A multi-season analysis of water quality trends from selected water sources in Kyeizooba sub-county, Bushenyi District, Uganda

Water is a priceless resource that is important for the continuation and growth of life on Earth and the different ways that humans utilize water in their daily lives and for economic activities (Kadyampakeni et al. 2017; Safari et al. 2020; Kanyesigye et al. 2023). These required functions are the production of energy, agriculture, industry, drinking, and sanitation. Water not only serves as the most basic human right but is also a significant element for the actualization of global development goals (Elangovan et al. 2018; Husain et al. 2020). In view of this, water safeguarding and quality assurance entail a complete monitoring and assessment system with continuous activity, as contaminating substances have significant impacts on both human and environmental health. Water quality is a great component for ensuring social progression. Environmental health and public welfare of women and men depend on water quality; for this reason, it is significant to measure and track its physiochemical features (Daworiye 2015; Atwebembeire et al. 2019). Such monitoring and analysis of the physico-chemical elements provide a very practical determination of the quality of water chemically and physically, the identification of contaminants, the assessment of health safety, and the creation of effective management approaches (Okonko et al. 2008; Patil 2012; Izah et al. 2018; Adesakin et al. 2020; Turibamwe & Wangalwa 2020). Unveiling this research employs a variety of water quality research traditions and contemporary techniques, comprising field, sampling, laboratory, and remote sensing.

Recent scientific discoveries have underscored the positive outcomes of interfacing the techniques of remote detection and machine learning with normal detection mechanisms. Ikonen et al. (2025) showed proactive measures like using a thermal infrared (TIR) band for TIR imaging to make water source monitoring more difficult by taking less time and resources. Similarly, Takahashi et al. (2025) used Sentinel-2 satellite imagery for temporal and spatial analysis to quantify spatiotemporal variability of turbidity, which was important in the examination of seasonal pollution trends and areas at high risk. The implication of this research shows that remote sensing might be useful in providing comprehensive water quality assessment on a large scale that can be done in parallel with field sampling, as there is a need to pursue the development of a more AI-driven hyperspectral imaging system. Estrada et al. (2025) reported that such a system has shown the potential to detect small water pollutants with high precision, a critical aspect in improving monitoring. Additionally, Dhal & Kar (2025) is an additional technique that identifies the level of pollution in the water and thereby helps to formulate the relevant water management politics. These innovative solutions increase the exactness and availability of water quality assessments, creating a broader context for monitoring and decision-making.

A broad array of attributes is embodied in the term ‘physicochemical parameters,’ such as temperature, pH, turbidity, dissolved oxygen (DO), electrical conductivity (EC), total dissolved solids (TDSs), and some special chemicals like organics and nutrients, which are important for the measure of water quality (Patil 2012; Lukubye & Andama 2017; Gorzel et al. 2018; Izah et al. 2018; Atwebembeire et al. 2019). These indicators include all environmental elements, human impact, and natural processes that affect water quality (Guardiola et al. 2015). Candidates for these features can be gathered by tracking datasets using systems that can be programed to flag instances of unwanted characteristics, enhancing situational awareness (Elangovan et al. 2018; Madhav et al. 2018; Saalidong et al. 2022). However, with growing water scarcity and increasing human activities, evaluating freshwater resources and their quality is becoming even more critical as poor water management poses severe risks to global economies and sustainable development (Gorzel et al. 2018). Despite clear evidence of declining water quality worldwide, the issue has not received adequate attention from policymakers and the public. Instead, societies have often adapted to deteriorating conditions at significant financial and ecological costs, resulting in millions of preventable deaths. Alarmingly, four billion people worldwide either lack access to safe drinking water or perceive their water sources as contaminated (Kibria 2016; Ebuete & Bariweni 2019; Marks et al. 2020). In response, researchers are increasingly integrating UAV-based remote sensing and geospatial modelling into traditional water quality assessments.

Building on these concerns, Guettaf et al. (2017) and Atwebembeire et al. (2019) certainly, two other papers focusing on the drinking water sources in Mbarara Municipality mention that the major sources are the shallow groundwater (boreholes, springs, and wells), as well as the roof rainwater collection harvesting. Lukubye & Andama (2017) highlighted that the areas around the water sources remained at high risk of pollution as these sources are highly significant waterholes as a result. The stability of the water sources, though, has been the extent of human activities such as agriculture, naturally occurring hazards, and waste that is improperly taken advantage of, as was also pointed out by Gorzel et al. (2018). Similarly, Marks et al. (2020) found the same water sources in Bushenyi, among others, including piped water, boreholes, SWASSCA taps, bottled water, rivers, rainwater, stream water, springs, spring tanks, shallow wells, mudholes, and mud wells. Although these studies point out that water quality in the areas is commonly higher than WHO safety limits, with bacteriological pollution, Escherichia coli that is specifically linked to domestic sewage and drain discharge. Addressing these concerns, Ikonen et al. (2025) were highlighted within the paper. The study covered the use of remote sensing, AI-driven analysis, and the application of geospatial models, leading to a class of tools designed for more efficient and improved water quality management purposes.

