Eight brands of domestic and imported bottled water were microbiologically analysed within three hours of purchase at a local supermarket. After 6 months of storage at room temperature (18-258c), few quantitative and qualitative differences were found in the microflora. Bottled water is often regarded as safer and healthier than tap water.
Original Description:
Original Title
A Survey of the Microbiological Quality of Bottled Water Sold in the Uk and Changes During Storag
Eight brands of domestic and imported bottled water were microbiologically analysed within three hours of purchase at a local supermarket. After 6 months of storage at room temperature (18-258c), few quantitative and qualitative differences were found in the microflora. Bottled water is often regarded as safer and healthier than tap water.
Eight brands of domestic and imported bottled water were microbiologically analysed within three hours of purchase at a local supermarket. After 6 months of storage at room temperature (18-258c), few quantitative and qualitative differences were found in the microflora. Bottled water is often regarded as safer and healthier than tap water.
International Journal of Food Microbiology 48 (1999) 5965
A survey of the microbiological quality of bottled water sold in the
UK and changes occurring during storage * A. Benito Armas, J.P. Sutherland Institute of Food Research, Earley Gate, Whiteknights Road, Reading RG6 6BZ, UK Received 23 June 1998; received in revised form 17 December 1998; accepted 10 February 1999 Abstract Eight brands of domestic and imported bottled water were microbiologically analysed within three hours of purchase at a local supermarket. Viable numbers of microorganisms were estimated on Plate Count Agar (PCA) and PCA diluted to quarter and tenth strengths (1/ 4 PCA and 1/ 10 PCA) and incubated at temperatures of 10, 15, 25 and 378C. Plate count agar diluted 4 21 to 1/ 4 and 1/ 10 incubated at 258C yielded the highest initial counts, up to 10 cfu ml . Pseudomonas spp. was the predominant species. After 6 months of storage at room temperature (18258C), few quantitative and qualitative differences were found in the microora. 1999 Elsevier Science B.V. All rights reserved. Keywords: Bottled water; Microbial growth; Storage 1. Introduction about tap water, since bottled water is often regarded as safer and healthier than tap water. During the past decade, there has been a consider- According to the European Community Directive able increase in the consumption of bottled water in (1980), which became law in the UK under the the UK, and this trend is expected to continue. Natural Mineral Water Regulations (Ministry of Moreover, uncarbonated water is now considerably Agriculture, Fisheries and Food, 1985), natural min- more popular than carbonated, having become a eral water is not sterilised, pasteurised or otherwise substitute for tap water in some households. (Anon, treated to remove or destroy microorganisms. The 1991, 1994, 1995). This reects consumer concerns number of bacteria recovered at the source is gener- 21 ally very low, around 10 cfu ml , but there are many reports that viable counts increase, notably in 4 5 21 uncarbonated water, to 10 10 cfu ml after 13 *Corresponding author. Present address: School of Health and weeks of storage (Schmidt-Lorenz, 1976; Gonzalez Sports Science, University of North London, 166-220 Holloway et al., 1987; Bischofberger et al., 1990; Mavridou, Road, London N7 8DB. Tel.: 144-171-753-7023; fax: 144-171- 1992; Mavridou et al., 1994; Tsai and Yu, 1997). 