Lively debate continues over whether marine reserves can lead to increased fishery yields when compared to conventional, quota-based management, apparently driven by differences in the complexity and biological richness of the models being used. In an influential article, Hastings and Botsford used an analytically tractable, spatially implicit, non-age-structured model to assert that reserves are typically incapable of increasing yields relative to conventional management, regardless of the type (pre- or post-dispersal, involving adults and/or larvae) or functional form (Ricker or Beverton-Holt) of density dependence present. A recent numerical (simulation) model by Gaylord et al. concludes that reserves can enhance yield compared to conventional management, a result the authors attribute to their spatially-explicit evaluation of stage-structured adult growth, survivability and fecundity; and intercohort (adult-on-larvae) post-dispersal density dependent population dynamics. Here we demonstrate that the increased model complexity is not responsible for the different conclusions. We analyze a spatially-implicit model without stage structure that incorporates intercohort post-dispersal density dependence. In this simple model we still find annual extirpation of adult populations outside reserves due to fishing to enhance larval recruitment there, allowing for increased yields compared to those achieved when harvest is evenly spread across the entire domain under conventional management. Consideration of neither spatially-explicit dispersal dynamics nor stage-structure in adult demographics is required for reserves to substantially improve yield beyond that attainable under conventional management. In contrast, consideration of within cohort post-dispersal density dependence among larva during settlement in an otherwise identical model generates equivalence in yield between the two management strategies. These results recast a common message characterizing the relative benefit of reserve versus non-reserve management from “equivalence at best” to “potentially improved”.