Building upon existing literature, this research aims to address the growing concern about quality water deterioration in Kyeizooba Sub-county, Bushenyi District, Uganda. This study offers a novel approach by implementing a range of evaluation tools that include a derived water quality index (WQI) and multiple linear regression (MLR) modelling to explore physicochemical and microbiological parameters. Contrary to classical research using huge sets of laboratory experiments, this research employs a simplified yet effective modeling approach that enhances predictive accuracy while reducing the need for frequent and costly lab experiments. This unique takeout of this research makes use of tools that are specially designed to target rural water sources, which are areas that usually have minimal data and limited resources. Doing this, the research is able to line out pragmatic recommendations that support the local authorities and policy help in the implementation of well-informed strategies that ensure clean water for all despite the plight of those soul-entrenched communities.

The study implemented a systematic experimental design to evaluate water quality in Kyeizooba, sub-county, Bushenyi District, Uganda. The methodological framework encompasses discrete phases, each designed to contribute to the integrity and validity of the findings. First, the study area was purposely selected based on its documented dependence on unprotected natural water sources for domestic and agricultural purposes. The reliance renders the population particularly vulnerable to waterborne diseases and necessitates a comprehensive assessment of water quality parameters. To account for temporal variability, water samples were collected from a network of 12 strategically distributed sites representing the primary water source, including streams, wells, springs, tanks, reserves, and impounded dams systems. Sampling was conducted during both the dry seasons (June–August) and the wet seasons (September–November). Fourth, physical and chemical parameters, including temperature, pH, turbidity, DO, EC, TDSs, nitrate, and phosphate, were quantified using calibrated analytical instrumentation. Fifth, the WQI was computed using the weighted arithmetic method (WAM). Sixth, to assess the predictive validity of the WQI, the resulting index values were subjected to MLR analysis. The MLR model was used to examine the relationship between the WQI and visual Cisco physicochemical parameters, thereby evaluating the extent to which the WQI accurately reflects the complex interplay of practice, influencing water quality in the study area.

However, these water sources are highly vulnerable to anthropological conditions due to their widespread utilization for these diverse activities, including brick manufacturing, livestock, husbandry, and construction. Therefore, the investigation seeks to address their knowledge defect by rigorously evaluating the physico-chemical and biological properties of these essential water sources. The study focused specifically on open wells, unprotected springs, and a system given their widespread prevalence of utilization by rural communities and domestic purposes. Additionally, the study sought to raise awareness among the local population about the risks associated with water contamination and promote preventive measures to safeguard public health.

A crucial factor in the strategic selection of a sampling location was the systematic assessment of anthropogenic activities agreeing within the intermediate vicinity of each water source. Understanding the nature of a human impact is essential for conceptualizing observed water quality parameters. Table 1 provides a comprehensive evaluation of these activities, meticulously documented through direct observation and standardized recording protocols at each sample site. The proximity of these activities to the water sources was measured using a tape measure to determine their potential impact on water quality. The number of open gardens in the vicinity was recorded, along with the classification of crops into annuals and perennials. Since agricultural practices significantly impact water quality through soil erosion, fertilizer runoff, and pesticide use, understanding the type and scale of cultivation provided valuable insights into pollution sources. Additionally, the presence and density of livestock, particularly cattle, were noted, as animal waste runoff and trampling near water bodies can contribute to microbial and nutrient contamination. These factors collectively helped in identifying areas where human activities exerted the highest pressure on water quality.

A systematic water sampling protocol was implemented to collect two samples for each site during both dry seasons (1 June–31 August 2023), between 8:00 AM and 11:00 AM. One sample was taken daily, resulting in a total of 90 samples per season. Each sample was obtained in a pre-sterilized 500 mL polyethylene container. To ensure a comprehensive characterization of water quality, covering multiple points within the water body. Specifically, samples were collected from multiple locations within each water source, representing various depths and proximity to potential contamination sources. To prevent contamination and maintain sample integrity, rigorous cleaning procedures were followed for all plastic containers prior to sampling. The bottles were first washed with a detergent solution, then rinsed thoroughly with dilute nitric acid to eliminate potential metal interference, and finally cleansed with distilled water. During sample collection, each container was carefully submerged and then raised vertically at a constant rate from the water surface to the sediment. Water interface, minimizing disturbance of bottom sediments and ensuring a representative sample of the entire water column. Immediately following collection, each container was hermetically sealed underwater to minimize atmospheric exchange and transferred to an insulated, high-cooled transport container for preservation. Following collection, all samples were transported to the NWSC laboratory in Mbarara for detailed physico-chemical and biological analyses. To ensure accuracy and prevent degradation, testing was initiated within two hours of collection. The specific locations and characteristics of the sampling sites, including geographic coordinates, water source type, and environmental conditions, are detailed in Table 2.

