Résumé
Aim: To determine the role of regional forcing on plot-level species diversity and composition, and to quantify the relative importance of biogeographical and climatic factors in explaining woody plant diversity and composition at the local-, island- and archipelago-scale. Location: Forty-one tropical islands of the Indo-Pacific region from Madagascar to Hawai‘i Island. Methods: We analysed the diversity and composition of tropical woody plant communities located across 113 plots, 41 islands and 19 archipelagos. We used generalized linear mixed-effects models and generalized dissimilarity models to determine the role of regional forcing at the island and archipelago scale and to assess the relative importance of biogeographical (area and isolation of islands or archipelagos, geographical distance between plots) and climatic factors in explaining differences in local diversity and composition (species turnover). Analyses were conducted at different geographical scales (local, island and archipelago) and taxonomic levels (species, genus and family). Results: Variation in local (plot-level) diversity (as species density, the number of species per 100 woody plants) was primarily explained by island and archipelago identity. Maximum species density was positively correlated with the area of an island (or archipelago) and negatively correlated with the isolation of an archipelago. Local climatic variability was also a significant predictor of species density, but less important than regional forcing. Climate variables explained < 20% of the variation in species turnover across all plots. The importance of geographical distance between plots relative to climate in driving species turnover decreased from the species to family level, and from the regional to island level. Main conclusions: Regional forcing was the key driver of local diversity and composition on islands. Island area and archipelago isolation are likely driving local diversity through their effects on the pool of island species. Geographical distance between plots is the main factor explaining species turnover, while at higher taxonomic levels, climatic factors and niche conservatism are the main drivers.