Isotope signatures and hydrochemistry as tools in assessing groundwater occurrence and dynamics in a coastal arid aquifer

The La Paz aquifer (1275 km²) is located in the valley of La Paz and is part of the state of Baja California Sur (Fig. 1) (Cruz-Falcón 2007). The climate is predominantly arid with an average annual temperature and rainfall of 23 °C and 265 mm, respectively. Rainfall occurs primarily during summer (August and September) and is mostly associated with tropical storms and cyclones (CONAGUA 2005; Cruz-Falcón et al. 2011). The mean potential evapotranspiration is approximately 2015 mm (CONAGUA 2009). The total number of production wells and annual groundwater extraction in the La Paz aquifer is 155 and 32 million m³, respectively. Sixty-eight percent (68 %) of the extracted groundwater is used for the public urban water supply, 27 % for agriculture, 3 % for livestock and 2 % for industry and other services. The well depths of the production wells vary between 3 and 200 m (Monzalvo 2010). Crops cover approximately 26,000 ha of the La Paz area and include agave, avocado, guava, cotton, garlic, pepper, quelite, tomato and forage, as well as other crops (INEGI 2005).

The geomorphological structure of La Paz consists of intermountain plains (valleys), table mountains (mesas) and mountain ranges (sierras) with elevations of up to 900 m above sea level (m a.s.l.). The valley of La Paz-El Carrizal (the valley of La Paz) represents a tectonic graben with a N–S orientation that is delimited on the W and E by the La Giganta and La Paz faults, respectively, and on the north by the Sea of Cortez (Fig. 1a, b). Hypsometric data and the presence of structurally controlled wadies (arroyos) suggest the existence of a series of secondary faults inside the graben (Álvarez et al. 1997; Cruz-Falcón 2007).

The valley of La Paz is the result of distention tectonics that created the Gulf of California, probably during the Pleistocene, and the subsequent accumulation of marine and continental granular sediments in the lower topography. The base of the lithologic column in the La Paz area is a metamorphic complex that consists of Mesozoic shales, phyllites, schists, sillimanite and gneiss (Cruz-Falcón 2007; CONAGUA 2009; Monzalvo 2010). This unit is overlain by Cretaceous intrusive rocks (gabbro, granite and granodiorite) in the Cordillera Oriental portion of the La Paz fault. The basin-fill material of the valley of La Paz consists of Miocene sandstones, conglomerates and fractured volcanic rocks (e.g., Comondú Formation, San Isidro Formation) in the lower part of the valley and alluvial fans (sands, silts, clays) in the upper part of the valley (Fig. 1a, b).

The hydrostratigraphy of the valley includes an uppermost unit that corresponds to non-consolidated, granular sediments and volcanoclastic material with little compaction and medium-to-high permeability. The underlying second unit matches the volcanic flows of the Comondú Formation, with high hydraulic conductivity as a result of fracturing, but is only of local importance due to its discontinuity. These two units represent an unconfined aquifer system approximately 300–400 m thick and having an average hydraulic conductivity of 1.2–14.3 × 10⁻³ m s⁻¹. Finally, the deeper third unit corresponds to the Pre-Tertiary intrusive rock formations that outcrop in the Cordillera Oriental portion of the study area and have low permeability. The basement of these units is located approximately 900 m below land surface and consists of a metamorphic complex (CNA 1997, 1999, 2002; CIGSA 2001). This study concentrates on the upper, unconfined aquifer units.

In La Paz the catchment recharge represents 15.9 % (42 mm) of the rainfall in the basin, surface runoff represents 3.6 %, and the rest of the rainfall is evaporated. Recharge occurs mostly in the Las Cruces and Novillo ranges located in the E and SSE, respectively, and groundwater flows in a NNW direction (Cruz-Falcón et al. 2011) (Fig. 1a). Groundwater extraction in excess of recharge has caused groundwater abatement. The water table currently slopes from 150 m a.s.l. in the southern portion of the study area, and from 0 m a.s.l. in the northern portion close to the sea, to 10 m below sea level in the central part of the study area, which is indicative of groundwater recharge from seawater intrusion in the central portion (Fig. 1a).

