When and where did troidine butterflies (Lepidoptera : Papilionidae) evolve? Phylogenetic and biogeographic evidence suggests an origin in remnant Gondwana in the Late Cretaceous

The age, geographic origin and time of major radiation of the butterflies (Hesperioidea + Papilionoidea + Hedyloidea) are largely unknown. The general modern view is that butterflies arose during the Late Jurassic/Cretaceous in the southern hemisphere (southern Pangea/Gondwana before continental breakup), but this is not universally accepted, and is a best guess based largely on circumstantial evidence. The extreme paucity of fossils and lack of modern, higher-level phylogenies of extant monophyletic groups have been major impediments towards determining reliable estimates of either their age or geographic origin. Here we present a phylogenetic and historical biogeographic analysis of a higher butterfly taxon, the swallowtail tribe Troidini. We analysed molecular data for three protein-encoding genes, mitochondrial ND5 and COI–COII, and nuclear EF–1α, both separately and in combination using maximum parsimony (with and without character weighting and transition/transversion weighting), maximum likelihood and Bayesian methods. Our sample included representatives of all 10 genera of Troidini and distant ingroup taxa (Baroniinae, Parnassiinae, Graphiini, Papilionini), with Pieridae as outgroup. Analysis of the combined dataset (4326 bp; 1012 parsimony informative characters) recovered the Troidini as a well supported monophyletic group and the monophyly of its two subtribes, Battina and Troidina. The most parsimonious biogeographic hypothesis suggests a southern origin of the tribe in remnant Gondwana (Madagascar–Greater India–Australia–Antarctica–South America) sometime after the rifting and final separation of Africa in the Late Cretaceous (<90 Mya). Although an ancient vicariance pattern is proposed, at least four relatively recent dispersal/extinction events are needed to reconcile anomalies in distribution, most of which can be explained by geological and climatic events in South-east Asia and Australia during the late Tertiary. Application of a molecular clock based on a rate smoothing programme to estimate various divergence times based on vicariance events, revealed two peculiarities in our biogeographic vicariance model that do not strictly accord with current understanding of the temporal breakup of Gondwana: (1) the troidine fauna of Greater India did not become isolated from Gondwana (Antarctica) until the end of the Cretaceous (c. 65 Mya), well after Madagascar separated from Greater India (84 Mya); and (2) the faunas of Greater India, Australia and South America diverged simultaneously, also at the K/T boundary. A recent published estimate of the time (31 Mya) of divergence between Cressida Swainson (Australia) and Euryades Felder & Felder (South America) is shown to be in error.