Aims
The general dynamic model (GDM) of oceanic island biogeography predicts how biogeographical rates, species richness and endemism vary with island age, area and isolation. Here, we used a simulation model to assess whether the isolation‐related predictions of the GDM may arise from low‐level process at the level of individuals and populations.
Location
Hypothetical volcanic oceanic islands.
Methods
Our model considers (a) an idealized island ontogeny, (b) metabolic constraints and (c) stochastic, spatially explicit and niche‐based processes at the level of individuals and populations (plant demography, dispersal, competition, mutation and speciation). Isolation scenarios involved varying the distance to mainland and the dispersal ability of the species pool.
Results
For all isolation scenarios, we obtained humped temporal trends for species richness, endemic richness, proportion of endemic species derived from within‐island radiation, number of radiating lineages, number of species per radiating lineage and biogeographical rates. The proportion of endemics derived from mainland–island differentiation and of all endemics steadily increased over time. Extinction rates of endemic species peaked later than for non‐endemic species. Species richness and the number of endemics derived from mainland–island differentiation decreased with isolation as did rates of colonization, mainland–island differentiation and extinction. The proportion of all endemics and of radiated endemics, the number of radiated endemics, of radiating lineages, and of species per radiating lineage and the within‐island radiation rate all increased with isolation.
Main conclusions
Our results lend strong support to most of the isolation‐related GDM predictions. New insights include an increasing proportion of endemics, particularly those arising from mainland–island differentiation, across isolation scenarios, as well as extinction trends of endemics differing from the overall extinction rates, with a much later peak. These results demonstrate how simulation models focusing on low ecological levels provide tools to assess biogeographical‐scale predictions and to develop more detailed predictions for further empirical tests.