Interactive comment on “ 17 O-excess traces atmospheric nitrate in paleo groundwater of the Saharan desert ”

Saharan paleo-groundwater from the Hasouna area of Libya contains up to 1.8 mM of nitrate, which exceeds the World Health Organization limit for drinking water, but the origin is still disputed. Herein we show that a positive 17 O excess in NO 3 − (Δ 17 O NO 3 = Δ 17 O NO 3 − 0.52 δ 18 O NO 3 ) is preserved in the paleo-groundwater. The 17 O excess provides an excellent tracer of atmospheric NO 3 − , which is caused by the interaction of ozone with NO x via photochemical reactions, coupled with a non-mass-dependent isotope fractionation. Our Δ 17 O NO 3 data from 0.4 to 5.0 ‰ ( n = 28) indicate that up to 20 mol % of total dissolved NO 3 - originated from the Earth's atmosphere ( x [NO 3 − ] atm ), where the remaining NO 3 − refers to microbially induced nitrification in soils. High Δ 17 O NO 3 values correspond to soils that are barren in dry periods, while low Δ 17 O NO 3 values correspond to more fertile soils. Coupled high Δ 17 O NO 3 and high x [NO 3 − ] atm values are caused by a sudden wash-out of accumulated disposition of atmospheric NO 3 − on plants, soil surfaces and in vadose zones within humid–wet cycles. The individual isotope and chemical composition of the Hasouna groundwater can be followed by a binary mixing approach using the lowest and highest mineralised groundwater as end members without considering evaporation. Using the δ 34 S SO 4 and δ 18 O SO 4 isotope signature of dissolved SO 4 2− , no indication is found for a superimposition by denitrification, e.g. involving pyrite minerals within the aquifers. It is suggested that dissolved SO 4 2− originates from the dissolution of CaSO 4 minerals during groundwater evolution.


Introduction
The accumulation of nitrate (NO − 3 ) in groundwater is a wellknown phenomenon occurring worldwide (Clark and Fritz, 1997;Kendall, 1998).Individual NO − 3 sources and mechanisms for its accumulation depend strongly on the environmental conditions during recharge, infiltration, and aquifer storage.High NO − 3 concentrations of paleo-groundwater from the Hasouna area (Libya) have been measured for decades, but the NO − 3 origin is still hotly debated (El-Baruni et al., 1985;Milne-Home and Sahli, 2007).Deciphering the source of NO − 3 for Saharan groundwater in Libya is highly challenging, as (i) the present arid conditions preclude appreciable recharge and (ii) groundwater is about 90 % of the source of the water supply of Libya (e.g.Salem, 1992;Edmunds, 2006;Abdelrhem et al., 2008).Several million m 3 of fresh water per day are transferred to the Mediterranean for agricultural and domestic use including drinking water supply.3 in modern groundwater, lakes, and wet depositions (Darrouzet-Nardi et al, 2012;Dejwakh et al., 2012;Li et al., 2010;Michalski et al., 2005;Michalski et al., 2004b;Tsunogai et al., 2011;Tsunogai et al., 2010;Nakagawa et al., 2013).
The aim of the present study is to trace the origin of NO − 3 in Saharan paleo-groundwater by using δ 15 N NO 3 values and triple stable isotopes of oxygen.This approach has never been used for ancient groundwater, and the major aim of this study is to test the applicability of 17 O NO 3 values in such surroundings.In addition, a reaction of NO − 3 with pyrite (FeS 2 ) minerals in the aquifers can act as a potential NO − 3 sink, which has already been shown for European shallow aquifers (e.g.Böttcher et al., 1990;Zhang et al., 2012).This requires the analyses of the stable isotope signatures of dissolved SO 2− 4 .Furthermore, ancient recharge and climate conditions are reconstructed by using a multi-element and isotope approach.

