Abstract Chronically high infestation of salmon louse (Lepeophtheirus salmonis) questions the sustainability of the Norwegian Atlantic salmon (Salmo salar) aquaculture industry. The confinement of millions of hosts, within hundreds of farms with overlapping larval dispersal kernels create the structure for extremely persistent parasite meta-populations. However, the processes regulating the temporal variation in cross-contamination of pelagic salmon louse stages among farms (i.e. connectivity), a vital process driving louse population dynamics, are not well described. Here, we employ a data driven biophysical dispersal model that reproduces three-and-a-half years of production histories of 132 salmon farms in western Norway and quantifies the connectivity of infective pelagic lice stages among the farms with the ocean currents. We show that although the complex geography of western Norwegian fjords governs the long-term topology of the connectivity network, there was a strong seasonal component to network fragmentation. The main de-structuring agent was the delayed infectivity of the pelagic lice stages at cooler temperatures increasing dispersal distances, enhanced by occasional large scale wind forcing events. Coordinated fallowing strategies and de-lousing treatments only played a marginal role in network fragmentation, suggesting that novel lice restraining strategies that consider the environmentally sensitive transport distances must be developed to successfully break up the connectivity network.