Occurrence of insecticides, fungicides, and herbicides in household drinking and irrigation water in an intensive floriculture region of Ecuador

Abstract

Pesticides are widely used in agriculture, floriculture specifically, posing significant ecological and health risks. Limited research has been conducted on the presence of neonicotinoid insecticides (NNI) and atrazine, in drinking and irrigation water in agricultural regions of Latin America. This study describes targeted and non-targeted analysis of concentrations of pesticides in drinking and irrigation water sampled in 2022 and 2023 in an industrial floricultural region in Pedro Moncayo, Ecuador. In targeted analysis, we found NNI in treated drinking water in 20.5% of household tap water samples (8 out of 39), in a community well, and in 57.7% of irrigation water samples (n = 15 out of 26). Imidacloprid and thiamethoxam emerged as the most frequently detected NNI in drinking water. Atrazine was detected in two household taps and one well. In exploratory non-targeted analysis, we found 63 compounds, including insecticides, herbicides, and fungicides, including azoles, in drinking and irrigation water samples. Fungicides accounted for 59% of all compounds (37 out of 63) detected. Intense pesticide use in floriculture in proximity to residential areas and aging piped water systems may allow pesticides to leach into treated drinking water pipes and irrigation water. The presence of a wide range of pesticides, especially NNI and fungicides (azoles, specifically), in drinking and irrigation water poses health risks to community members.

1. Introduction

Insecticides, herbicides, and fungicides are widely used pesticides in agriculture; some classes of pesticides are water soluble, persist in the environment, and can pose a risk to human health and the environment. Neonicotinoid insecticides (NNI), the most widely used group of insecticides, globally, are employed in modern crop protection against chewing pests (Jeschke et al., 2011). NNI are water soluble (Stehle et al., 2023), and can move from the application site via spray drift, runoff, and leaching (Alsafran et al., 2022). Especially high insecticide concentrations of NNI (imidacloprid, clothianidin and thiamethoxam) have been reported in surface waters globally (Thompson et al., 2020). In non-targeted organisms, NNI impact immune function, reduce growth and reproductive success (Stewart et al., 2014; Wu-Smart and Spivak, 2016; Lu et al., 2020). In humans, NNI are associated with various health risks, including oxidative stress, mitochondrial dysfunction, DNA damage, neurotoxicity (Xu et al., 2022), endocrine alteration, and reproductive damage (Oladosu and Flaws, 2025). In addition to NNI, atrazine is a widely used herbicide that has been found in surface and drinking water sources globally (Graymore et al., 2001). Similarly to NNI, atrazine is water-soluble and it can be transported to surface water via field run-off (Capel and Larson, 2001; Lazorko-Connon and Achari, 2009). In humans, atrazine exposure has been associated with developmental and reproductive effects, as well as respiratory, nephrological, and hepatological health risks (Gammon et al., 2005; Pathak and Dikshit, 2012). Fungicides, such as azoles, are also widely used in agriculture (European Centre for Disease Prevention and Control, 2013; Jørgensen and Heick, 2021). Triazoles, a class of azoles, are water-soluble, and can easily enter the soil and water systems through leaching and run-off (Bian et al., 2025). Evidence exists that azoles can cause liver toxicity, cardiotoxicity, and endocrine disruption (Chai et al., 2022; Bian et al., 2025).

The diversity of pesticides, the magnitude of use, and the varying solubility in water and persistence in the environment result in highly variable removal during treatment (Sadaria et al., 2016a, 2016b; Saleh et al., 2020; Thompson et al., 2020). NNI have been found in drinking water pre- and post-treatment, and removal is limited in conventional drinking water treatment methods, such as coagulation/flocculation and sand filtration with chlorination. In contrast, granular activated carbon (GAC) filtration is effective in removing NNI (Klarich et al., 2017). Conventional water treatment methods are also ineffective for removal of the herbicide, atrazine, from drinking water; however, similar to NNI, GAC filtration also removes atrazine (Lazorko-Connon and Achari, 2009; Ahmad et al., 2022). Fungicides exhibit a similar reliance on advanced treatment, and they can be efficiently removed via GAC (Liu et al., 2021); however, their complete degradation often requires advanced oxidation processes (AOPs), such as photocatalysis, Fenton reactions, ozonation, and peroxymonosulfate-based oxidation (Zhang et al., 2023; Bian et al., 2025).

Agriculture plays an important role in the economics of many Latin American countries; yet, pesticides in water are rarely monitored or monitored consistently (Hilber et al., 2024). Crop fields and greenhouses are frequently located in rural areas in proximity to residential areas, and can present a potential human exposure route in water – drinking, surface, and irrigation. Limited research has been conducted on NNI and atrazine in surface water in Latin America. A systematic review of studies published between 1998 and 2020 found 70 pesticides, including atrazine and glyphosate, in six regions of Latin America and the Caribbean. Pesticides were most frequently found in Brazil and Argentina (Grondona et al., 2023). Additionally, in 2020, Ecuador had the highest rate of pesticide application globally, with 3.74 kg/ha (Pizarro et al., 2024). Imidacloprid and atrazine were reported in the São Lourenço River in Brazil (Berton et al., 2018) and in Chile's Cachapoal River Basin in dissolved and particulate form (Climent et al., 2019). One study of pesticides along the Napo River in Ecuador, in an agricultural region, with high proximity to residential areas, found that imidacloprid (3233 ng/L) was present in 80% of the 40 river water samples examined (Cabrera et al., 2023).

