A microsatellite-based estimation of clonal diversity and population subdivision in Zostera marina, a marine flowering plant

We examined the genetic population structure in eelgrass (Zostera marina L.), the dominant seagrass species of the northern hemisphere, over spatial scales from 12 km to 10 000 km using the polymorphism of DNA microsatellites. Twelve populations were genotyped for six loci representing a total of 67 alleles. Populations sampled included the North Sea (four), the Baltic Sea (three), the western Atlantic (two), the eastern Atlantic (one), the Mediterranean Sea (one) and the eastern Pacific (one). Microsatellites revealed substantial genetic variation in a plant group with low allozyme diversity. Average expected heterozygosities per population (monoclonal populations excluded) ranged from 0.32 to 0.61 (mean = 0. 48) and allele numbers varied between 3.3 and 6.7 (mean = 4.7). Using the expected frequency of multilocus genotypes within populations, we distinguished ramets from genetic individuals (i.e. equivalent to clones). Differences in clonal diversity among populations varied widely and ranged from maximal diversity (i.e. all ramets with different genotype) to near or total monoclonality (two populations). All multiple sampled ramets were excluded from further analysis of genetic differentiation within and between populations. All but one population were in Hardy-Weinberg equilibrium, indicating that Zostera marina is predominantly outcrossing. From a regression of the pairwise population differentiation with distance, we obtained an effective population size Ne of 2440-5000. The overall genetic differentiation among eelgrass populations, assessed as rho (a standardized estimate of Slatkin's RST) was 0.384 (95% CI 0.34-0.44, P < 0.001). Genetic differentiation was weak among three North Sea populations situated 12-42 km distant from one another, suggesting that tidal currents result in an efficient exchange of propagules. In the Baltic and in Nova Scotia, a small but statistically significant fraction of the genetic variance was distributed between populations (rho = 0.029-0. 053) at scales of 15-35 km. Pairwise genetic differentiation between European populations were correlated with distance between populations up to a distance of 4500 km (linear differentiation-by-distance model, R2 = 0.67). In contrast, both Nova Scotian populations were genetically much closer to North Sea and Baltic populations than expected from their geographical distance (pairwise rho = 0.03-0.08, P < 0.01). A biogeographical cluster of Canadian with Baltic/North Sea populations was also supported using a neighbour-joining tree based on Cavalli-Sforza's chord distance. Relatedness between populations may be very different from predictions based on geographical vicinity.