Résumé
Aim: We map global patterns of taxonomic and functional change between past (pre‐human impacts) and present (after anthropogenic extinctions and introductions) in large oceanic island bird assemblages and investigate if these patterns can be explained by island characteristics and anthropogenic factors.Location: Sixty‐four oceanic islands (<100 km 2 ).Time Period: Late Holocene.Major Taxa Studied: Terrestrial and freshwater bird species.Methods: We compiled information on extinct, extant native and introduced bird species for all islands and used a probabilistic hypervolume approach to build a multi‐dimensional trait space and calculate several functional diversity metrics before and after extinctions and introductions. We identified which islands are global hotspots of human‐induced transformation by mapping multiple facets of biotic change and investigated intrinsic island characteristics and anthropogenic factors as drivers for these observed patterns.Results: The Hawaiian and Mascarene islands stand out as hotspots of taxonomic and functional change, but all islands changed taxonomically and functionally, mostly gaining species but losing functional richness. Taxonomic and functional changes vary across islands but are often consistent within the same archipelago. Island isolation and surface can explain some of the observed variations, but anthropogenic factors, namely human occupation, also shaped both taxonomic and functional changes. Islands with higher human pressure, as well as larger islands with high elevation ranges, tended to have greater losses in functional richness.Main Conclusions Most biodiversity change assessments are still largely based exclusively on taxonomic diversity, which is particularly worrying in the case of oceanic islands given that the magnitude of functional diversity change is often considerably larger. We call for comprehensive assessments of changes in both taxonomic and functional diversity across oceanic islands in order to better understand the drivers of these changes and, in turn, predict future trends.