Ecuador is the second largest exporter of roses in the world (Observation of Economic Complexity, 2023). Compared to traditional row crop agriculture, there is intense use of pesticides in floriculture, particularly in rose production (Pereira et al., 2021). The transport of water soluble pesticides, such as NNI, from floriculture to drinking water is understudied in this region and in industrial floriculture (Ecobichon, 2001). Results from the prospective cohort "Study of Secondary Exposures to Pesticides among Children and Adolescents" (ESPINA) in Pedro Moncayo, Ecuador found urinary metabolites of the NNI, imidacloprid, in children within 200m of floriculture greenhouses (Chronister et al., 2025). Additionally, children living closer to such greenhouses had lower AChE activity, reflecting higher exposures to cholinesterase inhibitor insecticides (Suarez-Lopez et al., 2012, 2020), higher systolic blood pressure (Suarez-Lopez et al., 2018) and neurodevelopmental deficits (Friedman et al., 2020). Lastly, NNI metabolites in ESPINA participants have been associated with endocrine disruption (Chronister et al., 2024). No studies have examined water as a possible exposure route in this rose production region.

Building on the above findings, the objective of this study was to quantify NNI and atrazine concentrations in drinking and irrigation water sampled in 2022 and 2023 in Pedro Moncayo, Ecuador using targeted analysis. We hypothesized that neonicotinoids in water would be elevated due to the extensive proximal floriculture activities in the region. Drinking and irrigation water samples were collected before and after water treatment, among households participating in the ESPINA study. Secondly, an exploratory non-targeted analysis (NTA) was performed to assess the presence of pesticides (insecticides, herbicides, and fungicides) in drinking and irrigation water samples, to evaluate a wide range of chemicals, including transformation byproducts, without prior selection.

2. Methods

2.1. Ethical Approval

The researcion was approved by the UC San Diego Institutional Review Board #804238.

2.2. Water management in Pedro Moncayo, Ecuador

In Pedro Moncayo, Ecuador, each parish's drinking water is managed by its own elected water committee (Junta Administradora de Agua Potable, JAP), which oversees the distribution and maintenance of the local drinking water supply systems (Secretaría Nacional de Planificación, n.d.). These drinking water supply systems, established in the 1970s, consist of gravity-fed piped water systems, including catchment tanks, drinking water treatment plants (DWTPs), storage tanks, and piped distribution systems that deliver water to household taps. The water used for this service is sourced from the Chiriyacu spring, which supplies the five parishes: Malchingui, Tabacundo, Tupigachi, La Esperanza, and Tocachi. A centralized treatment system, employing sand filtration and chlorination, is found in Tabacundo, which provides drinking water for all parishes. However, these parishes also require additional water sources, as the water provided by centralized supply is insufficient. In Tupigachi, additional water sources come from Cucupugro and Chinchiloma (Amagua Pillajo and Suárez Santiana, 2015) and one well centrally located in the parish serves as secondary drinking water sources when piped water is not available, particularly in the dry season. In Tocachi, water sourced from Mojanda's lagoons in the páramos (highlands), is treated with only chlorination (GAD Parroquial Rural de Tocachi, n.d.). In Malchingui, water from Cerro del Pueblo (Tangarán) is treated with both filtration and chlorination (Gammaconsul, n.d.). La Esperanza also supplies its water demand with Tomapamba, Chuquirahua, and Rayochupa sources (GAD Parroquial Rural de Tocachi, n.d.; GAD Parroquial Rural La Esperanza, n.d.).

An infrastructure challenge in the Pedro Moncayo region is that the DWTPs do not maintain full pressure at all times and provide intermittent service. Intermittent service can lead to negative pressure events, increasing the risk of contaminant intrusion through cracks or leaks, especially in older piped systems. Additionally, systems that rely only on chlorination may experience turbidity events, especially during the rainy season, requiring more chlorination.

2.3. Geospatial analysis and systematic water sampling

2.3.1. Drinking water

This study focused on household tap water in proximity to floricultural plantations and in the ESPINA participants' homes with a range of NNI and total pesticides in urinary metabolite samples of the children. Participant households in the water study were selected for household tap water sampling from the ESPINA study based on hotspot analysis for ESPINA children with high levels of pesticides in urinary metabolites together with ‘coldspot’ analysis of ESPINA children who had no NNI present in urinary metabolites in 2016 (see Supporting Information (SI) A). To ensure a comprehensive and representative dataset, sampling was conducted across all four piped drinking water systems and across the two irrigation canals in the parishes in Pedro Moncayo where ESPINA participants reside. Sample collection for irrigation and drinking water took place between July 25-August 4, 2022, and July 10-14, 2023 to coincide with the ESPINA study.

Water sampling was conducted in coordination with the ESPINA follow-up examination that occurred in the summer of 2022. Samples of the piped drinking water systems were taken before treatment (raw water), immediately post-treatment, and at various points along the distribution lines at household taps. This methodology allowed for the collection of water samples from locations nearest to and farthest from the treatment plant, as well as mid-system, ensuring diverse exposure assessments across the community water systems (see Fig. 1). A flushing protocol, in which tap water was allowed to run for 30 s to 2 min prior to sampling, was employed to minimize the risk from contaminated tap fixtures. This method is commonly used to improve the representativeness of water samples and reduce artifacts from local plumbing conditions (USGS, 2010).

Fig. 1.

Distribution of water sources and irrigation channels in Pedro Moncayo, Ecuador.

