Marine islands harbour a great part of our planet’s biological diversity and a high proportion of endemic species. However, island biota are particularly susceptible to anthropogenic threats like climate change, habitat loss and species invasions, making a detailed understanding of ecological processes on islands imperative if we wish to conserve their unique biotas. Due to their discrete and isolated nature, islands are useful model systems in ecological and evolutionary research. So far, however, most studies have concentrated on small-scale biogeographic patterns, and standardized global data on island biogeographic characteristics as well as a macroecological synthesis of their biotas are currently lacking. In this thesis, I provide a physical and bioclimatic characterization of the world’s islands and address the question of how abiotic island characteristics affect the diversity of island floras. I tackle two major aspects of this question: First, I investigate past and present bioclimatic and physical island characteristics as drivers of island plant diversity patterns, focusing on the spatial arrangement of islands and structure of archipelagos. Second, I investigate taxon-specific and trait-related differences in the response of plant diversity patterns to abiotic factors on islands. I present a database of past and present bioclimatic and physical characteristics including island area, isolation and geology for 17,883 islands larger than 1 km² worldwide. Using ordination and clustering techniques, I characterize and classify the islands in multidimensional environmental space. I also develop a set of ecologically meaningful metrics of island isolation and of the spatial arrangement of islands in archipelagic settings, including metrics related to stepping stones, wind and ocean currents, climatic similarity, inter-island distances and surrounding landmass area. These metrics account for different aspects of island isolation that influence immigration, in situ speciation and extinction on islands and turnover among islands. To link abiotic characteristics of islands to biotic attributes of island floras, I use a database of 1,295 island species checklists including c. 45,000 native vascular plant species, compiled for this thesis. It is the first global and most comprehensive dataset on island plant diversity, including species identities instead of mere species numbers. The global island characterization quantitatively confirms that islands differ from mainland areas in their bioclimatic and physical characteristics. Islands are, on average, significantly cooler, wetter and less seasonal than mainlands. I show that a thorough circumscription of the physical and spatial characteristics of islands and archipelagos, especially isolation, archipelago structure and scale, is needed to understand diversity patterns of their biotas. Isolation is the second most important factor after area determining vascular plant species richness on islands. Among the compared isolation metrics, the proportion of surrounding land area serves best for explaining vascular plant species richness. Also, accounting for stepping stones, large islands as source landmasses and climatic similarity of source areas increases the explanatory power of models compared to considering only the commonly used distance to the nearest mainland. Isolation is less important on large islands where in situ diversification counteracts the negative effect of isolation on immigration. At archipelagic scales, the intra-archipelagic spatial structure is particularly important for β diversity, i.e. species turnover among islands, and, indirectly through β, for γ diversity, i.e. the species richness of the entire archipelago. These findings highlight the importance of the spatial location of islands relative to each other for insular diversity patterns and indicate the necessity of considering islands in an archipelagic context in island research and conservation. For ferns on Southeast Asian islands, I show that the importance of physical island characteristics for diversity gradually decreases with spatial grain size from island level to plot level, where local environmental conditions are more important. Local communities may often be saturated, limiting the number of species that can immigrate from the regional species pool. To make predictions about local diversity on islands it is hence important to take the scale-dependence of species pool effects into account. Major plant groups differ in their dispersal abilities, levels of gene flow, speciation rates and adaptations to climate. Accordingly, comparative analyses among taxonomic plant groups reveal clear differences among taxa in the responses of species richness and phylogenetic diversity patterns to abiotic factors. I find varying island species-area relationships (SPARs), i.e. rates of increase in species richness with island area, among groups. The slope of SPARs is higher in spermatophytes than in pteridophytes and bryophytes, whereas the intercept is lower. In combination, phylogenetic trait and niche conservatism, environmental and dispersal filtering mechanisms and in situ speciation are expected lead to phylogenetically clustered assemblages. I show that physical and bioclimatic island attributes, linked to filtering and speciation, influence the phylogenetic structure and diversity of island floras. The strengths and directions of the relationships vary among taxonomic groups. Abiotic predictors explain more variation in phylogenetic diversity and structure for angiosperms and palms than for ferns, which is in accordance with the high dispersal ability and large range sizes of fern species. The abiotic characterization and regionalization of the world’s islands I present and the accompanying data should facilitate a more integrative consideration of islands in macroecological research. In this thesis, I provide the first predictions of insular vascular plant species richness and analyses of different diversity components (α, β, γ and phylogenetic diversity) of insular systems and their abiotic drivers at a global scale. I show that relationships between environmental drivers and species richness as well as phylogenetic assemblage attributes vary among taxonomic groups in dependence on their predominant dispersal and speciation-related characteristics. This is a new perspective in macroecological island research allowing inferences about the mechanisms underlying patterns of island plant diversity. Furthermore, understanding how the diversity of insular plant groups is shaped by immigration and diversification should promote our understanding of global diversity patterns in general.