Study area and sample sites
The Hasouna well field is located about 700 km south of Tripoli (Libya) within the Cambro-Ordovician Nubian Sandstone Aquifer System (Fig. 1).This system comprises one of the world's largest paleo-groundwater aquifers, which had been recharged at approximately 30 ± 10 and 10 ± 3 ka BP based on 14 C dating (Edmunds et al., 2003;Edmunds and Wright, 1979;Guendouz et al., 1998;Milne-Home and Sahli, 2007;Sonntag et al., 1978).Strong evidence exists for enhanced precipitation in North Africa during these time periods.However, individual epochs for humidity and groundwater recharge periods are still disputed; Pleistocene and Holocene humid periods in North Africa have been deduced from the records stored in lacustrine fresh-water sediments (Rognon, 1987), wadi river systems (Kuper and Kröpalin, 2006), travertine (Carrara et al., 1998), paleobotanic investigations (Doherty et al., 2000), rock art dating and varnish coating of petroglyphs (Dietzel et al., 2008), dating of aragonitic mollusk shells (Blackwell et al., 2012), together with cyclicity of geochemical signals observed in marine sediments of the eastern Mediterranean that are influenced by the Nile River hydrology (Wehausen and Brumsack, 1999;Lourens et al., 2001).
The aquifer consists predominantly of fractured sandstone ranging from ≈ 500 to 1500 m in thickness, with thin interbeds of marine limestone and marl.The sandstone itself is mostly quartzitic sandstone, but occasionally carbonate cement occurs.In the study area, the main Cambro-Ordovician sandstone aquifer is overlain by a shallow carbonate aquifer with a basal aquitard, predominantly composed of marly limestone, clay and shale.
Groundwater in the study area of Jabal Hasouna was discovered during oil exploration in the 1960s (Edmunds, 2006).A large number of wells have been installed as part of the Great Man-made River Project, which brings up to ≈ 6 Mm 3 of fresh water per day to the Mediterranean (Salem, 1992;Abdelrhem et al., 2008).The average spacing of the total 484 wells is ≈ 1.5 km, with individual discharges of ≈ 50 L s −1 covering ≈ 4000 km 2 (Binsariti and Saeed, 2000).

Materials and methods
Sampling of Hasouna groundwater from 28 wells was carried out in the year 2007, with sampling depths between 323 and 555 m.Coding and sampling site locations are summarised in Table 1 and shown in Fig. 1.

Chemical composition
The temperature and pH of the groundwater were measured in situ with pH meter WTW 330 and pH electrode WTW SenTix 41, having a precision of ±0.03 pH units (calibrated on-site using pH 4.0 and 7.0 WTW standard buffer solutions).Concentration of dissolved O 2 was analysed in the field using WTW Oxi 325 and WTW CellOx325, having an analytical precision of ± 0.2 mg L −1 .Sampled solutions were filtered in the field through 0.45 µm cellulose acetate membranes, and one aliquot was acidified.The samples were stored in polyethylene and hermetically sealed glass vessels for laboratory analyses.Concentrations of dissolved anions were analysed in non-acidified samples by ion chromatography (Dionex DX 500) with an analytical precision of about ±3 %.Alkalinity was obtained by potentiometric titration of non-acidified samples from the gas-tight glass vessels with an uncertainty of about ±5 %.In the acidified solutions (2 % HNO 3 ), cations were analysed by inductively coupled optical emission spectroscopy with a precision of better than ±5 % (Perkin Elmer Optima ICP-OES 4300; Merck multi-element standard).Calculation of saturation indices with respect to calcite (SI calcite ) and gypsum (SI gypsum ) was performed with the PHREEQC computer code (Parkhurst and Appelo, 1999) and the phreeqc.datdatabase.

