Integrated assessment of groundwater quality beneath the rural area of R'mel, Northwest of Morocco

1 This is an Accepted Manuscript of an article published by Elsevier in Groundwater for Sustainable Development, Vol. 14, August 2021, Article number 100620, available at: https://doi.org/10.1016/j.gsd.2021.100620. Cop. Elsevier 2021. En idUS: Licencia Creative Commons CC BY-NC-ND Integrated assessment of groundwater quality beneath the rural area of R’mel, Northwest of Morocco Otmane Sarti a,*, Emilia Otal b, José Morillo b, Abdelhamid Ouassini a a Department of Chemical Engineering, University of Abdelmalek Essaadi, 90000 Tangier, Morocco; *email: [email protected] b Department of Chemical and Environmental Engineering, University of Seville, Camino de los Descubrimientos, s/n. 41092 Seville, Spain ABSTRACT This study evaluates the effect of anthropogenic activities on the quality of groundwater beneath a rural area (R’mel area) located in the perimeter of Loukkos, Northwest Morocco. Ten public wells for drinking water were sampled and characterized for physicochemical parameters, heavy metals and pesticides occurrence. Multivariate Statistical Analysis (MSA), Water Quality Index (WQI) and geographic information system (GIS) were employed to identify the sources of groundwater pollution, the suitability for drinking purposes and the groundwater quality distribution within the R’mel aquifer. The results showed that almost all the samples had concentrations of NO3- and Ca2+ exceeding the WHO limits set at 50 mg/L and 75 mg/L, respectively. The heavy metal content was very low in the groundwater except for arsenic which exceeded the WHO threshold of 10 µg/L in four locations. Nine pesticides including Atrazine, Trietazine, Terbutryn, Chlorpyriphos ethyl, Endosulfan sulphate, Endosulfan alpha, Endosulfan beta, Metribuzin and Metolachlor were detected in four sites. However, no pesticides exceeded the standard of 0.1 µg/L set by the European Union. The Principal Component Analysis (PCA) demonstrated that the R’mel groundwater was exposed to two types of pollution that depend on groundwater redox conditions, the first being defined 2 as the release of geogenic/anthropogenic As under reducing conditions (minor pollution), while the second represents the nonpoint source pollution from agriculture (major pollution). The Hierarchical Cluster Analysis (HCA) has classified the samples into four clusters reflecting the rate of groundwater mineralization in the study area. Differently to HCA, the WQI has generated three groups. The comparison between WQI and HCA showed that the combined use of the two methods may lead to a better classification of groundwater samples. The spatial distribution of NO3- and WQI indicated the areas affected by irrigated agriculture, and the impact of anthropogenic activities on the R’mel groundwater has been assessed. Keywords: Drinking water; Heavy metals; Hydrochemistry; Pesticides; Multivariate Statistical Analysis; Water Quality Index 1. Introduction Groundwater is the most valuable water resource in the Earth. At the present time, there are areas where groundwater is considered one of the main sources of water supply for drinking, industrial and irrigation purposes (Liu et al., 2015; Li et al., 2016; Chen et al., 2018). Notably in arid and semi-arid regions where the groundwater plays a crucial role as the main source of water supply (Chen et al., 2017).Groundwater is generally of adequate quality for both drinking and agricultural activities (Rivett et al., 2018). However, potential sources of groundwater pollution have increased in recent years (Danielopol et al., 2003). In fact, throughout the world, groundwater is heavily polluted by chemicals and the situation is even more alarming in developing countries where groundwater shows traces of nitrates, nitrites, chlorides, ammonium, heavy metals, pesticides, fecal coliforms and other pollutants. Consequently, understanding the quality of the groundwater to ensure a reliable supply for various purposes is an essential step in maintaining the sustainability of these resources (Chen et al., 2019). The agricultural operations in high vulnerability areas can have an impact 3 on the environment, economy and human health (Pacheco et al., 2018). Fertilizers and organic domestic waste might generate high levels of nitrates in the soil (Kumar et al., 2009). Since their high mobility, nitrates easily pass through the soil and reach the groundwater (Gutiérrez et al., 2018). Thus, although the groundwater contamination by nitrate in shallow aquifers in areas characterized by agriculture, well-drained soils and oxic geochemical conditions is a growing concern threatening the groundwater quality (Chen et al., 2016, 2017), its importance in evaluating groundwater pollution is likely to be crucial due to its widespread occurrence in most agricultural systems (Qian et al., 2020). The dispersion of metals in the groundwater may originate from anthropogenic sources (agriculture, industry, mining activities, etc) but can also occur naturally driving from parent materials of the aquifer (e.g. rocks and sediments) (Chowdhury et al., 2016). The differentiation process between natural background level and contamination values of metals in groundwater could in fact carry the statue of threshold (Marandi and Karro, 2008). Gao et al., (2020) reported that the knowledge of Natural Background Level (NBL) plays a key role in indentifying and controlling pollution in groundwater. The consumption of drinking water can pose a risk if the concentration of metals exceeds standards (Chaturvedi et al., 2018) due to the toxicity of heavy metals