The ongoing decline of large marine vertebrates must be urgently mitigated, particularly under increasing levels of climate change and other anthropogenic pressures. However, characterizing the connectivity among populations remains one of the greatest challenges for the effective conservation of an increasing number of endangered species. Achieving conservation targets requires an understanding of which seascape features influence dispersal and subsequent genetic structure. This is particularly challenging for adult-disperser species, and when distribution-wide sampling is difficult. Here, we developed a two-step modelling framework to investigate how seascape features drive the genetic connectivity of marine species without larval dispersal, to better guide the design of marine protected area networks and corridors. We applied this framework to the endangered grey reef shark, Carcharhinus amblyrhynchos, a reef-associated shark distributed across the tropical Indo-Pacific. In the first step, we developed a seascape genomic approach based on isolation-by-resistance models involving circuit theory applied to 515 shark samples, genotyped for 4991 nuclear single-nucleotide polymorphisms. We show that deep oceanic areas act as strong barriers to dispersal, while proximity to habitat facilitates dispersal. In the second step, we predicted the resulting genetic differentiation across the entire distribution range of the species, providing both local and global-scale conservation units for future management guidance. We found that grey reef shark populations are more fragmented than expected for such a mobile species, raising concerns about the resilience of isolated populations under high anthropogenic pressures. We recommend the use of this framework to identify barriers to gene flow and to help in the delineation of conservation units at different scales, together with its integration across multiple species when considering marine spatial planning.