Islands offer exclusive prisms for an experimental investigation of biodiversity x ecosystem function interplay. Given that species in upper trophic layers, e.g., arthropod predators, experience a comparative disadvantage on small, isolated islands, such settings can help to clarify how predation features within biotic resistance equations. Here, we use observational and manipulative studies on a chain of nine Indonesian islands to quantify predator-mediated biotic resistance against the cassava mealybug Phenacoccus manihoti (Homoptera: Pseudococcidae) and the fall armyworm Spodoptera frugiperda (Lepidoptera: Noctuidae). Across island settings, a diverse set of generalist lacewing, spider and ladybeetle predators aggregates on P. manihoti infested plants, attaining max. (field-level) abundance levels of 1.0, 8.0 and 3.2 individuals per plant, respectively. Though biotic resistance-as imperfectly defined by a predator/prey ratio index-exhibits no inter-island differences, P. manihoti population regulation is primarily provided through an introduced monophagous parasitoid. Meanwhile, resident predators, such as soil-dwelling ants, inflict apparent mortality rates up to 100% for various S. frugiperda life stages, which translates into a 13- to 800-fold lower S. frugiperda survivorship on small versus large islands. While biotic resistance against S. frugiperda is ubiquitous along the island chain, its magnitude differs between island contexts, seasons and ecological realms, i.e., plant canopy vs. soil surface. Hence, under our experimental context, generalist predators determine biotic resistance and exert important levels of mortality even in biodiversity-poor settings. Given the rapid pace of biodiversity loss and alien species accumulation globally, their active conservation in farmland settings (e.g., through pesticide phasedown) is pivotal to ensuring the overall resilience of production ecosystems.