753-5081. E-mail address: jane.sutherland@bbsrc.ac.uk (J.P. Sutherland) Those organisms were mainly Gram negative non- 0168-1605/ 99/ $ see front matter 1999 Elsevier Science B.V. All rights reserved. PI I : S0168- 1605( 99) 00027- 6 60 A.B. Armas, J.P. Sutherland / International Journal of Food Microbiology 48 (1999) 5965 fermentative rods, frequently identied as Pseudo- chloride (PVC) bottles (Table 1) were purchased monas spp. and related genera, originating from the from a local supermarket. water itself (autochthonous microora) or contami- nants introduced during the bottling process (alloch- 2.2. Sampling thonous microora). The public health signicance of the high numbers of Pseudomonas spp. which One bottle of each type of water was sampled develop is unclear (Hunter, 1993). Many Pseudo- immediately after purchase and a second was sam- monas spp. recovered from water are resistant to pled after storage for six months in shaded daylight antimicrobial agents (Hernandez Duquino and at ambient temperature (18258C) in the laboratory. Rosenberg, 1987). Microbial numbers were estimated by decimal Bottled water has been the vehicle of transmission dilution in 1/ 4 strength Ringers solution (Oxoid of Vibrio cholerae, causing an outbreak of cholera in BR52) and plating on eight plates of each of the Portugal (Blake et al., 1977) and historically, re- following media: Plate Count Agar (PCA; Oxoid covery of Staphylococcus aureus (Leclerc et al., CM325) at full strength, quarter-strength (1/ 4 PCA) 1985) and Aeromonas hydrophila (Gonzalez et al., and tenth-strength (1/ 10 PCA). Duplicate plates of 1987; Manaia et al., 1990) from bottled water have each of the three agars were incubated at four caused concerns about its safety. temperatures: 10, 15, 25, and 378C. Plates were The purpose of this study was to evaluate the counted after 214 days of incubation, depending on quantitative and qualitative microbiological status of temperature of incubation. indigenous and imported bottled water sold in the If no colonies were recovered with the surface UK and to determine any changes occurring during plating technique, 1 ml pour plates were prepared storage at room temperature, with the aim of iden- with the same media, taking 1 ml from the same tifying optimal recovery conditions for the micro- bottle as the original sample. ora, with respect to media and temperature of incubation. 2.3. Statistical evaluation The following null hypotheses were examined: (a) There was no signicant difference between the 2. Materials and methods numbers of microorganisms recovered from the different brands of bottled water; (b) There was no 2.1. Bottled water signicant difference between the numbers of micro- organisms initially present in the water and those Three bottles, all with the same expiry date, of present after storage for 6 months at ambient tem- each of eight brands of domestically-produced and perature; (c) There was no signicant difference in imported non-carbonated bottled water in polyvinyl the numbers of microorganism recovered at tempera- Table 1 Analysis of eight brands of bottled water a b b b b b b b b Brand Origin pH Na K Ca Mg Cl SO NO HCO 4 3 3 d c A UK NR 10 3 102 24 30 60 6 NA B Belgium 5.8 3 0.5 3.5 1.3 5 6.5 1.9 11 d C France NR 8 5.4 10.4 6 7.5 6.7 4 64 c D UK 7.4 24 1 55 19 42 23 NA 248 E France 7.2 5 1 78 24 4.5 10 3.8 357 d F France NR 12 3.2 100 12 21 12 0.6 275 G UK 7.6 9 1 39 15 15 7 1 190 H UK 7.6 65 1 48 13 60 12 3.5 180 a Measured in laboratory. b 21 Information on label of bottle mg l . c NA data not available. d NR data not recorded. A.B. Armas, J.P. Sutherland / International Journal of Food Microbiology 48 (1999) 5965 61 tures of 10, 15, 25 and 378C; (d) There was no were used for further identication of microorga- signicant difference in recovery attributable to the nisms: API-20NE (Biomerieux, Basingstoke, UK) concentration of PCA used for recovery. for Gram-negative, non-fermentative organisms and To test these hypotheses, analysis of variance API-STAPH for Gram-positive cocci. Additional (ANOVA) of the microbiological data was carried tests, including lecithinase production, hydrolysis of out by the maximum likelihood estimation in SAS Tween 80, growth in 4% NaCl, growth on MacCon- Proc Lifereg. key agar (Oxoid CM115), growth at 428C, fermen- tation of sorbitol, acid production from xylose and 2.4. Bacterial identication motility (Gilardi, 1971) were