Assessment of heavy metal and other chemical pollution in Lake Tana along urban peripheries, Ethiopia Open Access

Lake water is an essential natural resource because it is the second-largest source of fresh water next to ice caps and groundwater (Gleick 1993). Lakes provide many social and economic benefits to the society. They have widely been used as a source of water for people, livestock, irrigation, industries, transportation, recreation, and others (Keeler et al. 2015; Zhang et al. 2018). Ecologically, lakes provide habitats for many species of plants and animals (Lamsal et al. 2015). Despite these many benefits, in developing countries, lakes are sinks for anthropogenic sources of pollution. These anthropogenic activities raise the levels of hazardous chemicals, trace metals, non-degradable organic pollutants, and nutrients in lakes (Githaiga et al. 2021). This results in eutrophication and poor water quality in many lakes (Dersseh et al. 2019), which significantly impact the ecological values and social and economic benefits of these lakes.

Lake Tana is the largest freshwater body in Ethiopia and represents 50% of the total inland waters of the country. It supports up to 67 different fish species of which 70% are endemic (UNESCO 2014). The status of lake nutrient water quality has been reported as poor in rural settings (Dersseh et al. 2019; Goshu et al. 2023). The lake receives chemical runoff from agricultural lands and suspended sediment from cultivated land. These represent 63% of the land use in the basin draining into the lake (Song et al. 2018). Nitrogen and phosphorus loads exported from the rural watersheds have increased nutrient concentrations respectively, in the lake to more than 2.4 mg/l TN and 0.18 mg/l TP concentrations and have resulted in the expansion of water hyacinth in the lake (Sishu et al. 2021; Derseh et al. 2022). The average suspended sediment concentration in the lake is 300 mg/l (Womber et al. 2021).

The lake also receives discharges from 17 urban catchments in the basin (Song et al. 2018), all of which are untreated (Wondie 2009). Lake water pollution in urban settings was assessed by Tibebe et al. (2019) and Goshu et al. (2010). This research focussed on physicochemical parameters in the waters of the Gulf of Bahir Dar. The observed nutrient concentration levels were comparable to the lake water quality in the sections of the lake receiving discharges from rural land use.

Anthropogenic fecal pollution was also assessed by Goshu et al. (2021) along the points of discharge from main rivers into the lake. The sanitation index of the city of Bahir Dar shows that over 46% of the population managed fecal sludge unsafely while 12% of the population relied on defecation in the open (Jumper et al. 2020). Despite heavy metals (HMs) being carcinogenic, mutagenic, and teratogenic and the potential for HMs to bio-accumulate in food chains (Yang et al. 2017; Njuguna et al. 2020), none of the studies investigated HMs in Lake Tana.

A comprehensive assessment of water quality is essential to improve management practices and to inform the development of strategies aimed at protecting water quality and safeguarding human health. Thus, the objective of this study was twofold:

• To establish a baseline of HM concentrations in Lake Tana and

• To assess the lake water quality at sites receiving urban discharges for safe and sustainable use of resources and to identify the main source of pollutants. A Water Quality Index (WQI) was used to assess the levels of pollution, and geospatial analysis, hierarchical cluster, and principal component analyses were often used to determine the probable sources of pollutants (Githaiga et al. 2021; Mekuria et al. 2021; Ustaoğlu et al. 2021).

The lake receives discharges from four (4) major rivers and more than 40 small streams. About 93% inflow water volume is from the major tributaries of Gilgel Abay, Gumara, Ribb, and Megech (Kebede et al. 2006). The annual estimated inflow of water is around 10,300 GL. The outflow from the lake is 3,700 GL, and around 6,600 GL is lost in evaporation (on average around 2.14 m of evaporation annually) (Kebede et al. 2006; Ligdi et al. 2010). The lake water level fluctuates between 1,785 and 1,787.5 (m ASL), i.e. over a 2.5 m range.

The land use and cover compiled by Song et al. (2018) indicated that the lake covers 20% of the catchment draining to the lake outlet which is about 15,000 km². The major land use is agriculture occupying 63% of the catchment. Urban land use accounts for almost 1% of the total catchment area. Three large cities and 14 small towns are currently located within the catchment. The large cities are Gonder Bahir Dar and Debre Tabor (see Figure 1(b)).