In 2022 and 2023, 54 drinking water samples were collected from Pedro Moncayo, including household (HH) taps, one well, and DWTP (immediately pre- and post-treatment). Specifically, 34 ESPINA HH taps were sampled across the parishes of Malchingui (6 HH), Tocachi (3 HH), La Esperanza (5 HH), Tabacundo (11 HH), and Tupigachi (9 HH). We sampled 39 HH taps over both years, with 14 HH taps in 2022 and 25 in 2023, including 5 HH sampled in both years. Additionally, 9 samples were taken pre- and post-treatment at DWTPs, and one was taken at the well in Tupigachi. This approach provided valuable insights into the drinking water quality throughout the parishes and their drinking water systems, from pretreatment stages at DWTPs (raw water) to household taps near these facilities and along the distribution systems at point of use. This showed a range of potential exposures and insights into localized environmental dynamics associated with flower plantation pesticide use (Fig. 1). To account for potential contamination in the taps, the water was run for 30 s to 2 min, and temperatures had to be stable, before taking samples, and field blanks were taken daily (See Fig. 1 for drinking water system sample locations in Pedro Moncayo.).

2.3.2. Irrigation water

The selection of irrigation sampling sites was based on their proximity to selected ESPINA households, where tap water samples were collected. These irrigation sites were located along both the new (“nuevo”) and old (“viejo”) irrigation channels that run through Pedro Mocayao with the parishes of Tupigachi, Tabacundo, La Esperanza, Tocachi and Malchingui. The open channels were sampled close to the water source, midway along the channels, and farthest from the water source. The old irrigation channel runs to the Upayaca and the Bobo rivers then into the Pisque River, and samples were taken at the end of the old channel near the outflow into these rivers. We collected 26 irrigation water samples: 19 samples were collected in 2022 and 7 in 2023, including 3 samples collected in both years. In total, 20 samples were collected in the old channel and 6 samples were collected in the new channel (See Fig. 1 for irrigation water system sample locations in Pedro Moncayo.).

The two main irrigation systems are locally known as the old canal and the new canal. Both are open canals extending approximately 70 km from their source at the Cayambe volcano (Fig. 1), traversing an elevation gradient from about 3500 to 2800 m above sea level. The new canal, constructed approximately 30 years ago, is located at higher elevations and is entirely lined with concrete; it runs predominantly through rural areas with a lower density of floriculture greenhouses. In contrast, the old canal, which has been in operation for over 100 years and is only partially concrete-lined, follows a lower-elevation route that passes through urbanized zones and areas with a high concentration of intensive floriculture.

2.4. Targeted pesticide analysis: NNI and atrazine analysis quantification

For targeted pesticide analysis of NNI and atrazine, we collected 1L of water sample (drinking and irrigation) in amber glass bottles (acid-washed or pre-baked at 550 °C) with minimal headspace. Samples were refrigerated at 11C and transported to the Aqua Bio Lab at Universidad San Francisco de Quito (USFQ). Each sampling day a field-blank was taken. Water samples were also taken and analyzed for physical (Temp, pH, DO, EC), microbiological (fecal and total coliforms), and heavy metals (See SI B&C) (Hladik et al., 2014; Bradley et al., 2018). For the sample preparation at USFQ, water samples were first filtered using an inert pre-ashed glass fiber filter (nominal pore size of 0.7-1.2 μm). This filtration step was performed to eliminate suspended solids. Next, deuterated imidacloprid-d4 was added as a surrogate into each sample to confirm the efficacy of the extraction procedure. One isotopically labeled neonicotinoid surrogate was deemed sufficient due to the small number of structurally/chemically related target compounds (i.e., neonicotinoids), based on prior work (Hladik et al., 2014, 2018; Hladik and Kolpin, 2016; Klarich et al., 2017; Webb et al., 2020; Thompson et al., 2021; Woodward et al., 2022). Because the final volume after extraction and reconstitution is 1 mL, 100% recovery corresponds to an expected concentration of 100 ng/L in the original water sample. The samples were subsequently subjected to solid phase extraction (SPE), using Oasis HLB cartridges. This process aimed for sample enrichment (1000-fold from 1L of water). The cartridges were subsequently sent to the LeFevre laboratory at the University of Iowa (UIowa).

At UIowa, samples were analyzed for NNI (imidacloprid, thiamethoxam, clothianidin, dinotefuran, acetamiprid, thiacloprid, and sulfoxaflor) and atrazine (Klarich et al., 2017). Following elution and evaporation under nitrogen gas, evaporated samples were reconstituted to a final volume of 1 mL in acetonitrile and deionized water (1:1). Then, 10 μL deuterated thiamethoxam were added as an internal standard. Extracts were preserved at −20 °C until quantification using liquid chromatography-tandem mass spectrometry (LC-MS/MS Agilent 6460 separated on a Zorbax C18 column) with an Agilent system operating in positive electrospray ionization mode. LC-MS/MS employed multiple reaction monitoring (MRM) with two transitions per compound to ensure precise results (i.e., quantifying/qualifying ion). Quality control measures were rigorous, encompassing the use of field, laboratory, and analytical blanks, as well as calibration curves conducted for each sample set to uphold stringent standards. The method's sensitivity and robustness in detecting trace amounts of NNI and atrazine have been previously reported (Klarich et al., 2017; Klarich Wong et al., 2019; Webb et al., 2020, 2021). The limit of detection (LOD) were calculated following Standard Methods (APHA, n.d.) based on the ability of the instrument to consistently detect (99% probability reported detection is real) very small concentrations (LODs are presented in and do not change with sample year because the value represents instrument capability). The method limits of quantification (LOQs), which herein represents the ability to reliability quantify within the linear range of the calibration curve without extrapolation beyond the lowest standard, for each pesticide differed slightly between 2022 and 2023, with lower LOQs in 2023. These detection limits and method LOQs are consistent with the theoretical level of detection and method detection limits, respectively, reported elsewhere in US Geological Survey studies (e.g. (Hladik et al., 2014, 2018; Hladik and Kolpin, 2016; Woodward et al., 2022)). These differences may influence year-to-year comparisons by affecting the classification of quantifications compared to detections at the lowest possible concentrations. Although somewhat variable, likely due to international shipping or differences in sample matrix effects, recoveries were generally robust at 105 ± 7% (mean ± SE). Values were not recovery-corrected, consistent with prior research (Hladik et al., 2014, 2018; Hladik and Kolpin, 2016).