Forty-six production wells and one spring were sampled for evaluation of ion chemistry and isotopes in August 2013. Physical parameters, including temperature, pH, electrical conductivity (EC), total dissolved solids (TDS) and dissolved oxygen (DO), were measured in situ using portable meters until measured levels did not vary, allowing for temperature compensations and calibrations with appropriate standards, such as calibration of the pH meter, which was performed at every sampling site. Alkalinity was determined in the field by titration with 0.02 N H₂SO₄. The groundwater samples were filtered using 0.45-µ membrane filters.

Activation Laboratories LTD, Ontario, Canada, performed the laboratory analyses of major ions and minor ions. Water samples were analyzed for cations (K⁺, Na⁺, Ca²⁺, Mg²⁺, Sr²⁺) using a Perkin Elmer ELAN 9000 ICP/MS and, if upper limits were exceeded, by a Perkin Elmer Optima 3000 ICP OES. Samples were spiked with internal standards to correct for matrix differences over the entire mass range. Additionally, a SLRS-4 and a NIST 1640 standard were run for every 32 samples for quality control. Anions (Cl⁻, NO₂ ⁻, NO₃ ⁻, Br⁻, SO₄ ²⁻ and PO₄ ³⁻) and SiO₂ were analyzed by ion chromatography (DIONEX DX-120 Ion Chromatography System). Nitrite concentration was absent (lower detection limit = 0.01 mg l⁻¹) in all except one sample (0.3 mg l⁻¹). This is because nitrite is usually rapidly transformed to nitrate. Phosphate concentration was absent (lower detection limit: 0.02 mg l⁻¹). A charge balance error of 10 % was accepted for use in further analyses and interpretations. Duplicates of samples LP10, LP20, LP30, LP40 and LP 41 were re-sampled for quality assurance and quality control.

Stable water isotope ratios in groundwater (δ²H, δ¹⁸O) were analyzed at Environmental Isotope Laboratory of Tecnológico de Monterrey. A volume of 0.2 ml of water was pipetted into an Exetainer vial and analyzed by a DLT-100 Liquid–Water Isotope Analyzer (Los Gatos). The results were determined and reported in the usual delta notation in ‰ with respect to the VSMOW standard.

Evaluations of carbon-13 (¹³C) ratios, radiocarbon (¹⁴C) activities and chlorofluorocarbon (CFC) concentrations were conducted for 26 representative samples. Carbon-13 determination was performed with a MM-903 spectrometer on carbon dioxide previously released by acidifying sample water. The results are reported in δ¹³C_DIC with respect to the VPDB (Vienna Pee Dee Belemnite) standard. The analytical precision was 0.15 ‰. Radiocarbon activity was measured by AMS at the NSF-Arizona Accelerator Facility and is reported as percent modern carbon (pmC). The analytical precision (1 − σ) ranged from 0.2 percent modern carbon (pMC) for pMC values near 30, to 0.5 pMC near 100 pMC.

Chlorofluorocarbon (CFC-11, CFC-12 and CFC-113) content was analyzed at Spurenstofflabor, Wachenheim, Germany, by gas chromatography with an electron capture detector after preconcentration using a purge-and-trap technique (Oster et al. 1996). The concentrations are reported in pmol l⁻¹. The analytical precision (1 − σ) ranged from 0.05 pmol l⁻¹ for pmol l⁻¹ values near 0, to 3 pmol l⁻¹ near 12 pmol l⁻¹.

A bivariate method, the Pearson correlation analysis, was applied for a preliminary evaluation of relation between two hydrochemical parameters. High correlation coefficients approximating 1 or −1 in combination with low significance levels (p < 0.05) represented a significant correlation. A hierarchical cluster analysis (HCA) grouped samples according to their similarity. Parameters were progressively merged into clusters, beginning with individual elements of a data set. Each agglomeration step used the distance between clusters from the previous agglomeration. A dendrogram helped to graphically identify the genealogical relationship between clusters and elements. The results were proved for geographic correspondence. All statistical analyses were performed using the Minitab 17.0 program (Minitab 2013).

Deuterium excess (d) was calculated with the formula d = δ²H–8δ¹⁸O. On the one hand, the values of d tend to increase when there is a high kinetic isotope fractioning during evaporation of sea water due high temperatures and dryness of the environment; on the other hand, the effect of evaporation before recharge will lead to a decrease of these values; therefore, this parameter was used as an indicator of vapor origins (Rindsberger et al. 1983; Jiménez-Martínez and Custodio 2008).

