High-elevation range limit of an annual herb is neither caused nor reinforced by declining pollinator service

1. Pollination failure has been proposed to be an important determinant of plant species' range limits if pollinator activity declines along an environmental gradient, directly limiting plant populations, or if plant populations decline along an environmental gradient and subsequently fail to attract sufficient visitation. Both mechanisms predict reduced pollinator visitation, increased pollen limitation and decreased seed production towards range limits, and the first additionally predicts declining pollinator abundance independent of any particular plant species. However, many self-compatible species have some capacity for autonomous self-fertilization, which may buffer reproductive success from declining pollinator visitation if inbreeding depression is mild. Thus pollinator-mediated limits may also predict selection for reduced reliance on pollinators towards range limits. 2. We tested these predictions towards the high-elevation limit of the self-compatible, bumblebee (Bombus)-pollinated Rhinanthus minor, along two elevation transects in the Rocky Mountains of Alberta, Canada. 3. Bombus abundance was highest at mid- (range-centre) and high-elevation (range limit) sites, so declining pollinator abundance is unlikely to impose high-elevation limits for bumblebee-pollinated species in this area. 4. Flowers per plant and per m2 declined at upper range limits, potentially rendering edge populations less attractive. However, visitation rate did not decline towards the range limit at either transect. Stigmatic pollen receipt declined with increasing elevation, but seed set did not, nor did outcross pollen supplementation increase seed set at any site. 5. Investment in floral attractiveness (corolla area/ovary area) increased towards range limits, but capacity for high-quality autonomous seed set and adult inbreeding coefficients inferred from genetic markers were uniformly high, suggesting frequent self-fertilization and weak inbreeding depression throughout the range. 6. Synthesis. We found no evidence for pollination failure towards the upper range limit of R. minor. Moreover, unlike some species with a capacity for autogamy, autonomous selfing makes a major contribution to R. minor's mating system and demography, and likely buffers reproductive success from stochasticity in pollination. Continued investment in floral attractiveness despite high autonomous selfing suggests some evolutionary benefit to pollinator-mediated outcrossing, rather than ecological