Abstract
Limestone aquifers provide the main drinking water resources of southern Italy. The groundwater is often contaminated by fecal bacteria because of the interaction between rocks having high permeability and microbial pollutants introduced into the environment by grazing and/or manure spreading. The microbial contamination of springwater in picnic areas located in high mountains can cause gastrointestinal illness. This study was carried out in order to analyze the interaction between Enterococcus faecalis and the soil of a limestone aquifer and to verify the influence of this interaction on the time dependence of groundwater contamination. E. faecalis was chosen because, in the study area involved, it represents a better indicator than Escherichia coli. The research was carried out through field (springwater monitoring) and laboratory experiments (column tests with intact soil blocks). The transport of bacterial cells through soil samples was analyzed by simulating an infiltration event that was monitored in the study area. Comparison of laboratory results with data acquired in the field showed that discontinuous precipitation caused an intermittent migration of microorganisms through the soil and produced, together with dispersion in the fractured medium (unsaturated and saturated zones), an articulated breakthrough at the spring. The short distances of bacterial transport in the study area produced a significant daily variability of bacterial contamination at the field scale.
Limestone aquifers provide the main drinking-water resources of southern Italy, producing an average of 3,700 × 106 m3 of water/year, for an average yield of about 0.027 m3/s/km2 (5, 7). Due to grazing and the high permeability of fractured rocks, significant quantities of microbial contaminants are rapidly transported into the subsurface and microbiological pollution of groundwater is often caused (4, 6). Significant contamination has also been detected where fractured limestone underlies soil media (6), even if a significant microorganism storage capacity characterizes the soil (20). Even though no persistent microbial contamination has been detected (6), cases of gastroenteritis have often been reported and associated with springwater in picnic areas located in high mountains. A thorough knowledge of the time dependence of microbial pollution and its relationships to several environmental factors (i.e., precipitation, temperature, soil, and fractured medium) are needed to predict the transport modalities of fecal bacteria through the aquifers and prevent gastrointestinal illness.
The purpose of the present study was to determine whether and how much the distribution of precipitation versus time and the migration of fecal bacteria through the soil influence the breakthrough of microorganisms at springs in limestone aquifers. The determination of the influence of soil on these phenomena required the development of flowthrough column tests in intact soil blocks by using a strain of Enterococcus faecalis. The breakthrough curves obtained through laboratory experiments were compared to results of daily microbiological monitoring at a spring. Due to the seasonality of the spring analyzed, the data were collected weekly or daily from October 2002 to May 2003. The study was developed through a detailed analysis of the effects produced in a small fractured aquifer during two different kinds of rainfall events.
Description of field site.
The study area is located in southern Italy and consists of calcareous deposits (Cretaceous-Oligocene; Monte Coppe, Coste Chiaravine, Monte Calvello and Monaci formations) (Fig. 1) (9). The rocks have very low primary permeability but are extensively fractured. The saturated aquifer comprises more than 100 m of fractured limestone, while the unsaturated medium is made up of fractured limestone and soil.
FIG. 1.
Groundwater flow scheme. 1, aquifer boundary; 2, spring; 3, well; 4, piezometer; 5, groundwater flow direction; 6, location of soil block extraction; 7, limestones; 8, pasture area.
In pasture areas, epilepti-vitric (mollic) andosols (11), which are characterized by an A/R profile (i.e., where the horizon, A, is directly superimposed on the bedrock, R), crop out. The A horizon generally ranges in thickness from 4 to 12 cm. It is made of sands (40%), silts (35%), and clays (25%). Its pH ranges from 5.1 to 5.4, while its organic matter percentage is between 9.1 and 10.7% and its total Kjeldahl nitrogen percentage is 0.5 to 0.6%. Grazing (by a few hundred head of cattle) is allowed from May to October in a zone which represents 35.7% of the study area, even though it is concentrated mainly within a radius of about 250 m from the spring; 64.3% of the aquifer is covered by beech woodland (Fig. 1).
