Islands have long been considered nature's laboratories. By offering a study system with known area, isolation and age, research into island ecosystems has pioneered our understanding of how species assemblages develop and persist over time in relation to abiotic factors. Yet a consistent eco-evolutionary understanding of how biotic interactions -- such as mutualistic interactions between plants, pollinators and seed-dispersers -- operate over contemporary and historical timeframes remains elusive. Here, first we identify challenges that hamper progress in understanding and predicting mutualistic interactions, such as spatio-temporal sampling resolution and the ability to make solid inferences on species' evolutionary trajectories and histories. We then propose how modern molecular approaches provide solutions to these long-standing challenges, opening the possibility of using genomics to advance island research to obtain a novel eco-evolutionary understanding of plant-animal mutualisms. Notably, by outlining empirically testable hypotheses and illustrating how molecular approaches can address long-standing questions when combined with island theory, we provide a roadmap for transforming island mutualism research into a predictive eco-evolutionary science. The need for transformative progress in this field is ever more pressing because already vulnerable island ecosystems continuously become more threatened by anthropogenic impacts, and mainland ecosystems are increasingly being fragmented into habitat islands.