Abstract Aim To determine whether different abundances of introduced species of Cinchona (Rubiaceae) affect species composition and facilitate species richness in managed tropical forests, to test whether any facilitative effects on understorey species depend on forest type, and to investigate whether facilitative effects can be attributed to the ‘substitutive facilitation model’. Location Makawao Forest Reserve on Maui, Hawai’i, USA. Methods Cinchona species (Cinchona pubescens and Cinchona calisaya) were mapped within various forest types. In three forest types (ageing Eucalyptus and Pinus plantations, and near-natural Acacia koa forests), we analysed environmental parameters (e.g. canopy cover, litter cover, pH value and soil depth) and the species composition of Cinchona-invaded and non-invaded plots; data were compared based on Cinchona cover and forest types. Habitat modelling for several endemic species and tree ferns was carried out to test whether Cinchona cover is an important variable for the probability of occurrence of these endemics. Results Cinchona species have naturalized mainly in Eucalyptus and Pinus plantations and Acacia koa forests and here add an additional shrub layer. In contrast to other studies, we revealed facilitative effects of Cinchona on native species within all forest types. Species richness is about 20% higher in invaded plots than in non-invaded plots, and these show a nearly 50% higher proportion of endemic species, including tree ferns. The proportion of endemics even increases with increasing Cinchona cover. For several endemics, Cinchona is found to be an important variable for the probability of occurrence, and the removal of Cinchona cover as an explanatory variable lowers the model fit. In addition to Cinchona, variables delineating vegetation structure and light availability have a strong effect on the model fit. Main conclusions In the structurally simplified Hawaiian forests studied, Cinchona facilitated endemic species in accordance with the ‘substitutive facilitation model’. This contrasts with the results of an earlier study in the naturally treeless Galápagos highlands, which revealed a sharp decrease in the abundance of endemics under Cinchona canopy. These results illustrate that, through the same structural change (addition of a vegetation layer), an invasive species may exert divergent effects across different ecosystem types. The facilitation of endemic understorey species by invasive tree species in managed forests leads to a dilemma in conservation but also to new perspectives for ecosystem restoration.