Canals, Contaminants, and Connections: Exploring the Urban Exposome in a Tropical River System

Abstract

Emerging and persistent contaminants (EPCs) were detected at high concentrations in Chiang Mai’s Mae Kha Canal, identifying urban waterways as important sources of pollution in the Ping River system in northern Thailand. Maximum levels of metformin (20,000 ng/L), fexofenadine (15,900 ng/L), gabapentin (12,300 ng/L), sucralose (38,000 ng/L), and acesulfame (23,000 ng/L) point to inadequately treated wastewater as a plausible contributor. Downstream enrichment patterns relative to upstream sites highlight the cumulative impact of urban runoff. Five compounds—acesulfame, gemfibrozil, fexofenadine, TBEP, and caffeine—consistently emerged as reliable tracers of urban wastewater, forming a distinct chemical fingerprint of the riverine exposome. Median EPC concentrations were highest in Mae Kha, lower in other urban canals, and declined with distance from the city, reflecting spatial gradients in urban density and pollution intensity. Although most detected concentrations fell below predicted no-effect thresholds, ibuprofen frequently approached or exceeded ecotoxicological benchmarks and may represent a compound of ecological concern. Non-targeted analysis revealed a broader “chemical cocktail” of unregulated substances—illustrating a witches’ brew of pollution that likely escapes standard monitoring efforts. These findings demonstrate the utility of wide-scope surveillance for identifying key compounds, contamination hotspots, and spatial gradients in mixed-use watersheds. They also highlight the need for integrated, long-term monitoring strategies that address diffuse, compound mixtures to safeguard freshwater ecosystems in rapidly urbanizing regions.

1. Introduction

Urbanization and inadequate wastewater infrastructure in tropical cities are driving widespread—and often underrecognized—chemical contamination of freshwater systems [1,2]. These water bodies receive complex mixtures of anthropogenic pollutants, including pharmaceuticals, industrial additives, food chemicals, and microplastics, via stormwater runoff, canal discharges, and untreated or poorly treated wastewater. Such chemical cocktails pose poorly quantified risks to aquatic ecosystems and public health throughout much of the tropical world [3,4]. Among these pollutants, emerging and persistent contaminants (EPCs) are of growing concern due to their ecological and human health impacts [5,6,7]. EPCs enter surface waters through multiple pathways, including leakage, illegal dumping, and especially the discharge of inadequately treated wastewater [8,9,10,11,12].

In urban areas, runoff frequently travels through modified streams and canal networks that discharge directly into rivers, creating efficient conduits for EPC transport. Wastewater treatment plants also contribute, as conventional technologies often fail to remove pharmaceuticals and other synthetic chemicals effectively [13,14]. As a result, surface waters commonly contain complex mixtures of pollutants—including pharmaceuticals, personal care products, agricultural inputs, microplastics, food additives, and industrial compounds—along with their metabolites and transformation products [15,16,17,18].

Understanding the pathways that deliver EPCs to rivers is essential for evaluating their environmental impacts across spatial and temporal scales [19,20]. Mapping these linkages supports pollution control, informs human health risk assessments, and enables long-term monitoring and management [21,22]. Urban centers—with high population densities, extensive impervious surfaces, and concentrated industrial and domestic discharges—are particularly important sources of EPCs and related pollutants to adjacent waterways [23,24]. Recognizing and quantifying these contributions is central to effective water governance and pollution mitigation.

Despite growing awareness of EPC-related risks, few studies in tropical cities have quantified the contribution of canal networks to downstream river contamination or characterized their internal pollutant burdens (reviewed in the discussion section). In northern Thailand’s Ping River watershed, we previously found that urban areas exhibited the highest concentrations of pharmaceuticals and daily-use compounds, with minimal seasonal variation—likely due to persistent anthropogenic inputs and hydrological mixing [4]. A follow-up study that combined targeted chemical analysis, non-targeted screening, and in vitro bioassays highlighted significant methodological challenges in resource-limited settings, reinforcing the need for improved analytical workflows and broader adoption of effect-based monitoring approaches [25]. Additionally, atmospheric deposition was identified as a diffuse contamination source, capable of delivering pollutants to rural surface waters even in the absence of direct human activity [26]. In one upstream agricultural catchment, atrazine concentrations spiked following wet-season runoff but generally remained below ecological risk thresholds [27]. Nonetheless, the complexity of these EPC mixtures raises concerns about potential ecological and synergistic effects that remain poorly understood.

This study advances regional understanding by explicitly examining the role of an urban canal system as both a contamination hotspot and a conduit for EPC transport to downstream waters. We focus on the export of EPCs from Chiang Mai’s Mae Kha Canal to the adjacent Ping River. Using a novel sampling strategy, we compare monthly EPC concentrations in the Ping River—measured upstream and downstream of the city—with data from seven sites across the canal network, including three canal–river discharge points. We hypothesize that EPCs transported through the canal system contribute to elevated concentrations downstream relative to upstream. A secondary objective is to evaluate the internal chemical burden of the Mae Kha Canal itself, a chronically degraded waterway that has received little systematic contaminant screening to date [28,29,30,31,32,33,34,35,36].

By directly linking contaminant concentrations in canals with downstream river conditions, this study provides one of the first empirical demonstrations of canal-to-river EPC transport in a tropical urban setting. It contributes to ongoing efforts to identify EPC sources and transport pathways in the Ping River catchment [4,25,26,27], now one of the most extensively studied systems in the region. This work also complements national studies of other Thai rivers [37,38,39,40,41,42,43,44,45,46,47,48,49,50] and aims to inform the identification of priority contaminants and the development of robust monitoring strategies to protect aquatic ecosystems and public health.

2. Study Area

Chiang Mai is a city with a rich historical legacy, a dynamic economy, diverse environmental settings, and increasing urbanization (Figure 1). Formerly the capital of the Lanna Kingdom, it is renowned for its ancient temples and enduring cultural identity. Today, Chiang Mai is Thailand’s fourth most populous province, with the city and surrounding boroughs forming a major urban center in the Ping River Catchment [31]. The area has evolved into a regional hub for technology, tourism, trade, and residential development. The urban landscape comprises residential neighborhoods, commercial districts, healthcare and educational facilities, light industry, and green spaces. These activities also extend to the surrounding boroughs, adding to the potential source areas for emerging and persistent contaminants.

Figure 2. Sampling sites in the Ping River and Mae Kha Canal system: (a) P-14, Ping River upstream of the city; (b) P-55, Mae Kha Canal before entering the urban core; (c) P-10, bridge near the canal’s outlet to the Ping River; (d) P-11, canal segment near the outer ring road; (e) P-08, Mae Kha Canal within the city; (f) P-05, Mae Kha Canal within the city; (g) P-41, canal outlet to the Ping River in the southern city area; (h) P-39, canal system outlet to the Ping River south of the urban boundary; and (i) P-40, Ping River downstream of Chiang Mai.

Figure 2. Sampling sites in the Ping River and Mae Kha Canal system: (a) P-14, Ping River upstream of the city; (b) P-55, Mae Kha Canal before entering the urban core; (c) P-10, bridge near the canal’s outlet to the Ping River; (d) P-11, canal segment near the outer ring road; (e) P-08, Mae Kha Canal within the city; (f) P-05, Mae Kha Canal within the city; (g) P-41, canal outlet to the Ping River in the southern city area; (h) P-39, canal system outlet to the Ping River south of the urban boundary; and (i) P-40, Ping River downstream of Chiang Mai.

The Ping River—one of the largest rivers in northern Thailand—originates in the highlands north of Chiang Mai and flows southward to join the Chao Phraya River at its confluence with the Nan River, over 450 km downstream. Its upper reaches drain mixed-use catchments, including forested, agricultural, and peri-urban areas, which likely contribute EPCs to the river prior to entering the city (Lee et al., 2024) [4]. The regio experiences a tropical savanna climate, with annual rainfall of 1000–1400 mm, largely concentrated during the southwest monsoon from May to November [51,52]. The climate can be divided into three distinct seasons: (1) a rainy season from June to late October, (2) a long dry season from November to March, and (3) a hot, dry season from March to May.

Running largely parallel to the Ping River, the Mae Kha Canal serves as a key drainage conduit, channeling surface runoff from the city into one of northern Thailand’s largest rivers [34]. Originally a natural stream, the canal has long been criticized for its malodor and pollution, with local authorities struggling to reverse its degraded state [30,32,33]. The pollution is thought to stem from untreated sewage and urban runoff [31], a condition likely exacerbated by rapid urban expansion and road development [35]. Past environmental studies of the canal have focused on a limited set of indicators—including potentially toxic elements, fecal bacteria, dissolved oxygen, basic water parameters (pH, turbidity, temperature), and benthic macroinvertebrates [28,29,30,31,32,33,34,35,36].