Bahir Dar city is located at the southern end of Lake Tana where this study was focused (see Figure 1(c)). The city is the second largest in the lake catchment. It is a center for tourism, commerce and industry as well as providing housing for a rapidly growing population. The city's population grew over the past decade by 2.4 times and is currently estimated to be 500,000 persons according to Ethiopian Central Statistical Agency (CSA 2021). Since 1995, the change in land cover indicates that the built-up areas increased by 250%, the paved areas by 40% while the area of wetlands reduced by 10% (Kindu et al. 2020).

Water samples were collected in 3 months during 2022 including the dry period (March) when the lake level was lowest, in the early rainy period (July), and at the end of the rainy period (September) when the water level reached its peak. Water sample collection points were located mainly at major drainage channel outfalls and in areas with high human activities such as hotels and lounges. Major drainage channels currently discharge into the lake at six locations as shown in Figure 1(c).

These channels are as follows:

• (1) Saint Gorge's church (SG): Drains wastewater discharged by hotels, and from pit latrines and runoff from open areas subject to defecation;

• (2) Kuriftu (KU): Drains wastewater discharged from lodges and runoff from markets;

• (3) Depo (DE): Drains domestic wastewater discharges and runoff;

• (4) Referral Hospital (RH): Drains hospital wastewater discharges and runoff;

• (5) SOS Scholl (SS): Drains runoff from residential areas; and

• (6) Airport (AP): Drains runoff from residential areas and cultivated lands during the rainy season.

Additional monitoring sites included:

• (7) Shumabo (SU): Drains wastewater discharges from the Shumabo recreation center. This site was a previously monitored location near a lounge (Wondie 2009; Tibebe et al. 2019)

• (8) and (9) Reference points: Located 2.1 and 2.7 km away from urban-impacted sites.

A triplicate sample was collected for analysis at each location. Water samples were collected in 500-mL plastic water bottles and transported to the laboratory at 4°C by standard procedures (Baird et al. 2017).

Following parameters were recorded in situ using a Palintest micro 800COND/TDS/pH multi-meter: temperature, pH, total dissolved solids (TDS), and electrical conductivity (EC). Dissolved oxygen (DO) was measured using a HANNA HI9145 probe. Parameters that were analyzed in the laboratory included ⁠, ⁠, P (dissolved), ⁠, Fe, Cr, Zn, Mn, Al, and Cu. These parameters were analyzed based on the standard method of the American Public Health Association (Baird et al. 2017) and were determined using a 7100 photometer. The details of the method used for the analysis are described in the Palintest Operation Manual (Palin 1960).

The WQI is the most widely used technique to evaluate water quality status (Kumar et al. 2019; Njuguna et al. 2020). It transforms water quality parameters and standard parameters into a single value that represents the status of water quality using a mathematical model and therefore makes complex data more understandable. Several WQIs have been developed since the initial development of the index by Horton (1965). Those indices include the National Sanitation Foundation Water Quality Index (NSFWQI), the Canadian Council of Ministers of Environment Water Quality Index (CCMEWQI), the Oregon Water Quality Index (OWQI), and the Weight Arithmetic Water Quality Index (WAWQI) (Chidiac et al. 2023).

This study adopted the WAWQI model which has been widely used. The index was computed in three steps. The first step is parameter selection. The number of parameters adopted for models typically varies between 4 and 26 and most models use 8–11 parameters (Chidiac et al. 2023). Based on their relative public health concerns, nine parameters were adopted for this study including temperature, pH, N-NO₃, Cr, Zn, Cu, Al, Fe, SO₄, and TDS. The second step is weighting the selected parameters. There are two ways of weighting parameters as follows.

• (1)

  The weight is assigned to the selected parameters based on expert opinion on a scale of 1–5 considering the risk posed (Horton 1965; Brown et al. 1970; Njuguna et al. 2020) or

• (2)

  The weight is calculated from the recommended standard guideline values (Nerae 2020; Ustaoğlu et al. 2021; Chidiac et al. 2023). This method was adopted for this study.

The third step is the aggregation of sub-indices into a single WQI score and its evaluation against a rating scale that categorizes water quality. The rating scale categories adopted for this study are as follows: WQI < 26, excellent; 26–50, good; 51–75, moderate; 76–100, poor; and >100, unsuitable.