To evaluate the quality of the waters of different sources, a study by means of two factors was carried out. The evaluation involved the physico-chemical parameters by employing the macro-invertebrate assessments. These investigations aimed to assess the temperature, pH, color, TDSs, total suspended solids (TSS), DO, nitrate (⁠⁠), phosphate (⁠⁠), ammonia (NH₃), chloride (Cl⁻) and EC. The temperature was summed directly on the site. However, others were analyzed inside the laboratory, in controlled conditions, so that the exact measurement was achieved. A HI series multimeter (HANNA Instruments) produced the necessary measurements of temperature, pH, and EC, ensuring that the readings were precise and repeated. The device, which was earlier assessed (Luvhimbi et al. 2022) for its accuracy in water quality assessments, was calibrated before each use according to the manufacturer's guidelines. The probe was fully submerged in the water sample until the readings stabilized, after which the values were recorded. Ensuring accuracy in water chemistry was kept calibrated according to the guidelines given by the manufacturer. The probe was plunged into the water, and it read the sample until the measurement values reached stable values, after which their readings were recorded. To ensure the integrity and prevent contamination of the probe, water was flushed through it whenever used with deionized water. The water color was also evaluated using the platinum-cobalt scale, which is a technique commonly used to determine the color intensity of water samples. A TDS meter operation, a method displayed previously by Dewangan et al. (2023) for its effectiveness in water quality assessment. Before use, the meter was carefully calibrated to ensure accurate readings. The probe was then immersed in the water sample, allowing the device to stabilize before recording the TDS value in milligrams per liter (mg/L). This process provided a precise measure of dissolved solid concentrations, offering valuable insight into the water's chemical composition. By following these standardized methodologies, the study ensured high-quality data collection, enabling a thorough and reliable evaluation of the selected water sources. Similarly, TSS were analyzed using the gravimetric method, a widely adopted technique in water quality assessment. Water quality control was employed for TDS measurements. The meter was perfectly adjusted prior to measurements in order to achieve accurate results. As a last step, the probe was immersed into the sample water; it entered the device weaker and then got a value of TDS in mg/L. After this process is complete, a precise assessment of the concentration of dissolved solids is obtained, helping to classify the sample water based on its dissolved solids content. By applying protocol procedures, the study procedure ensured that the measurements were of high quality, hence very helpful for thorough and reliable evaluation of the rest of the water sources. Equally, the technique for TSS analysis took gravimetric form, a broadly exercised technique in water quality control. Investigations reported by Wiyantoko et al. (2020) and Adjovu et al. (2023) demonstrated its competency in quantifying turbidity in the water bodies. This investigation used water filters to study the dynamics of suspended particles with water sampling. The dirty filter was alongside the trapped solid, dried in an oven with a fixed temperature of 103–105 °C until catalyst weight was reached. This process of cycling allowed determination of reusable particulate mass, which is the driving mechanism of bacteria mass. The filter was weighed again, after which the filter was weighed after drying, providing identification and quantification of the filter mass. On the other hand, the tare weight due to drying was recorded, and then the TSS concentration (mg/L) was computed.

A study showed the good performance of these devices in measuring DO fluctuations in the water bodies (Martin et al. 2012). The concept of getting optimum and genuine outputs was pertinently addressed by placing an electrochemical probe directly on the sampling sites to ensure constant contact with the water. The infiltrator was gently submerged to confirm the depth by the DO trend, which was used for the DO variation in water. To overcome all naturally dangerous effects, investigations were made between 8:00 AM and 11:00 AM, covering the time of active photosynthesis, respiration, and the change of temperature. This tool offered insights about oxygen diurnal dynamics in our water bodies and how it is shaped by both the environmental parameters and biological activity. In the same token, the nutrients were detected using the Gallery Plus discrete analyzer, the very sensitive device commonly applied in environmental research. In the field trials presented by Masindi et al. (2022) and Namatovu et al. (2023), the device was instrumental in the trace (correct determination of) concentration of water nutrients. According to this research, the analyser was used to specify nitrate (⁠⁠), phosphate (⁠⁠), ammonium (⁠⁠), and chloride (Cl⁻). The first step was the careful preparation of all reagents in a way that they provide the required nutrients, followed by calibration of the instrument using agreed benchmarks with a view to obtaining accuracy. First, the water samples are integrated into the sample tray, and an automated nutrient analysis program tests the samples, with reagents being dispensed, mixed, incubated, and absorbance measurement determined. The final concentrations of each nutrient were determined using a calibration curve, ensuring precision and consistency in the results.

