Combination of nitrate (N, O) and boron isotopic ratios with microbiological indicators for the determination of nitrate sources in karstic groundwater

Nitrate produced by anthropogenic activities is an important environmental issue, particularly in agricultural regions and especially when water is used as drinking water. Determination of the source of nitrate among the many potential sources is the first step towards water quality improvement.

Recent studies have proven the efficiency of the coupled use of nitrate (N and O) and boron isotope ratios to identify the origin of nitrate in water.^[1–3] Nitrate from mineral fertilisers is characterised by atmospheric values of δ¹⁵N (~0 ‰) and δ¹⁸O (~23.5 ‰), whereas nitrate derived from organic fertilisers, livestock effluents and sewage is usually enriched in ¹⁵N (δ¹⁵N varies from +8 up to +20 ‰) because of ammonium volatilisation.^[4–8] As animal and domestic waste δ¹⁵N overlap, additional tracers such as boron are used to discriminate between these two sources.^[3]

Boron is a co-migrant of nitrate and a tracer of domestic water input to the environment.^[9] Furthermore, it is not affected by processes affecting nitrate (e.g. denitrification).^[3] Although fertiliser δ¹¹B exhibits a wide range of values (–10 to 60 ‰), these ratios are significantly different from sewage (–2 to +2 ‰) and manure contaminations (5–50 ‰).^[1–3,9]

In addition to this dual isotopic approach, we tested microbiological indicators (bacteriophages and Bacteroidales) to characterise the origin of faecal contamination occurring in groundwater.^[10–14] F-specific RNA bacteriophages (FRNAPHs) are viruses, infecting bacteria of the gastrointestinal flora, with genogroups associated with animal (GI and GIV) or human (GII and GIII) faeces.^[10–12,15] Bacteroidales is an order of Bacteria present in high numbers in the gastrointestinal tract of warm-blooded animals with host-specific distribution (humans, ruminants and pigs).^[13,14,16] The use of microbiological markers allows a better discrimination among organic contaminants and gives complementary information on the transfer times within an aquifer (with a lifetime lasting a few days in the environment).

This new multidisciplinary approach was used to determine the origin of increasing nitrate concentrations in karstic groundwater located in an agricultural area. The borehole studied (F1) was dug up to 100 m deep in a Cretaceous limestone outcropping at the top of an anticline (Fig. 1). It is used for drink-ing water production for the South Aquitaine region (SW France), but nitrate concentrations have steadily increased from 5 mg L^–1 in the 1970s to 25 mg L^–1 since 2004 and traces of faecal contaminants have frequently been detected in the groundwater. The limestone aquifer is fed by deep groundwater but also by local recharge in the carbonate outcropping area (Fig. 1). The karstic nature of limestone potentially favours fast communication from surface to groundwater.

Fig. 1.  Geographic location of the F1 borehole (white star) and geologic structure of the anticline. Magnification of the F1 borehole and the conceptual alimentation model.

Water samples were regularly collected from this borehole from October 2010 to October 2012 at the pump outlet. Multilevel sampling was carried out twice (February and September 2011) using a mini pump within the borehole. All the samples were filtered through a 0.45-µm nylon membrane.

For nitrate concentrations, samples were stored frozen and were measured by high-performance liquid chromatography (HPLC Dionex, AS12 column). For the characterisation of the N and O isotopic composition of nitrate, samples were poisoned with HgCl₂ and analysed after reduction to nitrite through a granular cadmium-filled column. Nitrite was then converted into nitrous oxide by adding azide.^[17] The determination of δ¹⁵N and δ¹⁸O of N₂O was performed through a purge-and-trap and continuous-flow isotope ratio mass spectrometry system (DeltaVplus Thermo coupled with Gas Bench II). The method was calibrated with nitrate standards (USGS-32, δ¹⁵N = 180 ‰, δ¹⁸O = 25.7 ‰; USGS-34, δ¹⁵N = 1.8 ‰, δ¹⁸O = 27.9 ‰ and USGS-35 δ¹⁵N = 2.7 ‰, δ¹⁸O = 57.5 ‰). The linearity of the analysis was checked with an internal nitrate standard (IAEA, δ¹⁵N = 4.7 ‰, δ¹⁸O = 25.6 ‰). The precision was 1.2 ‰ for δ¹⁵N and 1.3 ‰ for δ¹⁸O.

For boron analyses, samples were acidified to pH 2 with HNO₃. Boron concentrations were determined by inductively coupled plasma optical–atomic emission spectroscopy (ICP OES/AES) JY2000. The ¹¹B/¹⁰B isotopic ratios were measured by a multicollector inductively coupled plasma mass spectrometer (Neptune, ThermoScientific) using a direct-injection nebuliser (d-DIHEN, Analab).^[18] Boron was first extracted from the samples using ion exchange chromatography. A volume of 5–30 mL of the samples at pH 9 was introduced into a column filled with 50 µL of resin (Amberlite IRA-743). Boron was retained on the resin and then eluted with dilute HNO₃ (0.1 and 0.5 N).^[18,19] δ¹¹B values were calculated by bracketing sample measurements with NBS-951 boron standard measurements. The external reproducibility of δ¹¹B measurements for natural water samples was 0.25 ‰ (2 s.d.).

