Filtering of epiphytic microbes can occur through plant traits and deterministic dispersal-mediated processes. This filtering can affect microbiome assembly yet their relative contribution to predictable variation in microbiome is poorly understood. We collected this data set to test the impact of host-plant filtering and dispersal on nectar microbiome presence, abundance, and composition. We inoculated bacteria and yeast into 30 plants across 4 phenotypically distinct cultivars of Epilobium canum. We compared the growth of inoculated communities to openly visited flowers from a subset of the same plants. 

We found evidence of host-selection when we inoculated flowers with synthetic communities. However, the host-plants with the highest microbial densities when inoculated did not have the highest microbial densities when openly visited. Host plants predictably varied in the presence or absence of bacteria, which was correlated with pollen receipt, suggesting a role for deterministic dispersal.  These findings suggest that host filtering could drive plant microbiome assembly in tissues where species pools are large and dispersal is high.  However deterministic differences in microbial dispersal to hosts may be equally or more important when microbes rely on an animal vector, dispersal is low, or arrival order is important.