Islands have traditionally been considered to be any relatively small body of land completely surrounded by water. However, their primary biological characteristic, an extended period of isolation from a source of colonists, is common also to many situations on continents. Accordingly, theories and predictions developed for true islands have been applied to a huge array of systems, from rock pools, to single tree species in forests, to oceanic islands. Here, we examine the literature on islands in the broadest sense (i.e., whether surrounded by water or any other uninhabitable matrix) as it pertains to terrestrial arthropods. We categorize islands according to the features they share. The primary distinction between different island systems is "darwinian" islands (formed de novo) and "fragment" islands. In the former, the islands have never been in contact with the source of colonists and have abundant "empty" ecological niche space. On these islands, species numbers will initially increase through immigration, the rate depending on the degree of isolation. If isolation persists, over time species formation will result in "neo-endemics." When isolation is extreme, the ecological space will gradually be filled through speciation (rather than immigration) and adaptive radiation of neo-endemics. Fragment islands are fundamentally different. In these islands, the ecological space will initially be filled as a consequence of connection to the source of colonists prior to insularization. Species numbers will decrease following fragmentation through the process of relaxation. If these islands become more isolated, species will eventually arise through relictualization with the formation of "paleo-endemics." Given sufficient time, this process can result in generic level endemism on ancient fragment islands, a phenomenon well illustrated in Madagascar and New Zealand. Recognizing the distinction between the different kinds of islands is fundamental for understanding emerging patterns on each, in particular speciation, biodiversity (e.g., neo-endemics versus paleo-endemics), and conservation (e.g., naiveté in interactions with alien species).
