From plots to islands: species diversity at different scales

Aim To investigate how plant diversity of whole islands (gamma) is related to alpha and beta diversity patterns among sampling plots within each island, thus exploring aspects of diversity patterns across scales. Location Nineteen islands of the Aegean Sea, Greece. Methods Plant species were recorded at both the whole-island scale and in small 100 m(2) plots on each island. Mean plot species richness was considered as a measure of alpha diversity, and six indices of the 'variation'-type beta diversity were also applied. In addition, we partitioned beta diversity into a ` nestedness' and a 'replacement' component, using the total species richness recorded in all plots of each island as a measure of ` gamma' diversity. We also applied 10 species-area models to predict the total observed richness of each island from accumulated plot species richness. Results Mean alpha diversity was not significantly correlated with the overall island species richness or island area. The range of plot species richness for each island was significantly correlated with both overall species richness and area. Alpha diversity was not correlated with most indices of beta diversity. The majority of beta diversity indices were correlated with whole-island species richness, and this was also true for the 'replacement' component of beta diversity. The rational function model provided the best prediction of observed island species richness, with Monod's and the exponential models following closely. Inaccuracy of predictions was positively correlated with the number of plots and with most indices of beta diversity. Main conclusions Diversity at the broader scale (whole islands) is shaped mainly by variation among small local samples (beta diversity), while local alpha diversity is not a good predictor of species diversity at broader scales. In this system, all results support the crucial role of habitat diversity in determining the species-area relationship.