Introduction

Fluoride contamination poses a serious concern on human health in regions were groundwater serves as the principal source of drinking water (Sajil Kumar 2014). Countries across the globe like Egypt, Libya, Syria Algeria, Jordan Sudan, Kenya, Turkey, Iraq, Iran, Afghanistan, India, Northern Thailand, America, and China are suffering from the ill effects of fluoride contamination in groundwater (Brindha and Elango 2011). It is clear that all these countries are largely depending on groundwater for drinking and domestic purposes. The permissible level of fluoride, in drinking water, recommended by the World Health Organization (WHO) is 1.5 mg/L. However, each and every country has its own standards set on the basis of regional climate and per capita water intake.

In India, groundwater is the major source of drinking water and this is the reason for its increased and adverse impacts on human health. India is well known for groundwater fluoride contamination and the associated health risks such as dental and skeletal fluorosis (Jolly et al. 1968). In India, most of the village people depend on their own dug/bore wells for daily water needs. The major issue is that there are no adequate water quality assessment programs in the rural areas. Moreover, the severity of the water-borne deceases will be high when identified in the last stages. In India, water quality standards are set and maintained by Indian standards and the recommended level of fluoride is 0.6 to 1.2 mg/L (BIS 1992). A certain amount of fluoride in drinking water is considered to be useful for dental health in some parts of the world. On the other hand, high intake of the same may cause serious health effects (WHO 2011).

Fluoride contamination in the groundwater is a natural hazard. The origin of fluoride is mostly from the geological formations like charnockites, granite, gneisses, etc. with substantial amount of fluoride-bearing minerals such as fluorite, apatite, mica, amphiboles, clay, and villiaumit (Apambire et al. 1997). Hadrochemical processes in the subsurface causes the mobilization of fluoride which results in an increase in F concentration in the groundwater. Consumption of water per day by the human being depends mostly on the loss of water through evaporation from the body. Tropical arid countries will have more water consumption than those of cold countries. This causes a notable difference in the intake of fluoride from case to case.

As the intake of fluoride varies in different geographical regions, it is very important to assess a background level of this ion in areas as a benchmark for future studies focusing on further contamination of the aquifer. In this context, estimation of fluoride levels in the groundwater and setting up of a background concentration are important for the water supply schemes depending mainly on groundwater. This can be done effectively by determining the natural background level (NBL). The estimation of NBL was carried out by many researchers in different parts of the world (Wendland et al. 2005; Walter 2008; Molinari et al. 2012). Most of the fluoride studies in India are focused on the normal estimation of the concentrations, geochemical evaluations, assessment of health impacts, potential of treatment, etc. (Sajil Kumar et al. 2014; Subba Rao 2011; Rafiquea et al. 2009; Kim et al. 2011; Gwala et al. 2014; Gogoi et al. 2015). However, there were no studies reported in literature for the estimation of a natural background level of fluoride contamination. With this aim, we employed hydrochemical and statistical methods for establishing the NBL of fluoride in Vellore District, Tamil Nadu. This must be a stepping stone for further research on fluoride contamination in the region.

Study area

Physiographical classification of Vellore District can be mainly identified as a hilly terrain in the E and SE parts and plain regions in the eastern part. The hilly terrain forms a part of the Eastern Ghats. Geology of the region comprises mainly of hard rocks of Archean age and sedimentary formations of recent age. Hard rock formations comprise mostly of charnockites, gneisses, and granites (Fig. 1). In these formations, groundwater occurs mainly in fractures and lineaments (CGWB 2009). Predominant trend of lineament is NE-SW, whereas in the central part, both NE-SW and NW-SE trending lineaments have been identified in the remotely sensed data. Schematic of the stratigraphic succession of study area is shown in Fig. 2.

