Studies of microbial communities from aquatic ecosystems provide important insights into relations between various aspects of ecosystem functioning and changes in biodiversity. Aquatic microbial systems provide a valuable counterpoint to studies of terrestrial systems, because patterns reflect consequences of interactions occurring over many generations of community development, and are unlikely to represent artifacts of the initial conditions established in experimental communities. In this paper we re-analyse our previously published data to separate the contributions of temporal and spatial variation to overall variation in ecosystem functioning. A new analysis based on re-sampling confirms a negative relationship between richness and the variability of one ecosystem process, carbon dioxide flux. The negative relationship reflects high variation among communities of low species richness, rather than high temporal variation within communities of low richness. We also review the various transformations and summary statistics proposed as alternate measures of variability in ecosystem functioning, to point out that different measures are often appropriate for different kinds of data. Finally, we conclude that arguments about the cosmopolitan distribution of microbes do not preclude the existence of important relations between microbial species richness and ecosystem functioning.