Regardless of plant population structure - continuous or fragmented, low or high density - the spatial scale at which dispersal can be measured or parentage inferred using genetic markers is limited by three factors; levels of available genetic information, the choice of statistical genetic model, and the experimental design. This talk will focus on the latter two of these factors with specific reference to quantifying dispersal in tropical trees. Although recent advances in parentage analysis models provide increasingly powerful means of estimating effective male and female reproductive success within local populations as well as of describing pollen and seed movement over much longer distance intervals, empirical studies have often failed to take full advantage of these procedures. Using studies of the Neotropical trees Alseis blackiana (Rubiaceae) and Spondias mombin (Anacardiaceae) as examples, I will describe how these models can be used to quantify pollen and seed flow in both continuous and fragmented forest settings. I will also discuss how the detection of long distance pollen dispersal events can be dramatically enhanced when the paternal genotype can be reconstructed from the progeny of singly sired fruit, as demonstrated by landscape level studies of pollen flow in South African and Panamanian strangler figs (Ficus sp., Moraceae). These and other recent studies demonstrate that integrated approaches utilizing unique features of a species reproductive biology, the most appropriate genetic models, and increasingly informative genetic markers, provide the most effective means of characterizing dispersal patterns and processes over the largest possible spatial area.

Key words: Allozyme, fragmented populations, parentage analysis, paternity analysis, pollen flow, seed flow