Advantages of dispersal on the scales that are possible in a long pelagic larval period are not apparent, even for benthic species. An alternative hypothesis is that wide dispersal may be an incidental byproduct of an ontogenetic migration from and then back to the parental habitat. Under this hypothesis, the water column is a better habitat than the bottom for early development. Because the parental area is often an especially favorable habitat for juveniles and adults, selection may even favor larval retention or larval return rather than dispersal. Where larval capabilities and currents permit, a high percentage of recruits may then be produced from local adults. Expected consequences of a high proportion of local recruitment are stronger links between stock and recruitment, greater vulnerability to recruitment overfishing and local modifications of habitat, greater local benefits from fishery reserves, and possibly more localized adaptation within populations. Export of some larvae is consistent with a high proportion of retained or returning larvae, could stabilize populations linked by larval exchange, and provide connectivity between marine reserves. Even a small amount of larval export could account for the greater gene flow, large ranges, and long evolutionary durations seen in species with long pelagic larval stages.