The unconfined aquifer is bordered by low-permeability marly rocks and/or normal faults which have produced significant cataclastic zones, characterized by low permeability. The groundwater flows from southeast to northwest, to the spring analyzed (1,012 m above sea level; Fig. 1), which has an average discharge of about 0.01 m3/s. Water levels in the limestone aquifer fluctuated by several meters due to the low effective porosity of the fractured rocks (Fig. 2).
FIG. 2.
Groundwater level fluctuations. a.s.l., above sea level; ott, October; nov, November; dic, December; gen, January; feb, February; mar, March; apr, April; mag, May.
Water level fluctuations in a nearby piezometer (10 m away) in response to pumping from a well showed that the limestone aquifer is laterally well connected in the subsurface. A pumping test in which the well was pumped for 4 h caused 0.51 m of draw-down in the piezometer. Results of this test yielded a transmissivity value of 5 × 10−4 m2/s and a storage coefficient of 3 × 10−4.
The Thornthwaite water budget method (19) was used to provide an estimate of net infiltration. An estimate of runoff was obtained by utilizing the experimental results of surface water monitoring in different catchment areas of the Italian limestone Apennines (3). The results obtained indicate that all groundwater recharge during an average year occurs from October to June. The major recharge event occurs from December to April. In July, August, and September, the weather is warm and evapotranspiration generally exceeds precipitation. The annual average rainfall level in the study area is 1,240 mm; the annual average net recharge is 630 mm (about 51% of the annual average rainfall level). These values were estimated on the basis of precipitation and temperature data recorded for a period of 80 years (1921 to 2000).
MATERIALS AND METHODS
Strain and medium.
A collection strain of E. faecalis (ATCC 29212) that is nalidixic acid resistant was aerobically cultured at 37°C in Luria-Bertani liquid or on solid medium supplemented with an antibiotic (20 μg of nalidixic acid/ml) (18).
Microbiological and pH monitoring.
Springwater samples were collected in sterile 1,000-ml bottles and transported in a refrigerated box to the laboratory weekly or daily from October 2002 to May 2003. Spring outflow was sterilized by flame to ensure that the water samples were not contaminated at the surface. Filtration processes for bacteriological analyses were carried out within 2 h or less after collection. Indicators of microbial contamination were determined by classic methods of water filtration (1,000 and 100 ml of water sample) with sterile GN-6 Metricel membrane filters (pore size, 0.45 μm; Pall), with incubation on m-Endo Agar LES (Biolife) for 24 h at 35°C for total coliforms, m-FC agar for 24 h at 44°C for fecal coliforms, and Slanetz-Bartley agar for 4 h at 35°C and 44 h at 44.5°C for fecal enterococci. pH was measured with a WTW Multi 340i pH meter.
Speciation of enterococcal isolates from springwater and rRNA gene amplification.
Taxonomic classification of fecal enterococci detected in the springwater samples was performed by use of API 20 Strep fermentation strips (bioMérieux, Marcy l'Etoile, France) and by sequence analysis of one of the 16S rRNA genes amplified with two universal oligonucleotides: P1 (5′-GCGGCGTGCCTAATACATGC) and P2 (5′-CACCTTCCGATACGGCTACC), annealing to nucleotides 40 to 59 and 1532 to 1513, respectively, of Bacillus subtilis rrnE.
Soil block extraction.
Three intact soil blocks of epilepti-vitric (mollic) andosols were extracted from pasture area at the study site (Fig. 1). To minimize the disturbance of samples, sod-covered blocks (181.36 by 181.36 by 11 cm) were carved from undisturbed soil directly by putting permeameter cells used for column tests into the soil itself. All blocks were covered with plastic and transported to the laboratory, where the experimental procedure started immediately.
Simulation of bacterial transport through soil blocks.