Lumped parameter models (LPMs) have been applied to support the interpretation of CFC tracers. The TracerLPM (USGS 2014) was used to evaluate groundwater age distributions and residence times with the LPMs most commonly used in previous studies (IAEA 1996; Plummer et al. 2001; Plummer 2003, 2005; Goody et al. 2006). Ages were corrected for a recharge temperature of 23 °C, a recharge altitude of 208 m a.s.l. and a salinity in ‰ (IAEA 2006). A recharge altitude was selected based on the study of Cruz-Falcón et al. (2011), which indicated that the recharge area is mostly in the El Novillo and Las Cruces ranges.

Radiocarbon data were used to evaluate older (>1 ka) groundwater in La Paz. Unadjusted ¹⁴C ages were calculated from measured ¹⁴C activities of dissolved inorganic carbon (DIC) using the Libby half-life (5568 years) and assuming an initial ¹⁴C activity A₀ of 100 modern carbon (pmC), except for samples of recently recharged water, whose initial activity was set at 105 pmC. Age adjustments were made using the formula-based inorganic adjustment models (Ingerson and Pearson 1964; Fontes and Garnier 1979—F&G). Alkalinity, pH, water temperature and carbon isotope (¹³C and ¹⁴C) data were used to make these age calculations. The model approach corrects the age for dissolution with “dead” carbon derived from the aquifer matrix, assuming the following parameters to account for the pristine arid sarcocauleschent scrub vegetation (Perea et al. 2005; Michener and Lajtha 2007; Rodríguez-Estrella et al. 1998): initial activity; isotope fractionation factors after Vogel et al. (1970), Mook et al. (1974) and Deines et al. (1974); δ¹³C carbonates of 0 ‰ VPDB and δ¹³C soil CO₂ of −19 ‰ VPDB. If a particular model resulted in an unreasonable (negative) age, the adjusted age was designated as either modern (recharge after the mid-1950s), pre-modern (recharge prior to the mid-1950s), or as a mixture of modern and pre-modern water with respect to ¹⁴C based on evaluating CFC tracers (Gardner and Heilweil 2014).

The physical, chemical and isotopic data for the groundwater samples are summarized in Table 1. Pearson method indicates that positive correlations exist between several hydrochemical parameters. The correlation between Mg²⁺, Ca²⁺, Sr²⁺, Cl⁻ and Br⁻ ions is highly significant (R > 0.9; p < 0.01). Meanwhile, Na⁺ is significantly correlated with Cl⁻ and Br⁻ (R > 0.9; p < 0.01). TDS present a highly significant correlation with alkali and earth alkali (Na⁺, Mg²⁺, Ca²⁺, Sr²⁺), as well as Cl⁻ and Br⁻ ions. None of the samples showed significant negative correlation.

The classification of water samples was performed using HCA with 17 variables (pH, temperature, EC, TDS, DO, alkalinity, Na⁺, SiO₂, K⁺, Ca²⁺, Sr²⁺, Cl⁻, Br⁻, NO₃ ⁻, SO₄ ²⁻, δ²H, δ¹⁸O and the distance of the sample location from the coastline). Ward´s linkage rule was applied for iteratively linking nearby clusters using a similarity matrix and performing an ANOVA to evaluate the distance between clusters (Ward 1963). The water samples merged into three groups at the phenon line shown in the dendrogram (Fig. 2) with geographical correspondence. Group 1 (23 samples) is located in the central part of the study area (south of El Centenario and Chametla) where agriculture is practiced and groundwater extractions from supply wells are high. Group 2 includes 16 samples located mostly in the urbanized area of La Paz. This group is similar to group 3 considering the statistical approach (Fig. 2); however, based on groundwater chemistry (Figs. 4, 5) it lies between group 1 and group 3. All samples of groups 1 and 2 are located in the 20-km fringe of the coastal area. Finally, group 3 consists of 8 samples that are located at higher altitudes and further away from the coastline. They are closer to the recharge areas (Las Cruces and El Novillo ranges). The three groups will be used and discussed in the subsequent hydrogeochemical evaluation.

