|Aim To examine the relationship between island characteristics (area, distance to the nearest continent, climate and human population size) and ant species richness, as well as the factors underlying global geographical clustering of native and exotic ant composition on islands.
Location One hundred and two islands from 20 island groups around the world.
Methods We used spatial linear models that consider the spatial structure of islands to examine patterns of ant species richness. We also performed modularity analyses to identify clusters of islands hosting a similar suite of species and constructed conditional inference trees to assess the characteristics of islands that explain the formation of these islandľant groups.
Results Island area was the best predictor of ant species richness. However, distance to the nearest continent was an important predictor of native ant species richness, as was human population size for exotic species richness. Native species appear slightly more modulated (i.e. well grouped in species assemblages that are present over a distinct cluster of islands) than are exotic species. Exotic species, while still exhibiting some modularity, tended to be widely distributed among island groups. Interestingly, ocean currents accounted for most of the variation in modularity and thus species composition for both native and exotic ant species.
Main conclusions Contrary to previous work, both native and exotic species appeared to be confined to particular island regions, and patterns in the distribution of both native and exotic species were limited by a similar suite of factors. However, the distribution of exotic ant species appeared to be more influenced by human-related variables and less structured relative to those of native ant species, perhaps due to the long-term (and increasing) influence of human-mediated dispersal that favours exotic species.|