A landscape genetics approach reveals ecological-based differentiation in populations of holm oak (Quercus ilex L.) at the northern limit of its range

The holm oak plays a relevant role in the functioning of Mediterranean forests. In the area north of Garda Lake, Italian Prealps, holm oak populations are at the northernmost edge of their distribution. Being peripheral, these populations are of particular interest for ecological, evolutionary and conservation studies. Through an explicit individual-based landscape genetics approach, we addressed the following questions: (1) are levels of genetic variation reduced in these marginal populations compared with central populations?; (2) despite the narrow geographical scale, do individual-based analyses have some power to detect genetic differentiation?; (3) do environmental and/or climatic factors exert a role in shaping patterns of genetic variation and differentiation? Through a Bayesian method, we identified three clusters whose genetic variability can be considered to be of the same order as that recorded in central Quercus ilex populations. Although being geographically very close (< 20 km), the differentiation was statistically significant (P < 0.05) with global F st and Φ Pt values of 0.019 and 0.038, respectively. Geography and phylogeography could not be invoked to explain this differentiation. A redundancy discriminant analysis revealed that relevant eco-pedological and climatic features, such as soil depth, aspect, elevation and humidity, were correlated with the observed pattern of differentiation. Toblino was ecologically separated from the other clusters, as it lies on deep soil with subhumid conditions. The differentiation of the Brione–Ranzo–Val Busa cluster appeared to be related to superficial soils and drier conditions, whereas the Nanzone–Padaro cluster was differentiated mainly according to its mid-elevation. Coupling spatial and genetic information on a local scale proved to be effective to investigate the evolutionary and demographic history of peripheral populations.