The study area exhibits a wide range of salinities, ranging from very fresh (479 µS cm⁻¹) to brackish (8920 µS cm⁻¹; σ = 2047 µS cm⁻¹), having slightly acidic to alkaline pH values between 6.8 and 8.3 (σ = 0.3). High EC values are primarily due to elevated Cl⁻, Na⁺ and Ca⁺ concentrations that increase from the recharge area (group 3) to the coastal area (group 1) between El Centenario and La Paz (Fig. 3a). Exceptions are water samples LP38 and LP39 from group 2, which are located in the coastal area but have low Cl⁻ concentrations (164 and 95.6 mg l⁻¹, respectively) in comparison to the rest of the group. The locations of the wells where these samples were collected coincide with an area where artificial recharge takes place, i.e. the wastewater treatment plant of the city of La Paz exchanges treated wastewater with farmers from Centenario and Chametla that, together with Paraiso del Mar, receive 6 million cubic meters per year (CONAGUA 2010). The pH–EC relationship indicates an inverse correlation that reflects the drainage of non-carbonate rocks and sediments. The temperatures range from 25.0 to 33.4 °C (σ = 1.7) Elevated temperature values were measured in samples LP05, LP06 and LP10 a subgroup of group three wells, and LP25 of group 2, possibly indicating a slight upconing of deeper, regional groundwater flows from lower, intrusive rock formations. An effect of groundwater warming due to direct influence of air temperature in the wells may be excluded because all wells were tube wells, i.e. not open.

The molar Na/Cl ratio of groundwater compared to seawater (0.86) varies from 0.19 to 1.49 (median 0.67) and indicates that some waters (19 %) are enriched in Na⁺, which means that molar Na⁺ concentration is in excess of Cl⁻, while most are depleted. The important depletion of Na⁺ reflects large evaporation effects from water recycling probably linked to agricultural practices. Moreover, the enrichment of Na⁺ indicates weathering of silicates. Another less important mechanism of Na⁺ enrichment or depletion is cation exchange in soils. Sodium is possibly enriched by exchange of Ca²⁺ for Na⁺ ions at clay surfaces, which produces Na–HCO₃ waters (e.g., LP9, LP10). Finally, Na⁺ is possibly depleted from solution by reverse ion exchange reactions of Na⁺ for Ca²⁺ and/or Mg²⁺ ions at clay surfaces in discharge areas (e.g., LP19, LP20).

Nitrate concentrations (NO₃–N) range from 1.4 to 48.8 mg l⁻¹ (σ = 9.7 mg l⁻¹). Eleven out of 47 samples (23 %) do not comply with the Mexican drinking water standard for NO₃ ⁻–N (10 mg l⁻¹). Samples with values exceeding the standard are mostly located in two agricultural zones south of El Centenario and Chametla, and south of La Paz (part of group 1) (Fig. 3b). Sulfate concentrations vary from 7.9 to 490 mg l⁻¹ (σ = 114 mg l⁻¹) and increase towards the coastal areas. Chloride concentrations and EC values indicate a similar trend. Two samples do not comply with the Mexican drinking water standard for SO₄ ²⁻ (400 mg l⁻¹). The values of alkalinity (HCO₃ ⁻ = 166–1290 mg l⁻¹; σ = 219 mg l⁻¹) and major cations tend to increase considerably towards the coastline.

The chloride-to-bromide mass ratio (Cl/Br) in seawater is typically 285. Cl/Br ratios in coastal groundwaters are typically close to 290 (Davis et al. 1998; Anders et al. 2014). Since Br is slightly more soluble than Cl, the Cl/Br ratio in evaporates (halites, gypsum) decreases due to evaporative effects. In consequence, the dissolution of these rocks may increase the ratio to 4000 (Davis et al. 1998; Anders et al. 2014). Wastewater loaded with NaCl may increase the ratio up to 655. Finally, the ratio may be decreased to 130 by the use of Br-based pesticides or leaching farm-animal or septic waste (Alcalá and Custodio 2008). The Cl/Br for groundwater samples ranges from 245 to 629 (mean = 364) (Fig. 4). Most samples close to sea have ratios higher than seawater; they are likely the result of evaporation of seawater. Some samples close to sea that have ratios lower than seawater may be indicative of animal farming. In general the samples are in the expected range of arid coastal aquifers.

