Abstract: In dryland cereals in NE Spain, the harvester ant, Messor barbarus L., is responsiblefor a high percentage of seed removal. A direct density dependent response of seed predators toseed patches may help regulate weed populations. In this study, we investigated if seed removalrate by harvester ants is influenced by weed seed density. For this reason, 60 circular areas of 1m2were created inside four 50×50m blocks after cereal harvest Lolium multiflorum L. seeds wereapplied at 1000, 2000, 5000, 10000 or 20000 seeds/m2, in ten randomly selected areas each; theremaining 10 were used to test the methodology used to retrieve the seeds. After 24h, theremaining seeds were collected using vacuum cleaners or a D-vac and seed removal rates wereestimated. In three of the four blocks, seed removal was extremely high (99-100%), and theresponse was therefore density independent Here, nests densities ranged from 468 to 900nests/ha. In the fourth block seed removal was 91%, and here nest density was lower (284nests/ha).