2.5. Non-targeted analysis (NTA)

To identify pesticides and metabolites not captured via targeted analysis, an exploratory NTA was conducted on water samples collected in 2022 to screen for compounds of interest (i.e., with a focus on pesticides/transformation products), using the same extracts as those analyzed by targeted methods. This integrated targeted/NTA approach enhanced the detection context of both known and unknown contaminants. Six water samples, which previously had quantitative levels of NNI/atrazine in the targeted analysis, were selected for NTA. Two drinking water samples, three irrigation water samples, and the well sample, representative of the different parishes, were analyzed at the High-Resolution Mass Spectrometry Facility (UIowa) and in collaboration with the USGS Pesticide Fate lab (Sacramento, CA). A Thermo Vanquish Flex UHPLC (Agilent Poroshell 120, EC-C18 2.7 μm, 2.1 × 100) coupled to a Thermo Q Exactive hybrid quadrupole Orbitrap mass spectrometer in positive ionization mode (each sample, 3 randomized-order replicate injections; scans: 75-750 m/z) was used (Klarich Wong et al., 2019). Compound Discoverer (CD) software was used for post-processing comprehensive compound analysis through data deconvolution and identification, incorporating pooled quality control (QC) samples, authentic standards spike mixtures, retention time prediction indexing, and spectral library matching (e.g., mzVault or mzCloud MS2 match scores >75%). To improve the reliability and interpretability of NTA results, only the compounds with a confidence level of 1 or 2a were selected (Schymanski et al., 2014). Compound identification details (e.g., mass defects, spectra matching) are presented in SI E.

3. Results/Discussion

3.1. Drinking water: targeted pesticide results

In 2022 and 2023, at least one NNI was detected in 20.5% of household tap water samples (8 out of 39). Atrazine was detected in 2 out of 39 samples, both in 2023 (see Table 1, and SI S-D1). Imidacloprid was detected in one sample in Malchingui (2022), with a concentration of 15 ng/L. All other imidacloprid detections were found in Tabacundo: in 2022 all 3 samples showed detection, in 2023, 2 of 8 samples had imidacloprid. Thiamethoxam was found in 12.8% (5/39 samples) in 2022 and 2023, four of the detections were in Tabacundo, while one was in Malchingui in 2022. As noted, one sample in Tabacundo exhibited elevated concentrations of imidacloprid (estimated: 11,000 ng/L) and thiamethoxam (estimated: 6000 ng/L). Although non-quantitative, we also note that the peak area measured for imidacloprid for this sample (HH17) in the NTA was over an order of magnitude greater than the imidacloprid authentic reference standard concentration (250 ng/mL; see SI E); thus, consistently indicating a very high neonicotinoid level in the given sample (Fig. 2). Clothianidin was detected in 2.6% of the samples (1 out of 39), also in the parish of Tabacundo. Acetamiprid was found in 7.7% (3 out of 39) in 2022 in Tabacundo and Malchingui. Lastly, atrazine was found in two samples only in Malchingui in 2023. The limited occurrence of atrazine might be attributed to the fact that atrazine is primarily used in crop fields, such as those for corn and sorghum, and less in floriculture (Secretaría Nacional de Planificación, n.d.; US EPA, n.d.).

Table 1.

Neonicotinoids and atrazine in household drinking water and irrigation channels, sampled 2022 and 2023.