A diffuse interaction between bacteria and soil blocks was obtained by developing column tests with a standard permeameter (catalog no. S248; MaTest, Treviolo, Italy) to minimize lateral flow within the gap between the soil block and the metal cell. The rainfall was applied to the tops of blocks. The outflow was collected at the bottom by using sterile plastic tubes. A peristaltic pump (catalog no. 505S/RL; Watson-Marlow, Wilmington, Mass.) was used to sustain a constant flow through the blocks.
The real precipitation monitored in the field from 6 to 9 January was simulated (Table 1). Due to the low frequency (about 3 mm h−1), complete infiltration into the aquifer was hypothesized.
TABLE 1.
Main characteristics of simulated infiltration
| Step | Infiltration (mm) | Frequency (mm h−1) | Length of dry interval (h) |
|---|---|---|---|
| 1 | 15.2 | 3 | |
| 2 | 10 | ||
| 3 | 10.8 | 3 | |
| 4 | 23 | ||
| 5 | 10.8 | 3 | |
| 6 | 11 | ||
| 7 | 20.4 | 3 |
Due to the significant clay content of the soil (25%), a solution with 0.001 M CaCl2 was used as rainwater to prevent the dispersion of clays within the soil and the column plugging (15).
Due to the results of field monitoring, which showed the high reliability of fecal enterococci as indicators of microbial contamination in the study area (see below), the interaction between fecal bacteria and soil blocks was analyzed through the utilization of a collection strain of E. faecalis (ATCC 29212), which is nalidixic acid resistant. No bacteria resistant to nalidixic acid were observed in the natural background of soil blocks collected in pasture areas.
At the beginning of the experiments, 0.75 × 109 E. faecalis cells (collected during the exponential growth phase) were applied to the top of each block in a 0.001 M CaCl2 solution.
Soil block drainage was collected in 10-ml sterile plastic tubes beneath the outflowing holes. Two hundred microliters of each water sample and relative serial dilutions were plated in triplicate on Luria-Bertani solid medium supplemented with an antibiotic (20 μg of nalidixic acid/ml) and incubated at 37°C. After 24 h, the number of E. faecalis cells (CFU) was estimated by utilizing only the plates for which the numbers of colonies ranged from 30 to 300.
RESULTS
Field monitoring.
Microbial pollution of groundwater was produced by cattle grazing, which is the only source of fecal microorganisms within the study area. The time dependence of fecal contamination showed series of peaks that were irregularly distributed (Fig. 3) according to the results of other studies of limestone aquifers of southern Italy (6). The existence of several spikes in bacterial concentration is well correlated with the sequence of different precipitation events which were able to produce effective infiltration (Fig. 2) and transport of microorganisms from the surface to the groundwater. The absence of spikes during the large precipitation event of late January 2003 is due to the fact that it was impossible to reach the spring every week. The spring is generally characterized by low pollution, even though the concentrations seem to show a general decrease from October 2002 through May 2003. This decrease is probably due to different factors, such as an increase in dilution and a decrease of microorganisms in the soil. The decrease of microorganisms in the soil is probably caused mainly by removal (transport of bacteria from the soil to the groundwater) and decay (produced by competition, freeze-thaw events [14a], etc.). The removal of bacteria can play an important role because no grazing is allowed from December through April. No fecal coliforms were observed in many contaminated water samples (number of fecal enterococci, ≥1) (Table 2). Hence, fecal enterococci are a more reliable indicator than fecal coliforms for the detection of microbial pollution at the study site. These differences may be due to different factors: (i) animal feces are characterized by a ratio of fecal coliforms to fecal enterococci below 0.7 (12), (ii) fecal enterococci are more resistant in the environment than fecal coliforms (13), and (iii) fecal coliforms and fecal enterococci vary considerably in terms of size, morphology, motility, and surface chemistry, which leads to substantive differences in their propensities for attachment to solid surfaces within soils and aquifers (2, 14).
FIG. 3.