Fig. 1
figure 1

Location map of the study are showing geology and sampling points

Full size image
Fig. 2
figure 2

Stratigraphic succession of the study area

Full size image

Soils are mainly sandy soil, sandy loam, red loam, clay, clayey loam, and black cotton soil. The red loamy soil is generally observed at the highest elevations whereas the black cotton soil is present in the valleys. Other types of soils are found at intermediate elevations. Geomorphology of the study mainly shows structural hills, residual hills, plateaus, valley fills, pediments, buried pediments, pediplains, and alluvial plains. The major source of groundwater replenishment in the study is monsoon rain from southwest and northeast monsoons. The contribution of these two seasons is 45 to 52% and 30–43% respectively towards the total annual rainfall of 949.8 mm. Summer is very hot and the temperature rises up to 37 °C during May and June. The mean daily minimum and maximum temperature are 18.2 to 36.8 °C. The relative humidity ranges from 37 to 85%. Groundwater occurs in crystalline aquifers and abstraction method is dug wells of depth ranging from 8 to 19.5 mbgl (CGWB 2009).

Materials and methods

Sampling and analytical procedures

A total of 68 groundwater samples were collected from the study area during June 2010. A strict reconnaissance survey was carried out and representative samples were selected. Location of the samples was marked using a handheld GPS (HC Gramin). Wells were pumped out till the in situ parameters were alleviated. Electrical conductivity (EC) and pH were measured on site with digital meters. Alkalinity was measured by titration with 0.02 N H2SO4 prior to the groundwater sampling. Groundwater was collected in polyethylene bottles (1 L capacity); bottles were sealed and brought to the laboratory for analysis and stored properly (4 °C) before analysis. Laboratory analysis was carried out as per the standard methods suggested by APHA (1995). Major ions like Ca, Mg, Na, K, Cl, SO4, NO3, and F were analyzed. Ca and Mg were analyzed using titration with EDTA. Cl concentration was determined using argentometric titration. UV visible spectrophotometer was used for analysis of sulfate. Sodium and potassium were analyzed using flame photometer. Fluoride concentration of water samples was determined using SPADNS method (APHA, AWWA, WEF 1998). The minimum detectable level of the SPADNS method was 0.02 mg/L. Nitrate was measured using an ion chromatograph. The analytical precision of the measurements of cations and anions is indicated by the ionic balance error, which has been computed on the basis of ions expressed in milliequivalent per liter (meq/L). The values were observed to be within the standard limit of ± 5%.

Multivariate analysis

Statistical analysis such as principal component analysis and geostatistical analysis were carried out using XLSTAT and Arc GIS 9.3, respectively. Principle component analysis was used to understand the correlation between each chemical parameter and to identify the factors responsible for the geochemical variations in groundwater. In the first step, data was standardized and correlation matrix was generated. In the next step, based on the correlation values, principal components were calculated. Among the total generated components, the significant complements (those with more variation in data) were selected based on the Eigenvalues (> 1). In the last step, by varimax rotation, factor loading of individual variable on one of the extracted factors was maximized and all remaining factors were minimized. With the rotated components, variables were grouped according to the hydrogeochemical condition of the study area.

Pre-selection and cumulative probability plots

In order to evaluate the background level of ions, we adopted cumulative probability plots (CPP), which was developed for the geochemical exploration (Sinclair 1974) and then used by the researchers in water resources (Park et al. 2005; Panno et al. 2006). In this method, maximum curvature of any two lines indicates the inflection points of two populations. The assumption is that values of log- or normal distributed populations will form a straight line. Number of inflection points is decisive of single or mixed populations, i.e., natural or anthropogenic or both.

In this study, the major focus is on determining the natural background level estimation of fluoride. As the major origin of this ion is from natural sources, we wanted to omit anthropogenic-influenced samples. All the samples were preselected based on a modified criterion suggested by Wendland et al. (2005) with slight modification. We adopted a criterion to distinguish the natural groundwater as,

  1. (i)

    TDS > 1000 mg/L; TDS represent overall water quality and the permissible limit is 1000 mg/L.

  2. (ii)

    Cl and NO 3  > NBL (in this study); Cl and NO3 are naturally unavailable in the study area and their presence in the groundwater is often indicative of direct or indirect human interventions

After the application of these two conditions, groundwater is expected to be free from anthropogenic inputs, which can rightly be considered as natural water for estimation of NBL of fluoride.