Chiang Mai’s centralized wastewater treatment plant serves only about 50% of the city’s population [4]. Roughly 80% of households rely on on-site systems—such as septic tanks and cesspools—even in neighborhoods covered by centralized infrastructure [48,49]. The fate of wastewater from commercial businesses, healthcare facilities, manufacturing sites, and residences remains poorly understood, particularly regarding whether it enters canals directly or indirectly via groundwater leaching. Limited treatment infrastructure, inadequate enforcement, and a lack of sewage collection systems amplify the challenges associated with decentralized wastewater management [50]. These conditions raise concerns about the diffuse entry of EPCs into surface waters through direct greywater discharges.

3. Methods

3.1. Overview

This study involved monthly water EPC data collection at nine locations within the Mae Kha Canal—Ping River system (Figure 1). The upriver sampling site (P-14) is 42 km above Chiang Mai City; the downriver site (P-40) is approximately 60 km downstream of P-14, 18 km downstream of the city. Three locations are where Mae Kha Canal exits into the Ping River (P-10, P-39, P-41). Four other locations are located within the canal system (P-55, P-11, P-08, P-05).

Samples were collected from June 2022 to June 2023, for a total of 13 months, although some sites may have 12 to 14 samples owing to replication or a later start. All 1000 mL grab samples were collected in new high-density polypropylene (HDPE) amber bottles rinsed with ultrapure water. Prior to analysis, the samples were spiked with 1 g/L of sodium azide to inhibit microbial activity. The samples were stored at 4 °C for less than one week and shipped cold to Singapore for analysis. The compounds selected for analysis in this paper include three agrochemicals (atrazine, 2,4-D, fenobucarb), three industrial compounds (4-Nitrophenol, TBEP, tolyltriazole), three daily-use compounds (sucralose, acesulfame, caffeine), nine prescription pharmaceuticals (atenolol, carbamazepine, fluoxetine, gabapentin, iomeprol, metformin, metoprolol, sulfamethoxazole, valsartan), five behind-the-counter pharmaceuticals (diclofenac, fexofenadine, gemfibrozil, ibuprofen, naproxen), and two over-the-counter pharmaceuticals (acetaminophen, diphenhydramine). Carbamazepine, diclofenac, gemfibrozil, fluoxetine, metformin, metoprolol, and naproxen were identified as prioritized EPCs in surface water in China or the UK [53]. After an initial data screening, fluoxetine and iomeprol were excluded from analysis because of low concentrations—leaving a total of 24 EPCs for analysis.

Our primary analysis focuses on median concentrations ± median absolute deviations, as well as concentration ranges across different water sample types. Due to the small number of samples, robust statistical analysis was not feasible (see discussion below). To assess the potential contribution of the urban system to downstream river contamination, we calculated enrichment factors (EF) using the following formula:

ERF = C_P-40/C_P-14 × 100

(1)

where C is an EPC concentration at either the downriver (P-40) or the upriver (P-14) sites. If a C was below detection limit, one-half the detection limit was used. Therefore, for samples with P-14 concentrations below detection limits, enrichment factors tend to be inflated, and caution is needed when interpreting the results.

3.2. Analytical Analysis

Determination of EPC concentrations followed the established targeted analysis protocols we employed in the past [4]. Briefly, filtered water samples (0.2 μm PES syringe filters) were spiked with an internal standard and analyzed by online solid-phase extraction (SPE) coupled with high-resolution LC-MS/MS (Agilent 6495 QQQ, Santa Clara, CA, USA, dual AJS-ESI source) using InfinityLab Poroshell 120 EC-C18, Dayton, OH, USA, columns. Chromatographic and mass spectrometry settings match those reported elsewhere [4].

Limits of detection (LOD) for each compound were determined as a signal-to-noise ratio of at least 10 above blanks, with most EPCs having LODs of 5 ng/L. Higher values were required for metformin (50 ng/L), caffeine (40 ng/L), gabapentin (25 ng/L), naproxen and TBEP (15 ng/L), and acesulfame and sucralose (10 ng/L). Four quality control samples (100 ng/L) were included in each batch, and replicates or blanks were reanalyzed if internal standards appeared inconsistent. All quantitative data were processed in Agilent MassHunter (v10.0) and post processed in R v4.2.2. QA/QC results (lab standards and field blanks) showed high consistency (see Supplement in reference [4]).

Non-targeted screening (NTS) followed modified protocols [54,55], using filtered extracts and the same chromatographic setup. Extracts and methanol blanks were analyzed in MS1 positive mode on an Agilent G6550A iFunnel Q-ToF LC-MS with three replicate injections per sample; methanol blanks were included after every 10 injections. Full LC-QToF parameters and post-processing methods are provided in Supplement S5 of our prior work in the study area [25].

4. Results

4.1. EPC Concentrations and Detection Frequencies

Table 1. Measured concentrations (ng/L) for 24 EPCs at upriver (P-14), downriver (P-40), urban canal, and canal exit sites.

	P-14	Canals	Exits to River	P-40	PNEC
Sample Number	13	53	39	13
Atenolol	bdl	122 [bdl–231] 0.85	46 [bdl–416] 0.64	bdl [bdl–10] 0.08	150,000
Carbamazepine	bdl	19 [bdl–79] 0.91	16 [bdl–59] 0.90	bdl [bdl–10] 0.23	2000
Estrone	bdl	bdl [bdl–22] 0.30	bdl [bdl–15] 0.46	bdl	<1
Metoprolol	bdl	151 [bdl–663] 0.92	92 [bdl–592] 0.87	8 [bdl–17] 0.69	8600
Diclofenac	bdl	59 [bdl–251] 0.74	33 [bdl–274] 0.72	bdl [bdl–7] 0.08	40
Ibuprofen	bdl	634 [bdl–1558] 0.98	203 [bdl–1727] 0.77	bdl [bdl–32] 0.31	11
Naproxen	bdl	bdl [bdl–294] 0.38	42 [bdl–192] 0.67	bdl [bdl–22] 0.23	1700
Diphenhydramine	bdl	71 [bdl–778] 0.53	26 [bdl–525] 0.51	bdl [bdl–30] 0.08	990
Acetaminophen	bdl	3526 [bdl–9908] 0.89	259 [bdl–3595] 0.64	bdl [bdl–30] 0.08	46,000
Fexofenadine	bdl [bdl–9] 0.08	1019 [bdl–15,932] 0.96	1015 [108–8883] 1.0	173 [bdl–1730] 0.92	200,000
Gabapentin	bdl	3727 [bdl–11,771] 0.94	2595 [bdl–12,287] 0.97	bdl [bdl–1337] 0.38	1,000,000
Gemfibrozil	bdl	648 [25–1318] 1.0	274 [35–2008] 1.0	28 [10–77] 1.0	500
Metformin	bdl [bdl–259] 0.08	915 [bdl–4145] 0.62	410 [bdl–19,977] 0.62	bdl [bdl–538] 0.38	160,000
Sulfamethoxazole	bdl	294 [bdl–1922] 0.96	137 [bdl–3223] 0.90	14 [bdl–157] 0.69	600
Valsartan	bdl	212 [81–1506] 1.0	190 [bdl–1086] 0.85	bdl [bdl–68] 0.38	560,000
Acesulfame	32 [bdl–72] 0.62	13,221 [1170–22,707] 1.0	4128 [403–21,469] 1.0	516 [165–1474] 1.0	72,400
Caffeine	bdl [bdl–1171] 0.31	4619 [102–11,556] 1.0	1588 [202–5750] 1.0	268 [121–1118] 1.0	1200
Sucralose	bdl [bdl–46] 0.08	4083 [bdl–38,335] 0.94	1395 [bdl–10,695] 0.87	375 [bdl–2048] 0.62	29,700
4-Nitrophenol	bdl [bdl–75] 0.31	441 [bdl–8727] 0.89	21 [bdl–1681] 0.74	bdl [bdl–61] 0.31	5000
TBEP	bdl [bdl–121] 0.31	112 [bdl–3223] 0.81	522 [bdl–3950] 0.85	131 [bdl–363] 0.85	44,800
Tolyltriazole	bdl	91 [bdl–2191] 0.81	41 [bdl–681] 0.87	8 [bdl–58] 0.54	8000 (m)
2,4-D	13 [bdl–322] 0.62	bdl [bdl–1303] 0.45	55 [bdl–1712] 0.85	33 [bdl–238] 0.92	600
Atrazine	150 [25–1612] 1.0	45 [bdl–240] 0.92	42 [15–157] 1.0	119 [23–963] 1.0	600
Fenobucarb	bdl [bdl–183] 0.31	bdl [bdl–1159] 0.42	bdl [bdl–493] 0.36	8 [bdl–997] 0.54	2130

Notes: Values are medians, range [min–max]; and detection frequency; bdl refers to below detection limit. Bolding indicates value exceeds the probably no-effects concentrations (PNEC), for freshwater, based on data in the Norman Toxicology Database [56]; (m) means the value is for marine water.

summarizes the median values, maximum concentrations, and detection frequencies of contaminants across four site categories: (1) the upstream river site (P-14; n = 13), (2) the urban canal sites within Chiang Mai city (n = 53), (3) the three Mae Kha canal outlets discharging into the Ping River (n = 39), and (4) the downstream river site (P-40; n = 13). Also reported are predicted no-effects concentrations (PNEC) [56].