and their ability to accumulate in the human body (Herojeet et al., 2015). Pesticides are considered emerging contaminants that can affect groundwater quality because they are bioactive, and therefore influence human health. Their extended usage in agricultural areas imply a risk of leakage of these compounds into the groundwater through soil leaching (Rojas et al., 2015; Akay Demir et al., 2019). The soil, in turn, plays a very important role in the control of groundwater quality. It stands as a barrier that, depending on its nature, prevents or allows the intrusion of undesirable substances. The comprehension of the processes that control groundwater quality is still an intricate operation that challenges researchers, given the huge amount of data collected during a given study. Recently, a better understanding has been achieved using Multivariate Statistical Analysis, notably Principal Component Analysis (PCA), Factor Analysis (FA), Cluster Analysis (CA) and Discriminate Analysis (DA) (Barakat et al., 2016). Multivariate 4 Statistical techniques have become an effective tool for many hydro-geochemical studies (Cloutier et al., 2008). They can be used as unbiased methods to find natural associations between samples and/or variables (Wenning and Erickson, 1994; Vega et al., 1998; Helena et al., 2000). PCA was used in several studies to identify anthropogenic and natural factors influencing the groundwater quality (Omo-Irabor et al., 2008; Güler et al., 2012; Bodrud-Doza et al., 2016; Meng et al., 2018), whereas HCA was employed for the arrangement of groundwater samples into clusters based on the similarities of hydrochemical data. This technique is one of the most widely used clustering techniques in the field of Earth science (Cloutier et al., 2008; Ghesquière et al., 2015). The use of a technique such as GIS could indeed solve these problems to a certain degree, especially when combined with a mathematical model that would reflect a clear image of the quality of this groundwater. In Morocco, the estimated volume of useful groundwater is around 4· billion m3, extracted annually, mostly allocated to agriculture (86%), industry (5.5%), and human consumption (8.5%) (Laamari et al., 2011). The country is adopting irrigated agriculture as a base for human development, in order to meet the needs of its growing population and to increase its agricultural exports (Ait Kadi and Ziyad, 2018). The deterioration of soil and water quality in irrigated areas is one of the highest-ranked concerns for the Moroccan Ministry of Agriculture. The agricultural operations in rural areas are mainly linked to the exploitation of groundwater resources, which require monitoring to maintain sustainability (Masoud et al., 2018). The Loukkos region is one of the major irrigated areas in Morocco. Indeed, it concentrates nearly 80% of the national production of red fruits (strawberries, raspberries and blueberries), of which 65% are exported to the European Union (Harbouze et al., 2019). Despite this interesting development the Loukkos area is still suffering from a lack of data concerning the groundwater quality especially since the irrigated area is surrounded by villages which supply directly from this groundwater. In fact, the Loukkos area has recently undergone intensive agricultural activity (El Bakouri et al., 2007, 2008). As a result, a great number of fertilizers and pesticides have been recently introduced in order to intensify agricultural production and to improve crop yields. Moreover, the characteristics of 5 the R’mel aquifer, which are manifested by shallow groundwater and sandy soils, make the situation more complex. This paper investigated the quality of drinking groundwater underneath the R'mel area characterized by sandy soils, shallow depth, and intensive agricultural activities. The main objectives of this study are 1) to conduct an exhaustive analysis of potential pollutants affecting the R’mel aquifer 2) to assess the suitability of groundwater for drinking purposes using the Water Quality Index (WQI), as well as to perform a comparison between the WQI and Hierarchical Cluster Analysis (HCA) for a potential combined use to improve the water quality classification 3) to investigate main activities leading to groundwater deterioration in the study area using Principal Component Analysis (PCA), and lastly, 4) to generate water quality maps indicating the level of pollution in the R'mel aquifer with the help of Geographic Information System (GIS). This study could be very helpful for the scientific community and local authorities who are in charge of the safety and security of the local inhabitants of the R’mel region. 2. Study site 2.1. Description of the study area Located in Northwest Morocco between Tangier and Rabat, the Loukkos region covers an area of 2560 km2. This region has a Mediterranean climate, with a wet and cool season from November to April and a hot and dry season from May to October. The average annual temperature ranges between 11 ºC in winter and 25 ºC in summer, and the average annual rainfall is approximately 700 mm between October and April (El Bakouri et al., 2008). A large part of the Loukkos region is dedicated to agriculture due to its topographic and edaphological characteristics. The region has a useful agricultural area of 147,500 Ha, 45,000 Ha of which are irrigated. The mountainous part of the Loukkos area belongs to the Rif domain, which has a complex geology, distinct from other Moroccan geological domains. In these ancient 6 formations, recent units of Tertiary and Quaternary are deposited. Agricultural land is made up of the crumbling material of these geological formations. The R’mel region, with an average altitude of 50 m to 80 m, is essentially made up of sandy formations of the Plio- Quaternary, which provides soil with a light texture that is easy to work and very favorable for agriculture. 