No abstract

Mediterranean refugial areas are generally underrepresented in large-scale genetic surveys of forest trees. In the case of silver fir (Abies alba Mill.), this has led to divergent hypotheses about the exact location of glacial refugia and the trajectory of recolonization routes. Based on the comprehensive sampling of Apennine populations, we aimed to reconcile discrepancies about the number and location of refugia for silver fir in the Apennines and test alternative demographic scenarios developed from palaeobotanical and genetic data. Location Mediterranean Basin; the Apennines and surrounding areas.Methods 1167 individuals from 16 Apennine populations, extensively covering the species’ distribution along the Italian Peninsula, and eight populations from the Alps and Eastern Europe were genotyped at 16 nuclear and three chloroplast microsatellite markers. The geographical distribution of genetic variation was explored using Bayesian clustering and multivariate methods. Based on the inferred genetic structure, the demographic history of A. alba was assessed by the approximate Bayesian computation (ABC) analysis.Results Two unexpected characteristics of genetic structure emerged: a sharp genetic boundary in the central Apennines and a tight genetic connection between southern Apennine and Eastern European gene pools. Two Apennine areas, corresponding precisely with refugial areas hypothesized in most recent palaeobotanical syntheses, have high genetic diversity on a par with Eastern European populations. ABC analysis showed an ancient separation between Apennine and Eastern European gene pools followed by an admixture event that, mainly through directional gene flow via pollen, might have established the genetic similarity between southern Apennine and Eastern European populations. In addition, there was evidence that the central Apennines acted as a small-scale, isolated refugium during the Last Glacial Maximum. Main conclusionsSilver fir rear edge populations have experienced a complex demographic history across several glacial-interglacial cycles, leading to unexpected genetic structure. Our study provides new insights into forest tree dynamics in the Mediterranean, showing the putative presence of multiple refugia for silver fir in the Apennines and a trans-Adriatic connection between silver fir populations in the southern Italy and the Balkans

BackgroundLocal adaptation is a key driver of phenotypic and genetic divergence at loci responsible for adaptive traits variations in forest tree populations. Its experimental assessment requires rigorous sampling strategies such as those involving population pairs replicated across broad spatial scales.MethodsA hierarchical Bayesian model of selection (HBM) that explicitly considers both the replication of the environmental contrast and the hierarchical genetic structure among replicated study sites is introduced. Its power was assessed through simulations and compared to classical ‘within-site’ approaches (FDIST, BAYESCAN) and a simplified, within-site, version of the model introduced here (SBM).ResultsHBM demonstrates that hierarchical approaches are very powerful to detect replicated patterns of adaptive divergence with low false-discovery (FDR) and false-non-discovery (FNR) rates compared to the analysis of different sites separately through within-site approaches. The hypothesis of local adaptation to altitude was further addressed by analyzing replicated Abies alba population pairs (low and high elevations) across the species’ southern distribution range, where the effects of climatic selection are expected to be the strongest. For comparison, a single population pair from the closely related species A. cephalonica was also analyzed. The hierarchical model did not detect any pattern of adaptive divergence to altitude replicated in the different study sites. Instead, idiosyncratic patterns of local adaptation among sites were detected by within-site approaches.ConclusionHierarchical approaches may miss idiosyncratic patterns of adaptation among sites, and we strongly recommend the use of both hierarchical (multi-site) and classical (within-site) approaches when addressing the question of adaptation across broad spatial scales.