Among the most widespread contaminants, the herbicide atrazine was detected at nearly 100% frequency across all sites, with elevated maxima of 1612 ng/L at P-14 and 963 ng/L at P-40. The insecticide fenobucarb showed high concentrations in urban canals (1159 ng/L), canal outlets (493 ng/L), and the downstream site (997 ng/L). Similarly, the herbicide 2,4-D peaked in urban canals (1303 ng/L) and canal outlets (1712 ng/L).

Among industrial compounds, detection frequencies ranged from 74% to 89% in the urban canals and canal exits. TBEP exhibited the highest median concentration among this group, reaching 520 ng/L at canal exits, with comparable maximum levels in urban canals (3223 ng/L) and exits (3950 ng/L). Industrial compound 4-nitrophenol was especially elevated in urban canals, with a median of 441 ng/L, a maximum of 8727 ng/L, and a detection frequency of 80% or more.

Three daily-use EPCs—acesulfame, sucralose, and caffeine—were present at high concentrations in both urban canals and canal exits. Acesulfame reached a median of 13,221 ng/L and a maximum of 22,707 ng/L in canals, with slightly lower values at the exits (4130 ng/L median; 21,470 ng/L max). Sucralose showed a similar pattern: 4083 ng/L (median) and 38,355 ng/L (max) in canals, compared to 1395 ng/L and 10,695 ng/L at exits. Notably, sucralose levels at the downstream river site exceeded 2048 ng/L. Caffeine was detected in 100% of samples from both urban canals and canal exits, with maximums of 11,556 ng/L and 5750 ng/L, respectively.

4.2. Patterns Within the Urban Canal Network

Several pharmaceuticals exhibited elevated median concentrations and high detection frequencies in both the urban canals and canal exits (

Table 2. EPC concentrations (ng/L) in the four urban canals.

	P-55	P-11	P-08	P-05	PNEC
Sample Number	12	13	14	14
Atenolol	8 [bdl–209] 0.5	117 [bdl–231] 0.92	134 [24–198] 1.0	102 [bdl–143] 0.93	150,000
Carbamazepine	23 [bdl–40] 0.92	16 [bdl–33] 0.92	24 [bdl–79] 0.86	20 [bdl–42] 0.93	2000
Estrone	bdl [bdl–12] 0.33	bdl [bdl–22] 0.46	bdl [bdl–16] 0.36	bdl [bdl–6] 0.07	<1
Metoprolol	201 [55–609] 1.0	109 [bdl–467] 0.92	215 [bdl–546] 0.86	155 [bdl–663] 0.93	8600
Diclofenac	29 [bdl–87] 0.75	81 [bdl–251] 0.69	79 [bdl–114] 0.71	40 [bdl–88] 0.79	40
Ibuprofen	85 [bdl–617] 0.92	712 [379–1558] 1.0	755 [101–1380] 1.0	467 [95–1263] 1.0	11
Naproxen	38 [bdl–168] 0.92	bdl [bdl–132] 0.23	bdl [bdl–294] 0.14	bdl [bdl–95] 0.29	1700
Diphenhydramine	bdl [bdl–72] 0.08	147 [bdl–500] 0.69	bdl [bdl–778] 0.43	73 [bdl–325] 0.86	990
Acetaminophen	284 [bdl–2857] 0.75	3330 [bdl–9908] 0.85	5423 [781–7663] 1.0	2695 [4884] 0.93	46,000
Fexofenadine	911 [361–4007] 1.0	1930 [519–15,932] 1.0	612 [bdl–12,460] 0.93	661 [bdl–4383] 0.93	200,000
Gabapentin	2900 [548–9202] 1.0	3552 [bdl–11,771] 0.92	4648 [bdl–6438] 0.93	2844 [bdl–4872] 0.93	1,000,000
Gemfibrozil	85 [25–701] 1.0	648 [362–1191] 1.0	821 [80–1318] 1.0	570 [161–1025] 1.0	500
Metformin	865 [bdl–3230] 0.67	1521 [bdl–4057] 0.69	518 [bdl–4145] 0.57	920 [bdl–3179] 0.57	160,000
Sulfamethoxazole	167 [5–2159] 1.0	294 [bdl–1067] 0.92	409 [bdl–1922] 0.93	99 [38–1786] 1.0	600
Valsartan	247 [89–481] 1.0	328 [109–1506] 1.0	246 [81–578] 1.0	262 [109–1326] 1.0	560,000
Acesulfame	2543 [1107–10,577] 1.0	15,087 [4769–22,707] 1.0	13,331 [2072–22,188] 1.0	8950 [2466–21,567] 1.0	72,400
Caffeine	867 [202–2832] 1.0	4804 [1569–6771] 1.0	4538 [685–9180] 1.0	4713 [1357–11,556] 1.0	1200
Sucralose	954 [bdl–3663] 0.75	6842 [1891–38,335] 1.0	4445 [1085–11,239] 1.0	3119 [bdl–7017] 1.0	29,700
4-Nitrophenol	21 [bdl–735] 0.67	296 [33–5055] 1.0	990 [bdl–18–5169] 1.0	209 [bdl–8727] 0.86	5000
TBEP	158 [bdl–2069] 0.83	101 [bdl–342] 0.85	81 [bdl–2864] 0.71	189 [bdl–3223] 0.86	44,800
Tolyltriazole	7 [bdl–71] 0.58	140 [bdl–2191] 0.85	90 [bdl–1235] 0.93	84 [bdl–1242] 0.86	8000 (m)
2,4-D	bdl [bdl–180] 0.33	28 [bdl–519] 0.54	bdl [bdl–1303] 0.29	28 [bdl–808] 0.64	600
Atrazine	17 [bdl–105] 0.75	42 [21–147] 1.0	50 [19–109] 1.0	45 [bdl–240] 0.93	600
Fenobucarb	bdl [bdl–56] 0.17	20 [bdl–294] 0.54	25 [bdl–1159] 0.57	bdl [bdl–411] 0.36	2130

Notes: Values are medians, range [min–max]; and detection frequency; bdl refers to below detection limit. Bolding indicates value exceeds the probably no-effects concentrations (PNEC), for freshwater, based on data in the Norman Toxicology Database [56]; (m) means the value is for marine water.

and

Table 3. EPC concentrations (ng/L) in the three canal exits to the Ping River.

