Biogeographic discordance of molecular phylogenetic and phenotypic variation in a continental archipelago radiation of land snails

In island archipelagos, where islands have experienced repeated periods of fragmentation and connection through cyclic changes in sea level, complex among-island distributions might reflect historical distributional changes or local evolution. We test the relative importance of these mechanisms in an endemic radiation of Rhagada land snails in the Dampier Archipelago, a continental archipelago off the coast of Western Australia, where ten morphospecies have complex, overlapping distributions.

Despite opportunities for distributional changes during periods when the islands were connected, there is no evidence that dispersal has contributed to the geographic variation of shell form at the broad scale. Based on an estimate of dispersal made previously for Rhagada, we conclude that the periods of connection have been too short in duration to allow for extensive overland dispersal or deep mitochondrial introgression. The result is a sharp and resilient phylogeographic pattern. The distribution of morphotypes among clades and distant islands is explained most simply by their parallel evolution.

We obtained partial mtDNA sequence (COI) for 1015 snails collected from 213 locations across 30 Islands, and used Bayesian phylogenetic analysis and Analysis of Molecular Variance (AMOVA) to determine whether geography or the morphological taxonomy best explains the pattern of molecular evolution. Rather than forming distinct monophyletic groups, as would be expected if they had single, independent origins, all of the widely distributed morphospecies were polyphyletic, distributed among several well-supported clades, each of which included several morphospecies. Each mitochondrial clade had a clear, cohesive geographic distribution, together forming a series of parapatric replacements separated by narrow contact zones. AMOVA revealed further incongruence between mtDNA diversity and morphological variation within clades, as the taxonomic hypothesis always explained a low or non-significant proportion of the molecular variation. In contrast, the pattern of mtDNA evolution closely reflected contemporary and historical marine barriers. Conclusions: Despite opportunities for distributional changes during periods when the islands were connected, there is no evidence that dispersal has contributed to the geographic variation of shell form at the broad scale. Based on an estimate of dispersal made previously for Rhagada, we conclude that the periods of connection have been too short in duration to allow for extensive overland dispersal or deep mitochondrial introgression. The result is a sharp and resilient phylogeographic pattern. The distribution of morphotypes among clades and distant islands is explained most simply by their parallel evolution.