A piper diagram elucidates the chemical patterns (Fig. 5). The general dominance of cations is Na > Ca = Mg > K, while the dominance of anions was Cl > HCO₃ >> SO₄. The central diamond in the diagram shows a clear distinction between the three selected groups. Group 1 (Ca–Cl and Na–Cl) near the coastline is characterized by seawater intrusion, and a trend towards elevated SO₄, Cl and NO₃ values indicates contamination by agricultural practices. Group 3 (Ca–HCO₃ and Na–HCO₃) is derived from rainwater and spreads along the HCO₃ facies type as it infiltrates the ground surface and undergoes weathering processes. Finally, group 2 represents a transition between both groups 1 and 3.

The δ¹⁸O and δ²H values vary from −11.6 to −7.1 ‰ (σ = 1.0 ‰) and from −82.0 to −52.1 ‰ (σ = 7.2 ‰), respectively (Table 1). Plotted in a conventional diagram, the general trend is δ²H = 6.28 (δ¹⁸O)–8.47; r ² = 0.78 (Fig. 6). It indicates that the samples are plotted mostly along and below the Regional Meteoric Water Line (RMWL) according to the Chihuahua IAEA GNIP station: δ²H = 7.1δ¹⁸O + 2.8 (Wassenaar et al. 2009). These δ¹⁸O and δ²H data indicate that groundwater in the coastal aquifer is of meteoric origin and a large proportion is subject to evaporation effects which is consistent with deuterium excess. Waters located at higher altitudes (group 3) plot along the RMWL, indicating direct infiltration of local precipitation with little evaporation. Deuterium excess for this group varies from 7.6 to 12.9 with an average of 10.6 ‰ (σ = 1.73 ‰) showing an increment of d values as temperature increases due to kinetic isotope fractioning related to high temperatures and dryness in the environment (Jiménez-Martínez and Custodio 2008). Group 2 shows a high variability of δ¹⁸O and δ²H values, which is similar to the values of the other two groups. A notable exception is a cluster of four samples representing the urban area of La Paz (LP1, LP2, LP3 and LP11). These samples show the most depleted δ¹⁸O and δ²H values of the study area, indicating groundwater recharge in higher altitudes (Figs. 1a, 6). Finally, group 1 shows the least depleted δ¹⁸O and δ²H values among all groups. The samples of group 1 and 2 have a similar distribution and deuterium excess values: from −0.1 to 15.2, and from −1.1 to 11.8 ‰, respectively. These samples are plotted on and below the RMWL (El Centerario and Chametla) which is consistent with the extensive evaporation before and during recharge in the sample area due to dry conditions and the heavy extraction and cycling of groundwater to supply the area with irrigation water. Waters plotted below the RMWL are the most mineralized. The area in which they are located is characterized by low infiltration rates and long residence times over land compared to other areas (Cruz-Falcon et al. 2011), which is congruent with their isotopic signature. Waters from the locality of El Centenario (LP22, LP23 and LP27) are derived from recharge of the western conglomerates (LP12, LP13). These samples show the most mineralogically enriched values in the study area, indicating lower altitude and/or less humid climate conditions during recharge (Fig. 6). The δ¹⁸O and δ²H values and water chemistry are affected in contaminated or saline groundwater. Brackish groundwater with enriched ¹⁸O values and high Cl⁻ or NO₃ ⁻ concentrations is indicative of evaporation effects in agricultural lands, as observed in samples LP18, LP27, LP28 and LP31. Groundwater from the urban area of La Paz (LP1, LP2, LP3 and LP11) exhibits relatively low Cl⁻ and NO₃ ⁻ values, suggesting little influence from domestic sewage.

Trace gases CCl₃F (CFC-11), CCl₂F₂ (CFC-12), C₂Cl₃F₃ (CFC-113) and SF₆ (sulphur hexafluoride) are increasingly used as tracers of young (e.g., up to decades old) groundwater. The principle of age dating with CFCs is based on comparing their concentrations in groundwater with the known worldwide atmospheric concentrations of these anthropogenic gases over the past 60 years to estimate the year of recharge at each site (Darling et al. 2012). However, several complications may arise. Local sources of trace gases from cities and industries, sorption or even the sampling process may bias the levels of CFCs (Plummer et al. 2001). Microbial degradation may also occur under reducing conditions (Cook et al. 1995; Plummer et al. 1998).