2.2. Land cover and agriculture The main types of soil, crops and agriculture in the Loukkos region are shown in Table 1 and presented on the cover map (Fig. 1). The R’mel area corresponding to the studied aquifer is characterized by a large irrigated zone (portions 3 and 4) in Fig. 1, which has sandy soil with coarse to very-coarse texture and cultivated for tea, watermelon, sugar cane, horticulture, cereal, groundnuts, fodder and citrus fruits (Ballouche, 2013), being the main crops that require pesticides in the study area strawberries, potatoes, tomatoes, citrus fruits, groundnuts, and sugar cane (Bakouri et al., 2007). The use of chemicals has an undeniable role in the development of intensive agriculture in the Loukkos region. The poor fertility of the R’mel soil requires an intensive intake of organic matter and fertilization (Ayadi-hajji, 2014). The main fertilizers used in the Loukkos land are NPK with a ratio of (14% N: 28% P2O5:14% K2O), urea 46%, ammonium nitrate 33.5%, and ammonium sulphate 21%, with an average use of 3000 t/y (Mourabit et al., 2002). Phytosanitary products have also helped to make this region one of the most important agricultural areas of Morocco due to the quality and quantity of its harvests (El Bakouri, 2006). 2.3. The R’mel groundwater The exploitable groundwater in the Loukkos region is divided into four aquifers, the R'mel aquifer (129 Mm3), the Drader-Souir aquifer (65 Mm3), the alluvial aquifer of Loukkos (54 Mm3), and the Ouled Ogbane aquifer (33 Mm3). The R’mel aquifer is situated in the southern part of Larache city, bordered by the Atlantic Ocean on the west and the Loukkos 7 River on the east. It covers an area of approximately 245 km2. The aquifer is mainly fed by the adjacent rivers, precipitations, and by the return of irrigation water from groundwater and/or surface sources. The R’mel groundwater circulates trough sandy-sandstone formations and fine sand dunes from plio-quaternary deposited on a substratum made up of Miopliocene blue marl. The soil characteristics in the R’mel aquifer (slight thickness with a sandy texture) makes it very vulnerable to the pollution by several types of contaminants, especially those derived from agriculture. The studied zone has likely been exposed to effects caused by inorganic fertilizers, pesticides, livestock, and some domestic activities. 3. Materials and methods 3.1. Sampling and chemical analysis In this study, sampling and analysis of groundwater at ten sites were conducted to investigate the quality of the R’mel aquifer (Fig. 2). The sampled public supply wells were destined specifically for domestic uses and a representative population relies principally on these resources to meet their domestics’ uses. All the samples were collected from the region of R’mel (R’mel aquifer). The level of the R’mel water table varies between 1.5 and 30 (m) representing a very low to relatively low depth in the southern (1.5 - 4 m), central (4 -9 m), and northern (9-15 m) sectors of the R’mel aquifer, respectively. While, near the coastal area and northwestern sector the depth below the land surface was between (15 and 30 m). The piezometric level was provided from the Loukkos Hydrological Basin Agency (ABHL). The analysis of the piezometric level of the R’MEL aquifer shows that the general flow occurs in two directions, one from the South to the North-East and the other toward the North-West (coastal zone). The piezometric level of the R’Mel aquifer reaches its maximum in the southern part of the aquifer (> 65 m). However, while the R’Mel aquifer has good hydrodynamic characteristics in the south-eastern sector resulting in a low hydraulic gradient, the south-western sector shows strong hydraulic gradient. In the coastal zone and to the north of the aquifer the level drops to reach -10 (m) which could explain the inversion of the flow direction due to the intense groundwater abstraction in this region (Fig. 2). A 8 considerable volume of groundwater was pumped just before sampling to avoid taking stagnating groundwater. Before their filling, the bottles were rinsed three times with the sampling waters. The temperature, pH and conductivity of the sampled groundwater were measured on the field using a CyberScan PC10 Multi-Parameter analyzer, while dissolved oxygen was determined using Dissolved Oxygen Meter Model 9500 analyzer. The sampling, transport and preservation were performed in accordance with Standard Methods for the Examination of Water and Wastewater (APHA, 2017). For the rest of water characterization, all samples were filtered using a filter with a pore diameter of 0.45 µm (Whatman, cellulose nitrate). Ion concentrations in groundwater such as nitrate, nitrite, phosphate, and ammonium were determined in accordance with Standard Methods for the Examination of Water and Wastewater (APHA, 2017) using PerkinElmer lambda 2 UV-Vis spectrophotometer. Chlorides were determined by the Argentometric Method (APHA, 2017). Whereas, Dissolved Organic Carbon (DOC) was analyzed using a Shimadzu TOC–VCSH analyzer, according to the manufacturer instructions, with a 5.0% coefficient of variation. Metals such Ca2+, Mg2+ and trace elements were analyzed by the direct injection of groundwater samples into an inductively coupled plasma-optical emission spectrometer (ICP-EOS) Agilent 5100. Five linear curves were prepared, and Relative Standard Deviation was