The water quality was categorized into four pollution levels based on the NPI. NPI values: <1 (no pollution), 1 ≤ 3 (moderately polluted), 3 ≤ 6 (considerably polluted), and >6 (very highly polluted) (Ustaoğlu et al. 2021).

where C_w = mean concentration of the HMs in the water samples (μg/l) collected over the study period at specified sampling site; DWC is the daily water consumption, which is 2 L d⁻¹ for adults, and 1 Ld⁻¹ for children according to WHO (2017). The adults' body weight (BW) was 70 kg, and 15 kg for children; EF = exposure frequency (days/year), which is 350 days per year; ED = duration of exposure and 70 years considered for adults, and 6 years for children; AT = average exposure time (365 days/year × 70 years for adults, and 6 years for children; RfD (mg/kg/d) is the oral reference doses and for each HM; Zn = 0.3, Cr = 0.003, Al = 1 Fe = 0.007, and Cu = 0.006. The RfD data were collected from the ECOTOX database of the USEPA (2019),

A one-way ANOVA test was employed to evaluate the temporal differences in water quality parameters during the sampling seasons. Hierarchical cluster analysis (HCA) was conducted to determine the relationship and similarities among the sampling points and grouping based on water samples' physicochemical parameters to understand pollution sources. Principal component analysis (PCA) was employed to assess and identify significant components that explain variations in water quality and source. The analyses were computed using R programming for Windows, version 4.2 at a significant level of 0.05.

A GIS was used to map the spatial distribution of water pollution status. Several models have been used to map the spatial distribution of physicochemical parameters in lakes and reservoirs. These include Inverse Distance Weighted (IDW), Kriging, and Spline (Krige 1951; Wang & Wang 2012). The IDW is simple and suitable when analyzing data with no outliers and was adopted in this study. The method is available in the Geo-Statistical Analyst Tool. The semi-variogram spherical model (Goovaerts 1999) had the smallest root mean square and was used in the analysis. The parameters that were mapped included nutrients and HMs.

The historical water quality trend was evaluated using the available dataset from the literature and unpublished secondary data. The available water quality parameters readings included EC, TDS, N-NO_3, and DP which have been monitored for almost three decades since 1995.

The lake water temperature ranged from 22 to 26 °C. The highest temperature was recorded in March, and the difference was significant compared to the lowest measurement in September (refer to Table 1). The pH is an essential factor in controlling biogeochemical processes in the lake ecosystem (Matta et al. 2017). The pH of the lake water sample was in the alkaline range. It was in a safe range for the aquatic environment and drinking (WHO 2017; USEPA 2022) with a pH > 8.5, except for the extreme measurements at sites near hospital wastewater discharges. Seasonally, the pH was highest in July. The recorded EC and TDS values ranged, respectively, between 85 and 200 μS/cm and 46 and 109 mg/l. The values indicated that salinity was lower than the guideline limit (TDS < 1,000 mg/l) given by the US Environmental Protection Agency (USEPA 2022). The in situ measurements recorded during this study are similar to the values recorded by Goshu et al. (2010), Tibebe et al. (2019), and Dersseh et al. (2019).

The measured DO levels ranged between 2.1 and 6.4 mg/l. The recorded DO levels are close to the minimum required level of 2 mg/l (USEPA 2022). High DO concentrations are attributed to low organic inputs from the surroundings, low turbidity and low suspended solids, photosynthetic activity of the green plants, and low temperature (Matta et al. 2017). The spatiotemporal trend in the DO levels indicated that DO concentrations were seasonally lower in March which is attributed to high temperature (Table 1). The DO levels in this study are consistent with the DO readings reported by Tibebe et al. (2019) (DO = 3.1–5.7 mg/l) and Goshu et al. (2010) (DO = 3.3–4.2 mg/l).

Urine from open defecation, pit latrines, and other domestic waste is a source of high nitrates that increase nitrate levels in the lake (Chua et al. 2022). Seasonally, the nitrate concentration is maximum in September when groundwater is discharged into the lake, and notwithstanding the lake water reaches its highest level in September (Ligdi et al. 2010). The nitrate concentrations observed in this study were higher than those reported by Tilahun & Ahlgren (2010) (0.0025– 0.003 mg/l in Lake Hawassa) and lower than those reported by Fetahi et al. 2014 (0.042 mg/l in Lake Hayq).

Ammonia nitrogen (N-NH₄) is a nitrogenous compound that can pollute fresh water. The ammonia concentration was below detection, except in one sample collected at the SG point area near the pit latrines.

It was hard to detect ammonia in the aqua phase in the lake because of the high water pH > 8. The ionic (NH₄⁺) form of ammonia exists only in acidic to neutral water (Reddy et al. 1984). Ligdi et al. (2010) and other unpublished data from the Ethiopian Ministry of Water Resource MoWR (2005) reported no ammonia detections in Lake Tana, the same as our observation, while Goshu et al. (2010) have reported the detection of ammonia.

