Ecological and health risk assessment of potentially toxic elements in Ewaso Nyiro River surface water, Kenya

Ewaso Nyiro basin covers an area of about 210,226 km2, 36.3%, of Kenya drainage area and bears 5.8% of Kenya water potential with an annual yield of 1469 million m3. The river is the principal source of domestic and irrigation water to the arid north of Kenya. To determine metal and nutrient concentration of Ewaso Nyiro River surface water, a total of 30 water samples, 15 samples each for dry (February) and wet (August) seasons of 2019, were collected. Chromium, lead, iron, manganese, cobalt, cadmium, mercury, selenium, molybdenum, boron, copper, zinc, arsenic, nickel, aluminum, total phosphorus and nitrate were analyzed in the two seasons. Ecological risk assessment was determined by calculating contamination factor, pollution load index and ecological risk index. Multivariate statistical analysis was used to infer pollutants association and identify their potential sources. Cadmium, arsenic, lead, molybdenum, mercury, selenium and nickel were not detected in both seasons, while manganese, iron and aluminum were the main pollutants identified. Ewaso Nyiro irrigation water had a manganese contamination factor of 9.17, implying it was very contaminated. Twenty-seven and 40% of sampled sites in dry and wet seasons, respectively, had more than 0.3 mg/L of iron that is recommended by USEPA in drinking water. Herbicides, leached fertilizer and fuel leaking into the river water were the primary sources of anthropogenic pollution.


Introduction
Water is an indispensable resource. It is used in agricultural production, domestic use and industrial activities since it is a universal solvent. Global freshwater distribution is uneven with some regions facing severe scarcity while others enjoying surplus [16]. It is estimated 4 billion people, more than half of global population, face severe water shortage at least one month in a year [4]. Water scarcity has resulted in regional and local conflicts [25,29]. Kenya is a case in point. Over a hundred people perished due to conflicts between herders and farmers in 2012 [1]. Water scarcity is a consequence of rapid population growth and changing weather patterns that are associated with erratic precipitation and prolonged droughts [5,18].
Immense pressure on water resources has resulted in pollution, jeopardizing human health [37]. Contaminated water causes over 5 million annual deaths worldwide [36].
Polluted water-related health complications are widespread and are not confined to developing countries, due to bioaccumulation and biomagnification in the food chain [2]. Incidences of ecological risk caused by zinc have been observed in Japan [30]. Irrigation water polluted with heavy metals contaminates crops under cultivation and poses risk to human health. Similarly, consumption of fish from polluted aquatic ecosystems may be a health threat [10].
High concentration of iron in drinking water has been reported to cause muscle crumps, severe constipation, headache, gastric ulcers, malignant tumor and even rapid heart pulses in some instances [9,33].
Kenya is classified as a water scarce country with 80% of land mass being arid and semiarid [33]. Majority of the population is concentrated in areas with adequate rainfall for agriculture, and vast areas are sparsely populated with nomadic pastoralists. Meager water resources that are available are heavily polluted with potentially toxic elements. High concentration of As, Hg, Ni, Cu and Cd was detected in lakes Bogoria and Elementaita [48]. Similarly, high levels of Mn, Fe and Al have been reported in Tana River [31].
Majority of Kenyan population is poor with limited access to improved drinking water [12]. Domestic use of polluted water is thus highly likely, and this may predispose innocent and poor people to deleterious effects of potentially toxic elements.
Regular monitoring of water sources used for domestic purposes is therefore paramount to safeguard human health. Previous documented studies of Ewaso Nyiro were based on water management, scarcity and conflicts [22,41]. The current study was conducted to (1) determine heavy metal and nutrient concentration in surface water of Ewaso Nyiro River, (2) assess probable sources of heavy metals and nutrients pollution, (3) evaluate whether Ewaso Nyiro River is under ecological risk and (4) determine whether domestic, irrigation and livestock use of Ewaso Nyiro River surface water poses any health risk to consumers.
Ewaso Nyiro drainage basin population has increased rapidly in recent past. For instance, Laikipia County population has tripled over the last three decades from 134,524 in 1979 to 399,227 in 2009 [22]. The upper reach has a population density of 100 people per km 2 and about 10 in the lower reaches that are arid and semiarid [8].
Water resources in the upper reach of the basin have been under immense pressure resulting in intermittent flow of Ewaso Nyiro below Archers Post [20].
Main economic activities in Ewaso Nyiro basin include large-and small-scale horticulture crops production in the upper reaches and livestock rearing downstream of the basin. The main challenges undermining Ewaso Nyiro ecosystem are catchment degradation, increased irrigation demand and groundwater salinity and saline intrusion [6].
Water sampling sites were distributed around the catchment and middle reaches up to Archers post since lower reaches had dried during the dry season. Sampling sites selection was based on land use pattern, economic activities and areas suspected to experience point source pollution.