calculated for each element. These analyses were carried out in duplicate or triplicate, from which the average was calculated. The analysis of pesticides in groundwater included 80 active ingredients of which (23 were organophosphorus, 34 were organochlorine and 23 were organonitrogen) (Table 2). For this reason, two methodologies were used for the analysis of pesticides. The first was used for the determination of semi-volatile organic compounds by Dispersive Liquid-Liquid Micro-Extraction (DLLME) and subsequent measurement was taken by the Gas Chromatograph Triple Quadrupole Mass Detector (Bruker Scion (GC‐436 + MSD TQ). This method is based on (EPA) Method 8270D revised July 5, 2014. The second method was used for the determination of organic pesticide residues by On-line Solid Phase Extraction- 9 Liquid Chromatography/Tandem Mass Spectrometry (On-line SPE-LC/MS/MS) using EVOQ triple Quadrupole for liquid chromatography (LC-TQ). This procedure allows the calibration of the Liquid Chromatograph/Mass-Mass for the subsequent identification, quantification and confirmation of the different pesticides by direct injection with the On-Line Solid Phase Extraction system (On-line SPE). Methodology based on EPA Method 543 revised March 2015. Extraction and analysis methods for each pesticide analyzed in the groundwater samples are listed in Table 2. All the solutions were prepared using ultrapure water (milli Q). All the analyses were performed at least in duplicate, from which the average was calculated. 3.2. Methods 3.2.1. Multivariate Statistical Analysis In this study, MSA was performed using IBM SPSS Statistics 22, and two different analyses were applied on the data set of 15 physicochemical parameters: EC, pH, Total Suspended solids (TSS), DOC, Cl-, PO43+, NO3- , NO2-, NH4+, Ca2+, Mg2+, Dissolved Oxygen (DO), As, Mn and Zn, to extract PCA, and HCA. Parameters such as (Cd, Co, Cr, Cu, Fe and Ni) were excluded from this study since they were not detected in most samples or were detected at negligible concentrations. PCA was generated by applying the orthogonal rotation factor of Varimax type using Kaiser Normalization (Kaiser, 1958). For HCA, Ward’s linkage method (Ward, 1963) with the Euclidian distance was applied as a measure of similarity in order to perform the cluster analysis (CA). Each cluster is based on a set of physicochemical parameters to find similarities between sampled locations. 3.2.2. Water Quality Index Nowadays, WQI has become a reliable tool for estimating the overall quality of groundwater (Adimalla et al., 2018; Adimalla and Qian, 2019). In most cases, among a group of physicochemical parameters analyzed in groundwater, some are within the allowed limits, but others are not, so the overall quality of groundwater is generally unknown. The WQI can summarize the set of physicochemical parameters into one single number easy to interpret 16 Correlation between physicochemical parameters serve to explain the relationship between parameters in a water sample, and identify the origin of dissolved compounds in water (Hamzaoui-Azaza et al., 2011).Table 10 shows the Pearson correlation coefficients between the physicochemical parameters. A significant correlation was observed between electrical conductivity (EC) and (NO3-, Cl-, and Ca2+) providing information about the principal parameters controlling the salinity of groundwater in the R’mel aquifer. A high positive correlation (r>0.8) was also observed between Cl- and the major cations Ca2+ and Mg2+. This significant relationships informs about chemical weathering and leaching of secondary salts in the groundwater (Prasanna et al., 2010). The reasonably good correlation between DOC and As-Mn (r>0.7) also suggests that the occurrence of As and Mn might be influenced at least partially by the concentration of DOC in the R’mel aquifer. Arsenic exhibits significant positive correlation with Mn (r = 0.62) which may indicates their common source in groundwater. Consequently, it is supposed that trace elements which are significantly correlated could be attributed to the same source of pollution (Mansouri et al., 2012; Towfiqul Islam et al., 2017). Despite the significant relationship between Ca2+ and NO3-, a low correlation was noted between NO3- and Mg2+. PO3-4 shows high negative correlation with Ca2+ and moderate correlation with Mg2+ and indicates that PO43- ions might decrease with the increase of hardness in the R’mel groundwater. The strong relationship (r=0.86) between NH4+ and NO2- could explain their weak concentrations in groundwater. Other significant correlations were also found between PO3-4 -pH, TSS-DOC, TSS-As and Cl -DOC. 4.3.2. Principal Component Analysis The output of the PCA allows the extraction of four rotated factors characterized by an eigenvalue greater than 1 and constitutes approximately 88.8% of the total explained variance. The results of PCA loading for the 15 physicochemical parameters are shown in Table 11. 