3.1. Spatial Variation of Heavy Metals in Water
The field parameters showed a variability of values depending on the season.
Figure 2 shows the results for temperature (a), pH (b), and electrical conductivity (c) in every sample point. The analysis of these results shows, on the one hand, a heterogeneity between the values of the conductivity according to the seasons and the sampling points (
Figure 2a). Indeed, the electrical conductivity values of water decrease from the dry season to the rainy season and oscillate from 378 to 1443 µs/cm in the rainy season, and from 370 to 2575 µs/cm in the dry season for groundwater. On the other hand, for surface waters, this parameter evolves from 50.3 to 179.56 µs/cm in the rainy season, and from 95.75 to 542.5 µs/cm in the dry season. This variability can be explained by the contribution of ions by runoff and infiltration water during the rainy period. The seasonal variations observed are on the same order as those obtained by [
52], a study carried out on River Benue water at Makurdi. A decrease in electrical conductivity in the rainy season would be linked to the dilution effect. On the other hand, there is a significant decrease in temperature from the rainy season to the dry season, with values that oscillate between 25.8 and 29 °C in the rainy season and from 28.5 to 35.7 °C in the dry season (
Figure 2c). Between 10 and 30 °C, the temperature has only a negligible effect on the mobility of metals. However, in a mining environment, a water temperature higher than 30 °C could influence acid mine drainage (AMD) because it can accelerate the oxidation reactions of minerals [
53]. The variation in temperature observed would be linked to the level of sunshine.
Figure 2b also presents the variations of pH according to the seasons and the sampling points. Analysis of graph (b) in
Figure 2 shows that pH is between 6.51 and 7.31 in the rainy season and 4.6 to 7.1 in the dry season. These results show that 26% of the samples in the dry season have a pH lower than 6.5, which represents the guideline value of the [
54,
55]. These waters are therefore acidic and can lead to the dissolution of metals [
11]. Similar seasonal variations in pH and conductivity have been obtained by other researchers [
37,
56]. Therefore, key field parameters such as temperature, pH, and EC are influenced by seasonal changes and sample points in the study zone, which can have an impact on HM behaviour in the environment.
Table 1 presents the content of each HM (Heavy Metal) analysed, in groundwater as well as surface water sample points. Additionally, the established limit values set by the EU and WHO are shown, to be able to contrast them. Analysis of this table shows that Fe, Sr, Sb, and Mn are the most important metals in the water. Concentrations of metals such as As (<4.5 µg/L), Cd (<0.5 µg/L), Pb (<6.5 µg/L), Cu (<61.7 µg/L), Cr (<20 µg/L), and Zn (<2852.25 µg/L) are all below the WHO and EU limit values; Al has not been detected in groundwater, but exists in trace amounts in surface water (<0.55 µg/L). Except for sample S5 (112.78 µg/L), all the other samples have values below 50 µg/L. The low levels found symbolize that the mine did not have such an impact on these waters for these metals. This testifies to the non-pollution of water by this metal. This result agrees with those of other authors [
34]. Indeed, these authors reported the non-detection of metals such as copper, nickel, and lead during their work in the same zone. However, their analysis device had a detection limit of 5 µg/L for copper and 30 µg/L for nickel and lead, not allowing them to obtain a value. No limit value has been set by the organizations for Strontium (Sr), although its concentrations are high (7.46 to 1359.70 µg/L). For cobalt, the concentrations of vary between 0.1 and 11.52 µg/L. Similar results were obtained by [
44] in western Côte d’Ivoire during the assessment of HM contamination of Zouan-Hounien groundwater.