The dissolved phosphorus (DP) concentrations ranged between 0.03 and 0.9 mg/l. All samples had DP values above the allowable limit of (p = 0.01 mg/l) given in the USEPA (2000) guideline. Concentrations above the limit are likely to lead to blue-green algal blooms. Spatially, the DP concentration was high in the site receiving agricultural runoff at the location AP in the western part of the lake. This implies that P sourced from fertilizer application is a more significant contributor than urban sources. Seasonally, the concentrations were significantly higher in July when fertilizer was applied to agricultural land, and early rainy season rainfall and runoff conveyed higher DP loads to the lake (refer to Table 1). Compared to other findings, the DP concentrations in the current study were higher than P concentrations in Ethiopian lakes such as Lake Hawasa (0.015 mg/l) and Hayq (0.002 mg/l) as reported by Tilahun & Ahlgren (2010) and Fetahi et al. (2014) respectively.

HMs pose environmental and public health concerns due to their toxicity, persistence, potential for bioaccumulation, and bio-magnification properties. HM levels in the study area are listed according to their average concentration values as follows: Cu (0.4 mg/l) > Fe (0.3 mg/l) > Cr (0.1 mg/l) > Al (0.03 mg/l) > Zn (0.01 mg/l) >Mn (not detected). Concentrations of Cr and Fe exceeded the criteria for drinking water and may pose health risks.

Compared to previous studies, the concentrations of HMs measured in this study were lower than the recorded concentrations in Lake Koka which receives wastewater discharges from a tannery. Lake Koka had a high concentration of Cr (6.5 mg/l), as reported by Tessema et al. (2020). The recorded Cr concentration in Lake Tana is similar to values reported for lakes in the Rift Valley of Ethiopia (0.10 and 0.12 mg/l) (Zinabu & Pearce 2003) but higher than the concentrations (0.0003–0.006 mg/l) measured in Lake Naivasha (Yang et al. 2017). The recorded concentrations of Cu are similar to the concentrations reported in lakes in Rift Valley where the source is from herbicide application (Zinabu & Pearce 2003).

The estimated mean NPI values varied from 3.6 to 6.5. The high value noted in July is attributed to fertilizer application on the agricultural land which is typically carried out from the middle of June to July (Sishu et al. 2023). Additionally, early rainy season runoff is also likely to have high phosphate concentrations (see Table 1). Overall, the calculated NPI of the lake was >3.0 in all seasons. It indicates that throughout the year, the lake is too enriched and is at high risk of eutrophication. Around the Gulf of Bahir Dar, duckweed and algae have already started to grow, as seen in Figure 2, and more than 50 ha of water hyacinth has bloomed in the northeast (Dersseh et al. 2019).

The effect of HMs (Cr, Cu, Fe, Zn, Al, Mn) on lake water quality was determined using HEI. The average HEI values were 3.0 (March), 2.8 (July), and 3.6 (September) (see Figure 7). In all seasons, the HEI values show low pollution levels (<10). However, the HEI values in the Gulf of Bahir Dar showed significant contamination compared with the previous studies. The reported HEI for six lakes in Kenya (Githaiga et al. 2021) was lower than the HEI observed in this study. The HEI observed in this study was lower than the HMs global index reported by Kumar et al. (2019).

The PCA, including retained components, their loadings, eigenvalues, percent variance, and the cumulative percentage and Pearson's correlation coefficient (PCC) are presented in Tables 3 and 4. The number of PCs to be retained was selected based on eigenvalues and scree plots (Figure 8(b)). Accordingly, the principal component with eigenvalue (>1) was retained.

The extracted eigenvectors of PCA of the water samples revealed that the first four principal components covered a 90.49% variance of the total data set. The variance explained by the principal component (PC1) is 39.25% as shown in Table 3; 25.71% of the variation was contributed by PC2; 13.06% of the variation was contributed by PC3, and 12.47% variation was contributed by PC4 with an eigenvalue of 4.32, 2.83, 1.44, and 1.37, respectively. The positively loaded parameter influences water quality while the negatively loaded parameter is not (Mekuria et al. 2021).