Pesticide	Imidacloprid		Thiamethoxam		Clothianidin		Dinotefuran		Acetamiprid		Thiacloprid		Sulfoxaflor		Atrazine
2022	DF	Range (ng/ L)	DF	Range (ng/L)	DF	Range (ng/L)	DF	Range (ng/L)	DF	Range (ng/L)	DF	Range (ng/L)	DF	Range (ng/L)	DF	Range (ng/L)
LOQ (ng/L)	10		9.9		9.9		10		10		10		10		10
LOD (ng/L)	1.43		2.23		4.97		2.24		1.90		4.0		4.0		2.3
Drinking water
Malchingui n = 3	1/3	0-15	1/3	0-22	-	-	-	-	1/3	0- < LOQ	-	-	-	-	-	-
Tocachi n = 1	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
La Esperanza n = 1	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
Tabacundo n = 3	3/3	21-11,000a	3/3	11-6,000a	1/3	0-21	-	-	2/3	0-31	-	-	-	-	-	-
Tupigachi n = 6	-	-	-	-	-	-			-	-	-	-	-	-	-	-
Irrigation water
Old channel n = 15	7/15	0-612	8/15	0-612	3/15	0-41	-	-	2/15	0- < LOQ	2/15	0-13	1/15	0- < LOQ	2/15	0- < LOQ
New channel n = 4	1/4	0- < LOQ	1/4	0- < LOQ	-	-	-	-	-	-	-	-	-	-	-	-
2023	DF	Range (ng/L)	DF	Range (ng/L)	DF	Range (ng/L)	DF	Range (ng/L)	DF	Range (ng/L)	DF	Rang (ng/L)	DF	Range (ng/L)	DF	Range (ng/L)
LOQ (ng/L)	5.1		5.8		7.0		7.6		8.3		7.9		8.5		4.2
LOD (ng/L)	1.43		2.23		4.97		2.24		1.90		4.0		4.0		2.3
Drinking water
Malchingui n = 6	-	-	-	-	-	-	-	-	-	-	-	-	-	-	2/6	0-175.8
Tocachi n = 3	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
La Esperanza n = 4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
Tabacundo n = 8	2/8	0-7.2	1/8	0- < LOQ	-	-	-	-	-	-	-	-	-	-	-	-
Tupigachi n = 4	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-	-
Irrigation water
Old channel n = 5	5/5	<LOQ-86.6	4/5	0-121.0	3/5	0-12.23	-	-	1/5	0- < LOQ	-	-	2/5	0-10.4	1/5	0- < LOQ
Newchannel n = 2	1/2	0-6.0	1/2	0- < LOQ	-	-	1/5	0-13.3	-	-	-	-	-	-	-	-
Total
Drinking watera n = 39	6/39	0–11,000 c	5/39	0–6,000 c	1/39	0–21	-	-	3/39	0–31	-	-	-	-	2/39	0–175.8
Irrigation waterb n = 26	14/26	0–612	14/26	0–121.0	6/26	0–41	1/26	0–13.3	3/26	0- < LOQ	2/26	0–13	3/26	0–10.4	3/26	0- < LOQ

NNI = neonicotinoid insecticides; DF = detection frequency: the ratio of samples with non-zero concentrations to the total number of samples for area and year; LOD = level of detection; LOQ = level of quantification; <LOQ = below the limit of quantification.

^aTotal number of samples taken in both years over all areas.

^bTotal number of samples taken in both years for both channels.

^cEstimated value.

Fig. 2.

Pesticide concentrations of drinking water and irrigation water in 2022 and 2023;

Panel A shows the concentration in ng/L for drinking water; Panel B shows the concentration in ng/L for irrigation water. Only concentrations above 0 are shown.

Note that concentrations are plotted on logarithmic scales.

In our study, water samples from the DWTPs and household taps across different parishes revealed that the presence of pesticide residues immediately pre- and post-treatment was largely undetected. We did find a cluster of NNI positive samples in Tabacundo, which is also where the greenhouse agriculture is most densely populated and built within residential areas. Additionally, the well and secondary drinking water source of Tupigachi, sampled in 2022 and 2023, had detectable NNI, specifically, imidacloprid (2022: 149 ng/L, 2023: 98.6 ng/L), thiamethoxam (14 ng/L, 28.6 ng/L), and clothianidin (not detected, 7.7 ng/L). Atrazine was also detected in this well, below the level of quantification in 2022 and 2023 (see SI D1).

Although studies of NNI in tap water are limited, especially in Latin America, studies of tap water in China and in the USA found greater detection frequency of NNI in tap water than our research; while imidacloprid was the most commonly detected in our research and in studies in China and the USA. A pilot study in China conducted on 84 tap water samples randomly selected from 38 cities frequently detected six NNI; imidacloprid (detection: 99%), clothianidin (92%), dinotefuran (90%), acetamiprid (94%), thiamethoxam (87%) thiacloprid (86%) (He et al., 2021). Similarly, a study of 884 drinking water samples including 789 tap water and 95 groundwater samples from 32 provinces of China in June 2019 found that acetamiprid was most frequently detected (93%) followed by imidacloprid (88%) and clothianidin (69%) (Mahai et al., 2021). In our study, we detected imidacloprid most frequently (53.8%), however, we detected clothianidin less frequently (2.6%). Overall, the detection frequencies (DF) in our studies ranged from ~0 to 54%, significantly lower than the ~69-99% reported in the Chinese literature.

A study of tap water in Hanoi, Vietnam detected the NNI imidacloprid (median: 0.15 ng/L), acetamiprid (0.07 ng/L), thiamethoxam (0.19 ng/L), and clothianidin (0.04 ng/L) (Wan et al., 2021), however, DF were not available for comparison. In Indiana, USA, five NNI (clothianidin, dinotefuran, imidacloprid, and thiamethoxam) were identified in tap water samples from 81 households; imidacloprid (DF: 75%) and clothianidin (43%) were most frequently detected (Xie et al., 2025). Similar to our study, studies in China and in the USA most frequently detected imidacloprid. Consistent with our findings, thiacloprid was not detected in drinking water. In Indiana, NNI were detected in a larger fraction of samples (~0-75%), compared to our study.

