Knowledge about migration potential is key to forecasting species distributions in changing environments. For many marine benthic invertebrates, migration happens during reproduction because of larval dispersal. The present study aims to test whether larval size can be used as a surrogate for migration potential arising from larval longevity, competence, sinking, or swimming behavior. The hypothesis was tested using larvae of three sympatric gorgonian species that release brooded lecithotrophic larvae in the same season: Paramuricea clavata, Corallium rubrum and Eunicella singularis. Despite different fecundities and larval sizes, the median larval longevity was similar among the three species. Free-fall speed increased with larval size. Nevertheless, the only net sinkers were the P. clavata larvae, as swimming was more common than free fall in the other two species with larger larvae. For the other two species, swimming activity frequency decreased as larval size increased. Interestingly, maximum larval longevity was lowest for the most active but intermediately sized larvae. Larval size did not covary consistently with any larval traits of the three species when considered individually. We thus advise not using larval size as a surrogate for migration potential in distribution models. The three species exemplified that different mechanisms, i.e., swimming activity or larval longevity, resulting from a trade-off in the use of energy reserves can facilitate migration, regardless of life history strategy. Forecasting the impact of climate change on biodiversity requires mechanistic models similar to those used for climate change predictions (IPBES 2030 work program, https ://ipbes .net/o4-suppo rting-polic y). Species, habitat or niche models should incorporate the mechanisms regulating species spatial distributions to project future changes in biodiversity and species richness. Since Darwin's early observations, the ecological niche concept has been key to explaining the spatial distribution of species under selection pressures that arise from the environment and interactions between species 1,2. The concept postulates that each species should have a unique set of conditions defining its persistence. Correlative approaches linking species occurrence data with environmental descriptors have been extensively applied to infer species niches and to project species spatial distributions (reviewed in 3). However, correlative approaches might fail to predict changes in species spatial distributions when environmental conditions evolve more rapidly than the species can cope with 4. Indeed, when local environmental conditions change, a species' persistence will depend on its ability to track suitable conditions for existence through migration and/or to adapt locally 5. Trait-based mechanistic models explicitly include migration processes through a migration parameter, i.e., expansion distance per unit time (meta-population 6 ; meta-community 7). However, spatiotemporal mechanistic models are still limited in their application due to knowledge gaps in migration parameters 8. For sessile species OPEN