Aims : To identify groups of species with similar geographic ranges in the non-endemic flora of the Socotra Archipelago, place them within a global biogeographic context, and examine how these chorotypes are reflected in Socotran vegetation types. Study area : The Socotra Archipelago (Yemen) and its surroundings, analysed against a 1° × 1° global grid to provide global context and a Socotra-centred regional focus. Methods : We assembled 4,493,566 GBIF records for 477 target taxa; 417 had usable occurrences and 389 met a ≥ 15-occurrence threshold for modelling. To obtain coarse-grained ranges, we fitted ensemble models and binarised predictions using conservative thresholds. Grid cells were grouped according to species turnover and geographic proximity using clustering methods, and statistical tests identified species associated with particular clusters. A second analysis focused on a Socotra-centred subset to resolve finer regional patterns. Species-level chorotypes were linked to synoptic vegetation tables to characterise vegetation types. Results : The global analysis identified ten spatially cohesive groups of species at the continent scale. The Socotra-centred analysis resolved finer regional structure. Socotra’s chorological spectrum is dominated by the Xeric Tropical-Subtropical chorotype (13.8%), with contributions from the Paleoarabian-Somali-Masai chorotype (10.1%) and Temperate-Subtropical Belt chorotype (9.2%). Vegetation types differed markedly in their chorological signatures: endemic-rich woodlands and shrublands showed high specialisation, whereas grasslands and halophytic communities exhibited more mixed spectra. Conclusions : Turnover-based partitions recover major Afro-Arabian structures and highlight the Red Sea-Gulf of Aden rim as a potential biogeographic hinge. The dominance of xeric Afro-Arabian chorotypes, together with a secondary signal from mesic Afrotropical and Indo-Pacific groups, suggests a flora structured by arid corridors and monsoon- or fog-related refugia. Linking chorotypes with vegetation types reveals an ecological gradient from endemic-rich communities to more open assemblages with broader biogeographic affinities. Despite known limitations, the approach provides reproducible co-distribution units connecting classical chorology with quantitative vegetation analysis.