necessary in some cases for nal identication. Randomly-selected colonies from each plate were checked for purity by streaking on fresh agar medium of the same strength as that from which they 3. Results and discussion were isolated. The streaked plate was incubated at the same temperature as that on which the original 3.1. Analytical characteristics of waters count was estimated. Isolates thus puried were initially screened by The analytical characteristics of the different Gram stain, cytochrome oxidase test, catalase test waters are shown in Table 1. and the ability to oxidise or ferment glucose (Hugh and Leifson, 1953; Gerhardt et al., 1981). Other microbial characteristics recorded included: colonial 3.2. Bacterial numbers morphology, production of water-diffusible pigment, 21 ability to grow at low temperatures and, for those Table 2 shows viable counts (cfu ml ) in the organisms isolated on lower-strength media, ability eight brands of mineral water sampled on the day of to grow on full-strength agar. The following systems purchase and after 6 months of storage, with various Table 2 21 Heterotrophic plate count (cfu ml ) in eight brands of uncarbonated bottled water on day of purchase (count 1) and after 6 months storage (count 2) using different media and incubation temperatures a Brand 108C 158C 258C 378C PCA 1/ 4PCA 1/ 10PCA PCA 1/ 4PCA 1/ 10PCA PCA 1/ 4PCA 1/ 10PCA PCA 1/ 4PCA 1/ 10PCA 21 21 21 21 21 21 21 21 21 21 21 21 cfu ml cfu ml cfu ml cfu ml cfu ml cfu ml cfu ml cfu ml cfu ml cfu ml cfu ml cfu ml b 3 4 4 A1 ** 7.7 310 ** ** 2.4 310 ** ** 2.4 310 ** ** ** ** 3 3 3 3 3 3 2 3 2 A2 1.7 310 1.0 310 1.5 310 1.5 310 2.0 310 1.2 310 ** 7.5 310 1.5 310 ** 7.5 310 ** 3 3 4 3 3 B1 ** ** 8.0 310 ** 8.0 310 1.6 310 ** 3.6 310 4.7 310 ** ** ** 2 2 2 3 3 B2 ** ** 5.0 310 ** ** 5.0 310 5.0 310 1.0 310 1.0 310 ** ** ** 4 4 4 4 4 4 4 4 4 C1 3.5 310 3.3 310 3.9 310 4.4 310 3.4 310 5.0 310 2.9 310 5.8 310 5.2 310 ** ** ** 3 3 3 2 3 3 3 3 3 C2 1.0 310 1.5 310 1.2 310 5.0 310 1.8 310 1.7 310 2.2 310 8.7 310 8.1 310 ** ** ** 3 3 3 3 3 3 3 3 3 D1 2.0 310 4.8 310 4.6 310 2.2 310 3.7 310 4.2 310 4.8 310 4.8 310 7.2 310 ** ** ** 3 3 3 3 3 D2 ** ** ** ** 2.7 310 3.5 310 3.0 310 2.0 310 1.0 310 ** ** ** 3 3 3 3 3 3 3 3 3 3 2 2 E1 4.0 310 3.6 310 4.5 310 4.7 310 4.2 310 4.2 310 3.0 310 4.6 310 5.6 310 1.5 310 5.0 310 5.0 310 2 2 3 3 3 3 E2 ** ** ** 7.5 310 5.0 310 1.7 310 2.0 310 1.5 310 2.5 310 ** ** ** c F1 * * * * * * * * * * * * F2 ** ** ** ** ** ** ** ** ** ** ** ** 4 4 4 4 4 4 4 4 G1 ** 2.6 310 2.8 310 1.2 310 3.1 310 3.2 310 2.3 310 2.2 310 2.5 310 ** ** ** 3 3 3 4 4 3 4 4 G2 ** 4.5 310 4.7 310 4.6 310 2.5 310 2.6 310 7.5 310 4.5 310 4.6 310 ** ** ** 3 3 3 3 4 4 4 4 4 H1 7.3 310 7.5 310 7.5 310 6.8 310 1.3 310 2.1 310 2.2 310 3.7 310 5.0 310 ** ** ** 3 4 4 3 4 4 4 4 4 H2 6.4 310 1.4 310 1.4 310 7.6 310 1.9 310 2.1 310 1.0 310 2.3 310 2.4 310 ** ** ** a AH represent different brands of water, sufxes 1 and 2 represent respectively numbers obtained in the initial sample and after 6 months of storage. b 21 ** ,500 cfu ml . c 21 * ,1 cfu ml . 62 A.B. Armas, J.P. Sutherland / International Journal of Food Microbiology 48 (1999) 5965 strengths of PCA and different incubation tempera- covered from water of this type, which is low in tures. nutrients and not heat-treated, absence of micro- 4 21 Initial microbial numbers of up to 10 cfu ml organisms may suggest the presence of toxic sub- were recovered at temperatures of 10, 15 and 258C stances in the water (Geldreich and Clark, 1965; from most of the bottles of water examined, on all Manaia et al., 1990). concentrations of PCA. At 378C, microbial numbers 21 were generally ,500 cfu ml , with the exception 2 21 of E1 (5.0 310 cfu ml on both reduced strength 3.3. Statistical evaluation 3 21 agars and 1.5 310 cfu ml on full strength PCA). After 6 months storage, there appeared to have Table 3 shows the outcome of the analysis of been an overall decrease in microbial numbers (apart variance applied to brands