Stable isotopes
Additional samples were gathered for analysis of the δD H 2 O and δ 18 O H 2 O values of H 2 O.The stable isotopes of hydrogen were measured using a Finnigan DELTA plus XP mass spectrometer working in continuous flow mode by the chromium reduction technique (Morrison et al., 2001).The oxygen isotope composition was measured with a Finnigan DELTA plus mass spectrometer using the classic CO 2 -H 2 O equilibrium method (Horita et al., 1989) where R s and R st are the respective analysed isotope ratios (D/H) for the measured sample and standard (VSMOW), respectively.The δ 34 S SO 4 and δ 18 O SO 4 values of dissolved SO 2− 4 were measured on BaSO 4 that was precipitated from acidified solutions by the addition of Ba(Cl) 2 .The solids were washed and dried and further analysed by means of continuous-flow isotope-ratio monitoring mass spectrometry (CirmMS) using a Thermo Scientific Finnigan Delta+ mass spectrometer at MPI-MM Bremen, Germany (according to Böttcher et al., 2001 andKornexl et al., 1999).The respective precisions of sulfur and oxygen isotope measurements were ±0.3 ‰ and ±0.5 ‰, for δ 34 S SO 4 and δ 18 O SO 4 .Results are reported in relation to the VCDT and VSMOW scales, respectively.
Separate samples were taken for isotope analyses of dissolved NO − 3 .Analyses were carried out using the bacterial denitrifier technique in combination with the N 2 O decomposition method (Casciotti et al., 2002;Kaiser et al., 2007;Morin et al., 2008;Savarino et al., 2007).Isotope ratios of nitrate were measured on a Thermo Finnigan MAT 253 isotope ratio mass spectrometer, equipped with a GasBench II and coupled to an in-house-built nitrate interface.Briefly, denitrifying Pseudomonas aureofaciens bacteria converts NO − 3 into N 2 O under anaerobic conditions.N 2 O is then thermally decomposed on a gold surface heated to 900 • C, producing a mixture of O 2 and N 2 which is then separated by gas chromatography and injected into the mass spectrometer for the dual O and N analysis (Erbland et al., 2013, and references therein).Isotopic data were corrected for any isotopic effect occurring during the analytical procedure by using the same approach as Morin et al. (2009) and Frey et al. (2009) were prepared in an identical way and followed the same analytical procedures.This identical treatment includes the use of the same background matrix as well as the same water isotopic composition for the standards and samples.The overall accuracy of the method is estimated as the reduced standard deviation of the residuals from the linear regression between the measured reference materials (n = 16) and their expected values.The δ 15 N NO 3 , δ 18 O NO 3 , and 17 O NO 3 values ( 17 O excess) of dissolved NO − 3 are given with respect to the atmospheric N 2 (AIR) and VSMOW standards, and were measured with uncertainties of ±0.5 ‰, ±2 ‰, and ±0.5 ‰, respectively.

