Water quality in earthen dams and potential health impacts: case of Nadosoito Dam, Tanzania

Nadosoito dam in Tanzania is faced with water quality issues that change through the year. It has been said that water from the dam is unpalatable (sour), especially during the dry season, and also has an odour (personal communication). Over 6,000 inhabitants of Arkatani and others from the surrounding villages depend on the dam for domestic supplies and watering animals (over 10,000 cattle, goats, sheep and donkeys), making it very important. It is reported that there are numbers of livestock deaths at some periods of the year. High turbidity harbours microorganisms including pathogens, thus posing a risk to human health (Senay et al. 2002). Despite the water's characteristic turbidity, the dam is shared by humans, livestock and wild animals, making it a potential source of zoonotic diseases. This also has the potential to add nutrients and organic matter when the animals drink from the dam.

Villagers report that the water situation in the area is terrible and worsens during dry season. Vaishali & Punita (2013) explain that waters with a high total dissolved solids (TDS) content are unpalatable and potentially unhealthy. Francis (1878) made the first report of cyanobacterial poisoning, which led to the deaths of cattle, sheep, dogs, horses and pigs after they drank water contaminated with cyanobacteria in Lake Alexandrina, Australia. Since then, many cases of farm animals, waterfowl and wild animals being poisoned by cyanobacterial water after swimming or drinking have been reported. The deaths of cattle reported in the Nadosoito dam area could well be associated with cyanobacteria.

Microorganism persistence in water is affected by several factors, including temperature. Their decay is usually faster at higher temperatures and may be enhanced by the UV radiation in sunlight acting near the water surface (WHO 2006). The WHO guideline for drinking water (2006) warns that, between 25 and 50 °C, a wide variety of microorganisms proliferate. The pH of the water affects the solubility of many toxic and nutritive chemicals, so that the availability of these substances to aquatic organisms is affected. As acidity increases, most metals become more water soluble and more toxic, while ammonia becomes more toxic with slight increases in pH (Vaishali & Punita 2013). The water's pH may affect its treatment and is an important parameter for the treatment system. For instance, more alkaline waters require longer contact time or higher free residual chlorine levels at the end of contact time for adequate disinfection (0.4 to 0.5 mg-Cl/l at pH 6 to 8, rising to 0.6 mg-Cl/l at pH 8 to 9); chlorination may be ineffective above pH 9 (WHO 2006).

The importance of Nadosoito dam for domestic and livestock use prompted this study. It was recognised that there was a need to understand the water quality variations and their health impacts, as well as selected microorganisms and pollutants (NO₃-N, NO₂-N, NH₃-N, P and Fe).

Nadosoito Dam is located near Arkatani Village, Sepeko ward, Monduli district, Arusha, Tanzania (Figure 1), with geographical coordinates, 03° 18′ S and 36° 27′ E. Two small seasonal streams deliver water to the dam, with one removing it during the rainy season.

A census in 2012 by the United Republic of Tanzania (URT) Bureau of Statistics recorded 158,929 people in Monduli district, of which, Sepeko ward recorded 16,720 people (URT 2012). The population (majority Maasai tribe) comprises mainly agro-pastoralists. They were previously purely pastoralist but agriculture is practiced nowadays as a strategy to cope with population increase and the effects of climate change on their livestock. Sepeko is largely grassy lowland with scattered shrubs, used for grazing.

Being a nomadic tribe, the Maasai constantly move their herds searching for better grazing and water sources. Some, especially women and children, stay at home and move only during severe droughts that lead to loss of water and grass. Most of the bean and maize farms, which are only rain-fed, are located in the lower part on the south-west side of the dam. There are hills north of the dam and residential areas scattered around it.

The area is frequently drought-stricken and people depend on the dam as the major water source. Average annual rainfall is 650 mm and the range in the long-term rainfall trend analysis (1984 and 2000) extends from <500 to >600 mm (Msoffe et al. 2011). The dam water is characteristically turbid with very fine clay particles, which settle slowly. An inclined plate settler (IPS) combined with a constructed wetland project by the Nelson Mandela African Institution of Science and Technology (NM-AIST) are located at the dam to treat turbid water.

