The sampling campaign was conducted in May 2021. A total of 36 samples were collected for Oxygen-18, Deuterium and Tritium analysis. Sixteen samples were collected from the River Niger and its main tributaries Alibori and Sota. Twenty samples were collected from groundwater. The samples were analyzed at the Laboratory of Radio-Analysis and Environment, University of Sfax, Tunisia. Two replicate samples were analyzed for δ¹⁸O and δ²H at the Laboratoire d’Hydrologie Appliquée, Université d’Abomey-Calavi, Benin. The stable isotopes of water (δ¹⁸O, δ²H) were measured using laser spectroscopy. The instrumental precision is approximately 1.5‰ for δ²H and 0.3‰ for δ¹⁸O. Rain samples were collected on a monthly basis (May to September 2021) for tritium ³H analysis. These rain samples were analyzed at Környezetanalitikai Laboratórium, Hungary.

The spatial distribution of the measured pH in the study area is presented in Figure 3 . pH values are mainly high in the aquifer discharge zones.

In the framework of this study, the software R 4.2.1, Rstudio 2022.12.0 + 353 and JAGS 4.3.1 were used to carry out data processing [12] - [14] .

The package MixSIAR [15] was used for Bayesian Mixing Modelling. The innovation of this study is that the model was applied to deuterium excess DEX [16] and tritium (³H) content to determine the relative contributions of each water source.

$\text{DEX}={\delta }^2\text{H}-8\times {\delta }^{18}\text{O}$ (1)

The package FactoMineR [17] was used for Principal Component Analysis (PCA) on a set of data composed of parameters with significant correlation (p-value < 0.05) assessed with the package metan [18] : In-situ field measurements (pH and Depth), water isotopes (¹⁸O, ²H, ³H), model output (mean surface water and groundwater percentage quantified for each of the sampling sites).

Figure 4 highlights signatures exhibited by stable isotopes of water in the study area. In rainwater, deuterium δ²H varies from −53.6‰ to 23.8‰ with a mean value of −16.77‰. Oxygen-18 δ¹⁸O varies from −8.03‰ to 1.93‰ with the mean value of −3.59‰. Tritium ³H varies from 1.8 T.U. to 5.2 T.U. with a mean value of 3.04 T.U.

In surface water, deuterium δ²H varies from −16.6‰ to 9.3‰ with a mean value of −2.41‰. Oxygen-18 δ¹⁸O varies from −2.86‰ to 4.12‰ with the mean value of 0.04‰. Tritium ³H varies from 0.8 T.U. to 3.83 T.U. with a mean value of 2.39 T.U.

In groundwater, deuterium δ²H varies from −41.2‰ to −20.6‰ with a mean value of −24.97‰. Oxygen-18 δ¹⁸O varies from −6.24‰ to −3.04‰ with the mean value of −4.14‰. Tritium ³H varies from 0.07 T.U. to 3.2 T.U. with a mean value of 1.77 T.U.

Four main groups of water have been identified ( Figure 4 ):

In order to have a better understanding of the identified groups of waters in the study area tritium (³H) was plotted against deuterium excess which revealed four different classes of water ( Figure 5 ). These classes are:

The comparative analysis of Figure 4 and Figure 5 shows that in the area of study, δ²H and δ¹⁸O are no longer enough to differentiate the water type, but their combination through the deuterium excess DEX together with their ³H content provides a clearer picture of the relationship within the aquifer system layers and between the aquifer system and the rivers.

The Bayesian Mixing modelling carried out confirmed the classes previously identified in Figure 5 . Moreover, based on the water type within each class, all of them except A have been divided in two subclasses. The average proportions of surface water and groundwater in the identified classes are presented in Table 1 and Figure 6 . Class A is composed of 95.47% of groundwater. For class B, groundwater (BG) is composed of 21.10% of surface water and rainwater, while surface water (BS; Sota River) is composed of 84.00% of groundwater. In class C, groundwater (CG) is composed of 73.77% of surface water/rain, while surface water (CS) is composed of 20.55% of groundwater. In the case of class D, groundwater (DG) is composed of 93.30% of surface water/rain, while surface water (DS; mainly Niger and Alibori rivers) is composed of 96.66% of surface water. Moreover, it is important to highlight that Class A’s surface water composition could be considered negligeable as well as subclass DS’s groundwater composition, because groundwater/surface water apportionment had been done based on them.

In Figure 7(a) , Pearson’s correlation coefficient was employed for in-situ parameters, isotope contents, groundwater and surface water proportions. pH, ²H, ¹⁸O and ³H have a positive correlation with each other. Water depth has a negative correlation with pH, ²H and ¹⁸O. ³H and %SW have a positive correlation but a negative correlation with %GW. Moreover, the Principal Component Analysis shows that the set of parameters selected can be used in the Kandi basin, to study surface water groundwater interactions. These parameters (mainly pH, ²H, ¹⁸O, ³H and water depth) explained about 84.86% of the processes involved in water fluxes exchange between surface water and groundwater in the study area ( Figure 7(b) ).

*GW = Groundwater & SW = Surface water.

Figure 8 . Shows a synthetic spatial overview of the surface water groundwater interactions in Kandi basin.

Surface water samples with the highest proportion of groundwater are from the Sota River (class BS; BK103, BK106, BK110, BK128, BK135). They are located in groundwater discharge areas. Moreover, groundwater samples (class DG; BK113, BK124, BK126, BK127) with the highest proportion of surface water are located in the discharge areas (northwest, around Alibori River). They are generally very shallow in that part of the basin (sampling depth less than 5 m). But it is important to highlight 3 groundwater samples (class A; BK105, BK111, BK117): even though they are located in discharge areas, their proportion of surface water is very negligible. Furthermore, they have the lowest tritium content (³H < 0.5 T.U.) and have among them the only artesian borehole identified in the basin with about 85 m depth. Also, it is important to highlight that tritium content doesn’t have a general spatial trend, this could be translating the complexity of the mechanisms involved in water flux exchange in the Kandi basin.

Previous studies in the Kandi basin indicated low mineralization of groundwater with less mineralized water in the recharge areas. Groundwater is mainly characterized by mixed water types. Moreover, recent groundwater recharge from modern precipitation, mixing between aquifers and recharge from surface water have influenced geochemical processes in the basin. Locations of probable groundwater discharge in the Sota river have been identified based on groundwater flow directions [9] [19] - [23] .

In this study, four main processes involved in groundwater-surface water interaction have emerged. They are:

The spatial locations of the occurrence of these processes suggest vertical hydraulic connections between deep and shallow aquifers, as well as complex hydraulic connections with aquifer systems and rivers. The detailed mapping of hydrodynamic properties of the studied aquifer system will enlighten more on the spatial extension of the identified processes, in order to quantify the groundwater reserves.

Having established the hydraulic connectivity between groundwater and surface water, it is imperative that detail assessment of the level of contamination of surface water and groundwater be carried out. It is also important to undertake in-depth assessment of groundwater vulnerability to pollution. This will enhance groundwater management in the study area.

The identified artesian borehole (in Madecali, extreme northeast) with high residence time must be monitored for abstraction to prevent groundwater mining. It is important to highlight that during the field campaign, the local communities have shown the location of another artesian borehole (located at Bodjecali, about 6 km westwards Madecali) that dried up. Therefore, in-depth studies are required to map the extent of this local artesian aquifer, in order to prevent it from drying up.

Overall, the findings improve the understanding of the processes governing surface water groundwater interactions in the Kandi basin.