Résumé
Islands are well known for their unique biodiversity, that is endemic species. Researchers have often assumed that island endemics are as old as the islands they occur on for calibrating molecular dating analyses. A reductio ad absurdum approach based on phylogenetic topologies is applied to New Caledonian biota in order to demonstrate that the age of an island does not necessarily correspond to the time of divergence of its endemic taxa. Our demonstration does not rely on any molecular clock inference and is, therefore, free of any flaws related to this method. We argue for further care when assuming that species and the biota they are restricted to (e.g. island, mountain and climatic region) have the same age. Finally, we review evidence on the age of islands and their biota radiations as well as discuss the effect of extinction on island biogeography/biota.