Two points were selected for sample collection from the dam to represent the water quality: the spillway (SW) and the waterway entry point (WWE). Samples were collected twice a month, in selected months in the wet and dry season in 2017, to ascertain water quality variation and treatability from the onset of rainfall to the end of the dry season, and the causes of animal deaths with respect to each season. The wet months selected were January through May, while the dry months were August and September. All physico-chemical samples were collected in sampling bottles, and preserved as per the standard methods for examination of water and wastewater (APHA 2012). Except for those containing preservatives (e.g. for Fe), the bottles were rinsed three times with the sampled field water before sample collection. They were then stored in iced cool boxes with blue ice cubes at 4 °C before delivery to the NM-AIST laboratory for analysis. Samples were taken at the WWE and SW to determine the changes occurring as water moves between the two. This was necessary to try to determine an appropriate location to draw water for treatment.

Most physical-chemical parameters were analysed in the field; pH, temperature, dissolved oxygen (DO), electrical conductivity (EC) and TDS, using a HANNA Multiparameter (HI 9829, HANNA Instruments, Woonsocket, RI, USA). Turbidity was determined with a HANNA Turbidimeter (HI 93703), after dilution of the samples (100 times, in most cases, in the wet season), as the raw water was too turbid for the equipment. TSS were determined by direct measurement with a spectrophotometer (HACH DR 2800, Hach, Loveland, CO, USA). Nitrogenous species (NO₂-N, NO₃-N, NH₃-N), P and Fe were determined at the NM-AIST laboratory with a HACH DR 2800 spectrophotometer. Chemical oxygen demand (COD) was determined by the reactor digestion method using a HACH COD digester (DRB 200) and a multiparameter photometer (HI 83099), as per the manufacturer's instructions and the standard methods for examination of water and wastewater (APHA 2012). Phycocyanin (PC) pigment unique to cyanobacteria was analysed in situ using AquaFluor model 8000-01 (Turner Designs, San Jose, CA, USA, serial # 803247). WHO water quality guidelines (Brient et al. 2008), were used to interpret the PC concentration on the basis that a concentration of 30 μg/L is equivalent to the WHO alert level of 20,000 cyanobacterial cells/mL, and less than 30 μg/L means that the number of cyanobacterial cells/mL is below the alert level.

The general trend was for most physical parameters (e.g. turbidity, TSS) to decline between the WWE and SW. Some parameters, like turbidity, increased with increasing rainfall (due to the high influx of suspended materials) while others, e.g. EC and TDS, declined with rainfall as a result of dilution. Nutrient concentrations (e.g. nitrate) were higher in the wet than the dry season, while nitrite and phosphate concentrations were higher in the dry than the wet season.

EC varied greatly between the wet and dry seasons, with maxima of 1,356 and 5,802 μS/cm respectively in the wet and dry seasons. The wet season maximum was very different from the other wet season results and was determined at the onset of rainfall (January 2017). This arose because of the transfer into the dam of organic and inorganic materials that had accumulated in the catchment. Other EC results from the wet season were in the range 243 to 412 μS/cm. These lower values arise mainly because of dilution by rainwater. The highest EC values were read, in both seasons, at the WWE sampling site. The wet season variation is due to rainwater dilution – only light rain had fallen by the time of the first sampling in January. The dry season variation was caused by the high rates of evaporation, which reduced the volume of water and concentrated the dissolved solids. EC is used only to give an indication of possible water quality problems.

TDS varied from 141 to 882 mg/l, and from 822 to 3,855 mg/l, in the wet and dry seasons respectively. The wet season TDS maximum was found at the WWE. The maximum TDS – 3,855 mg/l – was also obtained at the WWE at the final sampling (dry season), in September 2017. There was evidence of erosion on the sides of the dam during the rainy season, at that time, indicating that large amounts of sediment had been discharged into the water, leading to high TSS and TDS concentrations. Erosion is likely to increase the TDS in water because it enhances the supply and transport of potentially soluble ionic components from the catchment into the dam.

Water with TDS concentrations below 1,000 mg/l is usually acceptable for domestic use, although this may vary according to circumstances (WHO 2003). The TDS concentration in the dam was consistently below 1,000 mg/l throughout the wet season, and so within the acceptable limit. All dry season samples, apart from the first (August 1, 2017), had concentrations exceeding 1,000 mg-TDS/l, which might explain the objectionable taste of the water then, at least in part. Heavy rainfall diluted the dam water and the TDS concentration decreased in the wet season, while evaporation caused the TDS to increase in the dry season (Figure 2).

TSS concentrations ranged from 3,820 to 46,600 mg/l and 1,750 to 12,000 mg/l, in the wet and dry seasons respectively, at both sampling points. The general trend shows a decline from the wet to the dry season (Figure 3). TSS is a significant indicator of potential pollution because some pollutants and microorganisms can adhere to the suspended particles.

