Seagrasses are ecosystem engineers that offer important habitat for a large number of species and provide a range of ecosystem services. Many seagrass ecosystems are dominated by a single species, with research showing that genotypic diversity at fine spatial scales plays an important role in maintaining a range of ecosystem functions. However, for most seagrass species, information on fine-scale patterns of genetic variation in natural populations is lacking. In this study, we use a hierarchical sampling design to determine the levels of genetic and genotypic diversity at different spatial scales (centimeters, meters, kilometers) in the Australian seagrass Zostera muelleri. Our analysis shows that at fine spatial scales (<1 m), levels of genotypic diversity are relatively low (R (Plots) = 0.37 ± 0.06 SE), although there is some intermingling of genotypes. At the site (10’s m) and meadow location (km) scale, we found higher levels of genotypic diversity (R (sites) = 0.79 ± 0.04 SE; R (Locations) = 0.78 ± 0.04 SE). We found some sharing of genotypes between sites within meadows, but no sharing of genotypes between meadow locations. We also detected a high level of genetic structuring between meadow locations (F ST = 0.278). Taken together, our results indicate that both sexual and asexual reproductions are important in maintaining meadows of Z. muelleri. The dominant mechanism of asexual reproduction appears to occur via localized rhizome extension, although the sharing of a limited number of genotypes over the scale of 10’s of meters could also result from the localized dispersal and recruitment of fragments. The large number of unique genotypes at the meadow scale indicates that sexual reproduction is important in maintaining these populations, while the high level of genetic structuring suggests little gene flow and connectivity between our study sites. These results imply that recovery from disturbances will occur through both sexual and asexual regeneration, but the limited connectivity at the landscape scale implies that recovery at meadow-scale losses is likely to be limited.