Genetic evidence of Quaternary demographic changes in four rain forest tree species sampled across the Isthmus of Panama

Aim We examined and compared population genetic structure in a suite of four co-occurring Panamanian tree species and performed coalescent-based analyses of demographic history to evaluate hypotheses of tropical vegetation change during the Last Glacial Maximum (LGM). Location Isthmus of Panama. Methods Nuclear microsatellite variation was assayed in multiple populations (1179 trees, 21 locations, 613 locations per species) in Jacaranda copaia (Bignoniaceae), Luehea seemannii (Malvaceae), Simarouba amara (Simaroubaceae) and Symphonia globulifera (Clusiaceae). Population structure was analysed using FST-based statistics and a Bayesian clustering approach (baps). Bayesian coalescent methods (msvar) were used to infer demographic histories. Results High levels of genetic diversity were found in all of the species (HE range, 0.560.79). Jacaranda copaia and L. seemannii showed lower FST and fewer Bayesian clusters across similar spatial scales than did S. globulifera and Simarouba amara. For each species examined, the current effective population sizes (Ne) are much lower than ancient Ne, within all inferred baps demes. In light-demanding pioneer species J. copaia, L. seemannii and S. amara, estimates of the number of generations since the bottleneck events overlap with the end of the LGM (median site posterior estimates ranged from 16 to 19ka) while Symphonia globulifera estimates are consistent with earlier population declines (median 202ka) in the early Pleistocene and late Pliocene. Main conclusions The wind-dispersed deciduous species J. copaia and L. seemannii showed lower FST and spatially extensive demes, while the animal-dispersed evergreen Symphonia globulifera and Simarouba amara showed spatially restrictive demes and higher FST. Each deme examined shows evidence of historical bottlenecks. For three of the four species which are also light-demanding pioneer species, the mean estimated time and 95% highest posterior density of the bottleneck events coincides with the end of the LGM. These results suggest that these species have undergone historical bottlenecks as a result of reduced forest cover during the Pleistocene and provide evidence of shared demographic histories among co-occurring tropical forest trees.