17 The first component explains 28.31% of the total variance with high positive loading of DOC, TSS, Mn and As, and a moderate loading of Cl-. This component reports on the hydrogeochemical process leading to the release of As and Mn in groundwater. Indeed, several studies have attributed the release of As in groundwater to the presence of DOC (Anawar et al., 2002a; Halim et al., 2014). Anawar et al. (2002b) and Liu et al. (2003) reported that the high concentration of As in groundwater is mainly due to the high concentration of NH4+, DOC, and PO43- and inversely related to the concentrations of NO3-, DO, and SO4-, which reflects a reduced environment. Consequently, the mobilization of As and Mn in the R’mel groundwater could be due to the microbial reductive dissolution of Mn oxide (MnO), hydroxide Mn(OH)2, which could use the high level of dissolved organic matter (Anawar et al., 2002a, 2003; Liu et al., 2003; Halim et al., 2014). This reductive dissolution of MnO/Mn(OH)2 stimulates the release of As since it is adsorbed on MnO and Mn(OH)2 surfaces sediment (Zhang et al., 2020). The release of arsenic in the groundwater might be ascribed to two sources. The first being natural and mainly linked to rocks weathering and volcanic emissions, while the second represents the anthropogenic contribution and specifically linked to some industrial and agricultural activities (VITHANAGE et al., 2007). Grutzmacher et al.,( 2013) reported that in shallows aquifers the samples are commonly contaminated by both geogenic and anthropogenic sources. However, the geogenic contamination is defined as the exceeding of certain thresholds without the intervention of any human activity (Grutzmacher et al., 2013). This definition complicates the assertion of the source of As in the R'mel aquifer due to the large expansion of irrigated fields and the intensive use of fertilizers and agrochemicals. In addition, Bhattacharya et al., (2007) reported that despite the fact that As is often attributed to the geological setting, the contribution of anthropogenic activities should not be ignored. The second component (PC2) explains 23.49 % of variability and has a strong negative loading to PO43- and pH, and moderate loading to Ca2+ and Mg2+. This component represents the natural processes leading to the precipitation of calcium and magnesium carbonate 18 under alkaline conditions and suggests also that low PO3-4 concentrations could be due to the alkaline nature of groundwater samples. PC 3 represents 21.96 % of the total variability and is characterized by strong positive loading of EC, NO-3 and moderate loading of Ca2+ indicating the principal ions controlling the mineralization of groundwater in the study area. The high negative loading of DO toward this component could be an evidence for the stability of nitrates in the groundwater since all others inorganic nitrogenous forms were oxidized, thus causing an excess of nitrates in the groundwater. It has been reported that in the oxic groundwater nitrate is the most stable form (Ma et al., 2016; Hansen et al., 2017). On the contrary, nitrite can occur at concentrations comparable to those of nitrate under reducing conditions (Rivett et al., 2008). The moderate loading of Ca2+ in this component could explain the dominance of Ca2+ over Mg2+ in the groundwater samples and suggest that Ca2+ might have identical source such Nitrate in the groundwater. Generally, this component relates to the anthropogenic pollution caused by the intensive farming activities in the study area. Lastly, PC4 describes 15.03% of variability and has strong positive loadings to NH4+ and NO2- followed by a moderate loading of Zn. This last component illustrates the tendency of these compounds to be reduced under oxic groundwater conditions. This hypothesis is supported since the aquifer experiences high levels of NO3- ions. 4.3.3. Hierarchical Cluster Analysis The dendrogram generated from HCA shows four clusters (Figure 3). Cluster 1 contains a single sample (GW7). This sample is characterized by elevated concentrations of NO3-, Cl-, and Ca2+ reflects the highest dissolved valuesin the study area, the sample GW 7 is located in the central portion of the R’mel aquifer which is characterized by shallows depth and delineated by intensive farming activities. Cluster 2 encompasses two sub-clusters (GW6) and (GW1 and GW2) representing the central and northwestern portions of the R’mel aquifer, respectively. The water quality analysis in this area reveals higher dissolved values of Ca2+, NO3- and As, and displays pesticides traces. The cluster 2 tends to follow the same 19 characteristics of cluster 1 but with less sharpness. Cluster 3 also consists of two subclusters. The first is composed of (GW4 and GW9), while the second contains only GW8. This cluster represents both Northwestern and Southwestern parts of the R’mel aquifer displays the lowest nitrate concentrations and provides information on the region unaffected by agricultural practices. The placement of GW4 and GW9 in this cluster separately from GW8 was due to their high concentration of Ca2+, which likely occurs from the dissolution of bicarbonate rocks. GW8 was placed alone in cluster 3 as the cleanest sample since all physicochemical parameters, trace elements and pesticides were present at minor concentrations, and none exceedance of WHO standard was recorded. This sample was located away from agricultural fields and could reflect the natural background level of metals in the study area. Cluster 4 is formed by GW5, GW10, and GW3. These three sites have a low EC when compared to the other samples. However, high nitrate values (>100 mg/L) were noted, demonstrating also agricultural pollution. In general, the clustering results were based on the rate of groundwater mineralization and flow direction resulted in four classes in the following mineralization order: Cluster 1> Cluster 2> Cluster 4> Cluster 3. 4.3.4. Water Quality Index The calculated WQI score at each of the sampled sites are described in figure 4. The results show that the WQI range between 23.76 and 174.99. The sample GW8 contained water of an excellent quality and is therefore classified as suitable for consumption by the rural population. Samples GW3, GW4, GW5, GW9 and GW10 have scores ranging between 50 and 100 and were classified as good water for consumption. However, samples GW1, GW2, GW6 and GW7 were classified as poor water type as they exceed the score of 100, with a maximum value of 174 recorded at GW7. The highest values of WQI have been found to be specifically related to the highs concentrations of nitrates, calcium and to the arsenic values surpassing the standards in some locations. 