Indicators of Petroleum Hydrocarbon Biodegradation in Anaerobic

Geomicrobiology Journal, 23:45–58, 2006 c Taylor & Francis LLC Copyright
ISSN: 0149-0451 print / 1521-0529 online DOI: 10.1080/01490450500399813

Indicators of Petroleum Hydrocarbon Biodegradation in Anaerobic Granitic Groundwater Sara Eriksson and Lotta Hallbeck Department for Cell and Molecular Biology—Microbiology, G¨oteborg University, Sweden

Tobias Ankner and Katarina Abrahamsson Department of Chemical Biosciences—Analytical and Marine Chemistry, Chalmers University of Technology, Sweden

˚ Sj¨oling Asa Department of Medical Microbiology and Immunology, G¨oteborg University, Sweden

strikingly similar to bacteria in other hydrocarbon-contaminated environments.

The aim of this study was to find indicators of petroleum biodegradation in granitic groundwater. Both pristine and contaminated groundwaters from boreholes around petroleum storage vaults located approximately 40 m below the surface in granite and with storage capacities of up to 120,000 m3 were sampled. Total numbers of microorganisms, “most probable numbers” (MPN) of anaerobic bacteria, and chemical indications of microbial activity were determined in the groundwater. Hydrocarbon contaminants and metabolites were detected using gas chromatographymass spectrometry (GC-MS). In contaminated groundwater, the total number of microorganisms was 2–4 × 106 ml−1 , which was significantly higher than the 6 × 104 ml−1 found in pristine groundwater. This microbial abundance was also reflected in the MPN analysis. Up to 7 × 104 nitrate-, 2 × 103 iron-, and 3 × 104 sulfate-reducing bacteria were detected in contaminated groundwaters. In such groundwaters, depletion of anaerobic electron acceptors and detection of reduced species could be established. We also proposed using a high alkalinity/hardness of water quota (A/H quota) as an indicator of microbial activity. In contaminated groundwaters the A/H quota averaged 2.8, while in pristine groundwater the same was only 1.3. Moreover, the presence of 20 oxidized petroleum hydrocarbons, i.e., putative metabolites of which 9 were strictly intracellular, was detected in the contaminated groundwaters. Phylogenetic neighbor-joining analysis of 16S rRNA genes provided information about the bacterial communities. The bacteria in contaminated groundwater were found to be

Keywords

16S rRNA, bacteria, diesel, gasoline, GC-MS, fracture, iron, MPN, nitrate, petroleum, sulfate

INTRODUCTION Petroleum products such as diesel and gasoline can be stored on beds of water in underground vaults below the groundwater table. In Sweden, such storage vaults are located about 40 m down in the granitic bedrock and can hold up to 120,000 m3 of petroleum (Figure 1A). When the storage vaults are emptied (Figure 1B), there is a potential risk that remaining fuel will contaminate fractures in the surrounding bedrock. This can be catastrophic for the groundwater quality and for recipients in the vicinity of the storage facilities, where the fractures eventually discharge their water. The hydrology and geochemistry in granite is different from sedimentary rock and soil. Hydrocarbons do not penetrate the matrix of granite (Lor´en et al. 2001), so the transport of petroleum contaminants is restricted to fractures. Underground storage vaults in granite are difficult to access, so when it comes to purification treatments, biodegradation by intrinsic bacteria in groundwater-filled fractures is crucial. As groundwater is often depleted in oxygen (Banwart et al. 1996), anaerobic degradation processes are the most important. Suitable indicators make it easier to monitor biodegradation in contaminated subsurface areas. In enrichment cultures from granitic groundwater, a previous investigation has found a unique degradation of the monoaromatic hydrocarbon propylbenzene to propylphenols under anaerobic iron-reducing conditions (Eriksson et al. 2005). Some investigations report that organic acids and alcohols can serve as indicators of biodegraded petroleum (Ball et al. 1996; Rabus and Widdel 1995). Many other chemical indicators of the biological degradation

Received 16 May 2005; accepted 19 July 2005. We sincerely thank Lars-Olof Mostr¨om (Sweden) for his excellent help in sampling the groundwater, Professor Karsten Pedersen (Sweden) for his most appreciated comments regarding this manuscript, and Ernest Chi for his valuable help with the phylogenetic cladogram. This study was financially supported by the Swedish Geological Survey (SGU). Address correspondence to Sara Eriksson, CMB—Microbiology, Box 462, SE-405 30 G¨oteborg, Sweden. E-mail: sara.eriksson@ gmm.gu.se