Results and discussion

Major ion chemical characteristics

General expressions of the hydrogeochemical data of 68 groundwater samples are presented in Table 1. Total dissolved solids (TDS) can give a holistic value that can directly represent water quality of the area. In this study, TDS values range from 388 to 2686 mg/L (Avg. = 1059 mg/L). This shows that the groundwater in the study area is mostly contaminated with various ions. Among the major cations, sodium showed a range of 15 to 621 mg/L (Avg. = 204 mg/L). Considering the different origins of sodium in groundwater of Vellore District, natural and anthropogenic (predominantly tannery industries) can be found. Ions such as calcium and magnesium showed a range 18 to 240 mg/L (Avg. = 77 mg/L) and 24 to 199 mg/L (Avg. = 71.49 mg/L), respectively. Major anions bicarbonate, chloride, sulfate, and nitrate were in the range of 238 to 946 mg/L (Avg. = 474 mg/L), 25 to 1170 mg/L (Avg. = 274 mg/L), 7 to 394 mg/L, (Avg. = 251 mg/L), and 0 to 46 mg/L (Avg. = 11 mg/L). Very high concentration of Cl and SO4 ions in the groundwater is an indication of pollution from the tannery effluents (Thangarajan 1999).

Table 1 Hydrogeochemical parameters and water types of individual groundwater samples in the study area with corresponding geographical coordinates
Full size table

Fluoride geochemistry

This study which primarily focuses on fluoride, its concentrations, and the background levels will be discussed in detail. Groundwater has a concentration of F ranging from 0.2 to 3 mg/L with an average of 0.8 mg/L. In total, 35% of the samples have exceeded the permissible limit of 1.2 mg/L (BIS 1992). Certain amount of fluoride is good for dental and skeletal health. Thus, the acceptable limit is between 0.6 and 1.2 mg/L. Origin of fluoride in groundwater is mostly from geogenic sources, though some exceptional anthropogenic sources were reported in the literature. Mostly minerals like fluorite, apatite, mica, amphiboles, clay, and villuamite are rich in fluoride. Geology of the study area shows that crystalline rocks such as charnockite and gneisses are acting as the source for fluoride. The major controlling factors of fluoride geochemical dynamics in groundwater are solubility of F minerals, pH, temperature, anion exchange capacity of aquifer materials, type of geological materials and residence time, porosity, structures, groundwater age depth, concentration of carbonates, and bicarbonates in water (Chandra et al. 1981; Apambire et al. 1997; Sajil Kumar et al. 2014). Application of fertilizers and industrial activities can also increase the fluoride concentration in groundwater (Brindha and Elango 2011).

Geochemical facies

Geochemical facies are important indicators of the hydrogeochemical conditions of an aquifer and also playing a vital role in the mobilization of fluoride in groundwater. Two distinct water types were observed, i.e., Na-Ca-Mg-HCO3-Cl types and Ca-Na-Mg-Cl-HCO3 types (Fig. 3). The first type is favorable for fluoride enrichment in groundwater. It is frequently reported that high Na/Ca with elevated HCO3 concentration is determining factor of the groundwater fluoride concentration in F-rich aquifers (Reddy et al. 2010; Grützmacher et al. 2013). In case of Na-rich Ca-poor groundwater, precipitation of fluoride as calcium fluoride (CaF2), which is the most important controlling factor of the fluoride concentration in groundwater, will not occur. Moreover, Na-rich water will mobilize fluoride from the aquifer materials. Majority of the samples corresponding to the first group will be favoring this kind of hydrogeochemical conditions.

Fig. 3
figure 3

Piper plot showing water types in the groundwater

Full size image

Multivariate analysis of groundwater chemical data

All groundwater samples (n = 68) were analyzed and correlation coefficient among each parameter was obtained (see Table 2). Electrical conductivity showed very good positive correlation with TDS, TH, Ca, Mg, Na, Cl, SO4, and NO3, indicating that all these parameters are strongly contributing towards the total hydrochemistry of groundwater. Very high correlation for TH (total hardness), Ca (r2 = 0.75), and Mg (r2 = 0.85) clearly shows that temporary hardness in groundwater is mainly due to the increase of these ions. Significant positive correlation is observed from Na with Cl (r2 = 0.81) and SO4 (r2 = 0.75) showing the salinity origin from tannery effluents. Correlation of F with the other parameters was not so significant with any other parameters than pH (r2 = 0.35) and Mg (r2 = 0.30). This result shows that groundwater is considerably biased to the anthropogenic influences, like tannery pollution, agricultural activities and municipal sewage etc. Hence, we have done a more detailed study with the factor analysis.