Temperature of the water in the dam ranged from 19.55 °C (1 August) to 33.41 °C (20 March), with both measurements (max and min) taken at the SW. Temperature affects some physico-chemical properties of water (e.g. DO), which are discussed with the respective water quality parameters.

pH ranged from 7.01 to 9.81 across the seasons, with a wet season range from 7.01 to 8.8, and dry season range from 7.67 to 9.81 (Tables 1 and 2). The highest wet season pH (8.8) occurred in May, and the highest in the dry season (9.81) in August, both at the SW. The water was generally more alkaline in the dry season than the wet.

Turbidity is another indicator of the amount of material suspended in water, and is a measure of the amount of light scattered or absorbed as it passes through the water column. It is a measure of water clarity describing how far the light will go as it passes through water. Suspended silt and clay, organic matter, and plankton all contribute to turbidity. Turbidity at the dam ranged from 2,920 to 53,300 NTU in the wet season, with mean values of 15,381 NTU at the WWE and 11,737 NTU at the SW (Tables 1 and 2). In the dry season, the turbidity ranged from 204 to 4,330 NTU, with means of 1,265 NTU at the WWE and 1,613 at the SW. Such values indicate the potential presence of harmful contaminants.

Like TSS, turbidity was lower in the wet than the dry season, because settlement of solids is possible in the latter. WHO recommends <5 NTU as an acceptable turbidity range for drinking water (WHO 2006) while the Tanzania Standards for drinking water (TZS) recommends a range of 5 to 25 NTU (TZS 2003). For effective disinfection, it is recommended that turbidity is consistently below 5 NTU and preferably below 1 (WHO 1997). The values obtained are far above the recommended range and the water requires treatment before consumption. Turbidity varied in line with TSS at both sampling locations (Figure 4). In the context of the IPS it was noted that pre-treatment of such high turbidity water will require more settling/sedimentation time.

DO concentrations ranged from 2.54 to 8.8 mg/l in the wet season and 1 to 8.05 mg/l in the dry season. The maximum concentration of DO was found at the WWE in the wet season (January 2017) (Table 1) when the temperature was 25 °C. The lowest DO, also at the WWE, was observed at the end of dry season (September) (Table 1). DO is an important parameter in water quality assessment, and reflects the physical and biological processes prevailing in the water (Trivedy & Goel 1984). Organic pollution, for instance, tends to remove much of the DO in aerobic biological decay, decreasing the amount in the water (Chhatwal 2011). Other studies have related low DO concentrations to aesthetically displeasing colours, tastes and odours in water. DO depletion in water supplies can encourage microbial reduction of nitrate to nitrite and an increase in the concentration of ferrous iron in solution, with subsequent discoloration at the tap when the water is aerated (WHO 2006). A similar relationship was established in this study, whereby, when the DO concentration was at its lowest the nitrite concentration was at its maximum (290 mg/l) – at the end of the dry season (September). No health-based guideline value is recommended for DO (WHO 2006).

COD is a measure of the amount of oxygen required to oxidize dissolved organic and inorganic material, and is commonly used as an indirect measure of the concentration of organic compounds in water (Kumar et al. 2011). The COD level at the two sampling locations ranged from <0.7 to 87 mg/l. The minimum was obtained at both locations on 1 August, in the dry season, the maximum at the WWE in April (Table 1). The highest COD level at the SW was 29.1 mg/l, occurred on 7 March (Table 2). The maximum and minimum levels generally arise from the material loads brought into the dam by surface runoff and subsequent settling, respectively. This is due to the fact that much material is brought into the dam by the surface runoff and the rivers during the wet season, while no material is brought into it during the dry season, since the rivers to the dam are seasonal. The general trend for COD to decrease from the WWE to the SW, and from the wet to the dry season, is accounted for by several factors, including material sedimentation and possible removal of organic substrate that is coupled with the formation of the biomass and the consumption of DO (Orhon et al. 2009).

The most important source of ammonia in the water in the dam is the ammonification of organic matter. Animal and human excreta contain large proportions of nitrogenous matter, so their presence and decomposition in water tends to increase ammonia concentrations. Ammonia's presence in waters provides chemical evidence of organic pollution (Trivedy & Goel 1984). The concentration of ammonia (NH₃-N) ranged from 0.14 to 270 mg/l in the wet season and 2.2 to 27 mg/l in the dry. The concentration increased with increasing rainfall in the wet season at both sampling locations (Figure 5), with the highest value occurring at the SW in April (Table 2). The maximum concentration at the WWE was 213 mg/l, also in April (Table 1).

