Habitat fragmentation can impact plant populations directly through habitat loss as well as indirectly through changes in pollinator assemblages and pollination rates. To better understand the relationship between habitat fragmentation and pollinator-mediated reproductive success, we contrasted the pollination networks of a common dioecious species, Broussaisia arguta (Hydrangeaceae) between continuous and fragmented forests in Hawai‘i, where 14.8% of native plant species are dioecious and pollination is understudied. Within the system naturally fragmented forests on Hawai‘i Island, we studied adults of B. arguta from 8 fragments ranging from 0.42-15 ha and four transects within a nearby 1000-km2continuous forest. We estimated pollinator community assemblages and visitation rates through field observations, reproductive success of female plants through pollen tube and seed germination studies, and pollen limitation through a hand-pollination study.
Results/Conclusions
In 2010, we conducted ~ 67 daytime hours of observations on 51 plants and found different pollinator assemblages in the continuous and forest fragments. Pollinator assemblages were diverse with both bird and insect pollinators. In the forest, 70% of pollinator visits were made by honeybees while only one honeybee visited a plant in the forest fragments where moths accounted for 51% of the visitors. Pollinator visitation rates were low and not different between fragments and continuous forest. In 2011, ~ 67 hours of observation were conducted on 45 plants. Similar to 2010, moths were the dominant pollinator in the fragments, accounting for 47% of visits. The honeybees were absent from the continuous forest, and overall visitation rates were lower than 2010. Seed germination from open-pollinated fruits showed that medium-sized fragments had lower germination rates, while plants in the small and large fragments had comparable rates to the continuous forest. Examining pollen tube density from open-pollinated plants proved inconclusive with 81% of flowers showing medium or high tube density with no area effect. Pollen limitation was widespread and highly variable, with a mean PLI of 0.41 in the fragments compared to 0.34 in the forest. Other metrics of fragmentation beyond forest area were considered and failed to explain the observed variation in reproductive success. We conclude that fragmentation has impacted B. arguta by altering pollinator assemblages and but not visitation rates. Further, the unexpected, lower seed germination observed for open-pollinated fruits collected from medium-sized fragments may suggest that plant-pollinator relationships do not obey linear models. The spatial and temporal variation in pollinator communities observed in this study demonstrates the need for longer-term studies.