4.3.5. Hierarchical Cluster Analysis and WQI 20 Figure 5 illustrates a comparison between HCA and WQI classifications. This comparison allows the evaluation of the similarity between the two methods. It has been noticed that the samples with the closest WQI scores were classified in the same cluster and then sub-cluster. By comparing the clusters 1 with 2 that include GW1, GW2, GW6, and GW7 it appears that the samples were classified according to their WQI results. For e.g. sample GW7 which is characterized by the highest WQI (174) was placed alone in cluster 1. While GW1, GW2 and GW6 which scored (102), (108) and (117) respectively, were classed in cluster 2. Within cluster 2, GW1 and GW2 have closest WQI and were grouped in the same sub-cluster separately from GW6. The samples GW4, GW9 and GW8 were characterized by the lowest WQI values (65), (63), and (26), respectively, and, in agreement, they were arranged within cluster 3. Sub-clustering of GW4 and GW9 separately from GW8 was also in agreement with their WQI values. Similarly, cluster 4 comprised GW3, GW5, and GW10, which had similar WQI scores (86), (85) and (80), respectively, although in this case HCA could not group GW3 and GW5 in the same sub-cluster different from that of GW10 because of the highly similar WQI values. On the other hand, the HCA analysis yielded four classes instead of three in the WQI classification which is normal since it depends on how the categories of water according to the WQI are established. The WQI of sample GW7 is clearly higher than the ones of GW1, GW2 and GW6 (Fig. 4). However, they are all within the same category (blue). All waters have a WQI above 100 in the same “poor quality water” category. Probably if there were another category of water quality for those having a WQI>150, for instance, GW7 would be in the “very poor-quality water” category. However, HCA is able to distinguish and separate them in two clusters because it establishes more categories. Despite this, both methods yielded similar results, and their merging may lead to a better interpretation of the groundwater quality rating. 4.3.6. Spatial distribution 21 The spatial distribution of nitrate in the R’mel aquifer showed that almost the entire area is affected by higher levels of NO3- that exceed the standard value of (WHO, 2011) (Figure 6). The contamination of the R’mel aquifer by nitrate is mainly due to the large expansion of irrigated fields surrounding the study area. The spatial distribution of WQI (Figure 7) indicates that the eastern region represented by GW8 has excellent water quality because it is not impacted by the agricultural activities of the region since it is located outside of the irrigated area. Hence, this area may reflect the original quality of the R’mel groundwater. Samples GW3, GW4, GW5, GW9 and GW10 representing both the southwestern and northeastern portions of the study area were classified as good water type. Despite this fact, the water quality of this area tends to worsen over time as they are close to the irrigated fields. However, a large central zone of the study area contains groundwater of poor quality and poses a serious risk to the health of the rural population, as it is directly subjected to the intensive agricultural activities. 5. Conclusions The large expansion of agricultural lands in the R’mel region constitutes a serious threat to the groundwater quality and therefore to its rural inhabitants. This study focuses on the assessment of selected pollutants that could affect the quality of drinking water in the R’mel area. The physicochemical characterization revealed that the R’mel aquifer was strongly affected by agricultural fertilizers and in some locations also showed the presence of As at concentrations exceeding the WHO standard. In the other hand, the chromatographic analyses of pesticides revealed the detection of nine pesticides in the groundwater samples distributed between four wells GW1, GW2, GW6 and GW9. Despite their detection in the study area, those pesticides do not exceed the Moroccan and European Commission recommended limits and therefore do not pose any danger to the health of consumers. In this study, PCA has helped to identify the principal sources of pollution in the R’mel aquifer. The extracted components demonstrate that the R’mel groundwater was generally 22 dominated by two types of pollution. The first one was minor and controlled by watersediment interaction and leaching of As under reducing conditions (PC1 and PC4). While the second report on a dominant pollution from anthropogenic origin mainly related to the intensive agricultural activities (PC3). HCA allowed the classification of groundwater samples into four clusters referring to aquifer mineralization. The results of the WQI in the study area were specifically dependent on NO3-, Ca2+, As and TH, and led to the following classification (GW8- excellent water), (GW3, GW4, GW5, GW9- good water) and (GW1, GW2, GW6, GW7- poor water). By comparing the results of WQI and HCA it was noticed that the two methods were consistent and have common points in their classification. This study has shown that merging and combining WQI and HCA can lead to a better understanding of the groundwater classification in future works. The spatial distribution map of