C. DEFIVES, S. GUYARD, M.M. OULARÉ , P . M AR Y AN D J . P. HO R NE Z. 1999. The changes in bacterial counts during the storage of a natural mineral water from a French spring were studied. Samples were taken from the spring and the bottling line. Viable cultivable (VC) bacteria were counted on R2A medium. Total counts, viable and dead bacteria were counted using the LIVE/DEAD ® Bac Light TM VIABILITY kit and epifluorescence microscopy. Viable but non-cultivable (VNC) bacteria were estimated by difference between viable and VC counts. Isolates were clustered by phenotype. The microflora in the spring water increased from ³ 10-3 × 10 5 bacteria ml −1 after 6 d in storage and then stabilized. Mechanical bottling increased the allochthonous bacteria in the water that stabilized at 10 5 bacteria ml. Maximal growth is controlled by the low concentration of nutrients in the mineral water and the lysis of dead cells. The allochthonous bacteria came from the aquifer and colonized the filling line. The changes in the VC and VNC populations showed that the bacteria used starvation-survival and entry into the VNC state to adapt to the bottling stress and the enclosed oligotrophic environment.

Fifteen strains from two emergent mineral waters were isolated and tentatively identified with API 20NE and BIOLOG GN systems. These strains were screened for their sensitivities to seven replication‐inhibiting antibiotics of the (fluoro)quinolone group (nalidixic and pipemidic acid, flumequine, norfloxacin, ofloxacin, pefloxacin and ciprofloxacin). It was shown that the direct viable count (DVC) procedure could be improved by using certain antibiotic cocktails, which were active against the isolates. Geometric bacterial features were successfully determined with image analysis and adapted software (ICONIX, Perfect ImageTM). Elongations were significant and allowed rapid discrimination of antibiotic inhibited and non‐inhibited strains. Particular isolates in a mixed culture were characterized and enumerated after only 14 h exposure with the appropriate antibiotic cocktail. This method can also be applied to other communities, such as mixed cultures in bio‐fermentors or in food with known microflora.