Concentrations of fecal enterococci at the spring (line) and rainfall (bars) versus time. ott, October; nov, November; dic, December; gen, January; feb, February; mar, March; apr, April; mag, May.
TABLE 2.
Distribution of fecal coliforms and fecal enterococci at the spring analyzeda
| Microorganisms | % of samples with 0 CFU/100 ml | % of samples with ≥1 CFU/100 ml | Range of contamination (CFU/100 ml) |
|---|---|---|---|
| Fecal coliforms | 78.0 | 22.0 | 0-100 |
| Fecal enterococci | 42.4 | 51.6 | 0-94 |
Fifty-nine samples were tested.
The effects produced by two different kinds of precipitation events were thoroughly analyzed through daily monitoring. One event was characterized by a sequence of rainy steps and nonrainy intervals and produced the breakthrough curve for the spring that is shown in Fig. 4. The contamination event detected at the spring from 6 to 9 January showed that the breakthrough of fecal bacteria started 1 day after infiltration and was finished after 10 days. The maximum concentration (33 CFU/100 ml) was observed after 2 days. A significant increase in concentration was detected after 5 days, with a new peak after 6 days (24 CFU/100 ml). Less articulated was the breakthrough curve produced by the second event (Fig. 5), characterized by just 1 day of rainfall without dry intervals.
FIG. 4.
Daily concentration of fecal enterococci at the spring (line) and rainfall (bars) versus time from 6 to 16 January 2003 (dates are given as day/month/year).
FIG. 5.
Daily concentration of fecal enterococci at the spring (line) and rainfall (bars) versus time from 26 to 30 March 2003 (dates are given as day/month/year).
Field-measured pH values of water samples ranged from 6.9 to 7.6, indicating neutral-alkaline conditions. This finding is similar to those for other aquifers where the geochemistry is dominated by carbonate-mineral dissolution (8). Hence, the observed contamination was not influenced by the carbonate dissolution in the aquifer.
Identification of enterococcal species.
Bacterial colonies isolated after the membrane filtration of different samples collected from the spring were characterized with the API 20 Strep system; out of 100 isolates, 38 were identified as E. faecalis, 30 were identified as Enterococcus faecium, 23 were identified as Enterococcus gallinarum, and 9 were unidentified.
Chromosomal DNA was extracted from a few strains of each Enterococcus group, and PCR was performed to obtain amplification of the ribosomal 16S DNA genes. Upon BLAST comparison with the DNA GenBank (1), the sequences of the 16S genes revealed good agreement with the identification results obtained with the API system.
Column tests.
The simulated pulse infiltration caused a sequence of breakthroughs characterized by different concentrations of E. faecalis. Each breakthrough curve represented the breakthrough corresponding to each infiltration event (Fig. 6). Hence, the temporary halt of infiltration and the presence of dry intervals (several hours in the case studied) during a rainfall period caused different breaks in microbial transport through the soil. A new rapid increase in the concentration of E. faecalis closely coincided with the beginning of each rainy step. The concentration at the start of each breakthrough was significantly lower than that observed at the end of the previous breakthrough curve. On the whole, the pulse infiltration produced significant variation of the number of transported cells versus time and then an intermittent transport of bacteria to the “groundwater.”
FIG. 6.
E. faecalis breakthrough curves for the soil blocks.
The total number of bacteria eluted after effective infiltration with 57 mm of water (9% of the mean annual infiltration amount in the study area) represented 2 to 6% of the inoculated E. faecalis cells in the blocks. A certain amount of cell death during the course of the experiments should be taken into account. Anyway, the cell death does not seem to be supported by a control experiment in which the same amount of bacteria utilized for column charge was incubated for 3 days in a mix of CaCl2 solution and column soil. In this case, the number of bacteria after 3 days was not changed. These results confirm that the soil medium is characterized by a significant capacity for the retention of bacteria.
