Wastewater-impacted River Research Articles

Bioaccumulation and tissue distribution of pharmaceuticals and their transformation products in fish along the pollution gradients of a wastewater-impacted river

A field study on the occurrence and distribution of forty-three pharmaceutically active compounds (PhACs) in water and fish samples from anthropogenically impacted section of the Sava River (Croatia) was performed to estimate the importance of bioaccumulation for the environmental risk assessment of PhACs. The study was performed using a highly specific LC-MS/MS method, tailored to include the most prominent PhACs from different therapeutic categories as well as their major metabolites and/or transformation products (TPs). The results revealed a widespread occurrence of PhAC residues both in water and fish samples with a large spatial variability reflecting the distance from the dominant wastewater discharges. The most prominent PhAC categories in less polluted upstream part of the river were common psychostimulants caffeine and cotinine, therapeutic opioids and cardiovascular drugs, while in the river section affected by the local municipal and industrial wastewater inputs, antibiotic drugs became clearly predominant, especially in fish tissue samples. The apparent bioconcentration factors (BCFs) of investigated PhACs varied over several orders of magnitude, from 0.02 ± 0.01 Lkg−1 for O-desmethyl tramadol in fish muscle to 784 ± 260 Lkg−1 for terbinafine in fish liver, indicating rather large differences in their bioconcentration potential and affinity to different tissues, with the tissue-specific BCFs increasing in the following order: muscle < gills < gonads < heart < liver < kidneys. The bioconcentration potential of most of the PhACs included in this study was only low to moderate however moderately high BCFs of certain PhACs (e.g. sertraline, terbinafine, loratadine, diazepam and azithromycin) in some tissues should be taken into consideration when assessing their potential environmental risks. Moreover, it was shown that BCFs could be strongly affected by biotransformation in fish. Risk prioritization based on risk quotient (RQ) and ToxPi index, revealed antibiotics, in particular azithromycin, and therapeutic psychoactive substances as the most hazardous pharmaceutical contaminants in the Sava River.

Spatial and temporal metagenomics of river compartments reveals viral community dynamics in an urban impacted stream

Although river ecosystems constitute a small fraction of Earth’s total area, they are critical modulators of microbially and virally orchestrated global biogeochemical cycles. However, most studies either use data that is not spatially resolved or is collected at timepoints that do not reflect the short life cycles of microorganisms. To address this gap, we assessed how viral and microbial communities change over a 48-hour period by sampling surface water and pore water compartments of the wastewater-impacted River Erpe in Germany. We sampled every 3 hours resulting in 32 samples for which we obtained metagenomes along with geochemical and metabolite measurements. From our metagenomes, we identified 6,500 viral and 1,033 microbial metagenome assembled genomes (MAGs) and found distinct community membership and abundance associated with each river compartment (e.g., Competibacteraceae in surfacewater and Sulfurimonadaceae in pore water). We show that 17% of our viral MAGs clustered to viruses from other ecosystems like wastewater treatment plants and rivers. Our results also indicated that 70% of the viral community was persistent in surface waters, whereas only 13% were persistent in the pore waters taken from the hyporheic zone. Finally, we predicted linkages between 73 viral genomes and 38 microbial genomes. These putatively linked hosts included members of the Competibacteraceae, which we suggest are potential contributors to river carbon and nitrogen cycling via denitrification and nitrogen fixation. Together, these findings demonstrate that members of the surface water microbiome from this urban river are stable over multiple diurnal cycles. These temporal insights raise important considerations for ecosystem models attempting to constrain dynamics of river biogeochemical cycles.

Bioaccumulation and bioamplification of pharmaceuticals and endocrine disruptors in aquatic insects