	P-10	P-41	P-39	PNEC
Sample Number	13	13	13
Atenolol	54 [bdl–416] 0.92	61 [24–110] 1.0	bdl	150,000
Carbamazepine	14 [bdl–59] 0.85	19 [bdl–39] 0.92	15 [bdl–37] 0.92	2000
Estrone	6 [bdl–14] 0.69	6 [bdl–15] 0.54	bdl [bdl–9] 0.15	<1
Metoprolol	194 [45–592] 1.0	123 [29–263] 1.0	8 [bdl–26] 0.62	8600
Diclofenac	41 [27–274] 1.0	34 [bdl–83] 0.77	bdl [bdl–15] 0.38	40
Ibuprofen	262 [141–1727] 1.0	257 [78–558] 1.0	bdl [bdl–39] 0.31	11
Naproxen	55 [bdl–192] 0.69	74 [bdl–171] 0.85	bdl [bdl–27] 0.46	1700
Diphenhydramine	37 [bdl–525] 0.85	33 [bdl–173] 0.69	bdl	990
Acetaminophen	533 [bdl–3595] 0.92	511 [bdl–891] 0.85	bdl [bdl–245] 0.15	46,000
Fexofenadine	1177 [378–8883] 1.0	1530 [574–6445] 1.0	471 [107–1415] 1.0	200,000
Gabapentin	3624 [1292–12,287] 1.0	3076 [1045–6541] 1.0	1763 [bdl–3719] 0.92	1,000,000
Gemfibrozil	294 [201–2008] 1.0	372 [123–584] 1.0	48 [35–158] 1.0	500
Metformin	771 [bdl–19,977] 0.69	1460 [bdl–6815] 0.69	bdl [bdl–493] 0.46	160,000
Sulfamethoxazole	276 [59–3223] 1.0	187 [95–456] 1.0	14 [bdl–75] 0.69	600
Valsartan	201 [bdl–1086] 0.92	510 [bdl–861] 0.92	53 [bdl–121] 0.69	560,000
Acesulfame	4751 [2735–21,469] 1.0	4629 [2001–7779] 1.0	611 [403–1342] 1.0	72,400
Caffeine	2220 [1004–5750] 1.0	2526 [1207–3682] 1.0	278 [102–909] 1.0	1200
Sucralose	1216 [bdl–10,695] 0.85	1665 [bdl–10,358] 0.92	861 [bdl–2556] 0.85	29,700
4-Nitrophenol	30 [bdl–1681] 0.85	84 [5–1189] 1.0	bdl [bdl–53] 0.38	5000
TBEP	645 [bdl–2265] 0.85	1649 [bdl–3950] 0.85	214 [bdl–636] 0.85	44,800
Tolyltriazole	15 [bdl–681] 0.77	60 [17–278] 1.0	63 [bdl–143] 0.85	8000 (m)
2,4-D	36 [bdl–115] 0.69	55 [bdl–408] 0.85	74 [15–1712] 1.0	600
Atrazine	41 [16–105] 1.0	53 [25–157] 1.0	40 [15–136] 1.0	600
Fenobucarb	bdl [bdl–110] 0.38	bdl [bdl–62] 0.31	bdl [bdl–493] 0.38	2130

Notes: Values are medians, range [min–max]; and detection frequency; bdl refers to below detection limit. Bolding indicates value exceeds the probably no-effects concentrations (PNEC), for freshwater, based on data in the Norman Toxicology Database [56]; (m) means the value is for marine water.

). Maximum concentrations were especially high for metformin (19,977 ng/L), fexofenadine (15,933 ng/L), and gabapentin (12,287 ng/L). Both fexofenadine and gabapentin were detected in 94–100% of samples from canals and exits and remained elevated at the downstream river site (>1000 ng/L), indicating persistence and downstream transport potential. Acetaminophen also stood out, with a median of 3526 ng/L, a maximum of 9908 ng/L, and a detection frequency of 89% in the canals.

Spatial trends revealed localized point-source inputs for several non-ubiquitous compounds. Tolyltriazole, 4-nitrophenol, and fenobucarb peaked at site P-11 before declining downstream, suggesting industrial or agricultural origins. For example, tolyltriazole rose from 7 ng/L at P-55—which is above most of the dense urban infrastructure—to 140 ng/L at P-11. These patterns suggest strong anthropogenic loading at P-11 and a weakening of source–canal connectivity downstream, perhaps due to dilution or in situ transformation. At the three canal exit sites (P-10, P-41, and P-39; Table 3), pharmaceutical concentrations were highest at P-10 and P-41—particularly for acetaminophen, metformin, and gabapentin—consistent with direct municipal inputs. At P-39, many EPCs dropped below detection limits, indicating attenuation from degradation, dilution, or reduced loading.

Marked spatial contrasts were also observed among the four primary urban canal sites (Table 2). Site P-11 consistently showed the highest maximum concentrations for multiple compounds, including acetaminophen (3330 ng/L), gabapentin (3552 ng/L), caffeine (4804 ng/L), acesulfame (15,087 ng/L), and sucralose (6842 ng/L). This pattern points to strong healthcare-related wastewater influence near P-11. Concentrations of several pharmaceuticals—gabapentin, fexofenadine, acetaminophen, and ibuprofen—decreased progressively downstream (P-11 → P-08 → P-05), consistent with dilution or transformation processes. For instance, gabapentin declined from 11,771 ng/L at P-08 to 4872 ng/L at P-05, yet remained somewhat elevated at the two “downstream” canal exits: 6541 ng/L (P-41) and 3719 ng/L (P-39).

Sucralose followed a similar pattern, decreasing from 38,335 ng/L at P-11 to 7000–11,000 ng/L downstream (P-08, P-05, P-41), then dropping below 2600 ng/L at P-09. In contrast, ibuprofen exhibited relatively stable concentrations across the three city canal sites. A concurrent rise in 2,4-D concentrations at P-39, paired with the decline of pharmaceuticals, suggests increasing agricultural influence in peri-urban areas above this site. Compounds such as tolyltriazole and TBEP exhibited more irregular distributions, likely reflecting diffuse, legacy, or non-point sources such as runoff from roads or other surfaces draining urban contamination sources, such as industrial parks, mechanical works centers, and manufacturing centers, and vehicle repair shops.

In recap, across all sites, the highest concentrations were recorded for four pharmaceuticals—gabapentin, metformin, fexofenadine, and acetaminophen—and three daily-use compounds: acesulfame, caffeine, and sucralose. Industrial contaminants like TBEP and 4-nitrophenol also reached elevated levels, particularly in urban canals and at canal exits. While some EPCs (e.g., fluoxetine, naproxen) were detected sporadically, others—such as gemfibrozil and diclofenac—were consistently present. Metformin stood out for its broad spatial distribution, high variability, and frequent detection. These findings underscore the persistent and spatially structured signatures of EPCs in the urban aquatic system, reaffirming the role of urban canals as major transport corridors and concentration zones for emerging contaminants within the broader river network.

4.3. Potential Enrichment

Enrichment factors (ERFs; Equation (1)) indicated higher median concentrations of ten compounds at downstream site P-40 compared to both the P-14 upstream locations (

Table 4. Enrichment factors (ERFs) based on P-40 > P-14 concentrations.

Compound	Samples Enriched *	Median ERF	ERF Range	Comment
Acesulfame	100%	13	5 to 590	Constant
Gemfibrozil	100%	11	4 to 31	Constant
Fexofenadine	92%	72	11 to 692	Near-constant
TBEP	85%	21	2 to 60	Near-constant
Caffeine	77%	4	2 to 7	Frequent
Sucralose	62%	70	20 to 273	Episodic
Sulfamethoxazole	62%	6	<2 to 31	Episodic
2,4-D	62%	7	<2 to 18	Episodic
Tolyltriazole	54%	10	3 to 23	Episodic
Fenobucarb	54%	8	3 to 32	Episodic

* Sample number is 13. ERFs are determined by Equation (1). Very high maximum ERF values should be interpreted with caution, as upstream concentrations were often < limit of detection.

). Elevated ERF values (>100 ng/L) should be interpreted with caution, as P-14 concentrations may have been below detection—this method is intended for indicative comparisons only and not for absolute quantification.

The pharmaceutical gemfibrozil and the artificial sweetener acesulfame exhibited constant downstream enrichment (Table 4): all P-40 values were greater than those at P-14. Acesulfame had a median ERF of 13, meaning that at least half of the concentrations at P-40 were at least 13-fold higher than those at the upstream site P-14. Similarly, gemfibrozil had a median ERF of 11, also reflecting substantial downstream enrichment.

Near-constant enrichment was observed for fexofenadine (ERF = 72) and TBEP (ERF = 21), with 85–92% of paired samples showing higher downstream concentrations. Caffeine showed frequent enrichment (77%), with a median ERF of 4. Five EPCs exhibited episodic enrichment patterns: sucralose (70), sulfamethoxazole (6), 2,4-D, (7), tolyltriazole (10), and fenobucarb (8). Enrichment occurred in 54–62% of the 13 sample pairs for these five EPCs (Table 4).

Among the five EPCs with frequent-to-consistent enrichment, four had median concentrations below detection limits at the upstream site (P-14) but were consistently detectable downstream at P-40 (Table 1): caffeine (BDL vs. 268 ng/L), fexofenadine (BDL vs. 173 ng/L), TBEP (BDL vs. 131 ng/L), and gemfibrozil (BDL vs. 28 ng/L). Although enriched, their absolute downstream concentrations remained relatively low, with maximum values ranging from 77 to 1730 ng/L (Table 1)—demonstrating that enrichment does not necessarily correlate with acute contamination.

4.4. EPC Time Series

Time-series patterns across the sampling period for selected EPCs further support the transport of contaminants from the urban canal system into the Ping River, particularly given the consistently high concentrations observed at sampling locations in the canals and at their exits to the Ping River (Figure 3, Figure 4 and Figure 5).

