The hand-dug wells in the study area have an average of 4.2 meters depth, and 1.0 to 1.2 m diameters. They have a circular concrete lid, exception for wells P31 and P39 that have improvised lids with roof tiles and sheet metal (Alencar, 2021).
The domestic septic tanks area made by bricks and cement, out of the standards recommendations of the United States Public Health Service (1975). The depths of the septic tanks have an average of 1.8 m and distances between wells and nearest septic tanks vary from 8.6 to 50 m. Only two septic tanks did not comply with the minimum distance of 15 m between wells and septic tanks, as recommended by United States Environment Protection Agency. (Ananth et al., 2018; USEPA, 2002).
5.1. Groundwater Flow
The hydraulic heads measured in the sampled wells in the rainy season (Figure 2a) and dry season (Figure 2b) reveal small variations of hydraulic heads during the year.
The groundwater flow net is topography controlled (Figure 2) showing northeast and southeast flows, towards Atibaia river and to lower areas. Groundwater discharges to the Atibaia river (north area) and to an area with paleo meanders (south area). In this paleo meanders area, the groundwater is flowing from east to west area, also discharging in the Atibaia river.
5.2. Inorganic Parameters and Caffeine
The results of 11 physical-chemical parameters (pH, Eh, EC, Cl-, NH4+, NO2-, NO3-, K+, B e DOC) analyzed in a total of 40 samples, and the caffeine analyzed in 20 samples, are shown in Tables 1 and 2.
The pH varied from 4.94 to 7.41 in the rainy season, and from 5.63 to 7.57 in the dry season, corresponding to shallow conditions and a short residence time (Freeze and Cherry, 1979; Hem, 1985), probably associated with the decomposition of organic matter (Mokhtar et al., 2008).
During the wet season the highest values of pH were in the central and south portion of the area. In the dry season, the highest values were in the central and east sections. The lower pH values are situated mainly at west portion of the area in both periods (Figure 3).
The Eh varied from 150.83 to 348.91 mV in the wet season, and 166.06 to 381.53 mV in the dry season, suggesting a reduced and anoxic aquifer environment (Figure 4). According to Landon et al. (2011), the anoxic conditions in areas with shallow depths of water tables can occur due to the recent water recharges that contain a high amount of organic carbon as an electron donator, resulting from interactions of the shallow water table with soil.
The Eh maps (Figure 4) show the distribution of the lowest Eh values in the south-southeast portions of the study area in the two sampling periods.
The pH and Eh maps indicate the existence of two different reduced environments: (a) northeast to west area (low pH and higher Eh) – occurrence of natural levee, and (b) south portion (high pH and lower Eh) is a swampy area, presenting lower elevation of the alluvial plain, with abandoned meanders, outcropping water table and high contents of organic material.
Chloride varied from 1.02 to 34.4 mg/L in the rainy season, and 1.17 to 52.50 mg/L in dry season. Chloride is a traditional wastewater marker, but also be related to the contribution of meteoric water recharge (Ibrahim et al., 2019; Kawo and Karuppannan, 2018).
NH4+ was detected in four samples in the wet season (<0.05 to 4.98 mg/L), and in five samples during the dry season (<0.05 to 15.7 mg/L). No maximum values were proposed by World Health Organization (WHO, 2017). The European Union, however, established 5 mg/L as the maximum value permitted for NH4+ (Di Lorenzo et al., 2014); thus groundwater sample P04 in the rainy period, and P04 and P31 in the dry season show values of NH4+ above the permitted value. Nitrite (NO2-) was detected in only two samples during the rainy season and three samples during the dry season, below the maximum permitted value of 3 mg/L (WHO, 2017).
Nitrate (NO3-) varied from <0.04 to 60.40 mg/L during the wet season, and <0.04 to 42.60 mg/L in the dry season. NO3- has been widely used as a contamination indicator of groundwaters in different studies (Huan et al., 2020; Lapworth et al., 2017; Samatya et al., 2006). In three wells of the study area the contents of NO3- were close or above the maximum value permitted for potability (50 mg/L or 10 mg/L N-NO3-, WHO, 2017); and 4 wells presented nitrate contents above 20 mg/L (5mg/L of N-NO3-) indicating anthropic contamination (WHO, 2017).