Ion content and mixing approach
The analysed compositions of the sampled groundwater from the Hasouna area are summarised in Table 1.The investigated Hasouna groundwater displays near-neutral pH values (pH = 7.0 ± 0.5) at temperatures between 27 and 35 • C. The concentration of dissolved oxygen is 5.5 ± 0.8 mg L −1 (n = 9).Dissolved cations and anions generally occur in the quantitative sequences of Na + > Ca 2+ > Mg 2+ > K + > Sr 2+ and Cl − > HCO − 3 > SO 2− 4 > NO − 3 , respectively (Table 1).The average deviation from electrical neutrality for the aqueous solutions is 3.6 meq%, with a maximum of about 5 meq %.Values less than 5 meq % indicate the good quality of ion analysis by considering the individual analytical precisions (see Sect. 3.1; Appelo and Postma, 2007).Except for NO − 3 , the chemical composition of the groundwater is within the limits for drinking water.In ≈ 85 % of the analysed Hasouna groundwater, the maximum contaminant level (MCL) of the World Health Organization (WHO, 2004) for NO − 3 in drinking water of 0.71 mM is exceeded (Table 1).
All analysed groundwater is undersaturated with respect to gypsum (SI gypsum ≤ −0.93).However, approximately 20 % of the groundwater (6 of 28 samples) is slightly supersaturated with respect to calcite (−0.50 ≤ SI calcite ≤ 0.18; Table 1).If the analysed Ca 2+ concentration is reduced by the proportion gained from CaCO 3 dissolution by considering the dissolution of Mg 2+ carbonates like dolomite ([Ca 2+  ] + ), a Ca 2+ (reduced) to SO 2− 4 ratio of 1 is obtained (Fig. 2).This relationship typically results from the uptake of gaseous CO 2 (e.g. in soil horizons) and the subsequent dissolution of limestone and/or dolostone (e.g.calcite and/or dolomite; Dietzel et al., 1997).Dissolution of calcite and dolomite during the evolution of the groundwater seems reasonable, as both minerals have been documented in rock core samples from Hasouna well fields (Sahli, 2006).The congruent dissolution of CaSO 4 , like gypsum, causes the Ca 2+ (reduced) to SO 2− 4 ratio of 1 (Fig. 2).The dissolved ions cover a broad concentration range, but the individual ion concentrations are well correlated.For in- stance, the correlation of Na + and Cl − concentrations is shown in Fig. 3a.As groundwater samples #165 and #152 bracket the full observed range, their chemical composition is used for a binary mixing approach according to the expression where x #165 + x #152 = 1 (refer to the solid green line in Fig. 3a).In this expression [i] denotes the concentration of the dissolved ion (i is Na + or Cl − ) of the respective solution.The terms x #165 and x #152 are the volume fractions of groundwater samples #165 and #152, which yield the ion concentration range of the Hasouna groundwater.Analogous relationships are found for all dissolved ions (e.g.Ca 2+ vs. NO − 3 in Fig. 3b).If seawater is assumed for a mixing endmember solution, the mixing approach of Eq. (3) fails, as in most cases the ion ratios of the groundwater are inconsistent with seawater.This is shown by the Ca 2+ (reduced) to SO 2− 4 ratio of the Hasouna groundwater of about 1 (Fig. 2), which differs significantly compared to that of seawater (≈ 60; assuming ancient seawater composition close to modern conditions; Drever, 2002).Additionally, an evaporation trend for #165 to reach the composition of #152 is not consistent with the measured data (dotted arrows in Fig. 3a and b).This is even valid if a decrease in Ca 2+ concentration through calcite precipitation (SI calcite = 0) is considered, which may be induced by evaporation (solid arrow in Fig. 3b).Thus, the solution chemistry of the Hasouna groundwater can be best explained by mixing of the two end-member solutions #165 and #152.
As an exception, the concentration of silicic acid is nearly constant in all groundwater samples ([Si(OH) 4 ] = 0.22 ± 0.02 mM; n = 28) and reflects the solubility of quartz at the  Rimstidt, 1997).Thus, the concentrations of silicic acid in the ancient Hasouna groundwater from the Nubian Sandstone Aquifer are clearly controlled by quartz-water interaction considering thermodynamic equilibrium, which is typically found for ancient groundwater in quartz-dominated aquifers (Rimstidt, 1997, and references therein).
In accordance with the overall shift of Saharan paleogroundwater to low δD H 2 O and δ 18 O H 2 O values compared to apparent local precipitation (in Sfax and Tripoli, Fig. 4), it is postulated that the Hasouna groundwater was recharged under paleoclimatic conditions, which were cooler and much more humid compared to the modern situation (Clark and Fritz, 1997;Edmunds, 2006).In accordance with the abovementioned mixing approach, an evaporation effect for Hasouna paleo-groundwater (e.g.subsequent to precipitation or during infiltration) can be ruled out, as the slope of the regression line for the isotope data is similar to the MWLs (Craig, 1961;Kendall and Doctor, 2004).