For microbiological analyses, specific sampling bottles with sodium thiosulfates were used. Standardised methods were used to determine the concentration of total coliforms and Escherichia coli, Enterococcus species and total aerobic mesophilic flora.^[20–22] FRNAPHs were enumerated by concentrating 3 L of a water sample using the membrane filtration–elution method.^[23] Infectious FRNAPHs were counted (double agar-layer technique).^[24] FRNAPHs contained in one plaque were collected, re-suspended in 1 mL of phosphate buffered saline (PBS) with 15 % glycerol and stored at –20 °C until genotyping, which was performed using a one-step real-time reverse-transcription polymerase chain reaction (RT-qPCR) kit (QuantiTech Probe RT–PCR, Qiagen) and previously described primers and probes.^[11] To search for Bacteroidales, 2 L of sample water were filtered through a 0.22-µm pore size polycarbonate membrane. The filter was immersed in a guanidinium thiocyanate (GITC) lysis solution and stored at –80 °C until DNA extraction with a Qiamp DNA minikit (Qiagen). Previously described primers and probes were used to quantify ruminant (Rum-2-Bac),^[14] pig (Pig-2)^[13] and human (HF183)^[16] Bacteroidales markers. Standard curves were calculated for plasmids containing the target sequence. The presence of PCR inhibitor factors was monitored by adding a known plasmid concentration to the sample. The results were expressed as several copies in 100 mL of water. PCR reactions were performed in duplicate for each sample with a Rotor gene 6000 thermocycler.

Nitrate concentrations measured in the borehole varied from 19.4 to 25.2 mg L^–1. The lowest values (19.4 and 19.6 mg L^–1) correspond to the deepest samples (85 m) from February and September 2011. This is consistent with the conceptual model of a double alimentation of the borehole: deep and slightly contaminated water and nitrate-contaminated surface water with a rapid infiltration rate.

The δ¹⁵N and δ¹⁸O ranges usually reported in the literature for the different potential sources of nitrate are presented in Fig. 2 with our isotopic data for the groundwater samples. Our δ¹⁵N and δ¹⁸O data fall within the organic boxes (sewage, manure or soil organic nitrogen-derived nitrate) with respective values of 5.6–9.4 and 3.0–6.1 ‰ (Table 1, Fig. 2).

Fig. 2.  Nitrate isotopic values of F1 water and comparison with those noted in the literature as potential sources of nitrate.^[1–3]

Table 1.  Nitrate and boron concentrations and associated isotopic compositions
F1, samples collected at the pump outlet; F1-x m, sample collected at a depth of x m as indicated. CFU, colony forming units; Inv. flora, Invasive flora; PFU, plaque forming unit

Boron concentrations and isotopic compositions were measured for six samples, chosen to be representative of temporal and spatial variations. B concentrations varied only slightly, with an average value of 10.6 ± 0.7 µg L^–1. The δ¹¹B of these samples are rather homogeneous with an average value of 25.3 ± 0.3 ‰ (Table 1, Fig. 3). Combined δ¹⁵N and δ¹¹B (Fig. 3) data correspond to values usually reported for nitrate originating from an animal source. It excludes nitrate derived from domestic effluents. This could be derived from manure spreading on maize fields or from localised livestock effluents.

Fig. 3.  Relationship between δ¹⁵N–NO₃^– and δ¹¹B of F1 water and comparison with those noted in the literature as potential sources of nitrate and its co-migrant boron.^[4–6]

The occurrence of the total viable count in the water of the F1 borehole ranged between <1 and 282 CFU mL^–1 (colony-forming units per millilitre). Low coliform bacteria concentrations were observed in most water samples. However, standard faecal indicators (E. coli and Enterococcus) were detected in only three campaigns – September 2011, August 2012 and October 2012 – with concentrations under 4 CFU per 100 mL. FRNAPHs were detected in February 2011 at different depths of the F1 borehole with 3, 1, 11, 9 and 10 plaque forming units (PFU per 3 L) at respective 30-, 41-, 52-, 73- and 85-m depths. These concentrations are actually the same order of magnitude and cannot be considered significantly different for the five depths. Genotyping methods concluded that all these phages belong to genogroup I, which is generally associated with animal faecal pollution. Nevertheless, more extensive FRNAPH genotyping would be necessary to corroborate these preliminary results. Moreover, none of the host-specific Bacteroidales markers (human, porcine or ruminant) were detected, suggesting that another animal effluent could be responsible for faecal contamination occurring in the spring (duck and chicken farming are dominant in the studied area). Furthermore, the very short life span of microorganisms in groundwater (<2 days for bacteriophages)^[25] proves that transfer from the surface to groundwater could be very fast, possibly favoured by karstic channels.

The determination of the origin of nitrate is an essential preliminary step for water resource management and remediation. In agricultural areas, where there can be multiple nitrate sources, the combination of nitrate and boron isotopic composition can be very helpful to differentiate the different sources. Moreover, in the case of faecal contamination, and when groundwater potentially receives direct infiltration from the surface, characterisation of microorganisms provides valuable information on the source of the contamination.

This new approach was applied here to the case of a karstic spring where nitrate concentrations have gradually increased over the last 40 years and where episodic faecal contaminations are measured.

The combined approaches indicate that borehole F1 water is contaminated with nitrate from animal effluents. The existence of rapid surface water transfer due to the karstic nature of the aquifer allows the detection of these specific microorganisms.