Persistent enrichment of gemfibrozil (Figure 3a) and acesulfame (Figure 4a) is evident in their downstream concentrations at P-40, which typically lie between elevated levels at the canal exits and lower concentrations upstream at P-14. Their profiles closely align with those at the P-39 canal outlet, suggesting this exit may be a major discharge point. Fexofenadine (Figure 3b) and caffeine (Figure 4c) also showed enrichment, potentially linked to discharges at P-39, though with less consistency. The P-39 exit tended to have higher concentrations than those at P-41, which ran through per-urban environments.

Episodic enrichment patterns were also observed. For instance, 2,4-D spiked during both wet and dry periods (Figure 5b), while compounds such as sucralose (Figure 4b) and TBEP (Figure 5c) displayed irregular concentration spikes throughout the year, suggesting diffuse or intermittent input sources. However, these interpretations should be tempered by the limitations of monthly sampling, which may miss short-term fluctuations that could be resolved with higher-frequency monitoring [57,58].

Notably, the herbicide atrazine exhibited persistently elevated concentrations at both river sites (P-14 and P-40), in contrast with lower levels observed within the urban canal system (Figure 3a). Intermittent upstream peaks suggest sustained agricultural inputs from upstream catchments, aligning with previous findings [27]. Similarly, the widespread detection of 2,4-D across river and canal sites highlights its ubiquity, with episodic enrichment observed along the urban–rural gradient (Figure 5b). The industrial compound TBEP also displayed episodic spikes, particularly at the downstream river site (P-40; Figure 5c). Its limited downstream persistence raises questions regarding source origin, in-stream attenuation, or transformation—factors influencing most EPCs [17,59].

In terms of seasonality, the distinct wet-season “flushes” of EPCs commonly observed elsewhere due to surface runoff [60,61] were not consistently detected here. In some cases, higher dry-season concentrations point to a “concentration effect,” whereby steady contaminant inputs—such as from septic systems—interact with reduced water volumes to elevate EPC levels [62]. Conversely, declining concentrations during the wet season may reflect dilution as continuous sources mix with larger volumes of relatively clean inflow from tributary sources with fewer EPC sources.

4.5. Ecological Risk

In assessing ecological risk, several EPCs exhibited concentrations that exceeded their predicted no-effect concentration (PNEC) thresholds at least once (Table 1, Table 2 and Table 3). These included 2,4-D, 4-nitrophenol, atrazine, caffeine, diclofenac, estrone, gemfibrozil, ibuprofen, sulfamethoxazole, and sucralose, with most exceedances occurring in canal and canal exit samples (Table 2 and Table 3).

Atrazine was notable for exceeding PNEC values at both upstream and downstream river sites, consistent with its widespread use in non-urban agricultural areas [27]. Although these exceedances suggest potential for localized ecological harm, their infrequency indicates a low likelihood of sustained, system-wide impacts.

In contrast, ibuprofen, caffeine, and gemfibrozil raise greater concern due to more persistent exposure patterns. Gemfibrozil exceeded its PNEC of 500 ng/L at three of the four canal sites based on median values, while caffeine also exceeded threshold levels across multiple sites (Table 1, Table 2 and Table 3). Ibuprofen exceeded its PNEC threshold at all four urban canal sites based on median concentrations (Table 2). Maximum values were also elevated at canal exits and the downstream river site (Table 1 and Table 3)—indicating both chronic and spatially extensive exposure, likely originating in the city. While the PNEC for ibuprofen was conservative (11 ng/L), its consistent detection and elevated concentrations suggest a genuine risk rather than an analytical artifact. The case of ibuprofen is examined in more detail later in the paper. Risks associated with gemfibrozil and diclofenac are discussed elsewhere for a site in southern Thailand [46].

4.6. Non-Targeted Analysis

Among the most concerning compounds identified in the non-targeted analysis were several previously overlooked or poorly characterized substances, flagged based on ECOTOX [63], TOXCAST [64], and hazard/exposure scores from KEMI [65]. These included the hormones 17β-estradiol and 17-methyltestosterone (active pharmaceutical ingredients), the UV filter ensulizole, the fungicide kresoxim-methyl, the herbicide terbutryn, and the antiparasitic drug albendazole. Additional compounds of concern included the industrial antioxidant di-tert-butyl-4-hydroxymethyl phenol and several high-scoring fragrance and flavoring agents, such as (E)-β-damascone, phenethyl anthranilate, and n-undecyl cyanide. We also detected four per- and polyfluoroalkyl substances (PFAS), spanning fluorinated silanes, purine derivatives, and phosphinic esters. These contaminants likely originate from a mixture of domestic, agricultural, industrial, and personal care product sources, reflecting the complex and diffuse nature of urban chemical inputs to surface waters. In this exploratory screening, EPC concentrations were not determined.

5. Discussion

5.1. Urban–River Linkages

The concentration data reveal six general “linkage patterns” that illustrate how EPCs are transported from the Mae Kha Canal system into the Ping River (Figure 6). These align generally with enrichment trends in Table 4, but other subtle patterns emerge.

The strongest evidence of urban-river linkage comes from five compounds—acesulfame, gemfibrozil, fexofenadine, TBEP, and caffeine—which exhibited frequent to consistent enrichment at the downstream river site (P-40) relative to the upstream site (P-14) (Figure 6a; Table 4). These compounds consistently displayed spatial and temporal patterns indicative of urban origin and downstream transport, forming a distinct urban enrichment profile. Their persistence suggests limited in-stream attenuation and strong surface water connectivity between the canal and river, particularly via exits Set P-10, P-39, and P-41.

A second group of compounds—sucralose, sulfamethoxazole, tolyltriazole, and fenobucarb—exhibited episodic enrichment, often linked to transient inputs or seasonal pulses (Figure 6b). For instance, sucralose was not consistently enriched at P-40 but maintained elevated concentrations within the canal system, suggesting its persistence may be masked by dilution at the river scale. The herbicide 2,4-D exhibited a different pattern, with elevated concentrations detected at the upstream site, pointing to additional contamination sources outside the urban canal system (Figure 6c). The herbicide atrazine showed an inverse distribution compared to most EPCs: it was consistently present at high concentrations upstream of the city but largely absent from the canal system, highlighting its rural and agricultural origins (Figure 6d). Similarly, 4-nitrophenol appeared to originate from upstream sources—likely associated with industrial or commercial activity in peri-urban areas—with little evidence of input from the urban core (Figure 6c).

In contrast, gabapentin and metformin displayed distinctly urban signatures, with elevated concentrations confined to the canal system and only sporadic detection downstream (Figure 6e). A broader group of ten lower-concentration pharmaceuticals were also elevated within the canal network but showed limited persistence and infrequent downstream transport into the river (Figure 6f).

Collectively, the widespread presence of many compounds in the canal system underscores the influence of wastewater discharges and limited treatment capacity in shaping aquatic contaminant profiles at the site. In several cases, elevated river concentrations appear to be linked to short flow-path distances, limited in-stream degradation, and continuous dry-season flow, which enabled sustained export of contaminants from urban canals. The five most enriched EPCs—acesulfame, gemfibrozil, fexofenadine, TBEP, and caffeine—function as practical tracers of urban influence in this river system. These spatial patterns reinforce the view that urban canals such as Mae Kha are not merely passive recipients but active conduits for the transport of potentially bioactive and persistent contaminants into tropical river systems.

5.2. Urban Canals as Contamination Connectors—An Asian Perspective

Urban canals worldwide frequently exhibit elevated concentrations of contaminants due to direct discharges of untreated or partially treated wastewater [66]. This pattern is especially pronounced in rapidly urbanizing regions with limited wastewater infrastructure—for example, in Lahore, Pakistan, Hanoi, Vietnam, and Sri Lanka [67,68,69]. Even where formal drainage networks exist, urban canals often function as dynamic and poorly regulated conduits for emerging contaminants [68,69,70].

A recent multi-compound study of the Saigon River system in Ho Chi Minh City, Vietnam, illustrates the dominant role of urban canals in contaminant transport [70]. The analysis found widespread contamination by pollutants such as pharmaceuticals, pesticides, and industrial chemicals, with the city’s urban canals identified as the main sources of these contaminants entering the river. The highest pollution and ecological risk levels were observed in the urban center. This comprehensive assessment of EPCs in a rapidly urbanizing tropical metropolis highlights urban canals as critical hot spots for future monitoring and pollution control efforts.

