Lake Nokoué and the Chenal de Cotonou are located in the southern part of Benin under the influence of the cities of Cotonou, Abomey-Calavi, Porto-Novo, lakeside villages and continental contributions from the northern part of the country [4] . It is indeed the main receptacle for lacustrine discharges of effluents contaminated by metal residues from anthropogenic activities [5] [6] . Following a rapid increase in cottage industry and urbanization in recent years, this area is characterized by a galloping population. Figure 1 represents the study area composed of Lake Nokoué, the Channal of Cotonou and the city of Cotonou, in particular the 6th district. Table 1 and Table 2 present the sampling points identified on the lake and the channel.

NB. The missing coordinates are very close to the others, but these stations were sampled because of the observations we had made.

NB. The missing coordinates are very close to the others, but these stations were sampled because of the observations we had made.

The water samples were taken in the two ecosystems at 9 stations for the Chenal de Cotonou and Lake Nokoué. In order to deepen a better understanding of the high levels of metallic contaminants observed in certain areas during previous studies (Youssao et al. 2011a and b) [5] [7] , we have brought closer points, particularly in the middle of Lake Nokoué (ML 1 to 4) and in the Dantokpa market area (DAN 1 and 2; GBo 1 and 2) on the Cotonou channel. The water samples were taken from the surface after rinsing with medium water and stored properly (Youssao et al., 2018) [8] . Water samples were extracted by hot acid digestion [8] [9] [10] . To do this, 2 ml of ultra-pure HNO₃ and 1 ml of hydrogen peroxide were added to 2 ml of juice samples in a Teflon container, closed loosely to let the vapors escape. The entire system was placed in a programmable oven with two temperature levels (45˚ and 90˚). This characteristic of DigiPrep made it possible to gradually bring the samples to 45˚C for 20 min, then maintain it at this temperature for 40 min, then gradually bring it to 90˚C for 30 min and finally maintain it at this temperature level for 160 min. MRC certified water standards are digested under the same conditions; which allowed us to obtain good recovery rates for the 21 chemical elements recorded with recovery rates between 80% and 105% maximum except for zinc and cadmium with SRM SLRS-5 which respectively give 224.9% and 138.3%.

We have used hierarchical clustering on principal components to identify similarities between stations based on metal concentration with R software packages “FactoMineR” and “factoextra”. The method is indicated for multiple continuous variables to reduce the dimension of the data into few continuous variables containing the most important information in the data before performing the cluster analysis [11] [12] . Based on the principal component analysis, we reduced metals to three linearly uncorrelated variables that explains 84% of the variance in the metal data. And then, we identified three groups of metals: 1) macro elements (Mg and Ca), 2) micro-elements associated with the second cluster (Mn, Fe, Ni, Cu and Zn) and 3) trace elements (Mo, V, Co, Cr). We use multivariate analysis of variance (MANOVA) with the test of Pillai to identify whether there is significant difference between the four ecosystems (Lake Nokoue, channel of Cotonou, the Well water and the drilling) regarding to their concentration in each metal. We use pairwise comparison test of Games_Howell in the case that any significant difference is detected at α value set to 0.05. The test is based on Welch’s degrees of freedom correction and uses Tukey’s studentized range distribution for computing the p-values. Since the micro and trace elements contain more than two metals, we have chosen two uncorrelated metals among them to perform the Manova as the result for a metal can be extrapolated to its correlated metal.

The results of the analyzes indicate an unequal distribution of metals in the different ecosystems. However, analysis of the following tables (Table 3 and Table 4) shows atypical concentrations at certain points in the lake and the channel. Magnesium, calcium and manganese have very high maximum values in Lake Nokoué respectively at Ganvié market station GAN M (2990 ± 105 mg/L), Ganvié center, station GAN C (4991 ± 177 mg/L) and ML4 station (10662 ± 17.03 µg/L). On the other hand, iron, aluminum and strontium have very high concentrations in the Cotonou Channel respectively at Agbato station AGB (5236 ± 103

and 8289 ± 519 µg/L) and at the estuary station EST (6118 ± 68 µg/L). In general, we can conclude that these metals have an anthropogenic source except strontium which could come from marine waters. Particular on the site of Agbato, this atypical value of aluminum exceeds more than 25 times that of the WHO (2017) [13] which is 200 µg/L. This site must be particularly followed with regard to this toxic metal.

Water from traditional wells was sampled in the 6th district of Cotonou in the area bordering the Chenal de Cotonou and Lake Nokoué. The results of this work have been published [14] .

Table 5 indicate that the concentrations of metallic elements in surface waters are generally higher than those of well and borehole waters except for magnesium and calcium which are respectively more abundant in well waters (7950 mg/L and 2.73 µg/L) and drilling (2.82 µg/L Pb). These two chemical elements could have a source related to the geochemical background.

The characterization of groundwater and surface water and the search for trace elements in these resources have been the concerns of many researchers in recent years [15] [16] .

Moreover, these waters are also contaminated by metals such as aluminum and iron. These two metallic elements presented an atypical observation in the well waters sampled [14] [17] , but these concentrations are lower than those obtained in surface waters respectively at Agbato station AGB (5236 ± 103 and 8289 ± 519 µg/L). These compounds produce very unpleasant effects on the aesthetic level, such as stains on clothing, sanitary appliances and household appliances, but do not necessarily have an impact on health. Even if the consequences linked to the consumption of water containing a high level of aluminum have not yet been fully proven on human health and living aquatic organisms, it is advisable to monitor this element closely in drinking water. Drink and in ecosystems [14] . Dovonou (2014) [15] had recorded the total iron concentration of 0.5 mg/L and 40% of his samples have total iron contents exceeding 0.3 mgl/L in the Godomey wells. These values are lower than those recorded in surface waters. The toxic effects of this element appear only in case of absorption of very large quantities. If the diet is rich in calcium, the bioavailability of other absorbed minerals may be disturbed.

Magnesium concentrations ranging from 5 to 10 meq/L (6 to 12 mg/dL) can alter heart rhythm. Skeletal muscle paralysis, reduced lung capacity, coma and death may occur at magnesium concentrations of 15 meq/L (18 mg/dL) in plasma.

Dissolved manganese can form blackish particles in water and cause similar colored stains on appliances. High concentrations of dissolved iron in the wellbore can lead to the growth of iron bacteria. These bacteria can coat the inside of the casing or any other submerged part of the plumbing in the well and can cause problems.

The health effects of silver are: digestive disorders, gum line and bluish-gray discoloration of the skin. This dyschromia predominates on the face and uncovered parts, as well as on the nails (Rodier et al., 1984) [18] .

Aluminum causes Alzheimer’s disease and parkinsonian syndromes as well as slight cerebral atrophy [19] . Acute aluminum poisoning can cause lip and mouth ulceration [20] .