Stable isotopes of nitrate, origin and potential overprints
The δ 18 O NO 3 and δ 15 N NO 3 values of the dissolved NO − 3 for the Hasouna groundwater range from 6.9 to 17.4 ‰ and from 6.6 to 9.1 ‰, respectively, and are shown in Fig. 5 in relation to the isotope ranges of different sources for NO − 3 (see Kendall, 1998 andBöhlke et al., 1997).Our measured δ 15 N NO 3 and δ 18 O NO 3 values are almost within the range of a microbial origin for NO − 3 (e.g.obtained by nitrification in soils), where 18 O/ 16 O is incorporated from the reacting H 2 O and O 2 , and the δ 15 N NO 3 value is derived from organic matter (Fig. 5).
In modern groundwater the isotope composition of NO − 3 might be modified within pyrite-bearing aquifers by micro-bial NO − 3 reduction via oxidation of iron sulphides according to the idealised overall reaction 5 FeS 2 + 14 NO − 3 + 4 H + → 5 Fe 2+ + 7 N 2(g) ( 5) (Böttcher et al., 1990;Zhang et al., 2010Zhang et al., , 2012;;Saccon et al., 2013).This overprint can be followed by a characteristic superimposition of the sulfur and oxygen isotope composition of dissolved SO 2− 4 (e.g.Zhang et al., 2011Zhang et al., , 2012)).The measured narrow range of δ 34 S SO 4 and δ 18 O SO 4 values in Hasouna groundwater of +10.5 ± 0.7 ‰ vs. VCDT and +10.4 ± 0.8 ‰ vs. VSMOW (n = 10; see Table 1), respectively, is surprising, considering the observed wide range in SO 2− 4 concentrations (1.3 to 4.3 mM, see Fig. 6).In addition, no significant correlation of the above isotope data with the NO − 3 concentrations exists.This behaviour indicates that the source of dissolved SO 2− 4 was not coupled to NO − 3 reduction.Thus pyrite oxidation-driven NO − 3 reduction is considered to be insignificant.In addition, the heavy sulfur isotope values exclude significant contributions, at least from biogenic pyrite oxidation.If isotopically light sulfur would have been modified by later microbial SO 2− 4 reduction under essentially closed system conditions (e.g.Hartmann and Nielsen, 2012) to reach heavy residual SO 2− 4 values, one would expect an equilibration of the oxygen isotope signature of SO 2− 4 with the surrounding bulk water (Fritz et al., 1989).Using the fractionation factors derived by Fritz et al. (1989) much heavier oxygen isotope values for water are expected.The isotope data of the dissolved SO 2− 4 are close to values reported for Permian evaporites (Böttcher, 1999;Mittermayr et al., 2012;Nielson, 1979; δ 34 S SO 4 = +11 ± 1 ‰), but differ significantly from the isotope values of modern Mediterranean seawater δ 34 S SO 4 = +21 ‰; Böttcher et al., 1998;Fig. 6).Unfortunately, details regarding the hydrogeological conditions of the groundwater remain unknown, in particular with respect to the regional occurrence and distribution of CaSO 4 horizons.However, our hydrogeochemical findings and isotope values support the dissolution of gypsum during subterranean groundwater evolution, as shown by the relationship in Fig. 2.Moreover, the correlation for the concentrations of dissolved Sr 2+ vs. Ca 2+ of Hasouna groundwater ([ 1) is typically found where evaporite gypsum deposits are dissolved (e.g.Dietzel and Kirchhoff, 2003).Thus, from the solution chemistry of the groundwater as well as from the known occurrence of evaporites on the Saharan platform (Turner and Sherif, 2007), and occasionally documented gypsum in core samples from Hasouna well fields (Sahli, 2006), we reasonably assume salt dissolution to be the dominant source of the dissolved SO 2− 4 .This conclusion is also supported by the above-mentioned narrow range of stable isotope data with respect to SO 2− 4 , with no correlation of isotope data and SO 2−  Böttcher (1999), Faure and Mensing (2005), Mittermayr et al. (2012), andNielson (1979).
