Determinants of functional connectivity of holm oak woodlands: Fragment size and mouse foraging behavior

• Low acorn dispersal may override extensive pollen flow, limiting genetic cohesiveness.
• Large forest fragments are key to maintain connectivity as they are pollen sources.
• Seed crops in small forest fragments may better capture regional genetic diversity.
• Scaling up fragment-level guidelines to landscape scale may not be straightforward.
• Management practices aimed at raising connectivity should not sacrifice local recruitment.

Demographic and genetic connectivity of fragmented plant populations will depend on effective propagule flow across the landscape. We analyze functional connectivity in a holm oak (Quercus ilex) fragmented landscape by considering three important stages driving recruitment: effective pollination, acorn production and acorn dispersal. We used a network approach to (1) determine if pollen-mediated gene exchange across the landscape was spatially structured; (2) estimate the effects of limited acorn dispersal on functional connectivity; (3) identify which landscape traits could drive source–sink dynamics of gene flow.Although long distance dispersal was relatively frequent, most effective pollen flow occurred over short distances (<100 m). This resulted in a significantly modular structure of the mating network, yielding higher gene flow among nearby fragments. Limited mouse acorn hoarding activity had a strong impact on landscape connectivity, decreasing male gametic immigration rates into forest patches by one order of magnitude Besides, our results show that big forest fragments (>10 ha) are the main pollen sources, while small ones (<1 ha) are important pollen sinks. Thus, big fragments are critical to maintain functional connectivity, while small forest fragments may provide acorn crops better representing regional genetic diversity. In addition to area effects, less isolated and more central fragments showed higher migration rates and exchanged effective pollen with more fragments. Hence, we expected that landscapes with uniform or clumped distribution of big forest fragments would show optimal connectivity traits. However, despite that simulated gene flow was more evenly distributed across the landscape, connectance and migration rates decreased. Our results call for caution before translating patch-level management guidelines to the landscape scale. They also show that the level of functional connectivity may change throughout the recruitment process, suggesting that large-scale conservation strategies may fail if local effective seed establishment is disregarded.