Gene flow is a relational process involving dispersal into a reference population. In practice, gene flow is measured as dispersal with respect to sample structure, so estimates are sensitive to the size and placement of plots. In studies of continuous tropical forest, logistical considerations set an upper limit on plot size and restrict plot placement, yielding a sampling structure that may be discordant with population structure - the size, shape, density and dispersion of populations, neighborhoods, or demes in a landscape. Consequently, high rates of gene flow (for example, m > 0.3) are not so much a biological anomaly as they are an indication that the plot does not include a substantial portion of the donor population for the period examined. In studies of fragmented tropical forest, plot size and placement typically are confounded with forest structure - the size, shape, density and dispersion of patches in a landscape. Yet, forest structure may or may not correspond with plant population structure; fragmentation studies of gene flow implicitly test this relationship. Further, a priori assumptions concerning process homogeneity hinder estimation of gene flow in fragmented landscapes since fecundity distributions and the relative importance of pollen versus seed dispersal can differ between habitats. Determining the spatial and temporal flux in area and membership of a population, or at least scale-conscious estimation of gene flow, would foster valid comparisons among sites and species.

Key words: deme, forest structure, gene flow, neighborhood, population structure, tropical plants