Computer simulations are an important tool for developing conservation strategies for forest species. This study used simulations to investigate the genetic, ecological, and reproductive patterns that contribute to the genetic structure of the tree Luehea divaricata Mart. & Zucc. in five forest fragments in the Brazilian Pampa biome. Using the EASYPOP model, we determined the selfing and migration rates that would match the corresponding genetic structure of microsatellite marker data (based on observed and expected heterozygosity parameters). The simulated reproductive mode was mixed, with a high rate of outcrossing (rate = 0.7). This was consistent with a selfing-incompatible system in this species, which reduced, but did not prevent, selfing. The simulated migration rate was 0.02, which implied that the forest fragments were isolated by distance, and that the inbreeding coefficients were high. Based on Nei's gene diversity analysis, 94% of the genetic variability was distributed within the forest fragments, and only 6% of the genetic diversity was caused by differences between them. Furthermore, the minimum viable population and minimum viable area genetic conservation parameters (which determine conservation potential in the short and long term) suggested that only the Inhatinhum forest fragment had the short-term potential to maintain its genetic diversity. However, in the long term, none of the forest fragments proved to be sustainable, indicating that the populations will require intervention to prevent a decline in genetic variability. The creation of ecological corridors could be a useful solution to connect forest fragments and enhance gene flow between them.