Higher nitrate concentrations were detected in P46, P29, P14 and P11 in the wet season and P31, P29 and P10 in the dry season, indicating a contamination source. The abandoned meander water sample presented lower concentration of NO3- , and it can be associated with the loss of some total nitrogen by denitrification or assimilation with organic nitrogen, in reducing environment (Schaider et al., 2016).
Potassium (K+) varied from 0.75 to 16.01 mg/L in the wet season, and from 0.77 to 9.85 mg/L in the dry season. In natural waters, it has a strong tendency to be reincorporated especially in certain clay minerals (Hem, 1985). The presence of potassium in domestic effluent can be a result of consumption of foods or from cleaning and disinfecting products (Arienzo et al., 2009).
Dissolved Organic Carbon varied from 0.63 to 7.7 mg/L in the wet season, and from 0.55 to 9.40 mg/L in the dry season, indicating an occurrence of dilution in the wet season.
Boron varied from 2.6 to 24.9 μg/L in the wet season, and 0.06 to 17.63 μg/L. Boron can be authigenic, or can come from soaps and detergents wastewater due to its conservative propriety in groundwater (Barber et al., 1988; Schaider et al., 2014; Schreiber and Mitch, 2006).
Caffeine was detected (Detection Limit = 0.23 μg/L) in four samples of the wet season (P07, P12, P46 and Meander) and in three samples of the dry season (P07, P46 and meander). Moreover, only the concentration of the water sample from the abandoned meander reached the quantification limit (QL = 0.7 μg/L) showing values of 0.87 μg/L and 1.16 μg/L for rainy and dry seasons, respectively. This lowest contents of caffeine during the wet season can be associated with the dilution process.
The presence of caffeine in groundwater samples shows an infiltration of residual domestic waters (Edwards et al., 2019). Most sampling points with detectable caffeine levels have relatively low values of nitrate, and higher values of K+, B and DOC, and are found in the west and southeast regions of the study area.
The absence of caffeine in the majority of the monitored domestic wells is consistent with the results obtained by Godfrey et al, (2007); Nitka et al., (2019); Schaider et al., (2014); Seiler et al., (1999), confirming the association of low levels of caffeine in groundwaters with a fast degradation, or in circumstances where the biodegradation is not significant (Knee et al., 2010). Godfrey et al., (2007) suggest that the physical (sorption) and biological processes (microbial degradation) active in the vadose zones are responsible for the absence or lower concentration of caffeine and PPCPs in shallow groundwaters.
Swartz et al. (2006) studies showed preferential loses of caffeine along the most oxic flow lines. The intensive degradation of caffeine can occur inside an anaerobic septic system or in an environment with aerobic lixiviation (Seiler et al., 1999).
Albaiges, Casado & Ventura (1986) add that the low values of caffeine can undergo dilution of leachates by the infiltration of non-polluted groundwaters, or simply removal by degradation or adsorption in the aquifer. Godfrey et al. (2007); Yang et al. (2017a) also attributed this low caffeine concentrations to the transport processes and transit time.
The Figure 5 (a,b) shows the spatial distribution of ions NO3- and caffeine. The higher concentrations of nitrate are evident in the west and east regions during the wet season and in the east area in the dry season, Chloride presented similar NO3- behavior.
Caffeine was not quantified in samples that had a high concentration of chloride and nitrate. However, caffeine was detected in the west portion area. The concentrations determined in the meander water suggest anthropogenic pollution.
Figure 6 (a, b) presents the spatial distributions of DOC and caffeine; K+ and B showed similar spatial distribution in wet and dry season, with higher values in the central and south region and lower values in the north area. Similarities in the spatial distribution of these parameters can indicate transport in the same relative speed (Barber et al., 1988). The interaction of caffeine with DOC in the vadose zone can influence its transport (Yang et al., 2017b).
5.3. Correlations and bi-variate plots between Inorganic Parameters
Tables 3 and 4 are shown the Pearson correlation matrixes with significance levels of 5% (p < 0.05) calculated for 19 samples of wet season and 20 samples of dry season. Parameters pH, Eh, EC, chloride (Cl-), nitrate (NO3-), Potassium (K+), DOC and Boron (B) were considered for this analysis. Caffeine was not included in the correlation matrix due to the low concentrations.
