Predators and prey interact at small spatial scales, but during their lifetime disperse at much larger spatial scales. Trophic metacommunity theory proposes that dispersal is a critical processes that determines food webs structure at small and large scales. The application of metacommunity theory to empirical systems remains elusive because key parameters such as dispersal and interaction strengths have been very difficult to quantify. Here we develop a novel approach that combines population genomics with mesocosm experiments to parameterize a metacommunity model. Using genotyping‐by‐sequencing, we characterized the dispersal kernels of a predator ‐ prey pair living in a phytotelm metacommunity. We found that the prey dispersed up to 25 km while the predator dispersed only 200 m. We then quantified a functional response for these species using feeding trials. Even without invoking differences in the abiotic niche, our empirically parameterized simulation model produced patterns of population survival and occupancy that were consistent with past observations on the natural system. Importantly we found that these patterns were more likely to be found with simulations based on our observed values than in other regions of potential parameter space. This suggests that the observed dispersal kernels contribute to the dynamics of these species in the metacommunity.