In our study in Thailand, acesulfame emerged as a robust indicator of urban wastewater contamination and potential transport to river systems. Median concentrations in all urban canals exceeded 13,200 ng/L (range = 1170–22,700 ng/L; Table 1). A recent comparison [71] revealed consistently higher acesulfame concentrations in urban canals than adjacent rivers across tropical cities including Jakarta, Bangkok, Phnom Penh, Manila, and Can Tho (Vietnam). In several cases—especially in Jakarta and Manila—canal samples exceeded 30,000 ng/L, likely due to their low flow volumes, sluggish flow, and direct connection to residential or commercial drainage networks (Figure 7). Similar trends were reported in Tianjin, China, where researchers [72] found acesulfame concentrations up to 17,500 ng/L in the Dagu Drainage Canal, nearly equivalent to levels in wastewater treatment plant effluents. In contrast, concentrations in the adjacent Haihe River declined downstream (maximum = 6000 ng/L), reflecting a gradient in wastewater influence from urban to rural areas.

The striking spatial contrasts in acesulfame concentrations across urban surface waters in selected Asian cities (Figure 7) demonstrate that canal concentrations frequently exceed those in adjacent rivers—often by more than an order of magnitude in both median and maximum values. In Chiang Mai, for example, median acesulfame levels in the Mae Kha Canal are more than ten times higher than those at the upstream location on the Ping River (Table 4). Similar trends have been reported elsewhere, including in Wuhan and the Dongjiang River Basin, confirming acesulfame’s persistence and spatial sensitivity as a molecular marker of untreated or poorly treated sewage [73,74,75].

In Hanoi, Vietnam, acesulfame concentrations were measured up to 21,000 ng/L and caffeine levels reaching 27,000 ng/L in the Kim Nguu Canal, which receives untreated sewage from approximately 400,000 residents [76]. Like many urban canals, the Kim Nguu is shallow, highly urbanized, and hydrologically distinct from nearby rivers—highlighting the importance of accounting for canal-specific dynamics in urban water quality assessments. Acesulfame concentrations closely matched those in our Thai dataset, while caffeine levels were more than twice our observed maximum (11,560 ng/L; Table 1). The Hanoi study also reported a significant correlation between acesulfame and caffeine, which supports their joint use as tracers of domestic wastewater inputs [76]. In our study, acesulfame also correlated with caffeine (r_s = 0.91), as well as ibuprofen (r_s = 0.97), atenolol (r_s = 0.92), and gemfibrozil (r_s = 0.96), where r_s values are Spearman rank order correlation coefficients.

Collectively, this regional comparison underscores the contrasting behaviors of rivers and urban canals in contaminant transport. Rivers tend to exhibit highly variable concentrations due to dilution, in-stream degradation, and the mixing of multiple inputs—ranging from canal discharges to upstream and downstream wastewater outfalls. In contrast, urban canals often carry low volumes of slow-moving water and receive concentrated inputs from nearby sources, resulting in persistently elevated contaminant levels.

In densely populated and heavily engineered environments, canals may function as chronic and underrecognized reservoirs of wastewater-derived contaminants—even in cities with relatively advanced sanitation systems. A striking example comes from Singapore’s Marina Catchment, where widespread detection of EPCs and their metabolites was reported despite a fully sewered infrastructure, suggesting contributions from leaking sewer lines, non-point runoff, or subsurface inflows [77,78]. Collectively, the findings highlight the need for high-resolution, cross-system sampling strategies when using acesulfame and similar compounds to monitor urban water pollution, particularly in developing regions.

5.3. Mae Kha Canal in Focus

When viewed alongside broader pollution patterns in urban canals worldwide, our findings offer a valuable lens through which to reassess the Mae Kha Canal system in Chiang Mai. Frequently portrayed as one of Thailand’s most polluted urban waterways [28,29,30,31,32,33,34,35,36], Mae Kha has become emblematic of local urban water quality challenges. To evaluate this perception with greater spatial resolution, we reanalyzed EPC data from a previous study [4], categorizing canal sampling sites into three types—other urban (OU), peri-urban (PU), and rural-remote (RR)—based on land use, population density, and distance from the city center (

Table 5. Median concentrations (ng/L; columns 1–4) of selected emerging and persistent contaminants (EPCs) and their enrichment factors (EFs) across urban, peri-urban, and rural-remote canals in the Chiang Mai area (not from Mae Kha canal or Ping River), based on prior data [4].

EPC	Mae Kha Canal (MK)	Other Urban Canals (OU)	Peri-Urban Canals (PU)	Rural-Remote (RR)	EF (OU vs. PU)	EF (OU vs. RR)
	ng/L	ng/L	ng/L	ng/L	-	-
Acesulfame	13,221	3275	257	45	13	73
Ibuprofen	634	84	bdl	bdl	34+	34+
Gemfibrozil	648	138	8	bdl	16	55+
TBEP	112	171	bdl	10	23+	16
Caffeine	4619	598	213	122	3	5
Metformin	915	401	157	50	3	8+
Sucralose	4083	1970	bdl	bdl	394+	394+
4-Nitrophenol	441	72	bdl	bdl	19+	29+
Tolyltriazole	bdl	7	bdl	bdl	3+	3+
Fexofenadine	1019	175	19	bdl	9	70+

Sample numbers are 18 for both OU and PU canals; 22 for R-R canals. The “+” means the PU or RR median concentration was below detection limit. Mae Kha values are presented for reference (Table 1). Enrichment factors (EFs) represent the fold-difference in median concentrations between urban sites and peri-urban (OU vs. PU) or rural-remote (OU vs. RR) sites. EF values < 1 indicate a higher concentration in the PU or RR canals than in OU canals.

). We focused on the ten EPCs most enriched in the Mae Kha Canal–Ping River system (Table 4). These samples do not overlap with our core dataset, offering an independent perspective for evaluating whether Mae Kha’s urban reach is truly exceptional or representative of a broader pattern of inadequate wastewater management.

Focusing first on the ubiquitous artificial sweetener acesulfame, we observed the highest median concentrations in other urban canals (3275 ng/L; n = 18), followed by peri-urban canals (257 ng/L; n = 18), and rural-remote canals and ditches (45 ng/L; n = 22). The OU concentration represents a 73-fold increase over rural-remote sites and a 13-fold increase over peri-urban canals (Table 5). However, the median acesulfame concentration in the Mae Kha Canal was an order of magnitude higher (13,221 ng/L; Table 5).

Similarly, sucralose, ibuprofen, 4-nitrophenol, and gemfibrozil were substantially elevated in OU canals compared to peri-urban and rural-remote sites but remained 3–7 times lower than levels observed in Mae Kha (Table 5). Sucralose stands out due to its high enrichment factor: while its median concentration in urban canals was 1970 ng/L, concentrations in peri-urban and rural-remote canals were below detection limits. Yet even this urban value was substantially exceeded in Mae Kha (4083 ng/L; Table 5). Concentrations of gemfibrozil and 4-nitrophenol were also higher in Mae Kha than in OU canals (648 vs. 138 ng/L and 441 vs. 72 ng/L, respectively).

TBEP exhibited a different pattern. Its concentration was higher in OU canals than in Mae Kha (Table 5), though not higher than at the Mae Kha outlet point P-39 (522 vs. 171 ng/L; see Table 3). Other EPCs discussed earlier in this paper but not listed in Table 5 did not have sufficient detectable concentrations across the comparison sites to inform this analysis.

These findings reinforce the widely held perception that the Mae Kha Canal is contaminated. However, this is difficult to contextualize without measuring the same EPCs in other canals in other dense urban locations such as Bangkok. In the comparison shown in Figure 7, the median concentration for acesulfame recorded for Bangkok was not enriched, but the maximum concentration (1819 ng/L) had an ERF of about 8 (river and canal locations are not known). In another Thai study, an acesulfame concentration of 2179 ng/L was found in a stream below a small town in the Thai southern province of Ranong—although canals were not directly implicated as sources [46].

Spatially, we found that median pollutant concentrations were highest in the Mae Kha Canal, lower in other urban canals, and progressively declined across peri-urban and rural-remote sites—reflecting the influence of urban population density, land use, and wastewater infrastructure. However, the “other urban” (OU) category includes a heterogeneous mix of canals, and while median values suggest a stepwise decline, some individual OU canals exhibited contaminant levels comparable to, or even exceeding, those in Mae Kha. For instance, previously reported concentrations of acesulfame (~36,000 ng/L), fexofenadine (~21,000 ng/L), and caffeine (~57,000 ng/L) were observed in non-Mae Kha canal locations [4]. These findings highlight the need for spatially and temporally explicit monitoring to identify additional contaminant hotspots across the city and underscore the importance of more differentiated assessments of wastewater management practices along the urban–rural gradient.