nitrate indicates that almost the entire study area is affected by NO3- concentrations exceeding the WHO level of 50 mg/L which can pose serious health problems for the local inhabitants. Whereas, The spatial distribution of WQI revealed that the central area of the R’mel aquifer characterized by intensive irrigation farms was heavily polluted by agricultural practices. In conclusion, the R'mel aquifer located in the Loukkos region suffers from a wide variety of pollutants mainly resulting from agricultural practices. Another point is that an excess of nitrate together with high TOC concentration and low dissolved oxygen may lead to enhance anaerobic bacterial activities leading to release of As and Mn from sediment to groundwater, which increase the health risk of the consumers. In this sense, it is recommended to monitor the use of this groundwater. At the same time, a depollution technique should be implemented in the study area to reduce the negative effects of intensive farming and ensure drinking water safety for the rural population of this region. 23 Acknowledgments The authors would like to thank the Loukkos Hydrological Basin Agency (ABHL) for their collaboration in sampling and providing technical assistance for this study. Likewise, the authors thank the farmers and the local inhabitants of the R'mel region who provided us with the necessary help for this study. This work was supported by the Oficina de Cooperación al Desarrollo, Universidad de Sevilla. 24 References Abbasnia, A., Alimohammadi, M., Mahvi, A.H., Nabizadeh, R., Yousefi, M., Mohammadi, A.A., Pasalari, H., Mirzabeigi, M., 2018. Assessment of groundwater quality and evaluation of scaling and corrosiveness potential of drinking water samples in villages of Chabahr city, Sistan and Baluchistan province in Iran. Data Br. 16, 182–192. https://doi.org/10.1016/j.dib.2017.11.003 Adimalla, N., Li, P., Venkatayogi, S., 2018. Hydrogeochemical Evaluation of Groundwater Quality for Drinking and Irrigation Purposes and Integrated Interpretation with Water Quality Index Studies. Environ. Process. 5, 363–383. https://doi.org/10.1007/s40710-018-0297-4 Adimalla, N., Qian, H., 2019. Groundwater quality evaluation using water quality index (WQI) for drinking purposes and human health risk (HHR) assessment in an agricultural region of Nanganur, south India. Ecotoxicol. Environ. Saf. 176, 153–161. https://doi.org/10.1016/j.ecoenv.2019.03.066 Ahoulé, D.G., Lalanne, F., Mendret, J., Brosillon, S., Maïga, A.H., 2015. Arsenic in African Waters: A Review. Water. Air. Soil Pollut. 226, 302. https://doi.org/10.1007/s11270-015-2558-4 Ait Kadi, M., Ziyad, A., 2018. Integrated Water Resources Management in Morocco. pp. 143–163. https://doi.org/10.1007/978-981-10-7913-9_6 Akay Demir, A.E., Dilek, F.B., Yetis, U., 2019. A new screening index for pesticides leachability to groundwater. J. Environ. Manage. 231, 1193–1202. https://doi.org/10.1016/j.jenvman.2018.11.007 Anawar, H.M., Akai, J., Komaki, K., Terao, H., Yoshioka, T., Ishizuka, T., Safiullah, S., Kato, K., 2003. Geochemical occurrence of arsenic in groundwater of Bangladesh: sources and mobilization processes. J. Geochemical Explor. 77, 109–131. https://doi.org/10.1016/S0375-6742(02)00273- X 25 Anawar, H.M., Akai, J., Mostofa, K.M.G., Safiullah, S., Tareq, S.M., 2002a. Arsenic poisoning in groundwater: Health risk and geochemical sources in Bangladesh. Environ. Int. https://doi.org/10.1016/S0160-4120(01)00116-7 Anawar, H.M., Komaki, K., Akai, J., Takada, J., Ishizuka, T., Takahashi, T., Yoshioka, T., Kato, K., 2002b. Diagenetic control on arsenic partitioning in sediments of the Meghna River delta, Bangladesh. Environ. Geol. https://doi.org/10.1007/s00254-001-0459-x APHA, 2017. Standard Methods for the Examination of Water and Wastewater , 23rd Edition, American Public Health Association, Washington, DC. Ayadi-hajji, H., 2014. Outils de gestion de la pollution phytosanitaire diffuse au niveau d’un territoire : cas d’application zone humide Ramsar de la Merja Zerga au Maroc. Badeenezhad, A., Tabatabaee, H.R., Nikbakht, H.A., Radfard, M., Abbasnia, A., Baghapour, M.A., Alhamd, M., 2020. Estimation of the groundwater quality index and investigation of the affecting factors their changes in Shiraz drinking groundwater, Iran. Groundw. Sustain. Dev. 11, 100435. https://doi.org/10.1016/j.gsd.2020.100435 Bakouri, H. El, Ouassini, A., Aguado, J.M., García, J.U., 2007. Endosulfan Sulfate Mobility in Soil Columns and Pesticide Pollution of Groundwater in Northwest Morocco. Water Environ. Res. 79, 2578–2584. https://doi.org/10.2175/106143007X184528 Ballouche, A., 2013. Contribution à l’histoire récente de la végétation du Bas-Loukkos (province de Larache, Maroc). Physio-Géo 67–82. https://doi.org/10.4000/physio-geo.3426 Barakat, A., El Baghdadi, M., Rais, J., Aghezzaf, B., Slassi, M., 2016. Assessment of spatial and seasonal water quality variation of Oum Er Rbia River (Morocco) using multivariate statistical techniques. Int. Soil Water Conserv. Res. 4, 284–292. https://doi.org/10.1016/j.iswcr.2016.11.002 32 Weaver, J.M.C., Cavé, L., Siep Talma, A., 2007. GROUNDWATER SAMPLING (Second Edition). A comprehensive guide for sampling methods. WRC Report No TT 303/07, 2007, pp 168. Wenning, R.J., Erickson, G.A., 1994. Interpretation and analysis of complex environmental data using chemometric methods. Trends Anal. Chem. 13, 446–457. https://doi.org/10.1016/0165- 9936(94)85026-7 WHO, 2011. Guidelines for drinking–water quality, 4th edn. World Health Organization, Geneva (2011). WHO, 2001. ARSENIC AND ARSENIC COMPUNDS (Environmental Health Criteria 224), 2nd ed., Geneva: World Health Organization, International