NO3- showed correlation with EC in the dry season, but no correlation was observed in the wet season (Table 3 and Figure 11a), due to dilution of groundwaters during the rainwater recharge. NO3- presented a positive correlation with Cl- in the dry season and wet season (Tables 3 and 4, Figure 7b), suggesting that both elements are being simultaneously added to groundwater, from domestic sewer sources (Ismail et al., 2020; Reddy, 2013). Although the nitrate values do not reach the potability standards in most of the samples and chloride seems to show low concentration values, the correlation between chloride and nitrate is significant. The septic systems offer nitrogen continuously to groundwaters, generating seasonal variations in the concentration of nitrate and in the mixing rate with groundwaters (Nitka et al., 2019). The high concentrations of NO3- detected in the area suggest that the septic systems are the main source of NO3- in groundwaters (Schaider et al., 2016).
The rate B/Cl- (Figure 7d) can be used for understanding the origin of B in water resources (Dotsika et al, 2006; Rodriguez-Espinosa et al., 2020). The slightly correlation with Cl- during dry season and no correlation with NO3- (tables 3 and 4), suggests origin of B from meteoric water and fresh water (Dotsika et al., 2006).
The positive correlations between B and K+ and K+ and DOC (Tables 3 and 4, and figures 8a and 8b) can be attributed to the partial dissolution of potassium feldspar and its association with authigenic boron (Rodriguez-Espinosa et al., 2020).
EC showed no correlation with DOC in the wet season and positive in the dry season (Table 4 and Figure 8d). The contents of DOC in relation to the meander area, rich in organic matter, are product of the biota living there.
5.4. Relationship between caffeine and other parameters.
Caffeine was detected in wells P07, P12, P46 and meander in the wet season, and in wells P07, P46 and meander in the dry season. They are situated in the west portion of the study area and have bad constructive characteristics, precarious septic tanks nearby and poor sanitation conditions in their surroundings. In the abandoned meander water during the dry season, the value of caffeine is higher than in wet season, and the Eh is lower.
It shows that the biodisponibility of caffeine is also widely controlled by sorption processes that are associated with the physicochemical properties of the contaminants, type of soil (Karnjanapiboonwong et al., 2010; Laws et al., 2011), and the volume of sewer discharge. Karnjanapiboonwong et al., (2010) suggests that the adsorption behavior of caffeine is hard to predict simply based in the type of sorbent, considering that caffeine possess a high-water solubility (2.16×104 mg/L a 20 oC; low Kow).
In relation of the four samples in discussion (P07, P12, P46 and Meander), the reducing environment and the presence of organic matter can favor the presence of caffeine in detectable levels (Schaider et al., 2016). On the other hand, this environment is rich in organic matter and bacteria and favors the biodegradation processes of caffeine. It stands out, however, that the tendencies are not conclusive due to the fact that concentrations were bellow analytical quantification limit for caffeine.
In all 20 monitored wells in the study area, nitrate showed lower contents in the four wells with detectable levels of caffeine (with the exception of P46, which showed nitrate values of 60.4 and 11.2 mg/L for wet and dry seasons respectively) and relatively higher values in the wells with non-detectable caffeine concentrations. The higher values of NO3- suggests the septic systems are the main source of NO3- in groundwaters of the study area (Schaider et al., 2016).
The presence of low levels of caffeine in the groundwaters containing high concentrations of nitrate does not discard the possibility that the residual domestic waters are a source of contamination. The rapid degradation of caffeine in groundwater can justify the absence of this substance in the aquifer. The detection of caffeine can indicate the aquifer recharge by residual domestic waters, even when nitrate is not present in the water (Seiler et al., 1999).
In this study, the low caffeine concentration (above the quantification limit) in groundwater and water samples (abandoned meander and domestic wells) can be related to a (bio)degradation processes in presence of DOC concentrations, which is found in higher concentrations. The reduced environment is favorable to the caffeine presence in waters. However, the biodegradation of caffeine is fast in tropical climate alluvial plain, in environments like swamps and marbles, rich in organic material and bacterial flora.