Environmental fate of emerging contaminants such as pharmaceuticals and endocrine disrupting compounds at the aquatic terrestrial boundary are largely unexplored. Aquatic insects connect aquatic and terrestrial food webs as their life cycle includes aquatic and terrestrial life stages, thus they represent an important inter-habitat linkage not only for energy and nutrient flow, but also for contaminant transfer to terrestrial environments. We measured the concentrations of pharmaceuticals and endocrine disrupting compounds in the larval and adult tissues (last larval stages and teneral adults) of five Odonata species sampled in a wastewater-impacted river, in order to examine their bioaccumulation and bioamplification at different taxonomic levels. Twenty different compounds were bioaccumulated in insect tissues, with majority having higher concentrations (up to 90% higher) in aquatic larvae compared to terrestrial adults (reaching 88 ng/g for 1H-benzotriazole). However, increased concentration in adults was observed for seven compounds in at least one suborder (41% of the accumulated), confirming contaminants bioamplification across the metamorphosis. Both, bioaccumulation and bioamplification differed at various taxa levels; the order (Odonata), suborder (Anisoptera and Zygoptera) and species level. Highest variability was observed between Anisoptera and Zygoptera, due to the underlying differences in their ecology. Generally, Zygoptera had higher concentrations of contaminants in both larvae and adults. Additionally, we aimed at predicting effects of contaminant properties on bioaccumulation and bioamplification patterns using the commonly used physicochemical and pharmacokinetic descriptors on both order and suborder levels, however, neither of the two processes could be consistently predicted with simple linear models. Our study highlights the importance of taxonomy in studies aiming at advancing the understanding of contaminant exchange between aquatic and terrestrial food webs, as higher taxonomic categories include ecologically diverse groups, whose contribution to “the dark side of subsidies” could substantially differ.

Quantitative Determination and Environmental Risk Assessment of 102 Chemicals of Emerging Concern in Wastewater-Impacted Rivers Using Rapid Direct-Injection Liquid Chromatography-Tandem Mass Spectrometry.

The rapid source identification and environmental risk assessment (ERA) of hundreds of chemicals of emerging concern (CECs) in river water represent a significant analytical challenge. Herein, a potential solution involving a rapid direct-injection liquid chromatography–tandem mass spectrometry method for the quantitative determination of 102 CECs (151 qualitatively) in river water is presented and applied across six rivers in Germany and Switzerland at high spatial resolution. The method required an injection volume of only 10 µL of filtered sample, with a runtime of 5.5 min including re-equilibration with >10 datapoints per peak per transition (mostly 2 per compound), and 36 stable isotope-labelled standards. Performance was excellent from the low ng/L to µg/L concentration level, with 260 injections possible in any 24 h period. The method was applied in three separate campaigns focusing on the ERA of rivers impacted by wastewater effluent discharges (1 urban area in the Basel city region with 4 rivers, as well as 1 semi-rural and 1 rural area, each focusing on 1 river). Between 25 and 40 compounds were quantified directly in each campaign, and in all cases small tributary rivers showed higher CEC concentrations (e.g., up to ~4000 ng/L in total in the R. Schwarzach, Bavaria, Germany). The source of selected CECs could also be identified and differentiated from other sources at pre- and post- wastewater treatment plant effluent discharge points, as well as the effect of dilution downstream, which occurred over very short distances in all cases. Lastly, ERA for 41 CECs was performed at specific impacted sites, with risk quotients (RQs) at 1 or more sites estimated as high risk (RQ > 10) for 1 pharmaceutical (diclofenac), medium risk (RQ of 1–10) for 3 CECs (carbamazepine, venlafaxine, and sulfamethoxazole), and low risk (RQ = 0.1–1.0) for 7 CECs (i.e., RQ > 0.1 for 11 CECs in total). The application of high-throughput methods like this could enable a better understanding of the risks of CECs, especially in low flow/volume tributary rivers at scale and with high resolution.

Combining passive sampling with toxicity testing to evaluate potential ecotoxicological effects of pharmaceuticals in wastewater-impacted rivers

A passive sampling method was employed for time-integrative monitoring of five pharmaceuticals and one transformation product (TP) in rivers impacted by sewage treatment plants, in parallel with traditional sampling methods. Target pharmaceuticals, other than naproxen, were detected through passive sampling, with average concentrations in the range of 0.2–5.8 ng/L, and through active sampling, with average concentrations in the range of 0.5–21.7 ng/L. Meanwhile, the ecotoxicological effects of pharmaceuticals and TPs were assessed, including the formation of zebrafish embryos and expression of target genes, upon exposure of zebrafish embryos to sulfadiazine (SDZ) and its TP sulfacetamide, as well as two artificial mixed rivers. The exposure results showed negligible impacts of environmental levels of SDZ, while mimic mixture exposure disturbed the development of embryos and led to the alteration of the socs3, TNF-α, and IL-1β genes. The findings of this study indicated that although pharmaceutical concentrations in rivers receiving treated wastewater are low, the potential ecological effects on the aquatic environment require more attentions.