Programme on Chemical Safety. Yidana, S.M., Yidana, A., 2010. Assessing water quality using water quality index and multivariate analysis. Environ. Earth Sci. 59, 1461–1473. https://doi.org/10.1007/s12665-009-0132-3 Zaidi, F.K., Mogren, S., Mukhopadhyay, M., Ibrahim, E., 2016. Evaluation of groundwater chemistry and its impact on drinking and irrigation water quality in the eastern part of the Central Arabian graben and trough system, Saudi Arabia. J. African Earth Sci. 120, 208–219. https://doi.org/10.1016/j.jafrearsci.2016.05.012 Zhang, Z., Xiao, C., Adeyeye, O., Yang, W., Liang, X., 2020. Source and mobilization mechanism of iron, manganese and arsenic in groundwater of Shuangliao City, Northeast China. Water (Switzerland). https://doi.org/10.3390/w12020534 33 Figures captions: Figure 1: Soil occupation in the Loukkos region Figure 2: Water sampling location Figure 3: Dendrogramme of groundwater samples based on physicochemical parameters (Ward method) Figure 4: WQI results in the study area Figure 5: schematic comparison and combination between HCA and WQI classifications Figure 6: Spatial distribution of nitrate in the R’mel aquifer Figure 7: Spatial distribution of WQI in the R’mel aquifer 34 Fig.1. Soil occupation in the Loukkos region 35 Fig. 2 water sampling locations and flow characteristics of the R’mel aquifer 36 Fig. 3. Dendrogramme of groundwater samples based on physicochemical parameters (Ward method) 37 Fig. 4. WQI results in the study area 38 Fig. 5. Comparison and combination between HCA and WQI classifications 39 Fig. 6. Spatial distribution of nitrate in the R’mel aquifer 40 Fig. 7. Spatial distribution of WQI in the R’mel aquifer 41 Table 1. Soil characteristics, agriculture mode and crops type in the Loukkos region. Type of development Soil type Soil texture (grain) Crop type irrigated area 1-Dehs fine to medium citrus sugar cane beet cereal 2-Tirs fine to ultra fine 3-R’mel coarse to very coarse watermelon sugar cane horticulture cereal fodder groundnuts 4-R’mel coarse to very coarse tea non-irrigated area (favorable) 5-Tirs fine to ultra fine sugar beet cereal sunflower melon legumes non-irrigated area (unfavorable) 6-Biad equilibrated to coarse cereal legumes olivier marginal area 7-R’mel coarse to very coarse cereal eucalyptus 48 Table 8. Results of organonitrogen pesticides residues in groundwater samples (ng/L) WELL 1. Ametryn 2. Atrazine 3. Atrazine-Desetil 4. Bromacil 5. Cimoxanil 6. Ciproconazole 7. Desetil Terbumetone 8.Terbutylazine-desethyl 9. Diflufenican 10. EPTC 11. Metamitrone 12. Metribuzine 13. Miclobutanil 14. Prometryne 15. Propazine 16. Simazine 17. Tebuconazole 18. Terbumeton 19. Terbutilazine 20. Terbutryn 21. Triallate 22. Trietazine 23. Trifluralin GW 1 - - - - - - - - - - - - - - - - - - - - - - - GW 2 - - - - - - - - - - - - - - - - - - - - - - - GW 3 - - - - - - - - - - - - - - - - - - - - - - - GW 4 - - - - - - - - - - - - - - - - - - - - - - - GW 5 - - - - - - - - - - - - - - - - - - - - - - - GW 6 - - - - - - - - - - - 41.1 - - - - - - - - - - - GW 7 - - - - - - - - - - - - - - - - - - - - - - - GW 8 - - - - - - - - - - - - - - - - - - - - - - - GW 9 - 12.1 - - - - - - - - - - - - - - - - - 5.5 - 12.7 - GW 10 - - - - - - - - - - - - - - - - - - - - - - - - Below detection limit 49 Table 9. Characteristics of the mass spectrum for active substances exceeding the limit of quantification in the R’mel aquifer Active ingredient CAS Number Retention time (min) Quantification ion Calibration equation Gamme de quatifica tion LoQ* (ng/L) Atrazine 1912-24-9 10.231 200 y= +0.0272x+7.657e-4 2.5-2500 Trietazine 1912-26-1 10.232 104 y= +0.0518x-0.0095 2.5-2500 Terbutryn 886-50-0 11.923 170 y= +0.322x+0.2589 2.5-2500 Chlorpyriphos ethyl 2921-88-2 12.237 258 y= +0.4346x+0.0083 2.5-2500 Endosulfan sulphate 1031-07-9 15.483 237 y= +0.1961x-0.0436 0.4-2500 Endosulfan alpha 959-98-8 13.752 160 y= +0.3095x-0.0237 0.4-2500 Endosulfan beta 33213-65-9 14.798 160 y= +0.2491x+0.0787 0.4-2500 Metribuzin 21087-64-9 11.402 82 y= +0.0699x-0.0152 2.5-2500 Metolachlor 51218-45-2 12.224 162 y= +0.3798x-0.0101 2.5-2500 *Limit of quantification 50 Table 10. Pearson correlation Matrix for physicochemical parameters (n=15) of groundwater samples (n=10) The significant correlations (alpha=0.05) are in bold face. Variables EC pH TSS DOC CLPO42- NO3- NO2- NH4+ Ca 2+ Mg2+ DO As Mn Zn EC 1 pH 0.13 1 TSS 0.03 -0.21 1 D.O.C 0.24 -0.02 0.65 1 CL- 0.68 -0.36 0.45 0.65 1 PO43- -0.16 0.7 -0.12 -0.23 -0.55 1 NO3- 0.62 -0.28 -0.44 -0.03 0.53 -0.61 1 NO2- 0.48 -0.14 0.35 0.32 0.53 -0.31 0.35 1 NH4+ 0.15 -0.07 0.39 0.37 0.3 -0.03 0.06 0.86 1 Ca 2+ 0.66 -0.45 0.16 0.49 0.88 -0.77 0.79 0.5 0.21 1 Mg2+ 0.35 -0.6 0.64 0.51 0.85 -0.62 0.25 0.52 0.3 0.67 1 DO -0.54 -0.46 0.45 0.02 -0.25 0.1 -0.6 -0.14 0.12 -0.38 0.1 1 As 0.17 -0.48 0.64 0.73 0.59 -0.25 -0.02 0.27 0.39 0.48 0.52 0.41 1 Mn 0.22 0.06 0.52 0.78 0.5 0.17 -0.2 0.3 0.42 0.19 0.42 0.2 0.62 1 Zn -0.03 0.52 -0.33 -0.53 -0.53 0.51 -0.11 0.31 0.38 -0.5 -0.5 -0.16 -0.61 -0.29 1 51 Table 11. Rotated factor loadings of principal components on physicochemical parameters: parameters Components PC 1 PC 2 PC 3 PC 4 EC 0.229 0.019 0.829 0.192 pH -0.062 -0.941 0.235 -0.019 TSS 0.777 0.164 -0.305 0.251 DOC 0.911 0.061 0.181 0.053 CL- 0.618 0.472 0.576 0.145 PO42- -0.020 -0.890 -0.278 -0.004 NO3- -0.247 0.475 0.798 0.087 NO2- 0.215 0.206 0.312 0.891 NH4+ 0.293 0.020 -0.013 0.915 Ca 2+ 0.324 0.629 0.687 0.102 Mg2+ 0.554 0.645 0.169 0.227 DO 0.237 0.221 -0.877 0.058 As 0.799 0.376 -0.099 0.086 Mn 0.874 -0.179 0.030 0.163 Zn -0.557 -0.538 -0.052 0.626 Variability (%) 28.31 23.49 21.96 15.03 Cumulative (%) 28.31 51.80 73.76 88.80