Geography plays an important role in the distribution of plants on islands. This is in part because of the diversity of places and associated environmental conditions in which the islands are located, but also because of how islands are positioned with respect to one another. This relative positioning enters explicitly into island biogeographical character and can be expressed through spatial models. Over the past 20 years, spatial techniques for the empirical analysis of biological datasets have been increasingly applied to investigate biogeographical phenomena, particularly toward a better understanding of spatially structured underlying causative factors. These might include dispersal and competition, as well as environmental and historical influences. This study investigates patterns in the number of plant species occuring on 43 islands of the Great Barrier Reef (GBR) at three different geographical sectors (whole GBR, northern GBR, and southern GBR). Measures of spatial autocorrelation are calculated to explore the relationship between the diversity of plant populations on a given island and those on neighbouring islands. The relationship between the number of island plant species and local geographical context (latitude, longitude, distance from mainland, island area, island length, depth of surrounding GBR lagoon floor and island isolation) is investigated using three different regression models (ordinary least squares, spatially lagged and spatial error). Findings indicate that the southern islands exhibit the strongest spatial autocorrelation of plant species number between islands. In this sector, geographical context also explained the highest proportion of observed plant species numbers. The distribution of the number of plant species and their autocorrelation characteristics indicate metapopulation dynamics that could be a response to the variable sea-level history of these islands through the Holocene. This controls the time that plant communities have had to reach and maintain a dynamic equilibrium with their local environmental setting. Consistent higher performance of spatial as opposed to classic regression models highlighted the importance of interactions between plant communities on neighbouring islands, providing a persuasive case for explicitly building geography into studies of island plant communities.