In this context, we adopted a systematic sample technique for macro-invertebrate collection. The microhabitats in each of the water bodies were applied when these techniques were employed. Samples were taken with the use of kick nets of a cell size of 500 am and a method that included hand-picking to acquire a broad spectrum of taxa. The submitted specimens were mentioned to have been preserved in 70% ethanol to be set aside for detailed analysis done in the laboratory. The establishment of a plan in consonance with the guidance provided by the US Environmental Protection Agency and the American Public Health Association led to all the species and their representatives being sampled on different sites. At the end of the day, through the taxonomic identification protocol in the laboratory setting, the collected taxa of the sampled site were identified at the family or the lower level, and their relative percentages of composition across the sites were recorded.

Several studies, including studies by Jafar et al. (2023) and other authors, have supported the use of scales to classify WQI values in ensuring standardized assessment and complete information on a water source's overall quality. This is reflected in the integration of information from previous studies in order to establish the reliability and accuracy of WQI categorization in Table 4.

Water quality parameter analyses demonstrated that the levels of many physicochemical indicators showed seasonal differences depicting the functioning of the environmental and anthropogenic factors. The samples were analyzed for key pollution parameters, such as temperature, pH, color, TDS, TSS, DO, nitrates, phosphates, ammonia, chlorides, and EC, as shown in Table 5. Each of these parameters fluctuated in a seasonal manner due to reasons like the ecosystem's equilibrium between temperature, humidity, and environmental factors. To give an example, oscillations of DO levels with temperature variations confirm closely its dependance, while nutrient levels (nitrates and phosphates) peaked during specific months, most likely due to agricultural runoff. As seen in the leading facts, monitoring the water parameters and their dynamics is a central aspect of water quality management, which is specially treated in this research.

Conversely, sites like L8, L9, and L10 maintained relatively higher pH levels in both seasons, likely due to their storage-based nature, reducing immediate exposure to seasonal fluctuations. The overall trend highlights the role of hydrological and environmental factors in governing pH dynamics, where increased precipitation exacerbates the leaching of acidic compounds into water bodies, while in the dry season, lower water flow and evaporation contribute to maintaining slightly higher alkalinity. The presence of bicarbonates in certain locations helps buffer pH fluctuations, but the dominance of chloride and sulfate ions in areas with extensive human and animal activity may shift the balance toward acidity. While most sites remain within permissible drinking water limits, continuous monitoring is essential to mitigate potential risks associated with extreme seasonal pH variations.

Conversely, locations such as L10 and L5 exhibited comparatively stable temperatures across seasons, suggesting buffering effects from deeper water retention, shading from vegetation, and minimal exposure to rapid thermal fluctuations. Storage-based sources like L8 and L9 showed significant declines, likely due to evaporative cooling and limited thermal insulation, leading to more pronounced temperature reductions in enclosed environments. The presence of human activities, such as washing, brewing, and livestock interactions at locations like L1, L4, and L11, may also contribute to localized thermal variations, affecting heat absorption and dissipation rates. Temperature plays a critical role in regulating water quality, influencing DO levels, microbial activity, and the solubility of minerals. The recorded seasonal fluctuations, while remaining within acceptable limits, highlight the need for ongoing monitoring, as temperature variations can alter the biological and chemical stability of the water. The interplay of climate, land use, and hydrological processes underscores the complexity of seasonal water temperature dynamics, necessitating adaptive management strategies to mitigate potential long-term impacts on water suitability and ecosystem health.

Interestingly, sites such as L10 and L11 maintained relatively high TDS levels even in the wet season, suggesting a persistent input of dissolved solids from surrounding land use, including livestock rearing and domestic activities. The significant drop in TDS at L12, from 264.27 to 155.66 mg/L, indicates strong seasonal influences, where heavy rains flush out accumulated solutes, momentarily reducing mineral content. However, the relatively stable readings at L8 and L9, both tank storage systems, suggest limited seasonal influence due to minimal interaction with direct surface runoff. Conversely, wells like L6 and L7, positioned in swampy environments, recorded moderate declines, implying that subsurface flow and organic decomposition still contribute to maintaining higher mineral concentrations. TDS levels are crucial indicators of water quality, as excessive values may affect taste, palatability, and potential scaling in water distribution systems. While all recorded values remain within acceptable limits, seasonal shifts highlight the dynamic interaction between rainfall, land use, and water retention properties. The observed trends emphasize the necessity for continuous water quality monitoring, particularly in locations prone to agricultural runoff and human-induced contamination, to prevent long-term degradation of water resources and ensure sustainability.