A, B, C, D, E, G and H at from A, which showed a slight increase) recovered at temperatures of 10, 15 and 258C to test the null the three lower incubation temperatures (Table 2). At hypotheses. Brand F was omitted from the analysis 2 21 378C, a count of 7.5 310 cfu ml on quarter- and there were insufcient data from any of the strength PCA was obtained from sample A2, while water samples at 378C for inclusion. the organisms originally recovered from sample E1 The rst null hypothesis, that there was no signi- 21 had declined in number to ,500 cfu ml in sample cant difference between the numbers of microorga- E2. nisms recovered from the different brands of bottled Some brands of bottled water (C1, D1, E1 and water, is not true since there were signicant differ- H1) yielded similar counts at temperatures 10258C ences in microbial numbers between the brands of on the three strengths of PCA used to recover the water. This was also observed by Tsai and Yu (1997) organisms; but in others, bacteria failed to grow on who recorded heterotrophic plate counts ranging 2 21 5 21 full strength media (A1, B1 and G1), especially at from ,10 cfu ml to .10 cfu ml from a lower temperatures. This suggests inhibition of or- selection of domestic and imported bottled waters in ganisms in these brands by high concentrations of Taiwan, and Bischofberger et al. (1990) who iden- nutrients in the recovery medium as observed by tied qualitative and quantitative differences between others (Schmidt-Lorenz, 1976; Schwaller and bottled water from two springs. Schmidt-Lorenz, 1980). One initial bottle and one bottle of water stored for In brand F, a French import, no bacteria were 6 months for each brand were examined, which may recovered initially or after storage for 6 months, even appear to limit the reliability of the conclusions that from pour plates prepared with 1 ml of undiluted can be drawn from the results of the rst null water in the three different strengths of agar. Since it hypothesis. However, this was a small scale pre- is anticipated that bacteria would normally be re- liminary experiment with the aim of developing the Table 3 Analysis of variance of bottled waters. Term Degrees of Chi-squared Signicance freedom intercept 1 741.3 0.0001 brand of water 6 115.5 0.0001 a age (new or 6 months) 1 2.3 NS strength (of PCA) 2 6 0.0507 temperature (of incubation) 2 0.4 NS age 3brand 6 100.9 0.0001 temperature 3strength 4 3.8 NS age 3strength 2 2.7 NS age 3temperature 2 12 0.0025 brand 3strength 12 89.9 0.0001 brand 3temperature 12 18.7 0.096 a NS: not signicant. A.B. Armas, J.P. Sutherland / International Journal of Food Microbiology 48 (1999) 5965 63 methodology to be employed in a future, more undiluted or tenth-strength PCA, while for brand B, comprehensive, study. undiluted PCA gives a signicantly lower recovery Since there was no signicant difference in micro- than quarter-strength, which in turn was signicantly bial numbers between the primary and six month lower than tenth-strength PCA. For brands C, G and samples, it would initially appear that the second null H, undiluted PCA gave lower recoveries than quar- hypothesis is true, as there was no general trend for ter- and tenth-strength agar, but the latter did not the numbers in any brand of water to be higher at differ signicantly. For brands D and E there was no one of the two time points (age effect, Table 3). signicant effect of the dilution of the agar. However, there was a signicant interaction between age and brand of water (Table 3), indicating that for 3.4. Preliminary differentiation of isolated bacteria some brands of bottled water there was a signicant difference in microbial numbers attributable to stor- After screening more than 700 isolates from the age time. Further analyses revealed that brands B, C, initial sampling and a further 300 after 6 months D and E had signicantly lower counts after 6 storage, a considerable predominance of Gram nega- months storage, while for brands A, G