Microorganisms attain high population densities, which has led microplankton ecologists to assume that samples of a few tens of millilitres suffice to characterize the assemblage of species. However, the observation that microbial plankton communities contain a large pool of species with low population densities casts doubt on the validity of estimates based on conventional sampling methods. By standardizing estimates of species numbers, we show that marine phytoplankton communities have been undersampled more severely in ecosystems of low productivity, thus leading to bias in the patterns of diversity reported previously. We found that phytoplankton communities from unproductive, subtropical waters fit to right-skewed, lognormal species-abundance distributions, which has long been interpreted to arise from incomplete censuses. The sampling-standardized estimates of species richness show no relationship with ecosystem productivity, arguing against the idea that phytoplankton diversity peaks at intermediate levels of primary production. These results suggest that these 2 fundamental properties of marine phytoplankton communities, viz. diversity and productivity, might not be linked mechanistically.