Contribution of wastewater- versus non-wastewater-derived sources to haloacetonitriles formation potential in a wastewater-impacted river

Population growth and urbanization have led to the increasing presence of treated wastewater effluents in downstream drinking water sources. Drinking water sources influenced by organic matter from upstream wastewater treatment plant (WWTP) effluents are thought prone to the formation of haloacetonitriles (HANs), a group of nitrogenous disinfection by-products (DBPs) that can exhibit higher toxicity than currently regulated carbonaceous DBPs. We develop a framework for studying the HAN formation potential (HAN-FP) considering the WWTP and non-WWTP related sources of HAN precursors, and apply this framework to a representative WWTP-impacted river, the Illinois River, USA. A spatiotemporally-resolved river hydrodynamic and water quality model is developed using HEC-RAS to quantify the contribution of WWTP versus non-WWTP sources of HAN-FP precursors. Results show that non-WWTP sources of HAN-FP are considerable, accounting for up to 78% of HAN-FP concentration. Moreover, the contribution of the two sources varies due to streamflow discharge variability. During lower flows, the contribution of WWTPs drives the high concentration of HAN-FP and during higher flows, the contribution of non-WWTP sources becomes dominant. As a result, a high risk of HAN-FP may exist persistently (HAN-FP concentration is always larger than 9.7 μg/L in this study), not only during low flows but also during high flows due to both wastewater- and non-wastewater-derived HAN-FP sources.

Modelling the interplay of future changes and wastewater management measures on the microbiological river water quality considering safe drinking water production

Rivers are important for drinking water supply worldwide. However, they are often impacted by pathogen discharges via wastewater treatment plants (WWTP) and combined sewer overflows (CSO). To date, accurate predictions of the effects of future changes and pollution control measures on the microbiological water quality of rivers considering safe drinking water production are hindered due to the uncertainty of the pathogen source and transport variables. The aim of this study was to test an integrative approach for an improved understanding of these effects, i.e. climate change and population growth as well as enhanced treatment at WWTPs and/or prevention of CSOs. We applied a significantly extended version of QMRAcatch (v1.0 Python), a probabilistic-deterministic model that combines fate and transport modelling with quantitative microbial infection risk assessment. The impact of climatic changes until the period 2035–2049 was investigated by a conceptual semi-distributed hydrological model, based on regional climate model outputs. QMRAcatch was calibrated and validated using site- and source-specific data (human-associated genetic microbial source tracking marker and enterovirus). The study showed that the degree to which future changes affect drinking water safety strongly depends on the type and magnitude of faecal pollution sources and are thus highly site- and scenario-specific. For example, if the load of pathogens from WWTPs is reduced through enhanced treatment, climate-change driven increases in CSOs had a considerable impact. Preventing CSOs and installing enhanced treatment at the WWTPs together had the most significant positive effect. The simultaneous consideration of source apportionment and concentrations of reference pathogens, focusing on human-specific viruses (enterovirus, norovirus) and cross-comparison with bacterial and protozoan pathogens (Campylobacter, Cryptosporidium), was found crucial to quantify these effects. While demonstrated here for a large, wastewater-impacted river, the approach is applicable at other catchments and pollution sources. It allows assessing future changes and selecting suitable pollution control measures for long-term water safety planning.

Pharmaceutical transforming microbes from wastewater and natural environments can colonize microplastics

Treated wastewater effluents are a source of emerging contaminants, including microplastics and pharmaceuticals, that impact rivers and streams. As microplastics are transported from wastewater treatment into the environment, pharmaceuticals can sorb to the surface and also be colonized by microorganisms. To investigate the microbial communities that are important in pharmaceutical transformation on microplastic surfaces, we used a culture-based approach with naproxen as the model pharmaceutical. Microplastic beads served as a solid substrate for delivering naproxen to anaerobic cultures inoculated with either anaerobic digester sludge or sediment from a wastewater-impacted river. After demonstrating naproxen transformation activity within the cultures, we separated the bulk liquid culture from the colonized microplastic beads and transferred them into separate bottles of sterile media amended with naproxen. Naproxen transformation occurred in cultures that contained microbially-colonized microplastics. Results from DNA analyses of the microbial community from each treatment demonstrated a different microbial community structure on the colonized plastic compared to that of the planktonic cells, thus illustrating a selection of the microbial community by the microplastics. These findings demonstrate that the microbial community attached to microplastic beads can continue pharmaceutical transformation activity when the microplastics are transferred to new media, thus serving as a model for the potential transport of pharmaceuticaltransforming microbes from wastewater treatment to freshwater environments.