Water sampling
A total of 30 samples were collected during wet and dry seasons, as shown in Fig. 1, 15 samples for each season. Sampling in dry season was conducted in late February, while wet season samples were collected in August of 2019. Ewaso Nyiro River is shallow with homogeneous vertical water column. Sampling was done 15 cm below water surface using precleaned one-liter polyethylene bottles. Samples were later taken to the laboratory where they were filtered using 0.45-µm cellulose acetate membrane filter, acidified with nitric acid to pH > 2 and refrigerated at 4 °C before commencement of the analysis.

Sample analysis and quality assurance
Analysis to determine concentration of heavy metals in water was determined using inductively coupled plasma-optical emission spectrometry, ICP-OES (5100, Agilent Technologies), while nitrate (NO 3 − ) was determined by calorimetric method. Total phosphorus (TP) and iron (Fe) were determined by spectrometric method (UV-1700, Pharmaspec, Shimadzu) at an absorbance of 880 and 515 nm, respectively.
To ensure analytical integrity and quality of the results were of acceptable level, blanks and sample replicates were randomly analyzed and had to tally to validate results. Standard reference material (CROP-1-REV1 obtained from Inorganic Ventures, USA) was used for quality control (one blank and one standard sample for each 20 samples). Method efficiency was determined by the use of recovery rates that ranged between 89 and 104%. Relative standard deviation from seven samples was applied to generate minimum detection limits (MDL) that were as follows: Cd = 0.002 mg/L, Cr = 0.004 mg/L, Pb = 0.009 mg/L, As = 0.007 mg/L, Ni = 0.003 mg/L, Hg = 0.001 mg/L, B = 0.01 mg/L, Cu = 0.01 mg/L, Se = 0.02 mg/L, Al = 1.5 mg/L, NO 3 − = 0.01 mg/L and P = 0.1 mg/L. Data were analyzed using SPSS (version 21), while map was prepared using Google maps.

Contamination factor (CF)
Contamination factor is used to express the level of concentration of an element in a sample compared to preindustrial reference value. The current research has employed recommended concentration in human and livestock drinking and irrigation water as reference values. Contamination factor is used as a measure of potential toxicity, as shown in Eq. 1 [27].
The CF values were classified into six categories, as given in Table 1, as recommended by Hakanson [40].

Pollution load index (PLI)
The PLI which is an integrated tool for determining pollution of a certain group of pollutants (metals) was also applied, as indicated in Eq. 2 [43].
where CF is the contamination factor for individual metals and n is the number of pollutants (metals). The PLI values are classified into four categories, as given in Table 1.

Ecological risk (ER)
The ER is used to assess degree of environmental damage or probability of ecosystems experiencing detrimental ecological effects from pollutants [51].
Ecological risk was determined as recommended by Hakanson, as given in Eq. 3 [23]. where T i is the potential ecological risk coefficient. Potential ecological risk coefficient for Cu, Zn, As, Cd, Cr, Pb, Ni, Hg, Mn and Co is 5, 1, 10, 30, 2, 5, 5, 40, 1 and 5, respectively [46]. The ER values are grouped into six categories as shown in Table 1.

Heavy metals and nutrient concentration in water
There was a significant difference in concentration of pollutants in different sites and seasons as indicated in Table 2. Mean concentration of all detected pollutants, with exception of B, was higher in wet season compared to dry season. Only Al, Fe and Mn were detected in all sampled sites in the two seasons. As, Cd, Hg, Pb, Ni and Se were below detection limit in all sites in both seasons. Aluminum recorded the highest concentration in both seasons recording 22.5 and 16.90 mg/L in wet and dry seasons, respectively. Although World Health Organization (WHO) has not given maximum concentration of Al in drinking water, concentration above 0.2 mg/L is not acceptable to most consumers [13]. Eighty-seven percent and 67% of all sampled sites had more than 0.2 mg/L of Al in wet and dry seasons, respectively. Cobalt, Mn and Fe above WHO recommended drinking water guidelines were detected in one site for Co, three sites for Mn and six sites for Fe in wet season. Similarly, exceedingly high concentration of Mn and Fe was registered in three sites for Mn and four sites for Fe in the dry season. Median concentration of Fe in the two seasons was above, 0.3 mg/L, which may stain laundry and plumbing fixtures besides exceeding USEPA drinking water guideline [13], Njuguna et al. [33]. Majority of the sampled sites had relatively low concentration of NO 3 − registering a median of 0.27 and 0.10 mg/L, in wet and dry seasons, respectively. Although TP maximum concentration in both seasons was above 0.2 mg/L and likely to cause eutrophication, median concentration was below detection limit. High concentration of Al, Fe and Mn has been recorded in other water bodies in Kenya. Aluminum and Mn concentrations of 8.96 and 1.11 mg/L, respectively, were detected in Tana River surface water in 2018 [31]. Similarly, Fe and Mn, 3.26 and 0.98 mg/L, respectively, were registered in Sasumua reservoir that supplies water to Kenya's capital, Nairobi, besides 2.82 mg/L of Mn in Lake Elementaita [28,35]. Andisol soils that are prevalent in Kenya highlands are rich in Al, Fe and Mn [33].