5.4. Potential Risk

5.4.1. Single-Compound Concerns

In assessing ecological risk, some contaminants exceeded their predicted no-effect concentration (PNEC) thresholds at least once—most notably ibuprofen, diclofenac, gemfibrozil, atrazine, 2,4-D, sucralose, caffeine, estrone, sulfamethoxazole, 4-nitrophenol, and—primarily in canal and canal-exit samples. Atrazine was the only compound to exceed PNECs at both upstream and downstream river sites, consistent with its widespread use in non-urban agricultural areas [27]. While these exceedances suggest potential for localized ecological harm, their sporadic nature lowers the likelihood of sustained, system-wide effects. In contrast, caffeine and ibuprofen exceeded median PNEC thresholds across canal sites and were detected year-round, indicating persistent exposure and a greater likelihood of chronic impacts.

Ibuprofen, in particular, consistently exceeded its conservative NORMAN PNEC of 11 ng/L [56] (Table 2). As a widely used analgesic that is only partially metabolized and incompletely removed by conventional wastewater treatment, ibuprofen is frequently detected in surface waters [79,80,81]. Global reviews confirm its prevalence in wastewater-impacted systems, often at concentrations above ecotoxicological thresholds [82,83], with some describing it to be “medicating the environment” [84].

Regional studies reinforce these concerns [85,86,87]. In Thailand, ibuprofen is commonly found in Bangkok’s canals and rivers—a pattern linked to decentralized and inefficient wastewater infrastructure [40]. In Malaysia, its occurrence in rivers and drinking water sources correlates with population density and proximity to sewage outfalls [86,87]. Hydrodynamic modeling in the Philippines highlights its downstream persistence shaped by river flow dynamics, while reports from Indonesia document its co-occurrence with other pharmaceuticals, reflecting broader infrastructural gaps [88,89]. Similar observations in China, Japan, India, and Taiwan affirm ibuprofen’s persistence under diverse hydroclimatic and governance contexts [90,91,92,93,94].

Although ibuprofen degrades relatively rapidly under laboratory conditions, it remains a concern in natural systems due to its continual discharge and potential biological activity at low concentrations [95]. Its moderate persistence does not preclude environmental risk, especially under chronic exposure or in complex chemical mixtures [96]. Recent studies indicate that reproductive, biochemical, and developmental effects may occur at concentrations well below traditional toxicity thresholds, underscoring the importance of incorporating more sensitive sublethal endpoints into risk assessments [97,98]. Ibuprofen is known to disrupt prostaglandin synthesis, osmoregulation, and inflammatory signaling across a range of aquatic taxa [96,98]. In Chironomus riparius, adverse effects such as reduced survival, hormonal disruption, and altered gene expression have been observed at concentrations as low as 24 ng/L [98]. Conversely, mesocosm trials in the Philippines reported low acute toxicity in mayflies even at 2000 µg/L [84], though sublethal and chronic effects were not evaluated. Experiments on Cyprinus carpio and Rhamdia quelen revealed oxidative stress, reproductive dysfunction, and damage to kidney and immune systems at concentrations between 1000 and 10,000 ng/L [99,100].

While advanced wastewater treatment can remove over 90% of ibuprofen under optimal conditions, it remains frequently detected in effluents and downstream aquatic ecosystems—often at levels exceeding derived PNEC values [95,97]. Reviews consistently identify ibuprofen as a contaminant of emerging concern due to its pharmacological potency, frequent occurrence, and potential to bioaccumulate in aquatic vegetation and biofilms [96,101]. Moreover, its co-occurrence with other pharmaceuticals raises concerns about additive or synergistic effects on non-target organisms [102].

Collectively, these findings suggest that current PNEC thresholds—though useful as screening benchmarks—may underestimate ecological risks under conditions of chronic, low-dose, multi-compound exposure. The frequent exceedance of even conservative thresholds, along with evidence of sublethal effects at environmentally relevant concentrations, highlights the need for more nuanced risk assessment frameworks, in general. Regionally tailored PNECs that reflect native species sensitivities and local environmental conditions would improve ecological relevance and strengthen regulatory protections.

5.4.2. Cumulative and Mixture Risks

Urban freshwater ecosystems are increasingly exposed to complex mixtures of pharmaceuticals and other anthropogenic contaminants, presenting significant challenges for ecological risk assessment and water quality management [103]. While our concentration data and PNEC-based assessments highlight the persistence and ecological risks of several pharmaceuticals, these approaches do not fully capture the potential for long-term and cumulative impacts [103,104]. Chronic, low-dose exposures to pharmaceuticals such as ibuprofen and caffeine can potentially induce sublethal effects across multiple trophic levels—especially when bioaccumulation, trophic transfer, or mixture toxicity are involved [105,106].

A growing body of research demonstrates that aquatic organisms are rarely exposed to single contaminants; instead, they encounter complex mixtures whose combined effects can be additive, synergistic, or antagonistic, resulting in outcomes that cannot be predicted from single-compound data alone [107]. Empirical mixture studies frequently show enhanced, unpredictable, or delayed impacts at physiological and ecological scales, influenced by compound interactions, exposure duration, and environmental context. Although most pharmaceuticals show limited potential for biomagnification across entire food webs, tissue-specific accumulation (for example, in fish liver and brain) remains ecologically relevant [98,100].

Recent advances underscore that long-term and mixture effects remain insufficiently studied relative to their ecological importance. Emerging methodologies—including mesocosm experiments, multigenerational assays, and omics-based approaches—have begun to reveal subtle or chronic impacts, such as behavioral alterations, reproductive disruption, and shifts in community structure, that are often missed by conventional acute or single-compound toxicity assessments [104,105,106,107]. Addressing these knowledge gaps is increasingly urgent to sustain freshwater ecosystem services amid intensifying natural and anthropogenic stressors. To address these challenges, we recommend the following actions:

• Implement long-term, high-frequency monitoring to capture both chronic background concentrations and short-term contamination pulses.
• Develop region- and species-specific chronic toxicity benchmarks that reflect local ecological sensitivities and community composition.
• Integrate mixture and multistressor effects into risk assessment frameworks using component-based approaches, whole-mixture testing, and next-generation tools such as omics [105].

Together, these strategies will improve the ecological relevance of risk evaluations and support more effective protection of biodiversity and ecosystem function in urban freshwater systems.

5.5. Priority Compounds

In prior work, we identified several pharmaceuticals—diclofenac, gemfibrozil, metformin, and naproxen—and the pesticide atrazine as priority emerging pharmaceutical contaminants (EPCs) of concern in Thailand [4,46]. Among these, gemfibrozil was consistently enriched in the Ping River downstream of the city, though concentrations were not alarmingly high (Table 1). As discussed in the previous section, ibuprofen can now be added to this concern list. Earlier research in Bangkok also flagged diclofenac and other EPCs—including tylsalicylic acid, ciprofloxacin, and mefenamic acid—as posing potential ecological risks in canals and rivers [40].

Additional compounds have been identified as hazardous based on predicted no-effect concentration exceedances or related risk indices. For example, roxithromycin was detected in wastewater effluent from treatment plants in Bangkok at levels exceeding PNEC thresholds [39]. In the Chin River in northeast Thailand, oxytetracycline and enrofloxacin used in tilapia aquaculture were found at concentrations above freshwater safety limits [108]. Other investigations have reported elevated concentrations of concern for amoxicillin, doxycycline, enrofloxacin, lincomycin, and tetracycline in surface waters [109,110]. Preliminary ecological risk assessments have also pointed to the potential hazards posed by certain biocides (e.g., clotrimazole, triclosan) and synthetic musks (e.g., xylene, 4-MBC, tonalide) to aquatic organisms [43,44]. Similar concerns have been raised regarding disinfection byproducts, persistent organic pollutants, and perfluorooctanesulfonic acid (PFOS) in Thai surface waters [110,111,112].

The purpose of this brief review is not to provide an exhaustive inventory of harmful EPCs in Thailand, but rather to illustrate the difficulty of determining which compounds should receive priority attention. Regulatory and analytical limitations often constrain monitoring efforts to a narrow set of target EPCs—typically those already well-known or analytically accessible. To underscore this challenge, we draw on results from our non-targeted screening of the Mae Kha Canal reported in Section 5.5. Again, several were flagged based on ECOTOX [63], TOXCAST [64], and hazard/exposure scores from KEMI [65]. Included were four PFAS compounds, two hormones, pesticides and related substances, and several industrial compounds.