3.2. Irrigation water: targeted pesticide results

Irrigation water samples (n = 26) were collected from the new and old channels; 19 in 2022 and 7 in 2023 (see Table 1 and SI D2). We found that from 2022 to 2023, at least one pesticide was detected in 15 out of 26 samples (57.69%). In 2022 and 2023, NNI contamination was found in 13 out of 19 samples (68.4%) in the old channel and 2 out of 7 samples (28.6%) in the new channel. The NNI, imidacloprid and thiamethoxam, presented the highest DF, both with 14 out of 26 total samples (53.8%). One old channel sample in Tupigachi had elevated concentrations of imidacloprid (612 ng/L) and thiamethoxam (108 ng/L) in 2022. Interestingly, the concentrations for imidacloprid and thiamethoxam were either null or below the level of quantification in the new channel, except for one sample in 2023, which exhibited detectable levels of imidacloprid (6.0 ng/L) (Fig. 2). Clothianidin was detected in 6 samples (23.1%), dinotefuran was detected in 1 sample of the old channel in 2023 (3.8%). Acetamiprid and sulfoxaflor were detected in 3 of 26 samples (11.5%), all in the old channel. Thiacloprid was found in 2 samples (7.7%) from the old channel. Lastly, we found atrazine in 2 samples of the old channel in 2022, and in one sample of the old channel in 2023; yet, they were below the level of quantification. The old channel passes through areas with a higher concentration of intensive floriculture than the new canal and provide some explanation for the more severe pesticide pollution in the old channel.

Little research has been conducted on NNI in irrigation water. Some studies have been conducted on NNI in surface water; however, few have been conducted in Latin America outside of Brazil and few have captured seasonal variability. Extensive research in China demonstrates a near-ubiquitous presence of neonicotinoids in surface waters, significantly exceeding the frequencies observed in our study. Water samples from the Qiantang River exhibited 100% DF for thiamethoxam, acetamiprid, and dinotefuran, with imidacloprid and clothianidin detected at 93.6% and 97.3%, respectively (Zhang et al., 2025). Similar trends were observed in the Yangtze River, where one study from December 2020 to June 2021 detected thiamethoxam at 100% of the 72 sampling sites, followed by near-universal detection of imidacloprid (98.6%), acetamiprid (98.6%), and clothianidin (95.8%). Other frequently detected parent compounds in that study included flonicamid (86.8%), dinotefuran (86.1%), and thiacloprid (85.4%) (Wang et al., 2023). Furthermore, a third study of surface water samples from the Yangtze River in 2021 reported that the DF of all measured NNI (including imidacloprid, acetamiprid, clothianidin, thiacloprid, thiamethoxam, dinotefuran, and sulfoxaflor) was 100%, except for acetamiprid in January (91%) (Li et al., 2022). In our study in Ecuador, while imidacloprid and thiamethoxam also had the highest DF (53.9%), these levels represent a much lower detection frequency than those found in either the Qiantang or Yangtze Rivers. Clothianidin was detected in only 23% of our samples, contrasting the >95% DFs reported in the Chinese studies. As compared to the Qiantang and Yangtze River data, dinotefuran (3.8%) and sulfoxaflor (7.7%) were also detected significantly less frequently in our irrigation water samples.

In Brazil, 55 samples in the Mogi Guaçu River basin detected imidacloprid (5.5%) and thiamethoxam (20%) between 2017 and 2018 (Monticelli Barizon et al., 2022). In 18 sampling sites from Brazil's Guapore River, imidacloprid DF was 100% in December 2024 and 67% in June 2025 (Rheinheimer Dos Santos et al., 2020). The reason suggested for the lower DF of imidacloprid in June was the lower pesticide application rate as compared to the previous year. Similarly, in our study, imidacloprid and thiamethoxam were the most frequently detected NNI, however, the DF from the Guapore River were lower than in our study. Compared to the Guapore River, we detected imidacloprid less frequently. One possible reason could be differences in agricultural intensity, as Brazil is one of the world's leading agricultural exporters and has high pesticide usage to support large-scale crop production (Perobelli, 2025). Seasonal NNI water sampling is needed in future research to capture variability in pesticide application and runoff, including temporal differences (e.g. between dry and rainy seasons) (Thompson et al., 2020).

Finally, in the United States, a longitudinal study of 77 rivers between 2013 and 2022 identified a persistent presence of imidacloprid, with a detection frequency (DF) of 44% across more than 12,000 samples. Notably, nearly half of the sampling sites (44%) exhibited increasing concentration trends over the decade. In our study in Ecuador, imidacloprid was detected with a slightly higher frequency (53.8%) than the U.S. national average, though our smaller sample size (n = 26) reflects localized irrigation conditions rather than broad riverine trends (Miller et al., 2025).

3.3. Non targeted analysis results

In six samples selected for NTA, a total of 63 compounds, 39 with a confidence level of 1 (confirmed presence by reference standard), and 24 with a confidence level of 2a (most probable presence, e.g., strong MS2 library spectra matching) were identified in water samples collected from drinking water, irrigation sources, and a well in Pedro Moncayo during 2022. These compounds included insecticides, herbicides, and fungicides.

In NTA, multiple insecticides were detected, including three NNI (imidacloprid, thiamethoxam, and acetamiprid), also quantified using targeted methods, two degradation products from imidacloprid (desnitro-imidacloprid and imidacloprid-urea), organophosphate insecticides (diazinon, malaoxon and malathion), and six other insecticides (carbofuran, dimethoate, methomyl, pyriproxyfen, spinetoram J and spinosyn A). Notably, based on peak area measurements relative to the standard spike (SI E), NNI imidacloprid in household sample 17, previously identified in the targeted analysis, exhibited a peak area approximately one order of magnitude greater than the standard spike in the NTA. The azole fungicide, tebuconazole, was also markedly elevated in household sample 17 and fluxapyroxad in the old irrigation channel in Tupigachi (Sample 12). These findings indicate further pesticide contamination originating from agriculture activities beyond NNI (See Table 2).