45

46

S. ERIKSSON ET AL.

certain physiological groups of bacteria dominate in the environments and whether the existing species are related to known degraders (Bekins et al. 1999; Kao et al. 2001; Watanabe et al. 2000). This field study is the first to survey indications of microbial activity and consequently possible petroleum hydrocarbon degradation specifically in contaminated groundwaters in granite. To do so, the abundances of nitrate-, ferric iron-, and sulfatereducing bacteria, as well as chemical indicators of biological activity, were examined in contaminated and pristine groundwaters from boreholes located around four different emptied diesel and gasoline storage vaults. Moreover, the microbial populations of two contaminated and one pristine groundwater samples were evaluated by sequencing the bacterial 16S rRNA genes and investigating their genetic affiliations. MATERIALS AND METHODS Chemicals Unless otherwise specified, all chemicals used in this study were purchased from VWR International (G¨oteborg, Sweden).

FIG. 1. Schematic picture of a storage vault in granite. (A) The high pressure from the groundwater confines the stored petroleum product to the vault. During operation, the groundwater is pumped out from the bottom of the rock vaults. This creates a depression cone in the groundwater level, which results in a constant flow of groundwater towards the vault and prevents petroleum migration from the vault. (B) At the time of decommission of the facility, pumping will stop and the depression cone will eventually disappear. Groundwater starts to flow through the vault. Hydrocarbons from remaining petroleum bodies in rock cavities and fractures and traces of petroleum in the storage vault can dissolve in the groundwater and be transported into fractures in the rock.

of petroleum in anaerobic environments have also been identified. The indicators include evidence of microbial activity, such as higher alkalinity, and depletion of electron acceptors, ammonium, and phosphate, compared to pristine groundwater (Beller 2000; Kao et al. 2001). It is also possible to examine the anaerobic electron-accepting processes and the 16S rRNA gene sequences of the microbial populations to determine whether

Study Sites The groundwater samples were collected in 2002 from boreholes located in the granite around four petroleum-contaminated storage vaults in Sweden, as summarized in Table 1. Dieselcontaminated groundwaters from borehole A in Murjek (67◦ N, 21◦ E) and KBH03 in Ludvika (60◦ N, 16◦ E) and diesel- and gasoline-contaminated groundwater from borehole KB11 in Bl¨adinge (57◦ N, 15◦ E) were examined. In KB11, a layer of petroleum was present on top of the water phase. From borehole BH1 in Sala (60◦ N, 17◦ E), gasoline-contaminated groundwater was sampled and examined. Pristine groundwater from borehole HB01 in Bl¨adinge, located approximately 200 m from the vault, was also investigated. In addition, another groundwater from a borehole in Bl¨adinge, KM01, was sampled. As was the case in Bl¨adinge KB11, the groundwater from Bl¨adinge KM01 was also contaminated with a layer of diesel and gasoline. Sampling Procedures Groundwaters from the contaminated boreholes Murjek A, Sala BH1, Ludvika KBH03, and Bl¨adinge KB11 were sampled using N2 -driven Integra bladder pumps made of stainless steel (Solinst Canada Ltd., Georgetown, Ontario, Canada) connected to a Model 460 manual control unit. The pump was installed at sampling depth in the boreholes and the groundwater was pumped from the sampler to the surface via polytetrafluoroethene (PTFE) tubing. The pumps were cleaned with detergent before installation and frozen in protective ice blocks to avoid contamination with petroleum from the surface water during submersion. The pumps were installed 18 months before sampling, except in the Ludvika KBH03 borehole where the pump was installed the day before the sampling. The first 500 ml of

47

21

40 Integra bladder pump Chemistry, AODC, MPN, hydrocarbon chromatograms, metabolites,

16

27 Discrete interval

sampler Chemistry, AODC, MPN, hydrocarbon chromatograms, metabolites, 16S rRNA gene sequencing

Analyses performed

June

June

Sampling month (2002) Depth of water surface in borehole (m) Sampling depth (m) Sampling equipment