Standard deviations of 74.967 and 95.257 mg/l was obtained at the WWE and SW, respectively, in the wet season, while the dry season values were 7.21 and 10.53 mg/l at the WWE and SW, respectively. The high deviation in the wet season was a result of a very high ammonia value obtained during April 2017 sampling as compared to other sampling events. The threshold odour concentration of ammonia at alkaline pH is approximately 1.5 mg/l, and a taste threshold of 35 mg/l has been proposed for the ammonium cation – beyond that, it can cause taste and odour problems (WHO 2006).

Nitrate (NO₃-N) was generally higher at the WWE than SW in samples taken in the wet season. The concentration was below 10 mg-N/l (the WHO guideline level for drinking water) in only two samples (both taken in January) from the SW. In the dry season, all determinations of NO₃-N, except one (in August), at the WWE – 4 mg/l – exceeded the guideline value. This means that infants and pregnant or nursing women are potentially exposed to methaemoglobinaemia (WHO 2006) since they currently use the water untreated. The highest level reported was 1,715 mg-N/l, which was taken at the WWE in March (Table 1). NO₃-N was higher at the WWE than SW for all but four sampling sessions – the latter were 13 February, 20 March, and 6 and 29 May 2017, all in the wet season.

The maximum NO₃-N concentration in the dry season was 512 mg-N/l, in the August sample from the SW. There was significant variation in NO₃-N concentrations between the sampling sessions, with standard deviations of 207 and 135 mg-N/l for the SW and WWE, respectively.

Nitrite is more toxic than nitrate. The WHO provisional long-term exposure guideline for NO₂-N is 0.2 mg-N/l (WHO 2006). As for nitrate, the presence of nitrite in water is associated with methaemoglobinaemia, especially in bottle-fed infants (WHO 2006). With a single exception – 3 April at the WWE –all wet season samples at the WWE reported NO₂-N below the WHO guideline. In the dry season all samples from both locations reported NO₂-N concentrations significantly above the WHO guideline. The highest concentration – 290 mg-N/l – was obtained at the SW on 25 September and the minimum – 0.209 mg-N/l – on 1 August. Clearly, the dam water will require effective treatment to remove nitrite to the level recommended for drinking – especially in the dry season.

Phosphate (PO₄³⁻) concentrations varied in the wet season from 0.06 to 10 mg/l at the two sampling points. PO₄³⁻ concentrations were generally fairly similar at the WWE and SW (Tables 1 and 2). The higher concentrations, 2 to 10 mg-P/l were recorded between March and May when there were long periods of rainfall. Phosphate concentrations were higher in the dry than the wet season. The minimum and maximum dry season concentrations were 0.31 and 15.2 mg-P/l respectively (Tables 1 and 2) – the latter from the SW on 25 September. The high phosphate concentration might be related to the elevated level of pH in the dry season (Gao et al. 2012), as this promotes desorption of sedimentary inorganic phosphorus.

The Iron concentrations ranged from 13.0 to 1,380 mg/l. The highest value was reported from the WWE in the first January sampling. Iron concentrations differed significantly between the wet and dry seasons, with the higher levels occurring in the wet season.

As all the water in Nadosoito dam is derived from surface runoff, the iron concentration in its water is likely to increase during rainfall. This is due to the possible release of soluble iron to water from natural deposits in soil, leaching from underlying rocks and stones on the catchment. Iron is involved in phosphorus dynamics in fresh water; Fe (III) oxides and hydroxides readily adsorb and precipitate phosphate. This could explain the differences between phosphate concentrations in the wet and dry seasons. In aerobic conditions, interactions between colloidal organic material and iron decrease the extent of phosphate adsorption onto Fe-organic complexes and increase the availability of P to phytoplankton (Koenings & Hooper 1976). Whenever human activities, e.g. pastoralism and cultivation, disturb the balance of soils rich in iron, the inflow of iron into the water system is likely to be enhanced (Heikkinen 1990). This is because the loose soil particles are prone to erosion by the surface runoff and are subsequently washed out into the dam.

In the wet season, the PC concentration varied from 38.1 to 495 μg/l. It was highest in January at the WWE and lowest in April at the SW. The mean PC concentrations recorded in the dry season were 94.8 and 104.0 μg/l at the WWE and SW, respectively; relatively much closer than in the wet season. The most likely time for a cyanobacteria bloom is after inflows deliver significant quantities of nutrients to the reservoir. The minimum concentration in the dry season was 19.9 and the maximum 175.7 μg/l. Since a PC concentration of 30 μg/L is reported as equivalent to WHO ‘alert level 1’ of 20,000 cyanobacterial cells/mL (Brient et al. 2008), this implies that the PC concentration in all but one sampling session in both seasons exceeded the WHO alert level. (In one case in the dry season the PC concentration was reported as 19.85 μg/l.) Without essential nutrients, principally nitrate and phosphate, algae will usually not reach bloom proportions and PC concentrations are expected to be low. Excessive nutrient input from land-based sources is a major cyanobacterial bloom promoting factor (WHO 2003).

