We examined how the relationships between geographic distance and similarity of community dynamics vary at different spatial scales. A nested hierarchical sampling design was adopted for rocky intertidal sessile assemblages along the Pacific coast of Japan to set three spatial levels (among rocks: 6-456 m, among shores: 2.7-40 km, among regions: 263-1840 km).

To presume community dynamics, transition probabilities between ecological states (coralline algae, crustose algae, foliose algae, sessile animal and vacant) were calculated from the data of two censuses with intervals of 3 months and arranged into matrix. Because community change is complex resultant of multi-processes, whole matrix and sets of elements as two processes (recruitment: vacant points become colonized, competition: points occupied by functional group replaced by different group) were examined.

Distance-similarity relationships were tested by regression analysis (X: geographic distance, Y: Bray-Curtis dissimilarity among transition probabilities) with bootstrap procedure.

Significant regressions were obtained only at among rocks level for whole matrix and sets of elements relating recruitment. Averages of similarities were smallest at broadest spatial level and largest at finest level for all three measurements of dynamics.