Gasoline

None

Sala BH1

Contamination

Pristine Bl¨adinge HB01

pump Chemistry, AODC, MPN, hydrocarbon chromatograms, metabolites,

51 Integra bladder

49

May

Diesel

Murjek A

pump Chemistry, AODC, MPN, hydrocarbon chromatograms, metabolites,

25 Integra bladder

15

June

Diesel

Ludvika KBH03

TABLE 1 Compilation of the sampled groundwater at different locations in Sweden

pump Chemistry, AODC, MPN, hydrocarbon chromatograms, metabolites, 16S rRNA gene sequencing

39 Integra bladder

33

Diesel and Gasoline June

Bl¨adinge KB11

sampler Chemistry, AODC, 16S rRNA gene sequencing

37 Discrete interval

32

Diesel and Gasoline October

Bl¨adinge KM01

48

S. ERIKSSON ET AL.

groundwater pumped from the boreholes was discarded; the ensuing fresh groundwater was collected and analyzed. The pristine groundwater from Bl¨adinge HB01 and the contaminated groundwater from Bl¨adinge KM01 were sampled using a hexane-rinsed N2 -pressurized Solinst Discrete Interval Sampler, Model 425 (Solinst Canada Ltd., Georgetown, Ontario, Canada), which was lifted to the surface after sampling as previously described by Eriksson et al. (2005). To avoid contamination from the surface during sampling, the sampler was equipped with an ice plug around the sampler orifice during submersion. Petroleum Hydrocarbon Analyses of the Studied Groundwater The groundwater samples collected directly from the PTFE tubing from the Integra bladder pump, or directly from the Discrete Interval Sampler, in hexane-rinsed aluminum jars and instantly frozen in ethanol-soaked dry ice at the sampling site. Thereafter, the samples were stored at −20◦ C until analysis with gas chromatography–mass spectrometry (GC/MS). Petroleum hydrocarbons from 1 l of groundwater were extracted with nhexane (3 ml l−1 ). Of the organic phase, 1 µl was injected splitless into a Varian 3400 gas chromatograph (Varian, Malm¨o, Sweden) equipped with a Varian Saturn 2000 mass spectrometer as previously described in Eriksson et al. (2005). The chromatograms for aromatic hydrocarbons were qualified with the fragments m/z = 77 and 91 by means of specific ion mass spectrometry in full ionization mode, and for alkanes with the fragments m/z = 57 and 71. Total Number of Microorganisms Duplicate samples for determining the total numbers of microorganisms were collected in 50-ml Falcon tubes and preserved by adding neutralized formaldehyde to a final concentration of 2%. The total numbers of microorganisms in the samples were determined using the AODC method (Hobbie et al. 1977). Samples were filtered onto pre-stained Sudan black filters (0.2-µm pore size, 13 mm diameter, Osmonics, Minnetonka, MN, US) and stained with 0.2 ml acridine orange (10 mg l−1 , Sigma-Aldrich, Stockholm, Sweden) for 5 minutes. The cells were counted in an epifluorescence microscope (Olympus BH-2, Olympus optical AB, Malm¨o, Sweden) using blue light (390– 490 nm). Significant differences between mean values at the 95% confidence level were established using oneway ANOVA with a subsequent Student–Newman–Keuls test (n = 2) using the SPSS 12.0.1 software (SPSS Inc., Chicago, IL, USA). Most Probable Number (MPN) Groundwater samples for the MPN analysis were collected directly from the PTFE tubing from the Integra bladder pump, or directly from the Solinst Discrete Interval Sampler, in sterile N2 -filled 500-ml infusion flasks sealed with gas-tight rubber stoppers (DUMA, G¨oteborg, Sweden). Samples were immedi-