Broadly, the phosphorus and nitrogen loads determine the rate and magnitude of cyanobacterial growth (PC concentration), which means that higher loads imply greater potential algal growth (Wetzel 2001). This was shown in this study when high PC concentrations in the wet season coincided with high nitrate concentrations. The phosphate trend opposed that of the PC concentration, being higher in the dry season than the wet. Medium- to long-term management measures for cyanobacterial control include identification of nutrient sources (phosphorus and nitrogen) and significant reduction of nutrient input, to reduce the proliferation of both cyanobacteria and other potentially harmful algae (WHO 2003). Carmichael (1994) compiled case studies on nausea, vomiting, diarrhoea, and fever, and eye, ear and throat infections after exposure to mass developments of cyanobacteria. Cases of deaths among cattle, sheep, dogs, horses and pigs after drinking water containing cyanobacteria were reported by Francis (1878).

Faecal coliform forming unit (CFU) concentrations generally declined from wet to dry season (Figure 6), with mean values of 3,213 and 2,029 CFU/100 ml at the WWE and SW, respectively, during the wet season. The maximum value recorded in the wet season was 8,810 CFU/100 ml at the WWE, while the minimum was 760, also from the same location (Table 1). The maximum value was recorded at the beginning of the rainy season in January, and arose because of the additional contamination from the catchment brought in by the surface runoff associated with rainfall. The minimum dry season value of 30 CFU/100 ml was obtained at the WWE in September, on the last sampling run of the season, while the maximum value in the dry season – 710.5 CFU/100 ml – was obtained at the SW. All values from all sampling events exceeded the WHO guidelines and the TZS value of 0 CFU/100 ml (TZS 2003; WHO 2006). The contamination is related to the transport – by runoff, etc – of animal and human faecal matter to the dam, and direct defecation by livestock as they drink from the dam.

The results obtained indicate high levels of turbidity (204 to 53,300 NTU) and high NO₃-N concentrations (0.6 to 1,715 mg-N/l), both of which require proper treatment to bring them to the TZS prescribed range (5 to 25 NTU and 10 to 75 mg NO₃⁻ N/l respectively).

The concentration of faecal coliforms (FCs) was highest in the wet season (8,810 CFU/100 ml) with a minimum value of 30 CFU/100 ml obtained in the dry season, both at the WWE. The highest and lowest FC concentrations at the SW were 7,380 and 110 CFU/100 ml respectively. All values in both seasons were above the TZS and WHO recommended values for drinking water (0 CFU/100 ml). The maximum value arose from the very high contaminant loading in the runoff into the dam during the wet season.

The observed PC concentration (19.9 to 495 μg/l) is a useful surrogate for the concentration of cyanobacteria present. The WHO guideline alert level for PC is 30 μg/l. The rather high concentrations of cyanobacteria, and hence PC, can probably be related to the incidence of diarrhoea among livestock and people, as well as livestock deaths reported in the area. The higher levels coincide with the wet season when there are significant runoff inflows that deliver large quantities of nutrients to the reservoir.

Nadosoito dam, like other earthen dams, has calm, nutrient-rich, warm and slow-flowing water, thus favouring the growth and proliferation of cyanobacteria. This is well depicted in the results. While there is little that can be done to control such proliferation in earthen dams, investigations into ways to reduce nutrient loading in the influent – e.g., runoff – could provide a solution that might reduce the associated human and livestock health impacts to a large extent.

The results obtained from this study will help to provide the necessary guidelines for the design and operation of the IPS constructed wetland system. The outcome will be to increase access to safe drinking water, thereby reducing the human health impacts associated with direct use of the dam water and reducing local medical costs.

As most of the parameters determined reported higher values at the WWE sampling point, it is recommended that water for treatment by the IPS constructed wetland system is drawn from the vicinity of sampling point SW.

This research was funded by British Gas (BG)/Shell to its completion. The Nelson Mandela African Institution of Science and Technology (NM-AIST) provided all necessary guidance, laboratory and other facilities, thus enabling accomplishment of this work. Finally, the support of the VLIR-OUS program in field and laboratory activities is acknowledged gratefully.