Like many oceanic islands Mauritius has lost the vast majority of its large-bodied vertebrate seed-dispersing fauna. The litany includes the dodo, two species of giant tortoises, parrots, pigeons, fruitbats, and giant lizards. This recent loss has left many Mauritian fleshy-fruited plant species stranded without efficient seed dispersal interactions. Within the framework of the Janzen-Connell model we present experimental evidence demonstrating that seed dispersal is crucial for one of the many critically endangered plant species in Mauritius, Syzygium mamillatum (Myrtaceae) – but no animals currently eat the fruits and disperse the seeds. We asked the question: can we use extant ecological analogues to resurrect the functional aspect of seed dispersal of S. mamillatum? We successfully used giant Aldabran tortoises as ecological analogues. Effects of gut-passage were negative at the seed germination stage, but seedlings from gut-passed seeds grew taller, had more leaves, and suffered less damage from natural enemies than seedlings from non-gut-passed seeds. We thus consider the Aldabran giant tortoise to be an efficient and useful analogue species that can replace extinct endemic seed dispersers of S. mamillatum, and potentially revitalise seed dispersal interactions and other ecosystem functions in many currently dysfunctional ecosystems in Mauritius. In contrast to recent controversy about the use of non-indigenous extant megafauna in North America and elsewhere, we argue that Mauritius and other oceanic islands are ideal study systems in which to empirically explore the use of ecological analogue species in restoration ecology.