Our mention of these compounds—identified through initial screening of a single sample, with a full report forthcoming—highlights that most of the myriad EPCs in the environment are rarely included in routine water quality monitoring. Their detection underscores the importance of wide-scope surveillance approaches capable of capturing a broad array of chemical classes with potential ecological and human health implications [113,114]. Several of the substances identified, particularly fluorinated compounds, exhibit structural features associated with persistence, bioaccumulation, and toxicity. These findings reinforce the value of broad-spectrum chemical screening in uncovering previously unrecognized risks in urban surface waters.

Together, our targeted analysis of 24 EPCs and complementary non-targeted screening suggest the presence of a much broader and more dynamic contaminant mixture in the urban canal system—one that includes many poorly understood or unregulated compounds. Previous research in the Ping River system has shown that contaminant composition can shift abruptly over time, complicating efforts to establish a stable set of priority substances. Such variability highlights key limitations in conventional EPC monitoring frameworks, which often rely on outdated or narrowly defined target lists that may fail to capture region-specific risks [115,116,117]. As recently shown [53], these lists frequently omit regionally relevant compounds—particularly in developing or tropical settings where chemical usage patterns and environmental behavior differ significantly from temperate norms. Integrating regional consumption trends, analytical feasibility, and ecotoxicological thresholds into prioritization schemes remains a critical but underutilized approach in Southeast Asia [117].

Recent advances in non-targeted and suspect screening provide a powerful complement to traditional monitoring strategies [118]. These techniques enable the detection of previously unmonitored but potentially hazardous compounds, as demonstrated across urban surface waters in South and Southeast Asia [117]. Moreover, as non-targeted analysis evolves toward quantitative and risk-integrated frameworks, it offers not only the ability to identify “unknowns”, but also to evaluate their toxicological relevance and regulatory significance. Expanding the analytical lens beyond well-established contaminants is thus not only scientifically interesting—it is essential for developing more adaptive, equitable, and precautionary water quality monitoring systems.

5.6. Caveats and Limitations

While this study advances understanding of the spatial distribution and enrichment of emerging and persistent contaminants in a tropical urban canal–river system, several important caveats warrant consideration. First, the temporal resolution was limited. Monthly sampling over a 13-month period captured broad seasonal trends but was insufficient to detect short-term or episodic contamination events, such as those driven by stormwater surges, infrastructure overflows, or illicit discharges. Although both wet and dry seasons were represented, finer-scale or diurnal fluctuations in contaminant fate and transport likely went undetected. As a result, transient spikes linked to rainfall, unauthorized discharges, or daily practices may have been missed.

A second limitation concerns the limited comparability with other urban canals. Most other canals in the study area were sampled only once, providing “snapshot” data. This restricts our ability to assess whether Mae Kha is consistently the most contaminated canal or to evaluate intra-annual variability across systems with differing hydrological or anthropogenic contexts. The restricted spatial and temporal coverage thus constrains robust inter-canal comparisons.

Source attribution was also limited. While concentration patterns in Mae Kha suggest contributions from inadequately treated wastewater, the absence of facility-level discharge data and limited information on sewer connections precluded definitive linkage of specific contaminants to particular sources. We did not measure wastewater flows or use isotopic or compound-specific tracers, contributing to uncertainty regarding the relative roles of residential, commercial, hospital, or industrial inputs. Moreover, our pharmaceutical-focused analysis may skew interpretation toward healthcare-associated sources [119,120], potentially overlooking other relevant pollutants—such as endocrine disruptors, microplastics, antimicrobial resistance genes, and transformation products.

We were similarly unable to assess differences in wastewater treatment plant (WWTP) technology, either locally or across comparison sites. WWTP coverage, operational performance, and enforcement vary widely—not only between cities but even across neighborhoods within a city—introducing uncertainty about how infrastructure influences contaminant persistence in urban canals, which are known for pollution issues in Thailand [121,122,123,124]. Sparse and inconsistent information on this issue in the region likely contributes to the observed spatial heterogeneity in contaminant concentrations both within canal networks and between study regions. Future work incorporating fine-scale assessments of sewer connectivity and treatment technologies could clarify infrastructure–contaminant linkages in urban surface waters.

Statistical interpretation was constrained by a small sample size (n = 13), non-normal data distributions, and significant spatial and temporal autocorrelation among sites. These factors violate key assumptions of many inferential tests, limiting the reliability of p-value-based conclusions [125,126,127]. As a result, we emphasized practical significance through visualization of spatial and temporal patterns and calculation of enrichment factors to compare upstream and downstream concentrations. These approaches allowed for meaningful semi-quantitative comparisons of contaminant behavior despite a modest and autocorrelated dataset [128]. Additionally, we did not quantify contaminant loads or hydrological fluxes due to the absence of concurrent flow measurements.

Findings are also context-specific and may not generalize to other tropical cities with differing land use or infrastructure. Biological effects were not evaluated, as they were the focus of a previous study [25]. Higher-frequency sampling—daily, diurnal, or automated—would greatly enhance detection of rapid contaminant changes, improve source attribution, and better characterize hydrological variability. Such high-resolution strategies are increasingly recognized as best practice for capturing event-driven dynamics in urban water systems [129,130,131]. Again, future sampling designs should include more frequent and strategically timed measurements—for example, weekly campaigns to detect short-term variation, daily or diurnal sampling to capture behavioral cycles and infrastructure discharge patterns, and sub-daily or storm-triggered sampling to observe contaminant pulses during rainfall events. These efforts would substantially improve our understanding of the timing, sources, and ecological implications of EPCs in dynamic tropical water systems.

Despite these limitations, our study provides a valuable baseline for understanding EPC dynamics in the Mae Kha Canal and its urban watershed—an area where long-term, high-frequency monitoring is rare. This assessment suggests the canal system is indeed highly polluted. However, we are unable to assess its toxicity or determine whether similarly contaminated conditions exist in smaller, unsampled canals—or even other canals in the region.

Nonetheless, by visualizing contaminant patterns and applying enrichment factors, we adopt a robust approach well-suited to data-limited environments. These methods offer actionable insights into contaminant trends, intervention priorities, and broader implications for urban water quality management. Our findings underscore the need for more temporally resolved, source-informed monitoring to support future risk assessments and policy interventions in rapidly urbanizing tropical regions [132], while also demonstrating how exploratory, practical tools can guide decision-making where high-resolution data remain unavailable.

6. Implications and Conclusions

This study demonstrates the power of wide-scope surveillance to uncover spatially complex patterns of contamination in tropical urban water systems. By profiling pharmaceuticals, personal care products, and other anthropogenic EPCs across a canal–river network, we documented signatures that inform on hydrological connectivity (linkage) and contaminant enrichment. Several compounds—most notably acesulfame, gemfibrozil, fexofenadine, caffeine, and TBEP—were consistently elevated at downstream river sites, indicating sustained urban inputs and limited in-stream attenuation. Ibuprofen, diclofenac, and gemfibrozil occasionally exceeded ecotoxicological thresholds, raising concerns about chronic ecological risks.

In contrast, the herbicide atrazine was persistently detected in the river upstream of the city, pointing to agricultural inputs from surrounding rural catchments, consistent with prior findings [27]. The variable presence of 4-nitrophenol, likely from industrial or peri-urban sources, further underscores the complexity of source attribution in mixed-use watersheds. While enrichment factors proved valuable for detecting spatial gradients and prioritizing compounds of concern, future work should complement these with load-based estimates to capture total pollutant fluxes and account for hydrological variability.

Urban sources of contamination are likely diverse. Though definitive attribution remains limited by data gaps, the presence of multiple pharmaceutical residues and personal care products suggests major contributions from public health infrastructure, households, and the service industry—likely via diffuse and potentially illicit discharges. Addressing these challenges requires a more complete understanding of sewer connectivity, wastewater treatment variability, and infrastructure–pollution linkages at the neighborhood scale.

Although constrained by limited temporal resolution, this study provides a baseline for contaminant dynamics in an under-monitored tropical urban watershed—a plague of most urban centers in the region. Future efforts could pursue higher-frequency sampling—including weekly, diurnal, and event-based monitoring—to detect rapid contaminant fluctuations, support source identification, and better characterize the influence of stormwater and behavioral cycles on urban water quality.

Ultimately, managing the urban exposome in tropical cities in the region will require bridging wide-scope chemical surveillance with hydrological analysis and infrastructural insight. By linking contaminant enrichment, source complexity, and urban–rural hydrological connections, our approach provides a practical, scalable framework for informing policy and protecting ecosystem and public health in rapidly changing tropical environments.