There is growing evidence based on field studies of extensive seed dispersal movement among distant populations for plant species dispersed by vertebrates. Population genetic theory predicts an isolation by distance pattern where genetic variance increases with geographic distance among populations. However, we virtually ignore to what extent landscape complexities (altitudinal differences, river flows, or physical barriers) disrupt the isolation by distance patterns imposing a more complex distribution of genetic variance over the landscape. In this study we aim at gauging the role of seed dispersal in shaping the spatial genetic diversity across the landscape for a plant species dispersed by vertebrates, Prunus mahaleb. We combine a detail survey of the genetic composition of 8 focal populations based on microsatellite markers with a network analytical approach where topological traits of a population graph inform us about the processes governing network assemblage. We expect that network topology capture observed seed dispersal patterns among populations. Specifically, we assessed population centrality, population degree, clustering, modularity, and network diameter. Furthermore, we obtain different population graphs based on distinctive dispersal scenarios where the extent and amount of seed dispersal varies among populations. 
Results/Conclusions

We found that isolation by distance pattern explains genetic variation at landscape level, but at local scale population size and phenological overlapping become more relevant. Furthermore, congruence graphs show significant edge compression suggesting a non-linear dispersal processes. Compression edges (i.e., shorter geographical distance than topological distance) are strongly determined by asymmetrical seed dispersal patterns between population pairs as a result of selective foraging movement by frugivorous birds. Finally, a restricted seed dispersal movement (where most of propagules tend to disperse to nearby populations) couple with asymmetrical dispersal patterns result in a population graph characterized by a significant modularity, where a set of populations are more connected among them than expected by random and whereas they are scarcely connected to other populations belonging to different modules. Technically, our results highlight the need of combining robust estimates of contemporary seed dispersal movements with spatial-explicit analytical techniques disentangling the role of landscape features shaping the genetic composition in plant populations. Conceptually, these results document the strong influence of other factors, besides distance, in shaping the spatial distribution of the genetic variation among plant populations.