and H there tive over Gram positive bacteria was evident, with was no signicant change with storage time. Thus it percentages of Gram negatives of 97.5% and 93.4% appears that microorganisms in the different brands in the rst and second sampling respectively, con- of water are affected differently by ageing. Numbers rming other reports (Quevedo-Sarmiento et al., of microorganisms did not increase signicantly in 1986; Hunter and Burge, 1987; Mavridou, 1992; any sample during storage. Mavridou et al., 1994; Tsai and Yu, 1997). Gram The third null hypothesis, that there is no signi- negative oxidative bacteria, possessing cytochrome cant difference between recovery of organisms at oxidase and catalase, constituted more than 50% of temperatures of 10, 15 and 258C is also apparently the isolates on both occasions of sampling. Gram true (incubation temperature effect, Table 3). Never- negative fermentative organisms were not identied theless, there is a signicant interaction between age at any stage. Gram positive, catalase positive, fer- and incubation temperature for fresh samples, since mentative bacteria were found occasionally in some recovery at 108C was signicantly lower than at 15 brands of water. or 258C. In the samples stored for 6 months, this effect became more pronounced, since there was a 3.5. Identication of selected bacteria with API signicant interaction between time of storage and systems incubation temperature. The numbers recovered at 108C were signicantly lower than at 158C which in The results of identication of some isolates with turn were signicantly lower than those at 258C. This API-20NE and API-STAPH systems are shown in suggests that there may have been a change in the Table 4. Pseudomonas was the predominant genus composition of the microora to a population whose identied with API-20NE and was recovered from growth is favoured by the ambient storage tempera- all waters tested except brand F, from which no ture. organisms were isolated. Ps. uorescens and Ps. The fourth null hypothesis, that there is no signi- vesicularis were the most frequently isolated species cant effect of agar strength on numbers of micro- (Table 4). Mavridou et al. (1994) recorded Ps. organisms recovered, is not true. There is a marginal- uorescens and Ps. stutzeri as the most commonly ly signicant effect of agar strength (Table 3), isolated organisms from Greek bottled water. Hunter indicating that the agar strength affects recovery (1993) recovered predominantly Ps. stutzeri and Ps. from all brands of water under all conditions. Further putida from bottled waters from Germany and analysis showed that recovery on undiluted PCA is America, with Portuguese water having an abnormal- signicantly lower than on either of the more dilute ly high proportion (45%) of Ps. aeruginosa. Morais agars, which are not signicantly different. This and Da Costa (1990) recovered a large proportion of conrms the observation made above. The brand of Pseudomonas spp. from a single type of Portuguese water also has an effect to the extent that, for brand water, but with the exception of Ps. vesicularis, none A, quarter-strength PCA gave a higher recovery than could be speciated with certainty. Bischofberger et 64 A.B. Armas, J.P. Sutherland / International Journal of Food Microbiology 48 (1999) 5965 Table 4 Percentage of identied isolates from different brands of mineral water isolated on day of purchase (sample 1) and after 6 months storage (sample 2) A1 A2 B1 B2 C1 C2 D1 D2 E1 E2 G1 G2 H1 H2 Pseudomonas uorescens 14.3 100 33.3 73 8.7 26.1 66.6 9.1 vesicularis 71.4 77.8 16.7 40 4.2 23.5 17.4 42.8 paucimobilis 16.7 5.9 7.1 diminuta 33.3 40 8.3 5.9 36.4 mesophilica 8.7 7.1 Burkholderia cepacia 8.3 Sphaeromonas paucimobilis 4.3 Flavobacterium oryzahabitans 5.9 4.3 Comamonas sp. 14.3 13 Acinetobacter junii 6.6 CDCgr.IVC-2d 8.3 28.6 4.3 7.1 Moraxella 4.2 Staphylococcus spp. 16.7 4.2 8.7 8.7 other Gram negative 14.3 22.2 33.3 20 12.5 27 57.1 69.6 58.8 26.2 35.9 26.8 54.5 