ately transferred from the infusion flasks, using 50-ml syringes and sterile needles, into N2 -filled 100-ml serum bottles sealed with butyl rubber stoppers (Bellco Glass, Vineland, NJ, USA), and stored at 4◦ C until used as inocula. All inoculation took place within 6 hours of collection, except for the sample from Murjek A, which was used as an inoculum 48 hours after sampling. MPN was investigated for three types of anaerobic microorganisms: nitrate-reducing (NRB), iron-reducing (IRB), and sulfate-reducing bacteria (SRB). The synthetic salt medium used in MPN analysis consisted of a basal medium, which contained (l−1 double-distilled water): 0.2 g KH2 PO4 , 0.3 g NH4 C1, 0.1 g MgCl × 6 H2 O, 0.1 g CaCl2 × 2 H2 O, 0.05 g KNO3 , 0.1 g MgSO4 × 7 H2 O, and 0.05 mg resazurine. The medium was autoclaved and cooled under a 95:5 gas mixture of N2 :CO2 . The following solutions were added to the medium according to Widdel and Bak (1992), except as otherwise indicated (l−1 medium): 1 ml non-chelated trace element mixture (Widdel et al. 1983), 1 ml selenite-tungstate solution, 1 ml vitamin mixture, 1 ml thiamine solution, 1 ml vitamin B12 solution, and 30 ml 1M NaHCO3 ; pH was then adjusted to between 7.0 and 7.2 in the medium and 5.5 ml 1−1 sodium lactate (0.05% w/w) was added as a carbon source. The medium was portioned into anaerobic 1000-ml infusion flasks, sealed with butyl rubber stoppers (DUMA, G¨oteborg, Sweden) and aluminum screw caps, and placed in an anaerobic box (COY Laboratory Products, Grass Lake, MI, USA) under an atmosphere of 2% H2 , 5% CO2 , and 93% N2 . In addition, for the nitrate-reducing medium (NRM), 0.9 g l−1 NaNO3 was added as an electron acceptor before autoclaving and 0.5 g l−1 L-cysteine hydrochloride × H2 O was added as a reducing agent when the medium had cooled. In the ferric iron-reducing medium (ERM), 100 ml l−1 of approximately 0.1 M sterile poorly crystalline 2line ferrihydrite (FeOOH) (Eriksson et al. 2005) was added as an electron acceptor inside the anaerobic chamber after autoclaving. Sulfate-reducing medium (SRM) was prepared by adding 3.0 g l−l Na2 SO4 as an electron acceptor before autoclaving and 0.2 g 1−l Na2 S as a reducing agent when the medium had cooled. Inside the anaerobic chamber the medium was portioned in 9-ml aliquots into anaerobic, sterile Hungate tubes, and sealed with sterile butyl rubber stoppers and aluminum crimps. The first step in the MPN analysis of NRB and SRB was decimal dilution of the groundwater samples by two to five times and in the MPN analysis of IRB was decimal dilution by zero to three times. One ml of each dilution was inoculated in eight replicates of the medium. All transfers of bacteria to tubes were done using sterile, N2 -flushed 1-ml syringes and sterile needles. The MPN tube samples were incubated at 17◦ C for at least 10 weeks in the dark without shaking. Medium without inoculum served as the negative control. MPN tubes were analyzed for total numbers of microorganisms, reduced species and electron acceptor consumption; they were considered positive if the total numbers of microorganisms increased and the amounts of metabolic products were double or the electron acceptors were half that of

HYDROCARBON BIODEGRADATION IN GRANITIC GROUNDWATER

the negative controls. Sulfate was measured using SulfaVer 4, sulfide using sulfide reagents 1 and 2, nitrate using NitraVer5, and nitrite using NitriVer3 as directed by the manufacturer (all were obtained from HACH Europe, D¨usseldorf, Germany). Separate standard curves were made and absorbance was measured spectrophotometrically with a Novaspec II spectrophotometer (Amersham Pharmacia Biotech, Uppsala, Sweden). Ferrous iron was detected using the ferrozine method of Stookey (1970), modified as previously described in Eriksson et al. (2005). The total number of microorganisms was determined with the AODC method described above. Finally, MPN for NRB, IRB, and SRB were calculated using the Klee program for determining most probable numbers and associated 95% confidence levels (Klee 1993). Chemical Characterization of the Groundwater Groundwater was collected directly from the PTFE tubing from the Integra bladder pump, or from the Solinst Discrete Interval Sampler, in special vessels provided by ALcontrol (J¨onk¨oping, Sweden). This laboratory, accredited by the Swedish Board for Accreditation and Conformity Assessment (SWEDAC), also performed the analysis. The alkalinity and total hardness of water, and the amounts of sulfate, sulfide, nitrate, nitrite, ammonium, total iron, and phosphate were determined. The ferrous iron content in the groundwater was determined using a modified version of the analysis method of Lovley and Phillips (1987), in which 0.2 ml Ferrozine solution (0.25%, Sigma-Aldrich, Stockholm, Sweden) and 0.8 ml ammonium acetate (4M) were added to 2 ml groundwater immediately after sampling. A purple complex showed that ferrous iron was present in the groundwater. The same procedure was performed in another sample to which 0.1 ml hydroxylammonium chloride (10%) was also added as a reducing agent. In this sample all iron was converted to ferrous iron, thereby forming the purple complex. Ocular comparison of the two samples showed if ferrous iron was present in the groundwater. Detection of Putative Metabolites The groundwater was collected directly from the PTFE tubing from the Integra bladder pump, or directly from the Solinst Discrete Interval Sampler, in hexane-rinsed aluminum jars and frozen instantly at the sampling site. Thereafter, the samples were stored at −20◦ C until analysis. Organic acids and alcohols, i.e., putative metabolites, were detected using a modified version of the protocol of Fogelqvist et al. (1980). Both intraand extracellular metabolites were examined and the samples were treated in different ways. Extracellular metabolites were extracted with n-hexane (3 ml l−1 ) directly from 1 l groundwater by means of vigorous shaking for 45 min. The samples were acidified prior to extraction with 0.5 ml concentrated sulfuric acid. After extraction, the organic phase was placed in a vial and concentrated by N2 evaporation to dryness. To ensure that only extracellular metabolites