Iron and manganese play an important role in regulating the biochemical cycle of plants and animals [
57]. Iron is one of the most abundant metals in the waters sampled. The Fe concentration varies from 8 to 20,475 µg/L (mean = 1144.87 µg/L) for groundwater and from 40 to 835,843 µg/L (average of 115,548.15 µg/L) for surface waters. This high value of Fe is a pivotal point for our study, since Fe is well-linked with mining activity. Sampling points where the iron content exceeds the desirable limit value of [
54] (300 µg/L), respectively, represent 56.52% (13/23) for groundwater and 80% (4/5) for surface water; but the same groundwater pollution rate for Fe is 66.52% compared to EU standards (2022). S3 and S5 have iron levels lower than that obtained for some underground water (6307.66 µg/L and 40 µg/L, respectively); this difference would be linked to the aerobic oxidation of iron and precipitation. These waters, with iron concentration exceeding the value, are likely to present a bad appearance, an unpleasant taste, or promote the growth of iron-oxidizing bacteria [
57]. This significant pollution by iron of water resources in the canton would come from the mining waste. The Mn concentration varies from 5.76 to 431.72 µg/L (average 143.77 µg/L) in groundwater. For surface water, Mn varies from 11.27 to 2693.48 µg/L (
Table 1). A total of 52.17% of groundwater samples and 59% of surface water samples have Mn levels exceeding the WHO guideline value. Similar results have already been obtained in the field by other researchers [
34]. Excessive levels may result in nervous system breakdown and psychological problems in humans [
26]. Several studies have shown that these two metals are the most abundant in the earth’s crust; their presence in water comes mainly from soil leaching, the dissolution of rocks and ores, and industrial discharges [
58,
59]. In general, the highest concentration of Mn and Fe being found in hand pump water samples indicates the poor quality and maintenance of hand pumps [
60]. Since the mine stopped, the solid waste and ore, left on the site on the side of a steep mountain, are left to the rain and wind, which can drain them towards water resources. In addition, the work of [
33] showed that the soils of the mining area, the mining waste, and the sediments of the stream receiving drain water (sampling point S3) are polluted by Fe and other metals (Mn, Pb, Cd, Al, Cr, and Cu). Indeed, the study area is located in the Bassar structural unit with silico-ferruginous rocks (40–70% de Fe
2O
3) [
61]. Strong leaching of these soils by the rains will lead to infiltration and accumulation of heavy metals in the waters [
44]. This pollution may also be due to the corroded nature of the iron pipes of hand pumps, or due to the presence of more iron in the ground strata bearing the ground water. On the other hand, the antimony concentrations greatly exceed the guideline value of the WHO [
55] for all samples. The Sb content varies from 25.79 to 401.63 µg/L for groundwater and from 26.23 to 516.47 µg/L for surface water (
Table 1). Most antimony is found in the soil, where it binds strongly to particles that contain iron, manganese, or aluminium [
62]. The pollution of this metal in the study area is too general, so its presence may be natural.
Therefore, this analysis reveals elevated concentrations of certain heavy metals in the water sources, notably Fe, Sr, Sb, and Mn, originating from a former mining site. While metals such as As, Cd, Pb, Cu, Cr, and Zn remain below established limits, others like antimony (Sb) greatly surpass guideline values. The potential health and environmental risks necessitate urgent intervention. Proposed countermeasures include continuous monitoring, remediation of the mining site, community education, infrastructure improvement, and the exploration of water treatment technologies. Collaboration with environmental authorities is crucial for effective implementation. Addressing these concerns is imperative to safeguard water quality and protect the well-being of the local population.
In relation to environmental aspects, the high concentrations of iron and manganese, particularly in surface waters, indicate pollution likely stemming from mining activities, posing a threat to water quality. Excessive levels of these metals can impact human health and ecosystem stability. Antimony concentrations exceeding guidelines suggest potential contamination, emphasizing the need for further investigation. The cumulative impact of heavy metals raises concerns about ecosystem disruption and underscores the importance of continuous monitoring and remediation efforts. Addressing these issues is crucial for safeguarding both environmental integrity and human well-being in the study area.
3.3. Human Health Risk Assessment
The HQ, HI, and CR (cancer risk) for the drinking water pathway with respect to adults and children are quantified in
Table 3 and
Figure 4, using the arithmetic mean of the metal(loid) concentrations for all the drinking water samples of the study area. The results of the non-carcinogenic risks as depicted by HQ for children indicated on the one hand that Sb, followed by Fe and Mn, showed higher risks via the ingestion route, and Sb and Sr via the dermal route (HQ > 1). On the other hand, the values of HI for Sb, Sr, Fe, and Mn exceed 1. Those HMs exhibit the highest risks of contamination.
Except for Sb (24.40), the hazard quotient (HQ) and hazard index (HI) results for all HMs display the lowest human health risks through the dermal or ingestion route (HI < 1) for adults. The hazard index was found to be higher than 1 for Sb, indicating a detrimental effect of Sb on the health of both children and adults. Since the groundwater of the study area is not contaminated with a single metal, the potential risks of the combined effect of all the 12 metals (loid)s through ingestion of drinking water was assessed using HI and calculated to be 24.40 for adult and 63.67 for children. This suggests a potential risk to human health due to consumption of the waters of Bangeli district as drinking water.