With the onset of the wet season, EC values decreased across most locations, with the sharpest reductions observed at L4 (from 234.8 to 144.8 μS/cm) and L2 (from 210.2 to 151.7 μS/cm). This decline aligns with the expected dilution effect of heavy rainfall, which reduces ionic concentrations by flushing out accumulated salts and introducing fresher water into the system. Locations such as L6, L7, and L9 showed relatively moderate declines, suggesting that groundwater seepage, prolonged retention, or continuous exposure to dissolved solids maintained a higher mineral content despite seasonal shifts. Interestingly, L10 and L11 retained the highest EC values even in the wet season, likely due to persistent anthropogenic inputs from domestic wastewater, agricultural runoff, and livestock interactions, all of which contribute to increased ion loading. The WHO does not establish a strict limit for EC in drinking water but recommends that values below 1,000 μS/cm are generally acceptable for human consumption. The recorded values across all locations remain within this range, indicating no immediate concern regarding water salinity. However, elevated EC levels in locations such as L10 and L11, particularly during the dry season, suggest an increasing presence of dissolved solids that, if left unregulated, could lead to progressive mineralization, affecting taste and usability.

With the onset of the wet season, TSS levels increased across all sites, with the highest concentration recorded at L12 (181.86 mg/L), followed by L11 (160.90 mg/L) and L10 (141.84 mg/L). These locations, already characterized by high suspended solid loads, likely experienced intensified sediment transport due to heavy rainfall, runoff, and increased soil erosion from surrounding land use. Sites like L3, L6, and L7 also showed noticeable increases, reflecting the impact of rainfall-driven surface flow carrying organic matter and fine particulate matter into water bodies. Interestingly, some locations, such as L1 (106.71 mg/L) and L10 (141.84 mg/L), exhibited only marginal increases, suggesting that factors such as slower flow rates, deeper water retention, or buffering effects from vegetation may have played a role in stabilizing sediment resuspension. The WHO does not specify a strict guideline for TSS in drinking water but recommends that treated water should ideally have TSS levels below 50 mg/L to ensure clarity and minimize potential contamination. The recorded values at all locations far exceed this limit, indicating high turbidity and a strong likelihood of sediment-associated pollutants, such as heavy metals, pathogens, and organic matter. While moderate TSS levels may not pose immediate health risks, persistently high concentrations can degrade water quality, affect aquatic life, and reduce the efficiency of water treatment processes.

However, as the wet season arrived, color levels surged dramatically across all sites. Heavy rainfall triggered intense runoff, carrying organic debris, soil particles, and dissolved substances into water bodies, leading to sharp increases. L12 recorded the most extreme shift, soaring to 367.10 Pt-Co, followed by L10 (319.70 Pt-Co) and L11 (276.60 Pt-Co), indicating heavy pollution loads from agricultural runoff and livestock interactions. Similarly, locations such as L1 and L4 saw noticeable spikes, rising from 83.80 Pt-Co to 141.70 Pt-Co and from 72.50 Pt-Co to 132.90 Pt-Co, respectively, highlighting the impact of human activities like washing, brewing, and domestic wastewater discharge. Even traditionally lower-color sites like L6 (104.80 Pt-Co) and L5 (101.20 Pt-Co) experienced significant increases, further underscoring the widespread influence of seasonal changes. From a regulatory standpoint, the WHO sets a recommended limit of 15 Pt-Co units for drinking water, far below the levels observed in all sampled locations. While color itself is not necessarily a direct health risk, elevated values often indicate high organic matter content, potential microbial growth, and dissolved metals such as iron and manganese, which can affect taste, odor, and overall water quality. The stark seasonal contrast calls for urgent intervention, particularly during the wet season, through filtration, sedimentation, and proper water treatment processes to maintain safe and esthetically acceptable drinking water standards.

With the onset of the wet season, ammonia concentrations decreased across most locations, likely due to increased water flow, dilution, and enhanced microbial oxidation processes converting ammonia to other nitrogenous compounds. The most significant reductions were observed in L10 (from 3.50 to 0.69 mg/L) and L7 (from 2.92 to 0.899 mg/L), suggesting that heavy rainfall and surface runoff played a major role in flushing out accumulated ammonia. Despite the overall decline, certain locations, including L11 (2.16 mg/L) and L12 (2.45 mg/L), retained relatively higher ammonia levels even during the wet season, indicating continuous inputs from organic matter decomposition, domestic wastewater discharge, or livestock activities in the surrounding environment. Additionally, L8 (1.67 mg/L) exhibited a smaller reduction, possibly due to its storage-based nature, limiting the immediate impact of runoff dilution. According to the WHO, ammonia levels in drinking water should not exceed 1.5 mg/L, as excessive concentrations can result in undesirable taste, odor, and potential microbial contamination risks. While most locations exceeded this threshold during the dry season, ammonia concentrations in the wet season generally improved, with some sites falling within acceptable limits. However, the persistent high values at L11 and L12 indicate ongoing nitrogenous pollution, necessitating proper water treatment measures. Given ammonia's role as a precursor for nitrate formation, continuous monitoring is essential to prevent long-term water quality degradation and potential health risks associated with nitrogen contamination.

