The interplay between scale, pollination niche and floral attractiveness on density-dependent plant-pollinator interactions.

Pollinators mediate interspecific and intraspecific plant-plant indirect interactions (competition vs. facilitation) via density-dependent processes, potentially shaping the dynamics of plant communities. However, it is still unclear which ecological drivers regulate density-dependent patterns, including scale, pollination niches (i.e., the main pollinator functional group) and floral attractiveness to pollinators. In this study, we conducted three-year field observations in Hengduan Mountains of southwest China. By gathering data for more than 100 animal-pollinated plant species, we quantified the effect (positive vs. negative) of conspecific and heterospecific flower density on pollination at two scales: plot-level (4 m 2 ) and site-level (100-5000 m 2 ). Then, we investigated how pollination niches and floral attractiveness to pollinators (estimated here as average per-flower visitation rates) modulated density-dependent pollination interactions. Pollinator visitation depended on conspecific and heterospecific flower density, with rare plants subjected to interspecific competition at the plot-level and interspecific facilitation at the site-level. Such interspecific competition at the plot-level was stronger for plants pollinated by diverse insects, while interspecific facilitation at the site-level was stronger for bee-pollinated plants. Moreover, we also found stronger positive conspecific density-dependence for plants with lower floral attractiveness at the site-level, meaning that they become more frequently visited when abundant. Our study indicates that the role of pollination in maintaining rare plants and plant diversity depends on the balance of density-dependent processes in species-rich communities. We show here that such balance is modulated by scale, pollination niches and floral attractiveness to pollinators, indicating the context-dependency of diversity maintenance mechanisms.