The carcinogenic risks determined for all HMs (
Figure 4) showed a low risk of cancer, either via the oral or dermal route pathway, for both children and adults. The CR index was estimated for only As, Cr, Cd, and Pb, due to the availability of carcinogenic slope factors (SFs) for these elements. The cumulative range of CR was 2.2 × 10
−5 to 0.87 × 10
−4 (mean 0.72 × 10
−4) for adults and 0.055 × 10
−4 to 0.21 × 10
−4 (mean 0.14 × 10
−4) for children. Moreover, the cumulative CR value did not exceed the threshold range of 10
−6 to 10
−4. According to [
28,
29,
30,
46], the mean of total CR values due to drinking the water from Bangeli canton was less than 10
−4, which supported our study. For every HM, the carcinogenic risks for different populations vary greatly, generally in the order of adults > children. The reason that the carcinogenic risk for children is less than that for adults lies in the shorter duration of exposure for children. The same results were obtained by many authors, e.g., Isa Baba [
46]. Hence, the studied values did not possess any threat to local residents. In view of the fact that the cancer risk did not exceed the target risk of 10
−4, it can be thus be considered ‘acceptable’ and suggests that the ingestion of these waters over a long lifetime will not increase the probability of cancer for the consumers. The human health risk assessment does not only depend on the concentration of metal(loid) in drinking water, but also on the water consumption rate. The adverse carcinogenic and non-carcinogenic risk that has been calculated for the study area can also be related to the higher consumption of water due to an inhabitant’s occupation and/or the climate. Due to the tropical climate of the study area, the daily water intake is higher as compared to regions in colder climates [
12,
64]. In addition, the exposure parameters employed in the study were taken from the USEPA 2005 and from other countries [
12,
27,
65]; they might be different for Togo conditions. Therefore, for further precise risk characterisation, risk assessment approaches may be modified according to the investigation of the risk levels in Bangeli canton. In fact, these significant findings contribute to the overall understanding of heavy metal variations in Bangeli canton, emphasizing the complexity of factors influencing health risks. The importance of considering both carcinogenic and non-carcinogenic risks in water quality assessments should be noted, providing valuable insights for environmental and public health management in the region.
3.4. Statistical Analysis
With regard to correlations, only two significant correlations were found, between Cu–Cd (r = 0.66 and
p-value = <0.001) and Sb–Pb (r = −0.6 and
p-value = 0.0003) (
Table 4). The high positive correlation between Cu–Cd may explain similar hydro-chemical characteristics or the existence of a common anthropogenic source/origin of these metals in the environment [
12]. On the other hand, the negative correlation between Sb–Pb indicates that originated from different sources [
21]. Additionally, it was observed that there are many negative correlations in the dataset in comparison with some works [
10,
12]. However, no significant correlations were found between the remaining HMs. According to [
8], this is an indication of multiple points of origin for the HMs, which makes sense considering the diversity of human activities and environments in the study zone. The fact that Cd and Pb are significantly correlated HMs in the dataset is interesting, because it emphasises that mining activity really generates an impact in the zone, since the presence of Cd and Pb suggests anthropogenic activity [
66]. Therefore, the Pearson correlation matrix has highlighted the interconnections between specific heavy metals and their potential sources, shedding light on the anthropogenic and environmental factors contributing to heavy metal variations in the study area. These results contribute significantly to the broader understanding of heavy metal contamination and can inform targeted environmental management strategies in regions with similar characteristics.
On the other hand, ANOVA was used to study the differences in relation to seasonal variations and type of water. In the case of seasonal variations, it was observed that only the concentration of Pb and Sb had significant differences between the dry and rainy seasons (
Table 5). Ref. [
67] also performed an ANOVA, but they filtered the data into Pre-Monsoon, Monsoon, and Post-Monsoon. They also obtained significant seasonal differences with only two HMs (Pb and Fe). However, the remaining HMs did not show a significant variability among seasons. Ref. [
68] also did not obtain significant differences with any HMs, which demonstrates that most HMs are available in the water in a stable way. In the case of Pb, high values during the dry season are attributed to the high evaporation rate of surface water followed by high temperature, and low flow condition of the water bodies leading the accumulation of HMs [
67]. On the other hand, the HMs with significant differences according to the type of water were Fe and Sr. In the case of Sr, the groundwater concentration is higher than in the surface, so this element is clearly affected by the filtration processes. This shows the dynamic nature of heavy metal concentrations in water, highlighting the interplay between environmental factors, hydrogeological processes, and seasonal variations.
