Marine species with pelagic larvae typically exhibit little population structure, suggesting long‐distance dispersal and high gene flow. Directly quantifying dispersal of marine fishes is challenging but important, particularly for the design of marine protected areas (MPAs). Here, we studied kelp rockfish (Sebastes atrovirens) sampled along ~25 km of coastline in a boundary current‐dominated ecosystem and used genetic parentage analysis to identify dispersal events and characterize them, because the distance between sedentary parents and their settled offspring is the lifetime dispersal distance. Large sample sizes and intensive sampling are critical for increasing the likelihood of detecting parent–offspring matches in such systems and we sampled more than 6,000 kelp rockfish and analysed them with a powerful set of 96 microhaplotype markers. We identified eight parent–offspring pairs with high confidence, including two juvenile fish that were born inside MPAs and dispersed to areas outside MPAs, and four fish born in MPAs that dispersed to nearby MPAs. Additionally, we identified 25 full‐sibling pairs, which occurred throughout the sampling area and included all possible combinations of inferred dispersal trajectories. Intriguingly, these included two pairs of young‐of‐the‐year siblings with one member each sampled in consecutive years. These sibling pairs suggest monogamy, either intentional or accidental, which has not been previously demonstrated in rockfishes. This study provides the first direct observation of larval dispersal events in a current‐dominated ecosystem and direct evidence that larvae produced within MPAs are exported both to neighbouring MPAs and to proximate areas where harvest is allowed.

Steelhead Oncorhynchus mykiss are the most widespread of the Pacific salmonids Oncorhynchus spp. and are found in nearly all basins within their native range around the northern Pacific Rim. Here, we elucidate genetic population structure of steelhead in coastal basins from most of their coastal‐California range using variation at 15 microsatellite loci. Juvenile fish from 60 streams in 40 river basins were sampled in a single year from a single cohort. As samples of juvenile salmonids often contain sibling groups, a method was implemented to identify and eliminate all but one member of larger sibships. This, in conjunction with a rigorous sampling protocol and hierarchical sampling design, provided substantially improved resolution for understanding patterns of migration and demography. A pattern of isolation by distance was evident, as indicated by both phylograms that were largely concordant with geography and a significant regression of genetic distance on geographic distance, indicating that population structure is largely determined by migration that is dependent upon geographic distance. Within‐basin genetic distances tended to be smaller than those between basins, although there was substantial overlap between them. Using a Bayesian clustering method to evaluate signals of population structure above the level of a river basin, four geographic sites were identified where genetic composition shifted abruptly. These areas largely correspond to major geographic features of the coastline: San Francisco and Humboldt bays and two extended sections of coast (the so‐called Lost Coast and Russian Gulch areas) with no streams reaching inland more than several kilometers. Only one of these boundaries is concordant with the current delineation of steelhead Distinct Population Segments designated under the U.S. Endangered Species Act. Finally, there was a strong correlation between latitude and genetic variation, with fewer alleles present in the south, a pattern consistent with generally smaller population sizes in the south.