Geographic trait variations are often caused by locally different selection regimes. As a steep environmental cline along altitude strongly influences adaptive traits, mountain ecosystems are ideal for exploring adaptive differentiation over short distances. We investigated altitudinal floral size variation of Campanula punctata var. hondoensis in 12 populations in three mountain regions of central Japan to test whether the altitudinal floral size variation was correlated with the size of the local bumblebee pollinator and to assess whether floral size was selected for by pollinator size. We found apparent geographic variations in pollinator assemblages along altitude, which consequently produced a geographic change in pollinator size. Similarly, we found altitudinal changes in floral size, which proved to be correlated with the local pollinator size, but not with altitude itself. Furthermore, pollen removal from flower styles onto bees (plant's male fitness) was strongly influenced by the size match between flower style length and pollinator mouthpart length. These results strongly suggest that C. punctata floral size is under pollinator-mediated selection and that a geographic mosaic of locally adapted C. punctata exists at fine spatial scale.

Plant–plant interspecific competition via pollinators occurs when the flowering seasons of two or more plant species overlap and the pollinator fauna is shared. Negative sexual interactions between species (reproductive interference) through improper heterospecific pollen transfer have recently been reported between native and invasive species demonstrating pollination-driven competition. We focused on two native Impatiens species (I. noli-tangere and I. textori) found in Japan and examined whether pollinator-mediated plant competition occurs between them. We demonstrate that I. noli-tangere and I. textori share the same pollination niche (i.e., flowering season, pollinator fauna, and position of pollen on the pollinator's body). In addition, heterospecific pollen grains were deposited on most stigmas of both I. noli-tangere and I. textori flowers that were situated within 2 m of flowers of the other species resulting in depressed fruit set. Further, by hand-pollination experiments, we show that when as few as 10% of the pollen grains are heterospecific, fruit set is decreased to less than half in both species. These results show that intensive pollinator-mediated competition occurs between I. noli-tangere and I. textori. This study suggests that intensive pollinator-mediated competition occurs in the wild even when interacting species are both native and not invasive.