PCA shows five principal components (PCs) with eigenvalues higher than 1, representing 72.1% of the total variance of the dataset. Therefore, according to the Kaiser criterion, when analysing the behaviour of metals with the physico-chemical variables, the number of correlations can be reduced using these five PCs. The representation of the variance and the affinity of each PC with heavy metals is shown in
Table 6. It can be observed that PC1 has a higher representation of the dataset variance (21.75%), and thus encompasses the highest number of heavy metals (HMs). Probably, the HMs of PC1 (Sb, Cd, Sr, Co, Cu, and Fe) are quite strongly related to mining activity in the study zone, since they share characteristics of formation or certain behaviours in correspondence with environmental variables. In fact,
Table 6 also shows the correlation with the EC, temperature, and pH of each PC, and PC1 is moderately related to EC (r = 0.63 and
p-value = 0.005). It means that the weathering and dissolution processes have a significant impact on these HMs [
69]. Perhaps, during the activity of the mining industry in the study area, there was an uncontrolled disposal of mining wastes, and, as a result of the climate conditions, there were processes of dissolution which eventually caused the HMs to be filtrated into the groundwater. This suggests that these metals in solution may precipitate or adsorb onto the surface of precipitated hydrated sulphates during alteration processes [
10]. In contrast, Ref. [
70] obtained a PC1 with a strong correlation with Fe and Mn, and they suggested that it could be controlled by geogenic factors. This is interesting, because in our PC1 the Fe and Co have a negative correlation, which may mean that these metals have a different source of origin than the rest, probably geogenic. In this sense, PC2 had a positive correlation with Pb and negative with Sr, so it is possible that PC2 is influenced by mixed factors (natural and anthropogenic). PC3 and PC4 are clearly affected by mixed factors, because Cr and As are associated with anthropogenic origins such as metal processing industries, leaching of e-wastes, fossil fuel combustion, and industrial influents [
69], whereas Mn and Zn have natural origins from weathering and dissolution of parent rocks. On the other hand, PC5 showed strong correlation with solely Ni, which is inversely related to the pH level (r = −0.53 and
p-value = 0.037). So, PC5 can purely be attributed to anthropogenic sources such as vehicle emissions, industrial effluents, and landfill pollution [
70]. This negative correlation with pH can be explained by the precipitation of HMs at higher pH and high solubility at low pH [
69]. Ref. [
10] also obtained negative correlations between pH and some elements, and attributed it to an origin of the elements based on pyrite oxidation.
Additionally, PCA shows two different plots: the distribution of variables (HMs) (
Figure 5a) and the spatial distribution of sampling points (
Figure 5b). Based on the distribution of variables with respect to PC1 and PC2, HMs that appear closer together share similar behaviours and requirements. The further apart they are, the greater the difference between the elements. Therefore, at first glance, it can be observed that Cr, Ni, and Sb are grouped together, indicating that their formation or retention in the environment may be caused by the same factor. The same pattern can be observed with other groups of elements such as Mn, Fe, and Co, which are located on the opposite side of the plot, indicating that their behaviour in the environment is completely different from the previous group. This association highlights the important role played by mineral dissolution (Fe/Mn oxides and hydroxides) in water chemistry [
13]. On the other hand, the arrangement of Cu and Cd is virtually identical. According to [
14], these HMs are commonly originated from the leaching of phosphate fertilizers in agricultural lands. In contrast
Figure 5a shows that Pb, Zn, and Sr do not share similarities with other HMs and are more isolated.
On the other hand,
Figure 5b groups the sample points in relation to their characteristics. The samples that appear closer together exhibit a greater similarity in their HM composition, and, therefore are more similar in their environmental characteristics. It can be observed that HCA automatically differentiates between four distinct groups based on the sampling points. It is true that some sampling points coincide in the same group due to their proximity, such as F4, F5, and F6, indicating that in that area HMs are influenced by the same factors. However, other sampling points are close on the factor map but do not coincide in reality. This may be because different areas are affected by the same factors. Ref. [
10] also conducted a factor map with their sampling points, and the dispersion of their points depended on the content of jarosite and carbonates in the soils, so analysing these compounds for future investigations could be interesting.
Table 7 shows the means of each HM in relation to the four cluster groups. Firstly, it can be observed that the first group consists of only two sampling points (P1 and F19), which are characterized by high concentrations of Fe and Co, as well as relatively low concentrations of Sr compared to the other groups. Secondly, the second group is the largest one, as it has the highest number of samples (F1, F2, F3, F7, F8, F9, F11, F12, F13, F16, F17, F18, F20 F21, F22, F23, and P2), and does not stand out for having extraordinary concentrations of any HM within the scope of this study. Therefore, it can be inferred that the mean concentrations exhibited by this group would be considered normal for the study area. On the other hand, the third group consists of four sampling points (F4, F5, F6, and F10) and is characterized by high mean concentrations of Sr, Zn, and Cr. Lastly, the fourth group consists of a single sampling point (F15) that exhibits a high concentration of Cu compared to the others. Ref. [
14] also obtained the same result and attributed it to the application of fungicides or algaecides in agricultural lands. Taking everything into consideration, these findings contribute to a more nuanced understanding of heavy metal variations in the study area, shedding light on the complex interplay between natural geological factors and anthropogenic activities.