Populations of broadcast spawning marine organisms often have large sizes and are exposed to reduced genetic drift. Under such scenarios, strong selection associated with spatial environmental heterogeneity is expected to drive localized adaptive divergence, even in the face of connectivity. We tested this hypothesis using a seascape genomics approach in the commercially important greenlip abalone (Haliotis laevigata). We assessed how its population structure has been influenced by environmental heterogeneity along a zonal coastal boundary in southern Australia linked by strong oceanographic connectivity. Our data sets include 9,109 filtered SNPs for 371 abalones from 13 localities and environmental mapping across ~800 km. Genotype–environment association analyses and outlier tests defined 8,786 putatively neutral and 323 candidate adaptive loci. From a neutral perspective, the species is better represented by a metapopulation with very low differentiation (global FST = 0.0081) and weak isolation by distance following a stepping‐stone model. For the candidate adaptive loci, however, model‐based and model‐free approaches indicated five divergent population clusters. After controlling for spatial distance, the distribution of putatively adaptive variation was strongly correlated to selection linked to minimum sea surface temperature and oxygen concentration. Around 80 candidates were annotated to genes with functions related to high temperature and/or low oxygen tolerance, including genes that influence the resilience of abalone species found in other biogeographic regions. Our study includes a documented example about the uptake of genomic information in fisheries management and supports the hypothesis of adaptive divergence due to coastal environmental heterogeneity in a connected metapopulation of a broadcast spawner. see also the Perspective by Lampert