Plant individuals rely on pollinator services for their reproduction and often have to share these services with co-occurring neighbours, creating complex indirect plant–plant interactions. Many current theoretical models focus on the effect of floral resources’ density on the net outcome of these indirect plant–plant interactions, often neglecting the identity of plant species in the communities and especially the species’ spatial pattern. To fill this gap, we created a spatially explicit model whose goal was to study the interplay between relative densities and spatial distribution patterns of two plant species differing in their attractiveness for pollinators. Since theory predicts that pollinator behaviour strongly governs the outcome of pollination, we allowed the pollinators to systematically change their plant preferences based on their foraging experience. Thus the interplay between density and spatial patterns of plants was tested over a continuum of behaviours from specialists to generalists. Our most striking finding was that reproductive success of the less attractive species was affected in an opposite way by spatial patterns depending on whether the species had relatively low or high densities. Namely, when the less attractive species was highly abundant, its survival was higher when aggregated in large monospecific patches than when uniformly distributed. On the other hand, when the attractive species was more abundant, the less attractive species survived better when uniformly distributed. These results were consistent as long as the scale of the plant spatial aggregation was similar to or larger than the pollinators’ detection range. Our results suggest that aggregated plant spatial patterns manipulate pollinator behaviour by trapping them within monospecific patches. This effect was sufficiently strong to enhance the survival of a competitively inferior species and hence to act in a way similar to the more familiar niche or temporal separation among plant species.