Species-area relationships on small islands differ among plant growth forms

Aim We tested whether species-area relationships of small islands differ among plant growth forms and whether this influences the prevalence of the small-island effect (SIE). The SIE states that species richness on small islands is independent of island area or relates to area in a different way compared with larger islands. We investigated whether island isolation affects the limits of the SIE and which environmental factors drive species richness on small islands. Location Seven hundred islands (< 100 km(2)) worldwide belonging to 17 archipelagos. Major taxa studied Angiosperms. Methods We applied linear and breakpoint species-area models for angiosperm species richness and for herb, shrub and tree species richness per archipelago separately, to test for the existence of SIEs. For archipelagos featuring the SIE, we calculated the island area at which the breakpoints occurred (breakpoint area) and used linear models to test whether the breakpoint areas varied with isolation. We used linear mixed-effect models to discern the effects of seven environmental variables related to island area, isolation and other environmental factors on the species richness of each growth form for islands smaller than the breakpoint area. Results For 71% of all archipelagos, we found an SIE for total and herb species richness, and for 59% for shrub species richness and 53% for tree species richness. Shrub and tree species richness showed larger breakpoint areas than total and herb species richness. The breakpoint area was significantly positively affected by the isolation of islands within an archipelago for total and shrub species richness. Species richness on islands within the range of the SIE was differentially affected by environmental factors across growth forms. Main conclusion The SIE is a widespread phenomenon that is more complex than generally described. Different functional groups have different environmental requirements that shape their biogeographical patterns and affect species-area and, more generally, richness-environment relationships. The complexity of these patterns cannot be revealed when measuring overall plant species richness.