While geologists suggest that New Caledonian main island (Grande Terre) was submerged until ca 37 Ma, biologists are struck by the presence of supposedly Gondwanan groups on the island. Among these groups are the Oreosycea fig trees (Ficus, Moraceae) and their Dolichoris pollinators (Hymenoptera, Agaonidae). These partners are distributed in the Paleotropics and Australasia, suggesting that their presence on New Caledonia could result from Gondwanan vicariance. To test this hypothesis, we obtained mitochondrial and nuclear markers (5.3 kb) from 28 species of Dolichoris, used all available sequences for Oreosycea, and conducted phylogenetic and dating analyses with several calibration strategies. All our analyses ruled out a vicariance scenario suggesting instead that New Caledonian colonization by Dolichoris and Oreosycea involved dispersal across islands from Sundaland ca 45.9-32.0 Ma. Our results show that successful long-distance dispersal of obligate mutualists may happen further suggesting that presence of intimate mutualisms on isolated islands should not be used as a priori evidence for vicariance. Comparing our results to a review of all the published age estimates for New Caledonian plant and animal taxa, we showed that support for a vicariant origin of the island biota is still lacking. Finally, as demonstrating a causal relationship between geology and biology requires independent evidence, we argue that a priori assumptions about vicariance or dispersal should not be used to constrain chronograms. This circular reasoning could lead to under or overestimation of age estimates.