an important consideration for the following section on designs is that the scientific question concerns one impacted site. the more true replicates taken at that site and the control site, the more likely an impact will be detectable. furthermore, conclusions or inferences must be limited to that site alone. however, many studies are performed to learn about the structure and functions of seagrass beds in general, or for a specific region. in these cases, the results are intended to apply more broadly than one impact to a specific bed, and several sites must be chosen to represent the many sites potentially available for inclusion in the study. thus, the resources available for data collection need to be appropriated carefully. for example, it is usually desirable to sample the same number of replicates at each site(balanced design). the selection of the number and locations of replicates and the number of sites is best informed by pilot studies that characterize variability within and between sites.