A pronounced trend toward higher δ 18 O NO 3 and lower δ 15 N NO 3 values is obvious for elevated NO − 3 concentrations (solid green arrow in Fig. 5).This trend cannot result from denitrification (dashed arrow; Böttcher et al., 1990;Kendall,1998;Granger et al., 2010), but might be explained by an impact of NO − 3 from atmospheric deposition (e.g.Böhlke et al., 1997).
The origin of NO − 3 can be constrained by considering the 17 O isotope, and evaluating both the δ 17 O NO 3 and δ 18 O NO 3 values (see Eq. ( 1)).All Hasouna paleo-groundwater measured here exhibits a positive 17 O excess up to 17 O NO 3 ≈ 5 ‰, which is attributed to NMDF during the conversion of atmospheric NO x and O 3 to NO − 3 (atm) (Table 1; Kendall and Doctor, 2004;Michalski et al., 2004a;Thiemens, 2001).In Fig. 7a, the 17 O excess vs. the traditional δ 18 O NO 3 value is displayed.A potential secondary (de)nitrification or reduction impact can be followed by a shift parallel to the terrestrial fractionation trend (horizontal arrows indicate the MDF effect).For the given isotope data of NO − 3 in Fig. 7a, an overall relationship with a slope of 0.342 is obtained from least squares regression, which fits well with the 17 O NO 3 end member, the atmospheric photochemical NO − 3 (Michalski et al., 2004a; see Fig. 7a).The deviation from the overall correlation of isotope data, which is given by the green dashed line in Fig. 7a, can be explained by MDF effects.The isotope data of Hasouna groundwater plot slightly below the dashed line.This behaviour may be caused by minor variability of the isotopic composition of atmospheric photochemical NO − 3 and/or by a slight impact on secondary MDF effects, e.g.induced by denitrification.For the present case the overall nonrelevance of denitrification is shown in Fig. 5.   : Hasouna groundwater, Libya (this study).
The formation of NO − 3 ,atm is linked to the most important oxidant in the atmosphere: O 3 .The respective 17 O of atmospheric NO − 3 , gained from the conversion of atmospheric NO x and O 3 , can be followed by three major reaction pathways involving NO 2 , O 3 , OH − and H 2 O (Gao and Marcus, 2001;Savarino, 2000;Michalski et al., 2003).The proportion of NO − 3 ,atm for the former reaction paths and corresponding precipitation rates can be obtained from global climate models, which finally results in 17 O NO 3,atm of about +25 ‰, a value well corroborated by atmospheric observations (Michalski et al., 2003;Morin et al., 2008;Alexander et al., 2009) For instance, for the Zimam aquifer with enhanced salinities and elevated NO − 3 concentrations up to 1.1 mM (El-Baruni et al., 1985;located close to #152), significant leakage into the underlying Cambro-Ordovician sandstone aquifer through fractures in limestone and dolostone is postulated (Binsariti and Saeed, 2000).To understand the groundwater evolution in more detail, a sophisticated modelling approach using more refined age models and all respective hydrogeochemical parameters must be considered.
An evaporation trend is reflected by neither solution chemistry nor the δD 3 concentrations traces longer lasting dry periods, and more barren soils corresponding to warmer climate conditions.Past alternating arid periods and infiltration events are suggested, which lead to the observed isotope values.The accumulated NO − 3 ,atm is washed out from plant and soil surfaces as well as below the near-surface and rooting zone downward through the vadose zone to the water table.A rapid infiltration through the subsurface without significant evaporation of water had recharged the Saharan paleo-groundwater of Hasouna.