Table 2 Correlation analysis of the groundwater chemical data (n = 68)
Full size table

Factor analysis was performed and the principal components were derived for the entire population. Though there are several components reported, significant components were chosen based on the eigenvalues (see Fig. 4). Four factors were found to have an Eigenvalue more than 1, and they were considered for the further analysis.

Fig. 4
figure 4

Eigenvalues obtained from the factor analysis

Full size image

All the four components were critically analyzed based on the individual factor loadings of the parameters. These four components effectively explained 80.30% of the total variance in the population and individually PCA1 = 42.38%, PCA2 = 13.08%, PCA3 = 13.01%, and 7.82%, respectively. Table 3 shows the results of PCA of groundwater samples.

Table 3 Results of factor analysis of the groundwater samples
Full size table

Factor shows a strong loading for EC (0.99), TDS (0.99), TH (0.89) Ca (0.68), Mg, (0.76) Na (0.85), HCO3 (0.56), Cl (0.91), SO4 (0.82), and NO3 (0.53). High factor loading for the parameters like Cl, SO4, Na, EC, TDS, etc. shows that the samples are affected by the anthropogenic influences like tannery industries (Sajil Kumar 2014). Additionally, presence of NO3, the first factor shows the agricultural and sewage pollution in groundwater. However, this factor has high loading for calcium and hardness, showing an impact of groundwater-aquifer interaction as well. Thus, this factor can be considered as affected by the anthropogenic as well as geogenic sources.

In factor 2, high loading was found for pH (0.65) and F (0.84), and a slightly significant loading for HCO3. Influence of pH and bicarbonate on the fluoride mobilization is well documented in the literature (Handa 1975; Jacks et al. 1993; Saxena and Ahmed 2001). All the three parameters listed in factor 2 have natural origin and represent natural processes that control the water chemistry. Factor 3 has high loading for K (0.77). This can be related to the K-feldspars in the gneissic formations in the study regions and also to the fertilizer usage in agricultural fields. Factor 4 could not show any significant factor loading that can be accounted to any specific geochemical process. The factor loading for the most prominent factors (1 and 2) is further presented in vector plot (Fig. 5).

Fig. 5
figure 5

Vector plot showing the factor loadings for factor 1 vs factor 2

Full size image

Natural background level estimation

Methodology for the pre-selection is presented in “Pre-selection and cumulative probability plots” section. With the criterion 1, groundwater samples were separated base on the TDS values, and 28 samples found to be exceeding the limit 1000 mg/L (fresh drinking water standard based on WHO 2011). Remaining 40 samples were used in the further analysis. TDS is representing the mixture of natural and human impacted groundwaters. By separating highly impacted samples, further analysis will give more reliable results.

NBL values were estimated for Cl and NO3 based on the remaining 40 samples. Frequency distribution for Cl concentrations was created and the values were converted into logarithmic scales. Log converted values were plotted against the cumulative probability percentages and the probability plots (CPP) were generated. Same methodology followed in the case of nitrates. In the CPP, inflection point indicated the threshold values. Calculated intersection points were 200 and 10 mg/L, for Cl and NO3, respectively. Chloride and nitrate were not available in natural groundwater of the study area and found to be sourced from tanneries and agricultural/sewage sources, respectively. We omitted those samples that crossed NBL values of these ions and the remaining 24 samples were used in the NBL estimation of fluoride. Figures 6 and 7 show the probability plots used in the estimation of NBL (Cl) and NBL (NO3), respectively.

