Dispersal is the primary ecological process underpinning spatial dynamics in motile species by generating flux in reproductive locations over time. In migratory species, dispersal can also occur around non-breeding ranges, but this form currently lacks a unifying theoretical framework. We present a novel conceptual model for dispersal in migrants that builds upon existing literature, differentiating ‘reproductive’ dispersal (i.e. changes in breeding locations) from ‘non-reproductive’ dispersal, which we define as movements resulting in inter-annual or inter-generational changes in non-breeding locations. Crucially, unlike reproductive dispersal where movement outcomes are naturally propagated between generations, the outcomes of non-reproductive dispersal can be non-heritable even if dispersers survive to reproduce. We simulate a non-social migratory population with a genetically-determined migratory programme to model how heritability of this program influences both migratory connectivity and range shift propensity. When exposed to spatially-uncoupled shifts in habitable ranges (i.e. seasonal climate niches shifting at different rates), long-term persistence of simulated populations required changes in migratory programmes to arise through heritable forms of non-reproductive dispersal (e.g. mutations in migratory gene complexes). By contrast, non-heritable dispersal mechanisms (e.g. weather drift, navigation errors) did not drive long-term shifts in non-breeding ranges, despite being a major component of realised dispersal and migratory connectivity patterns. Migratory connectivity metrics conflate these heritable and non-heritable drivers of non-reproductive dispersal and therefore have limited power in predicting spatial population responses to environmental change. Our models provide a framework for improving our understanding of spatial dynamics in migratory populations and highlight the importance of teasing apart the genetic or cultural mechanisms that drive inter-generational migratory variability in order to evaluate and predict range plasticity in migrants.