. Another possibility is the misconnection of toilet wastewater to the drainage system.

Implications for Pollutants in Galle City Canal Drainage
The WB criteria was applied for the two conservative compounds (ACS and CBZ) to evaluate the impact of hospital wastewater (Figure 4b). Assuming that the urban raw sewage in the Hikkaduwa STP represents the combined sewage discharge to surface water, we used the ratio of ACS concentration at each sampling site to the average concentration of Hikkaduwa STP effluent (ACS/effluent) to normalize the WB from urban sewage to the drainage system. As CBZ also occurred significantly in the hospital wastewater, we used the ratio of CBZ concentration in each sample to the mean CBZ concentration of hospital discharge (CBZ/hospital) to evaluate the potential WB burden at a single site. The WB factor was in the same range in the Kepu and Moda canals, indicating similar levels for both markers. Their trend was similar to that of ACT (Figure 4a), supporting the hypothesis that ACT, ACS, and CBZ were derived from black water of the same origin. However, ACS and CBZ were remarkably elevated at sampling sites MOR3 and MOR4 in the Moragoda canal, where the CBZ burden was approximately four times higher than that of ACS at that site. It is therefore likely that other sources, rather than black water, discharge into the Moragoda canal between sites MOR2 and MOR3. The foremost cause of this discrepancy is thought to be hospital discharge with a high concentration of CBZ. The hospital nearest to the elevated sites is the Karapitiya Hospital, where discharge was collected as a reference for hospital wastewater. The distance from this hospital and MOR3 is <2 km, as measured on Google Maps. Since CBZ has been recognized for its preservation, it is highly possible that CBZ from the hospital is transported to the mainstream via small tributaries of the drainage system. In fact, one study conducted in South Africa detected CBZ as far as 6-8 km downstream of discharge from a centralized STP [16]. Further data collection is required to determine exactly how seasonal variation affects the flow and concentration in the drainage system.

Conclusions
Our results highlight existing issues in urban sanitation in Galle City, Sri Lanka, where graywater is discharged untreated and black water is discharged after simple treatment, while proposing and testing a process for identifying wastewater sources. Consideration of the magnitude, persistency, and metabolism rates of eight target sewage makeres in urban wastewater (including black water, graywater, and hospital wastewater) suggests that CAF is useful for understanding the magnitude of graywater discharge and the presence of labile ACT is evidence of the mixing of raw black water in surface water. We recommend the widespread installation of well-functioning septic treatment systems that can manage both black water and graywater. Treated hospital wastewater included trace amounts of active pharmaceutical ingredients that could contribute to surface water pollutants. The use of persistent indicators, such as CBZ, to identify hospital wastewater discharges into the drainage network proved useful. Although the number of target compounds in this study was small, these were clearly identified in local sewerage and are among the most ubiquitous compounds in human wastewater.
Author Contributions: D.T.T.Q. was responsible for methodology, validation, investigation, formal analysis, data curation, and writing of the original draft; O.M. was responsible for supervision, project administration, and funding acquisition; Y.O. was responsible for supervision, survey design, and project administration; and T.C. was responsible for conceptualization and resources. All authors have read and agreed to the published version of the manuscript.
Funding: This research was funded by the Japan Society for the Promotion of Science, Grants-in-Aid for Scientific Research JP20H04391.
Data Availability Statement: Not applicable.