49

were extracted, the total numbers of microorganisms in the water phase before and after extraction of a representative sample were determined as described above. These numbers should not differ. To study intracellular metabolites, 100 ml of sample was filtered through a sterile 0.22-µm pore size PTFE filter (Osmonics, Minnetonka, MN, USA) using a sterilized glass syringe and a 23-mm diameter filter holder (Millipore Corporation, Bedford, MA, USA). The filters were placed in vials. Due to clogging of the filter, only 60 ml of the groundwater from Murjek was filtered. To all the vials, 0.5 ml dichloromethane with 0.05% pentafluorobenzylbromide (PFB-Br, Sigma-Aldrich, Stockholm, Sweden) was added as derivatization solution. After this, 0.5 ml NaOH (0.1M) and 50 µl tertiary-butylammoniumhydrogensulfate (0.4M, Sigma-Aldrich, Stockholm, Sweden) were added. The vial was tightly sealed and treated with heat (60◦ C) and ultrasonication for 30 min. Of the organic phase, 1 µl was injected splitless into the GC-MS as previously described in Eriksson et al. (2005). Oxidized hydrocarbons could be detected by screening for the m/z = 181 fragment of the derivatization molecule PFB and characteristic fragments of the various putative metabolites.

Microbial Community Structure The groundwater was collected directly from the PTFE tubing from the Integra bladder pump, or directly from the Solinst Discrete Interval Sampler, into sterile, N2 -filled 500-ml infusion flasks sealed with gas-tight rubber stoppers (DUMA, G¨oteborg, Sweden). Samples were immediately transferred from the infusion flasks, with 50-ml syringes and sterile needles, into N2 -filled 100-ml serum bottles sealed with butyl rubber stoppers (Bellco Glass, Vineland, NJ, USA), and stored at 4◦ C until processed. Filtration of microorganisms from groundwater, extraction of DNA, and amplification of the 16S rRNA genes with subsequent cloning and sequencing were performed to identify the bacteria at the different sites. To dissolve and remove iron, 20 ml of groundwater were mixed with 10 ml oxalic acid (0.1% in double-distilled water). The oxalic acid had been sterile filtered twice using Filtropur S filters (Sarstedt, N¨urnbrecht, Germany) and then autoclaved. After that, the groundwater was filtered through a sterilized 13-mm diameter filter holder (Millipore Corporation, Bedford, MA, USA) provided with a sterile 0.22-µm pore size PC filter (Osmonics Inc., Minnetonka, MN, USA) pretreated with UV light. The groundwater was filtered within 2 hours of sampling and the filters were immediately frozen until DNA extraction. DNA-free gloves were used at all stages of handling to prevent contamination. DNA extraction, PCR, cloning, and sequencing were conducted as previously described in Eriksson et al. (2005). Sterile controls with autoclaved double-distilled water were prepared for the DNA extraction and the PCR. DNA could not be extracted from or amplified in these negative controls.