Therefore, PC1 showed a positive loading on pH, TDS, and EC. Pearson correlation also indicates that pH, EC, and TDS showed a significant positive relationship with an R-value of greater than 0.65 (Table 5). This result can be interpreted as dissolved solids originating from sources such as untreated domestic wastewater, and urban runoff (Huang et al. 2013). PC2 has positive loading on Cr and Cu. A significant positive relationship exists between Cu and Cr as also shown by Pearson correlation and (R = 0.85). Painting and some household detergents and cleaning products contain Cr compounds. Copper compounds are used as an algaecide in swimming pools that directly discharge into the lake water. PC3 indicates positive loading of N-NO₃ and Zn but negative loading on P. In urban wastewater, zinc can be found in household cleaning products, and personal care items, and nitrate can leach from septic tanks. P sources are from agricultural runoff and the resuspension of bed sediment (Kebedew et al. 2020). PC4 shows positive Al and temperature loading. It implies increasing temperature can cause release of aluminum from sediments and suspended particles into the water column (Koshikawa et al. 2002).

The risk quotients of Zn and Al were found to be less than 1 indicating that the exposed population can experience no significant health risk (Table 5). For adults, Fe and Cu HQs were less than 1 in the three sample points. While Fe showed significant risk in all sample stations for a child population. The highest Fe HQ was registered in the area where wastewater was discharged from the RH site to the lake. Some pharmaceuticals such as iron supplements and chemotherapy drugs and medical devices such as surgical instruments and implants contain high levels of Fe. When they are disposed of or cleaned, they can release iron into clinical wastewater (Meo et al. 2014). Cr was found highly risky both for adults and children and in all sample sits. It indicates that the exposed population can experience significant health risks. A high value of HQ of chrome was also reported in the river, Nigeria (Anyanwu & Nwachukwu 2020). Greater risks of Cr were also reported in the Tai Lake, china (Liu et al. 2014).

Figure 8 shows historical trends of selected water quality parameters in the Gulf of Bahir Dar in Lake Tana. Additional data reported in Figure 8 were compiled from unpublished data collected by the Ethiopian Electric Light and Power Authority (EELPA 1996), the Ministry of Water Resources (MoWR 2005), the Abay Basin Authority (Haimanot 2015; ABA 2016), and from Goshu et al. (2010) and Tibebe et al. (2019).

Compared with the nutrient loading trend in freshwater bodies in developed countries including North America and Europe, the trend in Lake Tana is the opposite. For example, in the US, the nitrogen and phosphorus loadings from urban watersheds have generally decreased (Oelsner & Stets 2019). Similarly, nitrogen and phosphorus loadings in the EU have decreased by 14% for N and 20% for P (Grizzetti et al. 2021). In contrast, increasing nutrient loadings have been reported in developing countries (Fink et al. 2018). In almost all lakes in the Rift Valley of Ethiopia, nitrate and phosphorus loadings have increased by 6 and 10 times over the last two decades (Merga et al. 2020). The implementation of waste management practices, urban runoff controls, and improved agricultural practices that can improve water quality is urgently needed.

This study has evaluated the physicochemical water quality and the HM concentrations in the waters of the Gulf of Bahir Dar in Lake Tana in Ethiopia. The results have revealed that nitrogen and phosphorus concentrations exceed aquatic water quality criteria as recommended in USEPA guidelines. Concentrations of HMs such as Cr and Fe also exceed the guideline limit. Both nutrients and HM concentrations vary spatially and temporally in the waters of the Gulf of Bahir Dar. The concentrations of Zn, Cr, and Al and nutrients all increase in the rainy season which is attributed to uncontrolled urban runoff and discharges from agricultural lands. Increases in Fe and Cu concentrations in the dry season are attributed to a combination of the release of Fe from the bed sediment due to the redox process as well as point source sewer discharges. Spatial comparisons of concentration across sample locations against reference locations as well as GIS spatial analysis, PCA, and HCA confirmed uncontrolled urban runoff, wastewater discharges, and runoff from agricultural lands are pollution sources. The WQI indicates that the current lake water quality status is moderate, while the NPI reaches a highly polluted level. In the past years, since 2010, both DP and N-NO₃ have been exponentially increasing in lake water. The HEI shows low pollution levels in all seasons. Also, it was found that the HEI values in the Gulf of Bahir Dar showed significant contamination compared with previous studies of tropical lakes. It is concluded that the implementation of waste management practices, urban runoff controls, and improved agricultural practices are urgently needed to improve the water quality and to counter the alarming growth and expansion of phytoplankton and water hyacinth.

The authors want to thank the Bahir Dar Institute of Technology, Faculty of Civil and Water Resource Engineering for providing the laboratory facility that supported this research.

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

The authors declare there is no conflict.