During the wet season, nitrate concentrations increased across all locations, with L12 peaking at 9.19 mg/L, followed closely by L10 (8.92 mg/L) and L11 (8.66 mg/L). While still within the ‘Good’ range, these elevated values point to increased surface runoff, which mobilizes nitrogen-rich fertilizers, animal waste, and decomposing organic material into nearby water bodies. Locations like L2 (6.77 mg/L) and L3 (6.00 mg/L) also showed noticeable rises, emphasizing the influence of agricultural activities and soil leaching during rainfall events. Although all recorded values remained well below the WHO maximum limit of 50 mg/L for safe drinking water, the observed increase during the wet season reflects heightened nutrient loading, which, if sustained, could pose environmental risks such as algal blooms and ecosystem degradation.

In the wet season, phosphate concentrations increased significantly across most sites, with L12 peaking at 5.27 mg/L, followed by L10 (4.58 mg/L) and L11 (4.35 mg/L). This sharp rise is likely driven by rainfall-induced surface runoff, which carries fertilizers, organic waste, and soil particles into nearby water bodies. Locations such as L4 (1.82 mg/L), L6 (1.72 mg/L), and L7 (1.83 mg/L) also experienced considerable increases, indicating nutrient leaching from surrounding land use practices. Even previously low-phosphate locations like L2 (1.21 mg/L) and L3 (1.44 mg/L) saw noticeable rises, emphasizing the widespread influence of wet-season runoff in mobilizing phosphates across different landscapes. According to the WHO, phosphate levels above 0.5 mg/L in surface waters can accelerate eutrophication, promoting excessive algal growth and depleting oxygen levels, which threatens aquatic ecosystems. The recorded values in both seasons far exceed this threshold, signaling potential risks of algal blooms, reduced water clarity, and habitat degradation. The pronounced increase during the wet season underscores the need for improved nutrient management strategies, particularly in areas prone to agricultural runoff and domestic waste discharge, to mitigate the long-term impacts of phosphorus pollution on water quality and ecosystem health.

In contrast, sites like L7 (134.76 mg/L) and L8 (137.88 mg/L) exhibited comparatively lower chloride levels, likely benefiting from improved natural filtration or reduced exposure to contamination sources. With the arrival of the wet season, chloride concentrations experienced a general decline, a clear indication of rainfall's dilution effect. Notably, L12 dropped from 202.40 to 130.80 mg/L, while L11 decreased from 197.45 to 155.54 mg/L. This reduction suggests that heavy rainfall effectively flushed accumulated salts from these systems. Despite this overall decline, certain locations, particularly L10 and L9, maintained relatively higher chloride levels, suggesting that persistent contamination sources or slower drainage processes allowed salts to remain concentrated. From a regulatory perspective, all recorded values remained within the WHO-recommended limit of 250 mg/L, ensuring no immediate concerns for drinking water safety. However, the consistently elevated values in sites like L10 and L12 highlight potential long-term risks, particularly in regions prone to salinity buildup.

During the wet season, DO concentrations increased across all locations, with values ranging from 4.24 mg/L (L4) to 7.04 mg/L (L1). The most prominent increase was recorded at L1, rising from 6.237 to 7.04 mg/L, indicating enhanced oxygenation due to rainfall-induced turbulence, improved mixing, and increased flow rates. Sites like L6 (6.40 mg/L) and L8 (6.49 mg/L) also experienced noticeable improvements, reinforcing the positive influence of precipitation in refreshing oxygen levels. However, locations such as L2 (4.78 mg/L), L3 (4.42 mg/L), and L11 (4.34 mg/L) maintained relatively low DO values even during the wet season, suggesting persistent oxygen depletion linked to organic matter accumulation, microbial respiration, or limited water exchange. According to environmental water quality standards, DO levels above 6 mg/L are generally considered supportive of aquatic life, while values below this threshold may stress aquatic organisms. Although some sites like L5, L3, and L4 remained below this benchmark, the overall seasonal increase in DO suggests improved water quality conditions during the wet season.

During the dry season, WQI values in several locations indicated water quality concerns. Locations such as L1 (192.58), L2 (206.39), and L3 (207.85), which are associated with swampy environments and significant human activity, exhibited moderately high WQI values, suggesting water quality concerns. These sites are susceptible to contamination from agricultural runoff, localized pollution sources such as animal waste, and domestic practices like washing and brewing activities. Sites like L4 to L9 showed consistent WQI values between 203.60 and 229.78, placing them in the ‘Poor’ category, indicating substantial contamination. The presence of irrigation activities, washing sites, and exposed water storage systems at these locations likely contributed to increased nutrient loading, fostering microbial growth and organic contamination. Notably, L8 (204.63) and L9 (203.60), both tank storage sites, exhibited elevated WQI values during the dry season, which can be attributed to poor maintenance, exposure to atmospheric contaminants, and biofilm development. The highest WQI values were observed at sites L10 (318.20), L11 (324.32), and L12 (379.22), particularly during the dry season, where values significantly exceeded the safe threshold. These sites, mainly dams serving livestock and domestic activities, are prone to nutrient enrichment from agricultural runoff, leading to increased algal blooms and microbial proliferation. In the wet season, WQI values generally followed a similar trend; however, sites such as L10 (370.83), L11 (374.76), and L12 (407.99) showed further elevation, suggesting intensified contamination due to surface runoff. Conversely, some sites, such as L2 (185.00) and L8 (179.42), exhibited slight improvements in water quality during the wet season, potentially due to dilution effects and enhanced water flow.