DISCUSSION
The two types of breakthrough curves observed in springwaters (Fig. 4 and 5) are surely conditioned by the different distributions of rainfall versus time, because the media were unmodified between the first and the second events. Hence, the presence of several nonrainy hours during precipitation produces a sort of pulse infiltration and the transport of bacteria through the unsaturated medium. The reliability of this hypothesis has been verified by means of column tests of intact soil blocks collected from pasture area through the simulation of the real infiltration period characterized by four rainy steps and three dry intervals. These experiments confirmed that the pulse infiltration through the soil produces a transport of bacteria characterized by a sequence of breakthroughs. Each breakthrough represents the curve corresponding to each rainy step of the precipitation event. This behavior causes an intermittent migration of microorganisms from the surface to the groundwater and, together with dispersion in the fractured medium (unsaturated and saturated zones), influences the breakthrough of fecal bacteria at the spring. This breakthrough is then characterized by a sequence of peaks, observable through daily monitoring. At the moment, it is unclear why breakthroughs 1 and 3 show increases (flat breakthrough) different from those for branches 2 and 4 (Fig. 6).
In our column experiments, apparent differences in relative storage capacities among the three blocks may have resulted from natural heterogeneities, even though very similar transport behaviors (breakthrough curves) were observed. Hence, these heterogeneities should not produce a diversified breakthrough at the site scale and the results of flowthrough column tests can be reasonably extended to the entire pasture landscape of the research area. These differences should depend mainly upon different grain and pore sizes used for distribution into the columns. For example, when macropores have nonuniform paths with respect to soil depth, the transport of bacteria is concentrated in a few areas of the soil (10, 15, 21) and different levels of retention of bacteria can be identified at the core scale. In some cases, it was found that more than 50% of the total drainage was collected in <20% of the area beneath the soil blocks (16, 17). In the case studied, water chemistry and cell size and morphology, which were constant in all experiments, had no significant influence.
The important influence of the interaction between soil and the distribution of precipitation versus time on the breakthrough at the spring is probably due to the short distances of transport in the fractured medium (a few hundred meters in the case study). The effects of dispersion within the limestone aquifer could be predominant when microorganisms migrate longer distances in the carbonate rocks.
As a matter of fact, in small limestone aquifers that are extensively fractured, both precipitation and soil strongly influence the transport of fecal bacteria in the subsurface, and then it is possible to directly relate what is going on in the soil (at the laboratory scale) to what is happening at the springs.
Acknowledgments
This project was funded by the European Union (KArst waTER research program, INTERREG IIC, CADSES, grant 96/C200/07) and by the Research National Council of Italy (grant CNRG00D43F).
We thank Paolo Capuano and Vincenzo De Felice of the Università degli Studi del Molise for their thoughtful comments.
REFERENCES
- 1.Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990. Basic local alignment search tool. J. Mol. Biol. 215:403-410. [DOI] [PubMed] [Google Scholar]
- 2.Becker, M. W., D. W. Metge, S. A. Collins, A. M. Shapiro, and R. W. Harvey. 2003. Bacterial transport experiments in fractured crystalline bedrocks. Ground Water 41:682-689. [DOI] [PubMed] [Google Scholar]
- 3.Boni, C., P. Bono, and G. Capelli. 1982. Valutazione quantitativa dell'infiltrazione efficace in un bacino carsico dell'Italia centrale. Confronto con analoghi bacini rappresentativi di diversa litologia. Geol. Appl. Idrogeol. 17:437-452. [Google Scholar]
- 4.Celico, F., L. Esposito, and V. Piscopo. 1998. Interaction of surface-groundwater in fragile hydrogeological settings in southern Italy, p. 475-485. In J. Van Brahana, Y. Eckstein, L. K. Ongley, R. Schneider, and J. E. Moore (ed.), Gambling with groundwater—physical, chemical and biological aspects of aquifer-stream relations. Proceedings of the 28th International Congress. International Association of Hydrogeologists, Las Vegas, Nev.