Table 2.

Compounds detected through non-targeted analysis (NTA) in 2022 water samples from Pedro Moncayo, Ecuador

Name	Drinking Water			Irrigation Water
	HH15a	HH18a	Wellb	7a	13b	22b	Final Confidence Levelc
Insecticides
Neonicotinoids (NNI)
Admire/Imidacloprid	*	*	*		*	*	1
(E)-thiamethoxam	*	*	*		*	*	1
Acetamiprid	*	*			*	*	1
Clothianidin	*	*			*		1
NNI transformation product from imidacloprid
Desnitro-imidacloprid	*		*		*		1
Imidacloprid-urea	*	*	*		*		1
Carbamate
Carbofuran	*	*	*		*	*	1
Organophosphate
Diazinon	*	*	*		*	*	1
Malaoxon	*						1
Malathion	*						1
Other insecticides
Dimethoate	*		*			*	2a
Methomyl	*	*					2a
Pyriproxyfen	*	*			*		1
Spinetoram J		*					2a
Spinosyn A	*	*			*	*	2a
Herbicides
Atrazine herbicides
Atrazine	*		*		*	*	1
Transformation product from Atrazine
MFCD00078645/desisopropylatrazine	*		*				1
XY5850570/desethylatrazine	*		*		*		1
Other herbicides
Ametryn						*	2a
Linuron	*						2a
N-(3,4-Dichlorophenyl)-N'-methylurea/DCPMU	*						1
Fungicides
Azoles (Benzimidazole, Imidazolinone, Pyrazole, Triazoles)
Benzimidazole
Carbendazim	*	*	*	*	*	*	1
Metrafenone	*	*			*	*	2a
Imidazolinone
Fenamidone	*	*			*		1
Pyrazole
Ethaboxam		*			*		1
Isopyrazam		*			*	*	2a
Triazole
Cyproconazole	*						1
Cyprodinil	*	*	*		*	*	1
Difenoconazole	*	*	*		*	*	1
Fenpropidin		*	*		*		2a
Flutriafol	*	*	*		*	*	1
Fluxapyroxad	*	*			*	*	1
Myclobutanil [ANSI]	*		*	*	*		1
Penconazole	*	*	*	*	*	*	2a
Propiconazole	*	*	*		*		1
Tebuconazole	*	*	*		*	*	1
Triadimenol B	*	*	*		*		1
Spiroxamine	*	*			*	*	2a
Oxazolidinedione
Etoxazole	*					*	1
Phenylamide
Metalaxyl-M	*	*	*		*	*	1
Benalaxyl	*	*	*			*	2a
Furalaxyl	*	*	*		*		2a
Mandipropamid	*	*		*		*	1
Phenylureas
Octhilinone	*		*	*			2a
Pyrimidine
Bupirimate	*	*	*		*	*	2a
Strobilurin
Fluoxastrobin			*				1
Azoxystrobin	*	*	*		*	*	1
Transformation product from Azoxystrobin
Azoxystrobin acid	*	*	*		*		2a
Carbamate fungicide
Diethofencarb		*			*	*	2a
Propamocarb	*	*	*		*	*	2a
Morpholine fungicide
Dimethomorph	*	*	*		*	*	1
Other fungicides
Boscalid	*	*	*	*	*	*	1
Pinonic acid	*					*	2a
Procymidone	*		*		*		2a
Other pesticides (uncategorized)
3,5-Dichloroaniline	*	*	*		*		1
Carbendazim	*	*	*	*	*	*	2a
Fluopyram	*	*	*		*	*	1
Acetanilide	*					*	2a

^aTabacundo

^bTupigachi

^cConfidence Level: 1= confirmed presence, 2a= probable structure identification, equivalent to 2a in the Schymanski framework (library spectrum match)

^*Detected.

Further NTA is warranted to understand the extent and patterns of pesticide contamination in drinking water, considering that organophosphate exposures have been associated with neurocognitive alterations in children (Muñoz-Quezada et al., 2013). Notably, In the ESPINA study, targeted analysis of urinary metabolite concentrations of organophosphates, pyrethroids, and neonicotinoids in Ecuadorian adolescents identified hotspots of organophosphate and pyrethroid exposure in areas with intensive floriculture, while neonicotinoids were more common in regions dominated by non-floricultural agriculture (Chronister et al., 2025).

In our NTA analysis, the most abundant group of pesticides by detection were fungicides, corresponding to 59% (37 out of 63) of compounds present in all samples at the given identification confidence intervals. The results revealed 15 different azole fungicides from four fungicide families: benzimidazole, imidazolinones, pyrazoles, and triazoles. We note that 10 of the 15 azoles identified belong to the triazole group. Similarly, a study investigating surface water and wastewater from Pasto city, Colombia in the Andean highlands also detected various fungicides, including azoxystrobin, carbendazim and difenoconazole in surface water (Hernández et al., 2024). A study in the Pearl River Delta in China, also found azoles—clotrimazole, ketoconazole, and miconazole (Huang et al., 2010). Azoles have the potential to cause liver toxicity, cardiotoxicity, and endocrine disruption (Bian et al., 2025), which justifies future targeted analyses to measure potential concentration ranges for these fungicides, and the others detected in the NTA, in both drinking and irrigation water. For example, a study in Wuhan, China, utilized targeted analysis to show the consistent presence of fungicides across the tap water system. In 2019, matched pairs of source water and treated water samples were collected from ten drinking water plants during February and July to evaluate contaminants originating from the Yangtze and Hanshui Rivers. The targeted results revealed that nine fungicides, including azoxystrobin, tebuconazole, and difenoconazole, were present in 100% of filtered source water samples. Additionally, in 169 tap water samples collected between February and October 2019, several fungicides including azoxystrobin, difenoconazole, and tricyclazole showed a 100% DF, highlighting their persistence throughout the distribution system (Liu et al., 2021). This study further illustrated that targeted analysis is essential for understanding the environmental fate of these fungicides. Furthermore, more research is needed on direct and indirect health effects of fungicides (Zubrod et al., 2019).

