Tropical arid to semi-arid ecosystems are nearly as diverse as more humid forests and occupy large parts of the tropics. In comparison, however, they are vastly understudied. For instance, fog precipitation alone supports a unique vegetation formation, locally termed lomas, on coastal mountains in the Peruvian desert. To effectively protect these highly endemic and threatened ecosystems, we must increase our understanding of their diversity patterns in relation to environmental factors. Consequently, we recorded all vascular species from 100 random 4 × 4 m plots on the fog-exposed southern slope of the mountain Mongón. We used topographic and remotely sensed covariates in statistical models to generate spatial predictions of alpha diversity and plant species' distribution probabilities. Altitude was the most important predictor in all models and may represent fog moisture levels. Other significant covariates in the models most likely refer also to water availability but on a finer spatial scale. Additionally, model-based clustering revealed five altitudinal vegetation zones. This study contributes to a better spatial understanding of the biodiversity and spatial arrangement of vegetation belts of the largely unknown but highly unique lomas formations. Furthermore, mapping species richness and plant species' distributions could support a long-needed lomas strategic conservation scheme. © 2013 Association for Tropical Biology and Conservation Inc.