50

S. ERIKSSON ET AL.

The diversity of 16S rRNA genes in the groundwater samples was examined by sequencing regions from Escherichia coli position 357 and about 450 base pairs forward (5 → 3 ), and comparing these to the GenBank database using BLAST www.ncbi.nlm.nih.gov/BLAST The BLAST was performed in January 2005. When performing the BLAST, all bacterial sequences from hydrocarbon-contaminated areas that showed the greatest similarity to those of the clones examined in this study were imported, together with the new clones from this study, into the ARB software (version 2.5B; O. Strunk and W. Ludwig, Technische Universit¨at M¨unchen, Munich, Germany). The sequences were aligned using the ClustalW DNA alignment algorithm provided in the ARB software and subsequently checked manually. An evolutionary distance cladogram using the neighbor-joining method with Jukes-Cantor correction integrated in the ARB software was constructed with sequences from E. coli positions 404 to 836, after masking ambiguously alignable segments. GenBank sequences of all characterized bacteria having the greatest similarity to the new clones were included in the cladogram, as well as the 16S rRNA gene sequences of bacteria from the various phyla to which the clones in this study were affiliated, i.e., the Gram-positive Bacillus subtilis (X60646), the Bacteroidetes bacterium Flavobacterium aquatile (M62797), the β-proteobacterium Azoarcus tolulyticus (L33692), the γ -proteobacterium E. coli (X80725), the ε-proteobacterium Helicobacter pylori (U00679), the αproteobacterium Caulobacter fusiformis (AJ227759), the acidobacterium Holophaga foetida (X77215), the Chlorobi bacterium Chlorobaculum parvum (Y10647), the Verrucomicrobia bacterium Optitus terrae (AJ229246), and the archaean Sulfolobus solfataricus (X03235). The reliability of the branching points was determined by performing 2000 bootstrap replications of the data.

Nucleotide Sequence Accession Numbers The sequences of the clones have been deposited at GenBank under accession numbers AY996564 to AY996587.

RESULTS Petroleum Hydrocarbon Analyses of the Studied Groundwater The presence of dissolved n-alkanes and aromatic hydrocarbons was investigated in the groundwaters from the boreholes at the study sites. As shown in Figure 2, both alkanes and aromatic hydrocarbons were present in groundwaters from Bl¨adinge KB11, Ludvika KBH03, and Murjek A. In the groundwater from Bl¨adinge KB11, a particularly large amount of aromatic hydrocarbons could be found. Alkanes or aromatic hydrocarbons could not be detected in the pristine groundwater from Bl¨adinge HBH01 or in the groundwater from Sala BHl.

FIG. 2. Chromatograms of aromatic hydrocarbons and alkanes from hexane extractions of the pristine groundwater from Bl¨adinge HB01 and the contaminated groundwater from Sala BH1, Murjek A, Ludvika KBH03, and Bl¨adinge KB11. Sampling was done in May and June, 2002. The chromatograms of aromatic hydrocarbons were qualified with the fragments m/z = 77 and 91 and of alkanes with the fragments m/z = 57 and 71, using specific ion mass spectrometry in full ionization mode.

Total Number of Microorganisms The total numbers of microorganisms were determined as shown in Table 2. The highest numbers, 2–4 × 106 ml−1 , were found in the contaminated groundwaters from Ludvika KBH03, Bl¨adinge KB11, Murjek A, and Bl¨adinge KM01. Compared to these numbers, a significantly lower count was established in pristine groundwater from Bl¨adinge HB01, with 6 × 104 microorganisms ml−1 . The number in the groundwater from Sala BH1 was 9 × 104 ml−1 , which was considerable although not significantly higher than the number found in the pristine groundwater from Bl¨adinge HB01. MPN In the MPN analysis, the numbers of NRB, IRB, and SRB were determined (Table 2). NRB were present in all examined

51

HYDROCARBON BIODEGRADATION IN GRANITIC GROUNDWATER

TABLE 2 Microbial composition of the examined groundwater from Bl¨adinge HB01, Sala BH1, Murjek A, Ludvika KB03, Bl¨adinge KB11, and Bl¨adinge KM01

Total number of microorganisms ml−1 × 105 (±SD) NRB ml−1 × 102 NRBb 95% confidence limits ml−1 × 102 IRB ml−1 IRBb 95% confidence limits ml−1 SRB ml−1 × 102 SRBb 95% confidence limits ml−1 × 103

Pristine Bl¨adinge HB01

Sala BH1

Murjek A

Ludvika KBH03

Bl¨adinge KB11

Bl¨adinge KM01

0.6 ± 0.03a

9 ± 2a

30 ± 6a

20 ± 7a

40 ± 2a

30 ± la

0.1 0.03–0.6

6 0.6–20

10 3–60

200 50–900

700 200–2000

200 20–600