Selection of performance reference compound (PRC) for passive sampling of pharmaceutical residues in an effluent dominated river

A passive sampling device, a polar organic chemical integrative sampler (POCIS), was used to monitor 13 pharmaceuticals and 8 transformation products in upstream and downstream wastewater treatment plant effluent. A POCIS laboratory calibration study was performed to determine uptake behavior and the effect of water flow on the sampling rate. Most compounds showed a linear accumulation, and the sampling rate values ranged from 0.031 to 0.559 L/day. The developed POCIS samplers were used in field experiments in a wastewater-impacted river. Using the calculated sampling rates, the time-weighted average concentration values were measured by three different approaches: (1) laboratory calibration sampling rates (2) performance reference compound (PRC) correction sampling rates and (3) field calibration sampling rates. Nine deuterated compounds (acetaminophen-d3, antipyrine-d3, sulfamethoxazole-d4, carbamazepine-d10, diclofenac acid-d4, clofibric acid-d4, bezafibrate-d6, ibuprofen-d3 and naproxen-d3) were studied as PRCs. Antipyrine-d3 was successfully tested as a PRC for sulfamethoxazole, ibuprofen, 2-hydroxy ibuprofen, diclofenac acid, 4-hydroxydiclofenac acid, carbamazepin, carbamazepin 10,11-epoxide, sulfadiazine, 1-naphthol, antipyrine, naproxen and 4-chlorobenzoic acid. Finally, the POCIS was used to monitor target compounds in river water and measure their attenuation. For most compounds, the POCIS attenuation results were not significantly different from those of the spot samples, which demonstrated that a POCIS with a PRC correction can determine the attenuation of organic micropollutants in rivers.

Monitoring the Behavior of Emerging Contaminants in Wastewater-Impacted Rivers Based on the Use of Fluorescence Excitation Emission Matrixes (EEM).

This study investigated the applicability of fluorescence indexes based on the interpretation of excitation emission matrices (EEMs) by PARAFAC analysis and by selecting fluorescence intensities at a priori defined excitation/emission pairs as surrogates for monitoring the behavior of emerging organic compounds (EOCs) in two catchment basins impacted by wastewater discharges. Relevant EOC and EEM data were obtained for a 90 km stretch of the Simeto River, the main river in Sicily, and the smaller San Leonardo River, which was investigated for a 17 km stretch. The use of fluorescence indexes developed by these two different approaches resulted in similar observations. Changes of the fluorescence indexes that correspond to a group of humic-like fluorescing species were determined to be highly correlated with the concentrations of recalcitrant contaminants such as sucralose, sulfamethoxazole and carbamazepine, which are typical wastewater markers in river water. Changes of the fluorescence indexes related to tyrosine-like substances were well correlated with the concentrations of ibuprofen and caffeine, anthropogenic indicators of untreated wastewater discharges. Chemical oxygen demand and dissolved organic carbon concentrations were correlated with humic-like fluorescence indexes. The observed correlations were site-specific and characterized by different regression parameters for every collection event. Caffeine and carbamazepine showed correlations with florescence indexes in the San Leonardo River and in the alluvial plain stretch of the Simeto River, whereas sucralose, sulfamethoxazole and ibuprofen have always been well correlated in all the investigated river stretches. However, when data of different collection events from river stretches where correlations were observed were combined, good linear correlations were obtained for data sets generated via the normalization of the measured concentrations by the average value for the corresponding collection event. These results show that fluorescence based indexes can be used to monitor the behavior of some trace organic contaminants in wastewater impacted rivers and to track wastewater discharges in streams and rivers.

Real-time PCR assays for the detection and quantification of carbapenemase genes (bla KPC, bla NDM, and bla OXA-48) in environmental samples.

In this study, we have developed real-time PCR assays using SYBR Green chemistry to detect all known alleles of bla KPC, bla NDM, and bla OXA-48-like carbapenemase genes in water, sediment, and biofilm samples collected from hospital and wastewater treatment plant (WWTP) effluents and rivers receiving chronic WWTP discharges. The amplification of bla KPC, bla NDM, and bla OXA-48 DNA was linear over 7 log dilutions (R 2 between 0.995 and 0.997) and showing efficiencies ranging from 92.6% to 100.3%. The analytical sensitivity indicated that the reaction for bla KPC, bla NDM, and bla OXA-48-like genes was able to detect 35, 16, and 19 copy numbers per assay, respectively. The three carbapenemase genes were detected in hospital effluents, whereas only the bla KPC and bla NDM genes were detected in biofilm and sediment samples collected from wastewater-impacted rivers. The detection of bla KPC, bla NDM, and bla OXA-48-like genes in different matrices suggests that carbapenem-resistant bacteria occur in both planktonic and benthic habitats thus expanding the range of resistance reservoirs for last-resort antibiotics. We believe that these real-time PCR assays would be a powerful tool for the rapid detection and quantification of bla KPC, bla NDM, and bla OXA-48-like genes in complex environmental samples.

