Optimization of a sampling design and significance of bacterial indicators: Application to the bacteriological survey of the Ardeche river, France

Abstract

Difficulties encountered in monitoring the bacterial pollution of rivers were investigated during the bacteriological survey of one of the most popular recreational rivers of France. The problem of the representativeness of bacterial indicator sampling was first studied by analyzing preliminary sampling results using nested ANOVA and estimating the error percentage (confidence interval/mean) obtained with different simulated sampling strategies. The best strategy, i.e. giving the lowest error percentage of the bacterial mean, was applied to a survey of fecal and total coliforms and fecal streptococci counts at different sampling locations (11) along the river, repeated five times during the summer of 1984. Two opportunistic pathogens of nonfecal origin were also counted at the same time as the bacterial indicators. In addition, the taxonomic composition of the bacterial populations isolated as fecal and total coliforms was investigated. Generally, the confidence intervals around the count means were low, and allowed a clear separation of the polluted and the nonpolluted zones of the river based on European Community standards. The counts of the three standard bacterial indicators were significantly correlated along the whole river. In contrast, the variations of Aeromonas spp and Pseudomonas aeruginosa counts were generally not correlated with those of the indicators. All the bacterial strains isolated as fecal coliforms were identified as Enterobacteriaceae and 70.8% of them were Escherichia coli. Only 24.5% of total coliforms were identified as Enterobacteriaceae. The last results indicate that, despite the standards, the total coliform variable is not relevant, for assessing the recreational quality of river waters. Replacement of this variable by others, such as easily measurable pathogenic bacteria of nonfecal origin, is recommended.