Furthermore, DO levels below the ideal threshold at sites L2–L5 exacerbated water quality concerns, reinforcing the impact of organic pollution and reduced oxygenation during stagnant dry periods. Elevated ammonia concentrations may have further contributed to these trends. The persistent observation of ‘Poor’ to ‘Very Poor’ water quality at critical sites underscores the urgent need for targeted interventions. Particular attention should be given to locations L10–L12, where consistently elevated WQI values indicate severe contamination risks that require immediate remediation measures, such as enhanced wastewater management and community awareness initiatives. The overall pattern suggests that sites burdened by inadequate waste management, stagnant water conditions, and heightened anthropogenic pressure are particularly susceptible to seasonal water quality deterioration. This highlights the importance of adopting a comprehensive approach that integrates physicochemical parameters to unravel the complex interplay between seasonal variations, pollutant sources, and hydrological dynamics. To effectively mitigate these risks, strategic actions such as improved maintenance of water storage systems, regulated agricultural runoff, and proactive community involvement are crucial for ensuring sustainable water quality improvements across the region.

The WQI values obtained for both dry and wet seasons demonstrate significant variations across sampling locations. Sites L10, L11, and L12 recorded exceptionally high WQI values in both seasons, indicating poor water quality. This aligns with their environmental characteristics, which include higher anthropogenic activities and possible contamination sources. Conversely, sites L1–L5 maintained comparatively lower WQI values, suggesting better water quality conditions. Seasonal differences were notable, with higher WQI values observed in the wet season, likely due to increased runoff carrying pollutants into water bodies.

In addition to these sediment observations, the macro-invertebrate analysis provides the same information. Table 6 illustrates the number of macro-invertebrate families differentiated at the selected places, reporting the dominance of Culicidae and Drosophilidae through their presence at the sites with a higher WQI value, for instance, at L12, L11, and L10. Such families live in waters where conditions of oxygen deprivation and pollution are common, as they are capable of thriving under these circumstances. On the other end of this spectrum are the dull waters with good qualities, like L4 to L1, where one is able to see an increased diversity of the pollution-intolerant families, like body and L3 and L4 stood out among the sites by the even greater increase in diversity, which signalied the stable and less contaminated water conditions favorable to developing big variety compositions of macro-invertebrates. Closure on the areas L6, L7, L8, and L9 for the macro-invertebrate analysis was a well-planned move as the environmental factors of these sites made any biological evaluation unfeasible. These regions featured rocky and pebbly substrates as well as a few aquatic plants, leading to large fluctuations in the river current, conditions that prevented a high variety of macro-invertebrates from being established. As a result, the absence of contrasting sites maximized the resolution and hence allowed a focus on ecologically consistent macro-invertebrate collection locations. The physicochemical assessments and biological data chosen on the basis of this indicator give a detailed picture of the water quality status for the sampling zones. This, in essence, acts as a sign, which shows that pollution-tolerant species and WQI scores are related and calls for urgent targeted action where conditions are severe in order to bring about any changes to the currently prevalent water quality.

The key role of the variance inflation factor (VIFs) is making the variables less than 3, pointing to the absence of multicollinearity. This is achieved with the linear constraint of the encoded space, as this would ensure the independence of each predictor, making the model more reliable and accurate. The implementation of the two aforementioned techniques (i.e., dominant and subtle contributors), however, returns the most comprehensive information about the water quality assessment. This model is crucial in soil and water quality investigation, establishment of ecosystem health, and planning solid water management strategies.

Finally, the decision-makers, environmental scientists, and public health professionals are the last on the list. They are getting practical value from the research that is done, from the data, which can lead them to take safety measures for the aquatic ecosystem and to have the best clean water resources for generations.

This study assessed the characterization of water mainly in the selected sources in Kyeizooba Sub-county, Bushenyi District, Western Uganda. The outcomes obtained have indicated that contamination is the major threat in the water that creates a serious concern for both public health and environmental management. The final results of this study showed that the sites at L10, L11, and L12 controlled by different parameters consistently had the poorest water quality and the highest WQI values, which represented a polluted environment and unsafe conditions. The excess quantities of ammonia, nitrate, and phosphate in the water above the WHO limits point to the interplay of agriculture, livestock exploitation, and domestic waste disposal. These pollutants were most severe during the wet season, where intensified runoff increased the inflow of organic matter and soil sediments that, in turn, led to increased TSS and turbidity. In some cases, the situation favors the growth of microbial life and also includes DO reduction and disruption of the ecological balance. Such conditions make aqua life unsafe and also make the water unsuitable for human consumption.

