It is known that asynchronous temporal variations in local populations can contribute to the stability of metapopulations. However, studies evaluating the hierarchical organization of multiple spatial scales are rare for continuous marine landscapes, especially for marine vegetation such as seagrass beds. In this study, long-term observation (26 yr) of temporal changes and nested spatial analyses were combined for an extensive seagrass meadow in Tokyo Bay, Japan, using remote sensing and geographic information system technologies. We examined how the dynamics at the whole-bed scale (~1 km2) are related to those at a local scale (0.04 km2), and investigated the relationship between the seagrass dynamics and long-term changes in environmental conditions using data on oceanography, water quality, and sediment dynamics. The seagrass bed size fluctuated between a maximum of 1.28 km2 (in 1987) and a minimum of 0.39 km2 (in 2001), with an average of 0.90 km2. The temporal variation in seagrass bed size at the whole-bed scale correlated with sand movement within the seagrass bed related to changes in the position of a sandbar. Seagrass bed size fluctuated asynchronously at a local scale. Multivariate analyses recognized clusters of local areas showing similar patterns of fluctuation. Temporal patterns in the various clusters responded differently to changes in environmental factors, e.g. the position of the sandbar was highly correlated with seagrass bed size in shallow habitats but not in deeper areas. The magnitudes of the temporal variations for the local clusters were greater than that of the entire bed, suggesting that asynchronous fluctuation in different areas of the bed plays an important role in the overall stability of the seagrass bed. The results of the present study also highlight the importance of physical processes in regulating the temporal dynamics of seagrass beds in shallow sedimentary landscapes.