In many plant-pollinator systems, interactions present a high degree of generalism, so that coevolution should be studied at the community level. Indeed intraspecific trait variation in such systems may both lead to variation in the gains that individuals are drawing from their interactions, and to variation in their choice/attraction of interaction partners. In this contribution, I will study whether the structural properties of the network of interactions between plants and animals impact the magnitude of coevolution and the properties of coevolutionary outcomes through simulations. In these simulations, individuals interact following simple interaction rules based on a single trait, the interaction strength between two individuals being controlled by their trait values. The fitness of an individual is proportional to the sum of its interactions, and traits present a constant heritability. The distribution of trait values in every species as well as species abundances are monitored through time to capture the natural selection applying on each species as a function of its position in the simulated network of interactions.