Understanding the flexibility of interactions and network rewiring (i.e. reassembly of interactions due to partner-switching) is necessary to comprehend how future anthropogenic changes will affect interspecific interactions and the functioning of communities. A higher rewiring could be expected in more disturbed landscapes because these landscapes contain fewer and more generalist species with more homogeneous traits. We sampled pollination interactions in 20 wild Olea europaea communities along a disturbance gradient to evaluate the hypothesis that the loss of natural habitats increases seasonal (within-year) interaction rewiring in plant-pollinator communities, influencing their functional structure. For this, we particularly tested whether rewiring frequency was negatively related to the extent of natural habitats surrounding the communities, whether interaction rewiring influenced the static structure of networks (nestedness, network specialization –H2’–), and whether a high generalization (low specialization –d’–) and abundance of species in communities made them more prone to rewiring. We show that habitat loss increased seasonal interaction rewiring in networks. Changes in rewiring were related to changes in the cumulative static structure of pollination networks. Nestedness decreased and network specialization (H2’) also tended to decrease as interaction rewiring increased, suggesting an indirect effect of habitat loss on the robustness of networks through their dynamics. As expected, generalist insect and plant species were more prone to rewiring. However, flower abundance had different effects on the rewiring probability of plant species depending on the extent of habitat loss, with abundant species rewiring more in disturbed communities and rewiring less in more natural communities. Likely, this is related to the context-dependent foraging behaviour of pollinators, which may switch to more abundant species if the cost of searching for trait-matching resources is high in disturbed habitats. Our work shows the role of partner-switching in generalist species to adapt to new conditions. It also highlights the importance of going beyond general network metrics to understand the underlying processes of community-level interaction assembly, and predict and anticipate the effects of anthropogenic disturbances on pollination services.