Limited dispersal distances in plant populations frequently cause local genetic structure, which can be quantified by spatial autocorrelation. In clonal plants, three levels of spatial organization can contribute to positive autocorrelation; namely, the neighbourhood of (a) ramets, (b) clone fragments and (c) entire clones. Here we use data from an exhaustive sampling scheme on a clonal plant to measure the contribution of the neighbourhoods of each distinct clonal structure to total spatial autocorrelation. Four plots (256 grid points each) within dense meadows of the marine clonal plant Zostera marina (eelgrass) were sampled for clone structure with nine microsatellite markers ( approximately 80 alleles). We found significant coancestry (f(ij)), at all three levels of spatial organization, even when not allowing for joins between samples of identical genets. In addition, absolute values of f(ij) and the maximum distance with significant positive f(ij) decreased with the progressive exclusion of joins between alike genotypes. The neighbourhood of this clonal plant thus consists of three levels of organization, which are reflected in different kinship structures. Each of these kinship structures may affect the level of biparental inbreeding and the physical distance between flowering shoots and their outcrossing neighbourhood. These results also emphasize the notion that spatial autocorrelation crucially depends on the scale and intensity of sampling.