Greater pollen-mediated gene flow among than within populations of a bumblebee-pollinated forest herb despite habitat fragmentation

BACKGROUND: Pollen flow within and among spatially isolated forest herb populations in agricultural landscapes depends on the movement activity of their pollinators. For central-place and traplining foragers, such as bumblebees, this movement activity is expected to reflect how they exploit floral resources across the landscape. However, it remains unclear how the bumblebees' foraging behavior actually shapes the mating patterns of forest herbs within and among forest patches. METHODS: Here, we conducted a landscape genetic analysis of two generations of the forest herb P. multiflorum using microsatellites to investigate (A) the proportions of pollen-mediated gene flow within and among forest patches, (B) how the spatial position and size of shoot clusters influence the pollen-mediated gene flow within forest patches, and (C) how landscape composition affects pollen immigration rates into forest patches. RESULTS: We found that pollen immigration rates per pollen receptor plant were high, with only a few shoot clusters within a forest patch serving as pollen donors. Larger shoot clusters had higher probabilities of donating pollen. Both pollen immigration rates per pollen receptor and the genetic diversity of the offspring were significantly affected by maize cover and linear landscape elements in the surrounding landscape. We found that in landscapes dominated by land-use types with low floral-resource availability, higher pollen immigration rates into wild plant populations do not necessarily lead to higher genetic diversity. CONCLUSIONS: Our results suggest that trapline foraging promotes greater pollen flow among forest patches than within them, and that shoot clusters within a forest patch contribute unequally to sexual reproduction. In the studied agricultural landscape, semi-natural habitat patches with high floral-resource availability may either compete for or facilitate pollinator activity and the resulting gene flow among forest patches, while smaller patches act rather as stepping stone habitats, larger ones may retain pollinators and distract them from less attractive forest herb populations. Similarly, linear landscape elements oriented towards pollen receptors increased both the amount and allelic richness of incoming pollen, whereas those oriented orthogonally had the opposite effect. The study also emphasized that self-incompatible bumblebee-pollinated plant species in resource-poor landscapes may form metapopulation with regular pollen exchange which does not automatically lead to high levels of introduced new alleles.