Topsoil disturbance reshapes diaspore interactions with ground‐foraging animals in a megadiverse grassland

International audience; Questions: Anthropogenic disturbances are known to be followed by extremely poor recovery in edaphic grasslands. However, the role of interactions with diaspore predators and secondary dispersers, which compose the dispersal filter and modulate plant community recovery, has been overlooked. We performed field experiments to investigate how soil disturbances affect diaspore interactions with the ground‐foraging fauna to better understand how disturbance influences regeneration potential.Location: Campo rupestre vegetation, megadiverse edaphic grasslands, southeastern Brazil.Methods: We used diaspores from five native species to compare removal rates between disturbed (top soil removal) and preserved sites; we also controlled invertebrate and vertebrate access to determine their role. In addition, we assessed differences in the diaspore removal effectiveness (DRE) and the structure of animal–diaspore interactions through network based‐approach.Results: For three species, the average diaspore removal rates was relatively high (between 30% and 70%). Invertebrates were the most common removal agents in both disturbed and preserved sites. Interactions with foraging fauna and removal rates were 20% smaller in disturbed sites. Considering all diaspore removal events in disturbed sites, 24% resulted in the displacement from disturbed to preserved sites, but no diaspore was transported from preserved to disturbed ones. The animal–diaspore network was exclusively composed by ant–diaspore interactions and was more diverse and robust in preserved sites compared with disturbed ones. Seed predator ants (Pheidole and Dorymyrmex) were more common in disturbed sites. Furthermore, significant differences were found in the DRE between ant species and site types, suggesting specificity in the provision of dispersal services.Conclusions: Topsoil removal affected removal proportions, DRE and ant–diaspore interaction network structure. The lack of diaspore dispersal towards disturbed sites indicates that soil removal affects secondary seed removal dynamics, impeding recovery potential. Disturbance negatively affected diaspore fate by reshaping interactions with ground‐foraging secondary seed dispersers and predators, constraining the development of seed bank and thus impacting upon vegetation dynamics and resilience.