non-fermentative rods al. (1990) observed that the species of Pseudomonas method for examination of bottled water to obtain were characteristic of the source of water. total viable counts with only diluted media and A large number of isolates could not be identied incubation at 258C, conditions that yield the highest with the API system due to the limitations of the counts in these experiments. database. Although the API 20NE system has been Different opinions have been expressed about the used to identify bacteria from bottled water (Manaia public health signicance of Pseudomonas and re- et al., 1990; Mavridou, 1992), it is often unsatisfac- lated species in bottled water. They are reported to tory for identication of bacteria from this environ- be resistant to several antimicrobial agents (Her- ment, having been developed specically for identi- nandez Duquino and Rosenberg, 1987) and strains of cation of bacteria of clinical origin. Additional tests Burkholderia (Pseudomonas) cepacia (isolated from were therefore necessary in most cases for discrimi- the second sample of brand C; Table 4) have been nation of species. associated with infections in immunocompromised The types of bacteria recovered can depend on the individuals and are sometimes a cause of chest culture techniques used. The pour-plate technique infections in children with cystic brosis (Hunter, (the ofcial method of the European Community 1993). However, other studies have shown a minimal Directive, 1980) with incubation at temperatures of risk of infection from drinking these waters (Leclerc, 228C and 378C tends to favour recovery of fast- 1980). growing fastidious mesophilic and psychrotrophic Gram positive cocci were detected in waters B, C, bacteria. To ensure that all the viable bacteria in E, and G. The presence of Gram positive cocci in bottled water are recovered, surface culture tech- bottled water has also been reported by Hunter and niques with low-nutrient media and longer incuba- Burge (1987) and Mavridou (1992). Their signi- tion at lower temperatures are advisable (Bischof- cance is unclear, although Hunter and Burge (1987) berger et al., 1990). and Hunter (1993) concluded that since the human Our results, however, showed that although higher commensals Staph. epidermidis and Staph. hominis recoveries were obtained on more dilute media, there were among the Gram positive cocci isolated from were no differences between the composition of the the waters, they were probably contaminated by microora isolated at different temperatures, and people at the stage of lling and are thus part of the bacteria isolated on full strength PCA were generally allochthonous microora. similar in identity to those isolated on 1/ 4 and 1/ 10 It was not possible to determine in this study PCA. From these results, it is possible to simplify the whether the microora was autochthonous or alloch- A.B. Armas, J.P. Sutherland / International Journal of Food Microbiology 48 (1999) 5965 65 Gerhardt, P., Murray, R., Costilow, R., Nester, E., Wood, W., thonous. Further characterisation of the microora Krieg, N., Phillips, G., 1981. Manual of methods for general isolated from mineral water would be necessary in bacteriology, American Society of Microbiology, Washington, order to assess its origin and potential risks. DC. Gilardi, G.L., 1971. Characterization of nonfermentative nonfas- tidious gram negative bacteria encountered in medical bac- teriology. J. Appl. Bacteriol. 34, 623644. 4. Conclusion Gonzalez, C., Gutierrez, C., Grande, T., 1987. Bacterial ora in bottled uncarbonated mineral drinking water. Can. J. Mi- These results suggest that use of quarter-strength crobiol. 33, 11201125. or tenth-strength agar is more suitable for recovery Hernandez Duquino, H., Rosenberg, F.A., 1987. Antibiotic-resis- of microorganisms from bottled water then full tant Pseudomonas in bottled drinking water. Can. J. Microbiol. 