Patrick L. Kohl, Ingolf Steffan‐Dewenter

• Abiotic factors are generally assumed to determine whether species can exist at the extreme ends of environmental gradients, for example, at high elevations, whereas the role of biotic interactions is less clear. On temperate mountains, insect‐pollinated plant species with bilaterally symmetrical flowers exhibit a parallel elevational decline in species richness and abundance with bees. This suggests that the lack of mutualistic interaction partners sets the elevational range limits of plants via a reduction in reproductive success. We used theAbiotic factors are generally assumed to determine whether species can exist at the extreme ends of environmental gradients, for example, at high elevations, whereas the role of biotic interactions is less clear. On temperate mountains, insect‐pollinated plant species with bilaterally symmetrical flowers exhibit a parallel elevational decline in species richness and abundance with bees. This suggests that the lack of mutualistic interaction partners sets the elevational range limits of plants via a reduction in reproductive success. We used the bee‐pollinated mountain plant Clinopodium alpinum (Lamiaceae), which blooms along a continuous 1000‐m elevational gradient and has bilaterally symmetrical flowers, as a model to test the predicted parallel elevational decline in flower visitation and seed production. Although the community of flower visitors changed with elevation, the flower visitation rate by the most frequent visitors, bumble bees (33.8% of legitimate visits), and the overall rate of flower visitation by potential pollinators did not vary significantly with elevation. However, we discovered that nectar robbing by bumble bees and nectar theft by ants, two interactions with potentially negative effects on flowers, sharply increased with elevation. Seed set depended on pollinators across elevations and followed a weak hump‐shaped pattern, peaking at mid‐elevations and decreasing by about 20% toward both elevational range edges. Considering the mid‐ and high elevations, elevational variation in seed production could not be explained by legitimate bee visitation rates but was inversely correlated with the frequency of nectar robbing. Our observations challenge the hypothesis that a decrease in the availability of pollinators limits seed production of bee‐flowered plants at high elevations but suggest that an increase in negative interactions (nectar robbing and larceny) constrains reproductive success.…