Risk assessment
Human and livestock drinking water contamination factor was < 1 for all potentially toxic elements and therefore was in low contamination factor category in dry season, as given in Table 3. However, irrigation water Mn contamination factor was > 6 and was therefore very contaminated. River Ewaso Nyiro human drinking water was moderately contaminated with Mn but very contaminated for livestock consumption and considerably contaminated for irrigation purposes in wet season. Although Al and Fe human drinking water was under very high contamination, only Fe had considerable contamination for livestock consumption. Manganese had moderate contamination for human drinking, considerable contamination for irrigation and very high contamination for livestock consumption in wet season. Only Mn among all potentially toxic elements had potential of causing mild ecological risk.
Pollution load index was < 1 for human and livestock drinking and irrigation purposes in both dry and wet seasons, implying the water would be considered generally uncontaminated. Highest risk index recorded was 15.88 in livestock drinking water and was way below 50 that can pose mild risk, suggesting there was no risk posed to human and livestock.
Although risk index posed by potentially toxic elements under investigation was negligible, Al, Fe and Mn contamination was very high. High Mn in drinking and food stuff has been confirmed to cause health complications. High Mn in drinking water lowers children IQ. A study conducted in Bangladesh indicated children drinking water with high Mn, above 400 μg/L, scored 6.4% lower in mathematics compared to those whose drinking water had lower or no Mn [17]. Manganese has been noted to replace metabolic role of calcium and may accumulate in bone tissue weakening the bones besides causing adverse effects to the lung, liver and cardiovascular system [3,9]. Aluminum concentration above 0.2 mg/L is not acceptable to most consumers, based on aesthetic considerations, besides increased number of studies associating high concentration Al in drinking water with Parkinson and Alzheimer's disease [9,13]. High concentration of Fe in drinking water on the other hand is associated with malignant tumor and causes persistence of hepatitis B and C besides being linked to kidney, lung, liver and stomach cancers [18]. Cancer is the third leading cause of death in Kenya after infectious and cardiovascular diseases [26].
Numerous incidences of high Fe in Kenyan drinking water have been reported. Athi River that serves over four million Kenyans with drinking water recorded a mean of 2.5 mg/L against USEPA recommended concentration of 0.3 mg/L [33]. The use of polluted water for livestock and irrigation purposes is unsafe due to bioaccumulation and biomagnification effect in the food chain. Consequently, the use of Ewaso Nyiro River water for livestock and irrigation purposes should be done with caution due to considerable Fe and very high Mn contamination.

Multivariate analysis
Principal component analysis, hierarchical cluster analysis and Pearson correlation were employed to infer pollutants associations and identify their sources.

Pearson correlation
Pearson correlation analysis was done among and between heavy metals and nutrients as displayed in Table 4. A strong positive correlation, p > 0.01, was observed between Cu and Cr, Cu and Al, in wet season. Other significant correlations were between Al-TP, Fe-TP, B-Cu, Cu-Fe, Zn-Fe, Zn-Al, Cr-Al, Cr-Co and Al-Fe. Strong positive correlation between pollutants indicates a common source, similar behavior or mutual dependence [24]. Aluminum and Fe may have originated from Andisol soils that are common in the study area [31]. Cobalt may have emanated from agrochemicals since it was only detected at site 11, that was under cabbage and garden pea production. Glyphosate that is a common herbicide in Kenya has been noted to contain Co and Cr that may dissolve and contaminated aquatic ecosystems [7]. Strong positive correlation observed between Al-Cr and Cr-Co may imply Cr was from both natural and anthropogenic sources since Al and Cr are widely distributed in earth crust [13]. Manganese may have been from leaking fuel from irrigation pumps and motor bikes that were being washed in river water since Mn compounds are used as fuel-oil additives [45]. The highest concentration of Mn was detected in site 7 where water had been contaminated with oil and sites 11 and 15 where motorbikes were being cleaned in the river.

