Mutualisms benefit interspecific partners, yet realized benefits vary depending on a variety of contextual factors. Partner abundance affects the costs and benefits in specialized mutualisms, but its role in facultative partnerships is less clear. We address this question by tracing impacts of varying partner abundance in a small alpine pollination web. The web comprises two clovers (Trifolium dasyphyllum and T. parryi) that vary in specialization on a shared bumblebee pollinator, Bombus kirbyellus. We tested the hypothesis that increasing partner density increases plant benefits by adding 0, 1, 2, or 3 queen B. kirbyellus to mixed clover patches and measuring visitation rates, outcross pollination, and seed set. Adjacent bumblebee exclusion controls assessed the contribution of co-pollinators to seed production, while hand pollination controls assessed seed limitation by resources. We conducted experiments throughout the summer to account for differences in generalization between species as co-pollinator availability varied over time. We traced effects from the addition experiment to recruitment by measuring field germination rates of seed progeny the following year. To assess potential impacts on clover population growth we developed a matrix model for each species and calculated relative contribution of life history stages to annual population growth rate (λ).

Results/Conclusions

We found that increased bumblebee density enhanced visitation to both clovers, but this only translated into greater pollination for the B. kirbyellus specialist, T. parryi. Increased outcross pollen transfer early in the summer enhanced seed set for both clovers, despite resource limitation of fecundity in some patches. At this time, seed production was pollination limited in both species. Later in the season when co-pollinating insects were more abundant, there were no benefits derived from increased bumblebee density. We observed a seed ‘quantity’ effect from adding bees and an unexpected seed ‘quality’ effect on germination rates; T. parryi seeds from B. kirbyellus pollination germinated more often than T. parryi seeds from non-Bombus pollinators or T. dasyphyllum seeds from the same treatments. Disparity in seed quality may have arisen from ‘relatedness’ differences between pollen donors and receivers for the two species. The population growth model showed that, although adult plant stasis comprises the greatest proportion of λ, offspring production is also important. Overall, we found that variation in mutualist abundance has stronger impacts on benefits to specialized partners. At the population level, outcomes of mutualism also depend on the relative importance of life stages impacted by the relationship.