Baobab fruiting is driven by scale-dependent mediation of plant size and landscape features

Context Both plant size and distribution of plants and resources across landscapes are known to influence pollinator behavior and resulting plant reproductive success. However, the relative influence of these intrinsic and extrinsic factors is unknown. Objectives We evaluated the relative contribution of individual plant size and landscape variables to reproductive success in bat-pollinated baobabs ( Adansonia digitata) and determined if the interaction is scale-dependent. Methods We recorded fruit number per baobab of 741 baobab in south-central Kenya and measured size metrics of individuals. We georeferenced baobabs and relevant resources across 10 km 2 to generate landscape variables. Conditional inference forests determined scale-specific responses over 20 buffer distances (50-m to 1000-m) around baobabs and identified relative variable importance. We modeled presence of fruit, as not all trees produce fruit. For fruiting baobabs, we modeled whether there were few or many fruits. Results Conditional inference forests were significant at 50-m to 600-m buffer distances. Individual characteristics of baobabs were the primary drivers of fruit presence, with larger trees more likely to fruit. Fruit presence was modified by baobab height and landscape variables. Land use primarily drove baobab fruit production category, which was modified by baobab size and other landscape variables. The importance of distance to and density of alternate food resources changed with scale. Conclusions Individual characteristics and landscape variables both influence reproductive success in the bat-pollinated baobab system, and relative variable importance was scale-dependent. The pollinator landscape is complex and scale-dependent, encompassing not only the distribution of the baobab population but also attractants (pawpaws) and distractants (figs) that further influence reproductive success.