Optimised sample sizes for analysing the genetic heterogeneity of mammary pathogen isolates from environmental samples

Modern molecular techniques and more efficient culture techniques, which may be automated in the future, allow for characterization of increasingly Large numbers of microorganisms. Very often, the number of bacterial isolates analysed per sample is determined by habit, convenience, Laboratory capacity or financial resources. If too few isolates are analysed per sample, information will be missed, while analysis of too many isolates is a waste of resources. Statistical considerations and knowledge of the heterogeneity of mammary pathogens, e.g. Streptococcus uberis, in milk and the environment should Lead to optimised sample sizes, i.e. the number of isolates from one bacterial species that needs to be characterized per milk or environmental sample. Calculation of the optimal sample size for multiple isolates originating from one clinical, food or environmental sample is a common problem in microbiology and survey design. As an example, we present data for S. uberis isolated from the environment and characterized by automated ribotyping with PvuII. Based on a Bayesian mode of inference, results show that approximately 20 isolates of S. uberis per sample should be ribotyped in order to find all existing ribotypes in soil samples with a probability of 95%. This high number of isolates per sample is probably never tested in real-life due to financial restrictions, showing that investigators are not gaining maximal information on strain heterogeneity from the samples analysed.