Fig. 6
figure 6

Cumulative probability plots showing the NBL of Cl

Full size image
Fig. 7
figure 7

Cumulative probability plots showing the NBL of NO3

Full size image

After the NBL values were fixed for Cl and nitrate, we have plotted TDS vs Cl (Fig. 8), and TDS vs NO3 (Fig. 9) graphs. NBL values derived in the previous step was marked in the plot so that the natural samples can be selected as those below 1000 mg/L (TDS), 10 mg/L (NO3), and 200 mg/L (Cl). By combining the results of Figs. 9 and 10, a total of 22 groundwater samples, with concentrations TDS < 1000 mg/L, Cl < 200 mg/L, and NO3 (10 mg/L), were derived for the NBL estimation of fluoride. In case of NO3, similar values (10 mg/L) were reported by European Environmental Agency (EUROWATERNET-Groundwater 2002). In their classic work in Midwestern Unites states, Panno et al. (2006) reported 0 to 2.5 mg/L of nitrate (as nitrogen), which is comparably a small value. In case of chloride, a threshold value of 167 mg/L was reported by Wendland et al. (2008) from Upper Rhine Valley. (Kelly 2000) reported a background concentration of 10 to 20 mg/L, from northeastern Illinois.

Fig. 8
figure 8

Bivariate plot showing the relation between TDS and nitrate

Full size image
Fig. 9
figure 9

Bivariate plot showing the relation between TDS and chloride

Full size image
Fig. 10
figure 10

Cumulative probability plot showing the NBL of fluoride

Full size image

For the better visualization of samples, milligram per liter is converted to microgram per liter. After the pre-selection process, 22 groundwater samples were used in the analysis which showed a range of fluoride as 0.09 to 2.17 with an average of 0.78 mg/L. Based on this data, NBL was generated by CPP. As we have selected only fresh samples, no anthropogenic inputs were expected, and hence, only one inflection point (threshold value) was expected. From the inflection point, NBL was estimated as 0.75 mg/L (750 μg/L). This means that the normal expected concentration of F in the original groundwater in study area is ranging from 0.02 to 0.75 mg/L.

To understand the NBL of fluoride in the study area, total data set was classified into two categories, i.e., > NBL and ≤ NBL. Spatial distribution maps were created based on this classification (Fig. 11). It shows that majority of the samples in the western part of the study area exceeded NBL of fluoride, i.e., 0.75 mg/L. A cross check with aquifer chemistry of rocks suggested that the presence of F-rich minerals were present in the granitic-gneissic rocks in the study area. As it was already identified through multivariate analysis, a better visualization of the influencing processes was possible by spatial mapping of the factor scores (Figs. 12 and 13). It revealed that F1 was distributed in the eastern and northeastern parts, where F concentration was below the natural NBL. On the other hand, in the southern and southwestern parts, F2 has high loadings. This confirms the distribution of fluoride concentration within the study area and its correlative behavior with the natural/anthropogenic geochemical interventions.

Fig. 11
figure 11

Spatial distribution of fluoride concentration in groundwater based on the NBL analysis

Full size image
Fig. 12
figure 12

Spatial distribution of factor 1 scores

Full size image
Fig. 13
figure 13

Spatial distribution of factor 2 scores

Full size image

Conclusions

Hydrogeochemical data from Vellore District is analyzed and a NBL of fluoride concentration in the groundwater is estimated. Geochemical methods, multivariate analysis, pre-selection method, and cumulative probability plots were employed in the study and found to be very effective in grounding the NBL of fluoride for this region. Component 1 of the factor analysis indicated the natural and anthropogenic influences, whereas factor 2 showed natural process with very high factor loadings for pH and F. This result goes well in line with the two major water types Na-Ca-Mg-HCO3-Cl types and Ca-Na-Mg-Cl-HCO3. Mobility of fluoride is found to be increased with the increase in Na, HCO3, pH and decrease in Ca values. For the NBL estimation, groundwater samples were preselected initially with TDS values and then with Cl and NO3 to eliminate the human affected samples. In total, 22 natural groundwater samples were selected from 68 samples and the cumulative probability plot showed that natural threshold value for fluoride in the study are is 0.75 mg/L. Based on the NBL values, 29 groundwater samples were found to be exceeding this limit. Spatial variation mapping of NBL and factor scores showed that the Southern and Southwestern regions of the study area were more impacted with fluoride contamination. This observation is supported by fluoride-rich minerals in the hard rocks of this region. This study established the present natural background level of fluoride for the first time in the study area, which will be of great use in the future management of fluoride contamination, especially with human health perspectives. We suggest to use the research methodology in other regions of the country to make a benchmark for the contamination status. This will be helpful in the future water management strategies.