Occurrence and attenuation of pharmaceuticals and their transformation products in rivers impacted by sewage treatment plants

Pharmaceuticals and transformation products were attenuated in the wastewater-impacted river, and environment conditions influenced the attenuation process.

Relative importance of P and N in macrophyte and epilithic algae biomass in a wastewater-impacted oligotrophic river.

The role of nutrient loading on biomass growth in wastewater-impacted rivers is important in order to effectively optimize wastewater treatment to avoid excessive biomass growth in the receiving water body. This paper directly relates wastewater treatment plant (WWTP) effluent nutrients (including ammonia (NH3-N), nitrate (NO3-N) and total phosphorus (TP)) to the temporal and spatial distribution of epilithic algae and macrophyte biomass in an oligotrophic river. Annual macrophyte biomass, epilithic algae data and WWTP effluent nutrient data from 1980 to 2012 were statistically analysed. Because discharge can affect aquatic biomass growth, locally weighted scatterplot smoothing (LOWESS) was used to remove the influence of river discharge from the aquatic biomass (macrophytes and algae) data before further analysis was conducted. The results from LOWESS indicated that aquatic biomass did not increase beyond site-specific threshold discharge values in the river. The LOWESS-estimated biomass residuals showed a variable response to different nutrients. Macrophyte biomass residuals showed a decreasing trend concurrent with enhanced nutrient removal at the WWTP and decreased effluent P loading, whereas epilithic algae biomass residuals showed greater response to enhanced N removal. Correlation analysis between effluent nutrient concentrations and the biomass residuals (both epilithic algae and macrophytes) suggested that aquatic biomass is nitrogen limited, especially by NH3-N, at most sampling sites. The response of aquatic biomass residuals to effluent nutrient concentrations did not change with increasing distance to the WWTP but was different for P and N, allowing for additional conclusions about nutrient limitation in specific river reaches. The data further showed that the mixing process between the effluent and the river has an influence on the spatial distribution of biomass growth.

Fate of Pharmaceuticals and Their Transformation Products in Four Small European Rivers Receiving Treated Wastewater.

A considerable knowledge gap exists with respect to the fate and environmental relevance of transformation products (TPs) of polar organic micropollutants in surface water. To narrow this gap we investigated the fate of 20 parent compounds (PCs) and 11 characteristic TPs in four wastewater-impacted rivers. Samples were obtained from time-integrated active sampling as well as passive sampling using polar organic chemical integrative samplers (POCIS). Seventeen out of the 20 PCs were detected in at least one of the rivers. All the PCs except acesulfame, carbamazepine, and fluconazole were attenuated along the studied river stretches, with the largest decrease found in the smallest river which had an intense surface water-pore water exchange. Seven TPs were detected, all of which were already present directly downstream of the WWTP outfall, suggesting that the WWTPs were a major source of TPs to the recipients. For anionic compounds, attenuation was the highest in the two rivers with the lowest discharge, while the pattern was not as clear for neutral or cationic compounds. For most compounds the results obtained from active sampling were not significantly different from those using POCIS, demonstrating that the cost and labor efficient POCIS is suitable to determine the attenuation of organic micropollutants in rivers.

Nitrifying bacteria mediate aerobic ammonia oxidation and urea hydrolysis within the Grand River

Ammonia oxidation is an important process for decreasing ammonia concentrations in wastewater-impacted rivers. Ammonia-oxidizing bacteria (AOB) and archaea (AOA) are re - sponsible for ammonia oxidation, which is the first step of the nitrification process. Nitrification and urea hydrolysis were monitored in sediment and water column samples in the Grand River (Ontario), and nitrification inhibitors (allylthiourea and 2-phenyl-4,4,5,5-tetramethylimidazoline- 1-oxyl 3-oxide) helped identify the relative contributions of AOA and AOB to ammonia oxidation. Despite the presence of AOA, our results implicated AOB as the dominant contributors to ammo- nia oxidation, both directly and in association with urea hydrolysis.

Enhanced formation of disinfection byproducts in shale gas wastewater-impacted drinking water supplies.

