The idea that “everything is everywhere, but the environment selects” has been seminal in microbial biogeography, and marine phytoplankton is one of the prototypical groups used to illustrate this. The typical argument has been that phytoplankton is ubiquitous, but that distinct assemblages form under environmental selection. It is well established that phytoplankton assemblages vary considerably between coastal ecosystems. However, the relative roles of compartmentalisation of regional seas and site-specific environmental conditions in shaping assemblage structures, have not been specifically examined. We collected data from coastal embayments that fall within two different water compartments within the same regional sea and are characterised by highly localised environmental pressures. We used PCNM and AEM models to partition the effects that spatial structures, environmental conditions and their overlap had on the variation in assemblage composition. Our models explained a high percentage of variation in assemblage composition (59-65%) and showed that spatial structure consistent with marine compartmentalisation played a more important role than local environmental conditions. At least during the study period, surface currents connecting sites within the two compartments failed to generate sufficient dispersal to offset the impact of differences due to compartmentalisation. In other words, our findings suggest that, even for a prototypical cosmopolitan group, everything is not everywhere.

Morphological species identification

Seawater of 1 liter was immediately preserved with a mixture of neutralized formaldehyde and acidic Lugol’s iodine solution at a ratio of 6:1. Two methods were applied for the quantitative analysis of phytoplankton communities, (a) the standard method of cell counting under a Zeiss AxioVert 100 inverted microscope [1], preceded by serial sedimentations and final concentration of individuals into settling chambers of 25 ml, and (b) sample filtration (300 ml, 2 replicates) through nitrate cellulose membranes of 0.45 μm mesh [2] followed by transillumination of the filter with immersion oil and observation under a Zeiss Axiolab upright microscope. This approach was adopted due to the observed underestimation of the smaller fraction of phytoplankton (<10 μm) by the Utermöhl method. A graph of the cumulative number of species (y axis) over new fields of vision (x axis) was plotted for 30 random samples to estimate the optimal sample surveying effort. The levelling off of this accumulation curve indicated the minimum number of fields to be counted in the specific magnification to provide a good sample representation. The results for both methods were expressed as number of cells per liter and combined in a single dataset.