ABSTRACT
BACKGROUND: Flow cytometry is an invaluable tool for the analysis of large series of samples in aquatic microbial ecology. However, analysis of the resulting data is often inefficient or does not reflect the complexity of natural communities. Because bacterioplankton assemblages frequently fall into several clusters with respect to their cellular properties, these subgroups seem to be a promising level of abstraction. Image analysis was used to detect clusters from flow cytometry data. The method was tested on a bacterial community under heavy protozoan grazing pressure. METHODS: A bivariate histogram of flow cytometry data was transformed into a gray-scale image for image analysis. After low-pass filtration, regional maxima were delimited by a watershed algorithm. The resulting areas were then used as gates on the original measurements. RESULTS: Three clusters could be detected from the bacterial assemblage. Protozoan grazing had a strong impact on the bacterial community, which could be analyzed in detail at the level of individual subgroups. CONCLUSIONS: Investigation at the level of bacterial subgroups allowed a more detailed analysis than whole-community statistics and delivered essential and ecologically meaningful information. Image analysis proved to be an adequate tool to detect the subgroups without a priori knowledge.