33, 286289. strength agar, probably because organisms from an Hugh, R., Leifson, E., 1953. The taxonomic signicance of aqueous environment are frequently nutritionally fermentative versus oxidative metabolism of carbohydrate by depleted and growth can be inhibited by transfer to a various gram-negative bacteria. J. Appl. Bacteriol. 66, 2426. nutrient-rich environment. Secondly, an incubation Hunter, P.R., Burge, S.H., 1987. The bacteriological quality of temperature of 20258C for recovery of organisms bottled natural mineral waters. Epidemiology and Infection 99, 439443. from ambient stored water is recommended for Hunter, P.R., 1993. A review: The microbiology of bottled natural quantitative and qualitative recovery of microorga- mineral waters. J. Appl. Bacteriol. 74, 345352. nisms from bottled water, since they grow more Leclerc, H., 1980. Les qualites bacteriologiques de leau minerale rapidly at this temperature than at 158C and under- dEvian, Leau Minerale dEvian, Evian, France, 24-27. recovery is possible at an incubation temperature of Leclerc, H., Mossel, D.A.A., Savage, C., 1985. Monitoring non- carbonated (still) mineral waters for aerobic colonisation. Int. 108C. J. Food Microbiol. 2, 341347. Ministry of Agriculture, Fisheries and Food, 1985. The natural mineral waters regulations 1985. Statutory instruments 1985 Acknowledgements No. 71. London: HMSO. Manaia, C.M., Nunes, O.C., Morais, P., Da Costa, M.S., 1990. Heterotrophic plate cohunts and the isolation of bacteria from This work was supported by a COMETT student- mineral waters on selective and enrichment media. J. Appl. ship. The authors are grateful to Duncan Hedderley Bacteriol. 69, 871876. for the statistical analysis. Mavridou, A., 1992. Study of the bacterial ora of a non- carbonated natural mineral water. J. Appl. Bacteriol. 73, 355 361. Mavridou, A., Papapetropoulou, M., Boufa, P., Lambiri, M., References Papadakis, J.A., 1994. Microbiological quality of bottled water in Greece. Lett. Appl. Microbiol. 19, 213216. Anon, 1991. The market for bottled water. Soft Drinks Manage- Morais, P.V., Da Costa, M.S., 1990. Alterations in the major ment International. December 1991. heterotrophic bacterial populations isolated from a still bottled Anon, 1994. 2001: A soft drinks odyssey. Food Manufacture. mineral water. J. Appl. Bacteriol. 69, 750757. March 1994. Quevedo-Sarmiento, J., Ramos-Cormenzana, A., Gonzalez-Lopez, Anon, 1995. Soft drinks nd a niche. Food Manufacture. March J., 1986. Isolation and characterisation of aerobic heterotrophic 1995. bacteria from natural spring waters in the Lanjaron area Bischofberger, T., Cha, S.K., Schmitt, R., Konig, B., Schmit- (Spain). J. Appl. Bacteriol. 61, 365372. Lorenz, W., 1990. The bacterial ora of non-carbonated natural Schmidt-Lorenz, W., 1976. Microbiological characteristics of mineral water from the springs to reservoir and glass and natural mineral water. Annali dell Istituto Superiore Sanita 12, plastic bottles. Int. J. Food Microbiol. 11, 5172. 93112. Blake, P.A., Rosenberg, M.L., Florencia, J., Bandeiro Costa, J., do Schwaller, P. and Schmidt-Lorenz, W., 1980. Flore microbienne de Prado Quintino, L.D., Gangarosa, J., 1977. Cholera in Portugal: quatre eaux minerales non gazeiees et mises en bouteilles I II Transmission by bottled mineral water. Am. J. Epidemiol. Denombrement de colonies, composition grossiere de la ore, 105, 344348. et caracteres du groupe des bacteries Gram negatif pigmentees European Community Directive, 1980. Council Directive No. en jaune. Zentraalblatt fur Bakteriologie und Hygiene I Abt 80/ 777/ EEC of 15 July 1980 on the approximation of the laws Orig C 1 C1:330347. of member states relating to the exploitation and marketing of Tsai, G. J., Yu, S. C., 1997. Microbiological evaluation of bottled natural mineral waters. Ofcial Journal of the European uncarbonated mineral water in Taiwan. Int. J. Food Microbiol. Communities L229: 1-10. 37, 137143. Geldreich, E.E., Clark, H.F., 1965. Distilled water suitability for microbiological applications. J. Milk Food Technol. 28, 351 355.