The disposal and leaks of hydraulic fracturing wastewater (HFW) to the environment pose human health risks. Since HFW is typically characterized by elevated salinity, concerns have been raised whether the high bromide and iodide in HFW may promote the formation of disinfection byproducts (DBPs) and alter their speciation to more toxic brominated and iodinated analogues. This study evaluated the minimum volume percentage of two Marcellus Shale and one Fayetteville Shale HFWs diluted by fresh water collected from the Ohio and Allegheny Rivers that would generate and/or alter the formation and speciation of DBPs following chlorination, chloramination, and ozonation treatments of the blended solutions. During chlorination, dilutions as low as 0.01% HFW altered the speciation toward formation of brominated and iodinated trihalomethanes (THMs) and brominated haloacetonitriles (HANs), and dilutions as low as 0.03% increased the overall formation of both compound classes. The increase in bromide concentration associated with 0.01-0.03% contribution of Marcellus HFW (a range of 70-200 μg/L for HFW with bromide = 600 mg/L) mimics the increased bromide levels observed in western Pennsylvanian surface waters following the Marcellus Shale gas production boom. Chloramination reduced HAN and regulated THM formation; however, iodinated trihalomethane formation was observed at lower pH. For municipal wastewater-impacted river water, the presence of 0.1% HFW increased the formation of N-nitrosodimethylamine (NDMA) during chloramination, particularly for the high iodide (54 ppm) Fayetteville Shale HFW. Finally, ozonation of 0.01-0.03% HFW-impacted river water resulted in significant increases in bromate formation. The results suggest that total elimination of HFW discharge and/or installation of halide-specific removal techniques in centralized brine treatment facilities may be a better strategy to mitigate impacts on downstream drinking water treatment plants than altering disinfection strategies. The potential formation of multiple DBPs in drinking water utilities in areas of shale gas development requires comprehensive monitoring plans beyond the common regulated DBPs.

Uptake of human pharmaceuticals in bull sharks (Carcharhinus leucas) inhabiting a wastewater-impacted river

The presence of human pharmaceuticals in sewage-impacted ecosystems is a growing concern that poses health risks to aquatic wildlife. Despite this, few studies have investigated the uptake of active pharmaceutical ingredients (APIs) in aquatic organisms. In this study, the uptake of 9 APIs from human drugs was examined and compared in neonate bull sharks (Carcharhinus leucas) residing in pristine (Myakka River) and wastewater-impacted (Caloosahatchee River) tributaries of Florida's Charlotte Harbor estuary. The synthetic estrogen used in human contraceptives (17α-ethynylestradiol) and 6 of the selective serotonin/norepinephrine reuptake inhibitors (citalopram, fluoxetine, fluvoxamine, paroxetine, sertraline, venlafaxine) used in human antidepressants were observed at detectable and, in some cases, quantifiable levels in plasma of Caloosahatchee River sharks. Comparatively, only venlafaxine was detected in the plasma of a single Myakka River shark at a level below the limit of quantitation. These results suggest that sharks residing in wastewater-impacted habitats accumulate APIs, a factor that may pose special risks to C. leucas since it is one of few shark species to regularly occupy freshwater systems. Further research is needed to determine if the low levels of API uptake observed in Caloosahatchee River bull sharks pose health risks to these animals.

Abiotic influences on dissolved oxygen in a riverine environment

Dissolved oxygen (DO) is an important parameter in riverine health. Periphyton and/or macrophytes are frequently the drivers behind the fluctuation of DO levels in aquatic environments; however, the effects of abiotic and biotic factors on biomass and in turn on DO may be variable from river to river. The objective of this paper is to understand which factors govern DO levels in terms of daily minimum DO ( DO min) and daily DO variation (Δ DO) in a major wastewater-impacted river using statistical data analysis and modeling. Both climatic conditions (reflected in water temperature) and hydrometric conditions (flow) were major factors influencing DO min and Δ DO. The effect of flow on DO min and Δ DO was discontinuous, depending on the flow magnitude. Nutrient loading from wastewater effluent was not identified to have a significant impact on a daily basis; however, their role over large time scales is unclear. In the data-driven modeling approaches, the non-linear approach using multiple-layer perceptron neural network, which has very flexible architecture, was superior to the linear approach used (multiple linear regression). Although nutrients are likely related to the DO min and Δ DO, flow and temperature were sufficient to obtain robust prediction in the data-driven modeling. This approach is useful to model and understand complicated processes when their governing mechanisms are not well presented in conceptual- or physically based models.