To support the credibility of the results and to ensure an appropriate data interpretation, an MLR model was employed. The model was constructed to test the results of the WQI analysis and confirm a possible vital contributor to pollution or contamination. This model identified ammonia, nitrate, and phosphate as the key pollutants, as they significantly contribute to water quality deterioration. Appropriate location of preventive measures reduces the vulnerability zones and serves as the core of the response strategies.

To address these issues and ensure the long-term sustainability of aquatic ecosystems, the following recommendations are proposed: implement the individualized level of care system for highly polluted places such as L10, L11, and L12, which employs multi-stage filtration units, activated carbon filters, UV disinfection systems, chlorination systems, and target microbial pathogens and chemical pollutants in order to help the environment. There is also a need to provide inexpensive water-filtering gadgets that can be used at the household level. Enlarging areas next to the water bodies, applying controlled approaches to fertilizing land, and developing integrated farming ways will foresee a decrease in the amount of nutrients immigrating from the soil. The educational program will teach sustainable methods that prevent the soil from turning into the waterways by lowering the erosion level and the amount of chemicals added inadvertently. Operation and maintenance of wells, dams, and water tanks can be improved through tasks such as closing open systems, improving the drainage system, and also placing protective lids.

The global measures of preventing water contamination may be considered as part of the development of early warning systems for water body contamination through the use of artificial intelligence (AI) monitoring techniques and internet-connected sensors that are cheap to operate. Practically, low-cost IoT sensors, like turbidity and pH sensors, can be installed at critical points (e.g., L10, L11, and L12) for water quality indicators. These sensors will connect to AI-based platforms that are capable of real-time data analysis and detecting any abnormal situation like angle holes or sudden spikes of ammonia and nitrates. These platforms will direct the alerts to the authorities concerned for quick action. Furthermore, solar-powered, GSM-connected IoT gadgets can be placed in far-off places, to do surveillance without using the electric grid. Using mobile applications that are built by the community and make it the main feature, local residents will have a chance to notify each other about visible contamination, a need to repair the infrastructure, or the sources of the pollution. It is a delivery method that involves the local people through manual questionnaires, and site visits and creates real-time data collection. With this definition of inclusivity, effectiveness, and sustainability in water quality management can be achieved. The intermixing of such technologies as well as long-term screening methods and a community role comprises a centered, flexible, and interactive group of strategies. The interaction between authorities and the facility with accurate detection and prompt response to the introduction of negative factors into the system is the main element contributing to ensuring long-term water quality improvement. In addition to this, the management bases the decision on water quality standards compliance and the monitoring of important parameters such as TSS, ammonia, nitrates, and microbial content among others. To add to the technological solutions, outreach campaigns about safe water practices, promoting personal hygiene, and skills transfer on best waste management through community involvement can create a conducive environment and increase the participation of residents in safe water practices. Reimposing wetlands that act as a natural filter of pollutants may notably help to clean water, therefore a single individual call should be made locally in conservation activities. Gradually introducing garbage gathering places an approximate distance from the sensitively located resources, rotting of waste products, and a refined way for discharging already collected animal waste from the water bodies will have a ‘clean water’ effect. The environment should be policed using bright and healthy pollution codes, while the source of pollution such as agriculture, livestock, and domestic waste should be guarded by enforcing penalties for noncompliance and the incentives for those who comply. The major activities to be undertaken by the authorities, along with the NGOs, research organizations, and government bodies, include involving community members in water management programs through the provision of technical expertise, funds, and resources for sustainable solutions. The reason for using such an integrating model is the ability to get rid of the water quality problems, to protect the environment, and to get safe and sustainable water sources for the community of Kyeizooba sub-county. In implementing such a holistic approach, a major part of the improvement of community health and environment, agriculture sustainability, and ecological conservation will be played by the overall strategy.

All authors contributed significantly to the study's conception and design. K.C. performed the laboratory work and experimental processes, while P.P.N. led the conceptualization, writing, review, and editing of the manuscript, along with conducting the analysis, including the development of the MLR model and preparation of all related graphs. The supervisory roles were undertaken by K.S.K., C.S., and J.Y. P.P.N. also authored the first draft of the manuscript, with all authors providing valuable feedback on subsequent versions. The final manuscript was reviewed and approved by all authors.

The authors declare that no funds, grants, or other support were received during the preparation of this manuscript.

All relevant data are included in the paper or its Supplementary Information.

The authors declare there is no conflict.