1 Department of Genetics, University of Washington, Seattle (USA), 2 Zoologisches Museum, Universität Zürich (Switzerland), 3 Department of Ecology, Ethology and Evolution, University of Illinois, Urbana (USA)
Estimation of organismal dispersal rates is important for understanding biogeographic history, genetic population structure, and local adaptations, especially where climate-related large-scale extinction and recolonization processes occur. During each Pleistocene glacial period, the areas in which western Palearctic water frog populations could persist were restricted to southern, mostly Mediterranean, regions. We investigated two such refugia, the Balkan peninsula and Anatolia. Each refugium has geographic peculiarities and contains a distinct species assemblage: (1) Balkans: High mountains close to the coast restrict coastal lowlands inhabitable by water frogs to narrow, often interrupted areas, even during glacial times. This, together with the scarcity of major rivers and the smallness of plains associated with them, provides a series of barriers to dispersal. Moreover, the Peloponnisos is partly separated from the rest of the Balkans by a salt water barrier. Three water frog species occur in the southern Balkan lowlands: Rana shqiperica, Rana epeirotica, and Rana ridibunda. The last is sympatric with each of the first two, which are allopatric to each other. We included only the last two species in our analysis. (2) Anatolia: The Anatolian peninsula is a dry plateau, surrounded by high mountains, that was uninhabitable by water frogs during glacial periods. There are, however, extended regions of coastal lowlands that are directly or indirectly interconnected, especially in the west and southeast. Only one water frog species (Rana bedriagae) lives in western Anatolia, whence oursamples came. We examined up to 14 variable electrophoretic loci in frogs from 10 localities (Rana epeirotica: 5 loci, 3 populations; Rana ridibunda: 8 loci, 6 populations; Rana bedriagae: 14 loci, 4 populations) and used Slatkin's rare allele approach: occupancy curves give summary statistics for alleles occurring in one to the total number of sampled populations; p(1) values specify the average frequency of alleles restricted to single populations. All three species showed an intermediate amount of genetic exchange among the populations sampled. Judged by these summary measures, the geomorphological structure of the refugia thus appears to have little influence on the dispersal pattern in these frogs. For Rana ridibunda and Rana bedriagae, the number of populations sampled permitted us to test these findings using a modified jackknife approach that generates new data sets in all possible combinations of two up to the total number of populations. For Rana bedriagae from Anatolia, all resulting curves were close to the mean curve based on all populations. The p(1) values generated by removing each population in turn revealed an estimated Nm (effective population size x migration rate) between 0.17 and 0.53. Rana ridibunda from the Balkans, in contrast, showed a variety of curves, many of which indicate a lower rate of dispersal than the overall estimate. The jackknife estimates of Nm values range between 0.10 and 0.15, except for two outliers (1.56 and 6.10), which strongly indicate unequal migration rates among populations; two of the three localities on the Peloponnisos are responsible for the high values. Because only a single Rana ridibunda was sampled at the third Peloponnisos site, it is not certain whether dispersal was markedly reduced within this region, but migration between the Peloponnisos and more northern Balkan localities was certainly restricted. The contrast between the Balkan data and the more continuous change in allele distributions in Anatolia indicates that several thousand years after the last glaciation we can still detect considerable differences in dispersal patterns in Pleistocene refuge areas. This suggests that at least the most recent wave of large-scale dispersal from south to north (after the Würm glaciation), important in the context of hemiclonal reproduction in natural interspecies hybrid lineages of this group of frogs, may be analyzable using similar approaches in conjunction with data on geographic distribution of alleles.