Chlorofluorocarbon mixing ratios in the study area ranged from 45.7 to 2479 (CFC-11), 93 to 1587 (σ = 413) (CFC-12) and 10 to 318 (CFC-113) pptv (Table 2). The highest CFC-12 values were indicated for samples located in irrigation areas (LP28, LP42), although elevated values are also indicated in samples located towards the El Novillo range (LP4, LP8 and LP18) and in El Centenario (LP22) (Fig. 7a). CFC-11 values indicate a notable deviation from CFC-12 values and also from CFC-113 values. CFC-12 and CFC-113 values are more consistent. Disagreement between tracers is expected for groundwater samples because the amount of increase in CFC concentrations due to contamination is considered to be different for each CFC gas (Horst et al. 2008; Kusano et al. 2014).

The tracer pairs of CFC-12 and CFC-113 in Fig. 8a generally plot within the regions bounded by the piston flow and exponential mixing models. In Fig. 8b, the samples generally plot below the exponential mixing model curve. This lower CFC-11/CFC-12 ratio may be due to microbial degradation of CFC-11, as also observed in previous studies (e.g., Lindsey et al. 2003; Horst et al. 2008). It has been demonstrated that CFC-12 and CFC-113 are less degraded in waters (e.g., Plummer and Busenberg 2000). However, in our study no individual sample indicated dissolved oxygen concentrations of <3 mg l⁻¹, which indicates that aerobic conditions predominate. One explanation of this discrepancy could be that only part of the flow paths arriving at a sampled well underwent anaerobic conditions, or oxygenation occurred in the well tube. Moreover, the different CFC tracers have distinct temporal patterns of their input functions. The air-mixing ratios of CFC-11 and CFC-113 indicate a turnover in the 1990s, which leads to ambiguity in age dating with modern ages that span the 1990s. Thus, in this study the estimation of residence times was performed using only CFC-12 tracers.

The estimation of groundwater residence times (RTs) requires the knowledge of the aquifer geometry. The most commonly used models are piston flow and exponential mixing. Although hybrid models are also feasible, such as exponential piston flow models, they were not considered due to a lack of detailed knowledge of aquifer conditions at individual wells. The investigated area represents an aquifer that is primarily unconfined with well depths of up to 200 m and long screen lengths. This geometry indicates that the exponential mixing model may generally match the groundwater conditions rather than the piston flow model. The tracer plot in Fig. 9a confirms this assumption. The individual samples do not plot on the piston flow curve, with the exception of a few dispersed samples (LP4 and LP22).

The RTs calculated with the TracerLPM model range from 26 to 46 years and 18 to 206 years for PFM and EMM, respectively (Table 2). Samples with CFC concentrations greater than levels of possible atmospheric equilibrium for the northern hemisphere were considered to have been contaminated by excess air and were not used for RT calculations.

Carbon-13 values range from −12.3 to −9.5 ‰ VDBP, and ¹⁴C activities vary between 29.5 and 100.5 pmC (Table 2). Uncorrected RTs vary from recent to 10 ka. Radiocarbon data indicate a trend from elevated levels in the eastern mountainous flanks to lower levels in the western extremes (Fig. 7b). The highest values on the order of 100 pmC mostly represent groundwater recharged in the flanks of the El Novillo range/La Paz fault in the eastern portion of the study area (e.g., LP1, LP3 and LP44). These waters consistently indicate a depleted δ¹³C signature and low EC, Cl and DIC values (Fig. 9a, b). Two trends can be observed in Fig. 9b: group 2 and 3 waters reduce ¹⁴C activities with increasing DIC/salinity, and group 1 waters raise ¹⁴C values with increasing DIC/salinity. This rise in ¹⁴C values may be explained by the increased admixture of seawater and evaporation in group 1 that increases DIC/salinity.

The corrected values in Table 2 suggest that groundwater is modern with the exception of several sites (LP13, LP14, LP23, LP25, LP38 and LP42). These waters (0.9 and 4.7 ka) are located in the central and western lowlands (El Centenario/Chametla). There is no general positive correlation between groundwater age and temperature; however, the two oldest waters (LP13 and LP25 with 4.0 and 4.7 ka, respectively), located in the westernmost portion, coincide with two of highest temperatures in the study area (31.8 and 32.7 °C, respectively) which could indicate local upconing of older groundwater from the lower aquifer unit.