Principal component analysis (PCA)
PCA is a data reduction technique that reduces number of variables to a smaller set of scores called components without losing much of primary data [15]. Varimax rotation was employed in current analysis since variables were presumed to be uncorrelated [33]. Two principal components that explained 80.8% of total variance were extracted at eigenvalue > 1 in wet season, as given in Table 5. Component one explained 66%, while component two explained 14.8% of total variance. Al, Cu and Fe were in component one and had significant positive loading, > 0.9. Nitrate was the only variable in component two and had a high loading of 0.96. Al, Cu and Fe may have emanated from natural sources. Cases of these elements emanating from natural sources have been reported [39]. Nitrate may have originated from leached fertilizers from agricultural activities.
Dry season had two principal components with components one and two contributing 71.9% and 14.7% of total variance, respectively. Component one significant positive loading, > 0.9, was from Al, Cu, Zn and Fe. Palaeozoic and quaternary sediment that is widespread in Kenya is a major source of Zn, Cu, Al and Fe [11]. Nitrate was the only variable in component two with a high loading of 0.96 and was from leached fertilizers just like in wet season.

Hierarchical cluster analysis (HCA)
HCA which is a cluster analysis technique is often coupled with PCA to confirm and validate observed relationships [47,50]. Euclidean interval pattern and ward linkage were applied, while potentially toxic element concentrations were standardized using Z-score.
There were four clusters in both wet and dry seasons as indicated in Figs. 2 and 3. Clusters formed were based on element attributes and sources. Cluster A in wet season dendrogram was composed of Cu, Cr, Al and B. These  pollutants may be from both anthropogenic and natural sources. Copper and Cr may be from herbicides, while Al and B may be from both natural sources and waste water containing Al used in coagulation and B from detergents [7,13,42]. Custer B was composed of Zn, Fe and TP which could be from natural sources. Andisol soils prevalent in study area are rich in Fe, while cases of TP and Zn originating from natural sources have been reported [39]. Cobalt and Mn were in cluster C and may have been from herbicides and fuel leaking into the river water. The highest concentrations of Mn and Co were detected at site 7 that was under agricultural production and was experiencing fuel leak that was contaminating river water. Nitrate was the only pollutant in cluster D and was from leached fertilizer infiltrating into the aquatic ecosystem. Nitrate is readily soluble and may seep from agricultural land and pollute water bodies [13]. Cluster A in dry season dendrogram was composed of Cu, Pb, Zn, Al, Fe and TP and may have originated from natural sources. Lake Olbolosat was the only sampling site with detectable Pb and also recorded the highest concentration of Zn. Lead is naturally found in soil in low quantities and may emanate from natural weathering of rocks [10,32], while Zn is relatively mobile and may permeate through soil and end up in river water [38]. Only boron was in cluster B and may have been from anthropogenic sources. The highest concentration of B was detected at Archers Post where people were doing laundry and bathing on river banks. Nitrate and Mn were the only pollutants in clusters C and D, respectively. Nitrate may have been from leached fertilizers, while Mn may have been from leaking fuel since the highest concentration of Mn detected was at Rumuruti Swamp where fuel was leaking from irrigation pumps. Rumuruti Swamp is rich in Cyperus papyrus and other beneficial macrophytes such as Cyperus articulatus and Typha latifolia may be introduced to promote phytoremediation of Al and Fe [14,33].

Conclusion
There was a significant difference in concentration of potentially toxic elements among different sites. Heavy metals and nutrients were distributed from the catchment to the middle reaches without a properly defined pattern. Dry season had relatively better water quality compared to wet season probably due to river water contamination from runoff. Poor agronomic practices such as the use of Co-and Cu-based herbicides close to riparian zone may have contributed in Ewaso Nyiro River pollution. Leaking irrigation pumps and Fig. 2 Hierarchical dendrogram of nutrients and metals in Ewaso Nyiro River during wet season cleaning motor bikes on the river banks were observed to pollute the river with Mn. Twenty-seven and 40% of sampled sites in dry and wet seasons, respectively, had more than 0.3 mg/L concentration of Fe that is recommended by USEPA in drinking water. This is alarming bearing in mind high concentration of Fe in drinking water is associated with cancer and majority of residents within Ewaso Nyiro ecosystem use its water without any form of treatment. Proper agronomic practices should be encouraged to curb Mn and Co pollution besides introduction of macrophytes such as Cyperus articulatus and Typha latifolia in Ewaso Nyiro ecosystem for Fe and Al phytoremediation, respectively. Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.