Fig. 3 .Fig. 4 .
Fig. 3.Chemical composition of groundwate Cl -concentration.b: Ca 2+ versus NO 3 -conce green lines: Mixing between end-member so 15 highest ion concentrations, respectively.evaporation of solution #165, where 50 an respective loss of H 2 O by evaporation.Solid by considering calcite precipitation to reach NO 3 -in drinking water according to WHO (WH 20

Fig. 7 .
Fig. 7. a: 17 O-excess ( 17 O NO3 ) versus  18 O NO3 value of Hasouna groundwaters.b: 20 Mixing approach for #165 and #152 end-member solutions, where the 17 O-excess of dissolved NO 3 -and the reciprocal NO 3 -concentration (mM) can be followed by a linear relationship.Isotope values are related to the VSMOW standard.
and station, southern Tunisia (yellow line: D = 8  18 O + 13;Abid et al., 2012).Paleo MWL: Paleo Meteoric Water Line estimated from isotopic data of the present study by least square method (dashed green line: Eq.(

Fig. 7 .
Fig. 7. a: 17 O-excess ( 17 O NO3 ) versus  18 O NO3 value of Hasouna groundwaters.b: Mixing approach for #165 and #152 end-member solutions, where the 17 O-excess of dissolved NO 3 -and the reciprocal NO 3 -concentration (mM) can be followed by a linear relationship.Isotope values are related to the VSMOW standard.

Fig. 7 .
Fig. 7. a: 17 O-excess ( 17 O NO3 ) versus  18 O NO3 value of Hasouna groundwaters.b: 20 Mixing approach for #165 and #152 end-member solutions, where the 17 O-excess of dissolved NO 3 -and the reciprocal NO 3 -concentration (mM) can be followed by a linear relationship.Isotope values are related to the VSMOW standard.•:Hasouna groundwater, Libya (this study); ■ and ■: NO 3 --rich deposit of Atacama (Chile) and Mojave desert (California), respectively(Michalski et al., 2004a).25 H 2 O -δ 18 O H 2 O relationship.Low d excess as well as low δD H 2 O and δ 18 O H 2 O values indicate that the groundwater of the Hasouna basin was recharged under cooler and more humid climates compared to the current conditions along the coastal areas of Libya.The negative correlation between 17 O NO 3 values and reciprocal NO − . The δD H 2 O (analytical precision: ±0.8 ‰) and δ 18 O H 2 O values (±0.05 ‰) are given relative to the Vienna Standard Mean Ocean Water (VSMOW).As an example, the definition of δ values is given for the D/H distribution in H 2 O by the expression

www.biogeosciences.net/11/3149/2014/ Biogeosciences, 11, 3149-3161, 2014Table 1 .
Composition of Saharan paleo-groundwater from wells of the Hasouna area (Libya).No.: Well number (see Fig. 1).T: Temperature in • C. [ ]: Concentrations are given in mM, except for O 2(aq) in mg L −1 .eq : Derivation from electrical neutrality in meq %.Site positioning of the wells is given in coordinates X and Y and well depths, D, in metres.δ 18 O H 2 O and δD H 2 O values of the groundwaters are given in ‰ (VSMOW).δ 34 S SO 4 and δ 18 O SO 4 values for the isotope composition of dissolved SO 2− 4 are related to the VCDT and VSMOW standards, respectively, and given in ‰. δ 15 N NO 3 , δ 18 O NO 3 , δ 17 O NO 3 and 17 O NO 3 values of the dissolved NO − n.a.: not analysed . In contrast, microbially induced nitrification in soils results in 17 O NO 3,soil = 0 ‰.The individual 17 O excess of dissolved NO − 3 of the analysed Saharan groundwater, 17 O NO 3 , can be used to calculate the fraction of atmospheric NO − 3 vs. the total NO − 3 concentration according to the expression 17 O NO 3 .Considering the measured 17 O NO 3 values of the groundwater from 0.4 to 5 ‰ in Table 1 and 17 O NO 3,atm ≈ +25 ‰, individual proportions from 1.6 up to 20.0 % for atmospheric