Time to the most recent common ancestor and divergence times of populations of common chaffinches (Fringilla coelebs) in Europe and North Africa: insights into Pleistocene refugia and current levels of migration.

We analyzed sequences from a 275-bp hypervariable region in the 5' end of the mitochondrial DNA control region in 190 common chaffinches (Fringilla coelebs) from 19 populations in Europe and North Africa, including new samples from Greece and Morocco. Coalescent techniques were applied to estimate the time to the most recent common ancestor (TMRCA) and divergence times of these populations. The first objective of this study was to infer the locations of refugia where chaffinches survived the last glacial episode, and this was achieved by estimating the TMRCA of populations in regions surrounding the Mediterranean that were unglaciated in the late Pleistocene. Although extant populations in Iberia, Corsica, Greece, and North Africa harbor haplotypes that are basal in a phylogenetic tree, this information alone cannot be used to infer that these localities served as refugia, because it is impossible to infer the ages of populations and their divergence times without also considering the population genetic processes of mutation, migration, and drift. Provided we assume the TMRCAs of populations are a reasonable estimate of a population's age, coalescent-based methods place resident populations in Iberia, Corsica, Greece, and North Africa during the time of the last glacial maximum, suggesting these regions served as refugia for the common chaffinch. The second objective was to determine when populations began diverging from each other and to use this as a baseline to estimate current levels of gene flow. Divergence time estimates suggest that European populations began diverging about 60,000 years before present. The relatively recent divergence of populations in North Africa, Italy, and Iberia may explain why classic migration estimates based on equilibrium assumptions are high for these populations. We compare these estimates with nonequilibrium-based estimates and show that the nonequilibrium estimates are consistently lower than the equilibrium estimates.