- 5.Celico, F., P. Celico, P. De Vita, and V. Piscopo. 2000. Groundwater flow and protection in the Southern Apennines (Italy). Hydrogeology 4:39-47. [Google Scholar]
- 6.Celico, F., I. Musilli, and G. Naclerio. 19 February 2004, posting date. The impact of pasture and manure spreading on microbial groundwater quality in carbonate aquifers. Environ. Geol. [online.]
- 7.Celico, P. 1983. Le risorse idriche sotterranee dell'Appennino carbonatico centro-meridionale. Idrotecnica 1:3-18. [Google Scholar]
- 8.Cronin, A. A., R. G. Taylor, K. L. Powell, M. H. Barrett, S. A. Trowsdale, and D. N. Lerner. 2003. Temporal variations in the depth-specific hydrochemistry and sewage-related microbiology of an urban sandstone aquifer, Nottingham, United Kingdom. Hydrogeol. J. 11:205-216. [Google Scholar]
- 9.De Corso, S., D. Scrocca, and M. Tozzi. 1998. Geologia dell'anticlinale del Matese e implicazioni per la tettonica dell'Appennino molisano. Boll. Soc. Geol. Ital. 117:419-441. [Google Scholar]
- 10.Dexter, A. R. 1993. Heterogeneity of unsaturated, gravitational flow of water through beds of large particles. Water Resour. Res. 29:1859-1862. [Google Scholar]
- 11.Food and Agriculture Organization. 1988. Soil map of the world, revised legend, reprinted with corrections. World soil resources report 60. Food and Agriculture Organization, Rome, Italy.
- 12.Geldreich, E. E., and B. A. Kenner. 1969. Comments on fecal streptococci in stream pollution. J. Water Pollut. Control Fed. 41:336-341. [PubMed] [Google Scholar]
- 13.Gleeson, C., and N. Gray. 1997. The coliform index and waterborne disease. E and FN Spon, London, United Kingdom.
- 14.Harvey, R. W., and S. P. Garabedian. 1991. Use of colloid filtration theory in modeling movement of bacteria through a contaminated sandy aquifer. Environ. Sci. Technol. 25:178-185. [Google Scholar]
- 14a.Kibbey, H. J., C. Hagedorn, and E. L. McCoy. 1978. Use of fecal enterococci as indicators of pollution in soil. Appl. Environ. Microbiol. 35:711-717. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.McMurry, S. W., M. S. Coyne, and E. Perfect. 1998. Fecal coliforms transport through intact soil blocks amended with poultry manure. J. Environ. Qual. 27:86-92. [Google Scholar]
- 16.Phillips, R. E., V. L. Quisenberry, and J. M. Zeleznik. 1995. Water and solute movement in an undisturbed, macroporous column: extension pressure effects. Soil Sci. Soc. Am. J. 59:707-712. [Google Scholar]
- 17.Quisenberry, V. L., R. E. Phillips, and J. M. Zeleznik. 1994. Spatial distribution of water and chloride macropore flow in a well-structured soil. Soil Sci. Soc. Am. J. 58:1294-1300. [Google Scholar]
- 18.Sambrook, J., E. F. Fritsch, and T. Maniatis. 1989. Molecular cloning. A laboratory manual, 2nd ed. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, N.Y.
- 19.Thornthwaite, C. W., and J. R. Mather. 1957. Instructions and tables for computing potential evapotranspiration and the water balance, 5th printing. C. W. Thornthwaite Associates, Laboratory of Climatology, Elmer, N.J.
- 20.Trevors, J. T., J. D. Van Elsas, L. S. Van Overbeek, and M. Starodub. 1990. Transport of a genetically engineered Pseudomonas fluorescens strain through a soil microcosm. Appl. Environ. Microbiol. 56:401-408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Wildenschild, D., K. H. Jensen, K. Villholth, and T. H. Illangasekare. 1994. A laboratory analysis of the effect of macropores on solute transport. Ground Water 32:381-389. [Google Scholar]