3.4. Implications

In 2024, 5479 ha in Ecuador were used for rose plantation (ESPAC, n.d; Instituto Nacional de Estadística y Censos, n.d.), which is characterized by large-scale operations that tend to use higher amounts of pesticides, regardless of toxicological classification or biological target (Vasco et al., 2025). In Pedro Moncayo specifically, floricultural greenhouses are used to cultivate roses, with the area under production continuing to expand (Cachipuendo et al., 2024). Our research indicates that these greenhouse operations may contribute to drinking water contamination. In addition, prior research in the region found that children living near greenhouses more frequently exhibit urinary concentrations of organophosphate insecticides (parathion, malathion), and the NNI, imidacloprid (Chronister et al., 2025).

Potential exposure routes for NNI from greenhouses into piped drinking water systems could be leaching due to aging water systems with intermittent water service, intense use of pesticides in rose production, and the proximity of floriculture to household water pipes. This is further supported by the fact that we did not find NNI in pretreatment water systems but only after water treatment. Studies conducted in Ontario, Canada (Sultana et al., 2018), and South Korea (Kim et al., 2021) found consistently lower NNI concentration after water treatment compared to raw drinking water, indicating that the increase is not related to the water treatment process. Similarly, a study in China indicated that NNI enter the water system post-treatment, leading to concerns about their persistence and potential health impacts due to chronic exposure in drinking water (He et al., 2021).

Small drinking water systems in rural areas often lack the financial and technical resources to replace aging infrastructure at sufficient intervals to ensure distribution of safe water to household taps at point of use (Glade and Ray, 2022). Compounding this challenge is intermittent service that generates discontinuous pressure, which can lead to intrusion of contaminants through back-suction into the distribution system, compromising chemical and microbiological quality of household drinking water (Kumpel and Nelson, 2016). Additionally, as drinking water moves through the distribution system, free chlorine levels decline due to reactions with organic material, though remaining disinfection may still occur along the pipeline with sufficient chlorine residual (Islam, 2015; Onyutha and Kwio-Tamale, 2022). The presence of NNI alongside low free chlorine in Tabacundo tap water supports the hypothesis of back infiltration. We found residual chlorine levels below the Ecuadorian standard in 72% (8/11) of Tabacundo households (Supplemental Information (SI) SB-1). A possible solution could be adding water production capacity to avoid service breaks and concomitant pressure drops and back siphoning. Further, the implementation of additional point-of-use water treatment technologies at the household level could be a viable solution. As previous research has highlighted, more advanced water treatment methods, including GAC filtration, are required to successfully remove NNI, atrazine, and triazoles from drinking water (Lazorko-Connon and Achari, 2009; Klarich et al., 2017; Kim et al., 2021; Bian et al., 2025). However, the installation of additional treatment infrastructure in households also comes with operation and maintenance costs, including filter replacement (Bayer et al., 2005; Minnesota Department of Health, n.d.).

3.5. Limitations

This study has limitations that point to directions for future research. While there is spatial and temporal variability in our water quality sampling, there is limited temporal scope. The two measurement points separated by a year do not fully capture seasonal variations and long-term trends in pesticide contamination, necessitating continuous year-round monitoring, for a comprehensive understanding of seasonal and annual variability. This research does not represent water quality profiles for Ecuador generally, but only the systems we sampled in Pedro Moncayo, near floricultural fields and ESPINA participants' households, thus reducing the generalizability of our results to other regions of Ecuador. Additionally, given that limited research currently exists regarding fungicide presence in these water systems, we plan to conduct dedicated targeted fungicide analysis in future studies. Lastly, it is possible that pesticide contamination in drinking water samples originated from contaminated tap fixtures rather than the piped water supply. To minimize the risk of such contamination, we implemented a flushing protocol in which tap water was allowed to run for 30 s to 2 min prior to sampling.

4. Conclusion

In the present study, we conducted targeted analyzes of NNI and atrazine levels in both drinking and irrigation water in a floriculture region of Ecuador and conducted NTA to gain a broader understanding of chemical composition present in water samples. Our findings revealed that 20% of drinking water samples and 58% of irrigation water samples tested positive for NNI. Further, our initial exploratory NTA revealed the presence of many insecticides, herbicides, and multiple fungicides, many belonging to the azole family, with confirmed levels of confidence, warranting further targeted analysis to evaluate potential concentration ranges.

Pedro Moncayo's floriculture greenhouses were built in and around residential areas, and large volumes of pesticides are applied to roses. While source water may not be heavily contaminated, our findings indicate that pesticide infiltration may occur post-treatment, within the drinking water distribution system itself in proximity to floriculture. A combination of aging piped systems, intermittent service, and drinking water systems located in intense pesticide spray sites may allow pesticides to leach into treated drinking water and irrigation water. Intense agriculture spray sites should be located away from drinking water to reduce this exposure pathway for community members.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.envpol.2026.127757.

Data availability

Data will be made available on request.