Plants may protect themselves against herbivorous arthropods by providing refuges to predatory arthropods, but they cannot prevent herbivores from taking countermeasures or even from reaping the benefits. To understand whether plants benefit from providing self-made refuges (so-called domatia), it is not only necessary to determine the fitness consequences for the plant, but also to assess (1) against which factors the refuge provides protection, (2) why predatory arthropods are more likely to monopolise the refuge, and (3) how herbivorous and predatory arthropods respond to and affect each other in and outside the refuge. In this article, we focus on the last aspect by studying the dynamics of refuge use of a predatory mite (Typhlodromalus aripo) and its consequences for a herbivorous mite (Mononychellus tanajoa) on cassava plants in Benin, West Africa. The refuge, located in-between the leaf primordia of the cassava apex, is thought to provide protection against abiotic factors and/or intraguild predators. To test whether the predator waits for prey in the apex or comes out, we sampled predator-prey distributions on leaves and in the apex at 4 hour-intervals over a period of 24 hours. The predatory mites showed pronounced diurnal changes in within-plant distribution. They were in the apices during the day, moved to the young leaves during night and returned to the apices the next morning. Nocturnal foraging bouts were more frequent when there were more herbivorous mites on the leaves near the apex. However, the foraging predators elicited an avoidance response by mobile stages of their prey, since these were more abundant on the first 20 leaves below the apex during late afternoon, than on the same leaves during night. These field observations on cassava plants show that (1) during daytime predatory mites monopolise the apical domatia, (2) they forage on young leaves during night and (3) elicit avoidance by within-plant, vertical migration of mobile stages of the herbivorous mites. We hypothesize that cassava plants benefit from apical domatia by acquiring protection for their photosynthetically most active, young parts, because predatory mites (1) protect primordial leaves in the apex, (2) reduce the densities of herbivorous mites on young leaves, and (3) cause herbivorous mites to move down to less profitable older leaves.