Print PageEcological networks allow invasive plant species to interact indirectly with native plant species. Insect-pollinated invasive plant species become integrated into native pollination networks, but it is unclear whether they alter network topology systematically. An understanding of the roles of invasive species in pollinator networks can help to inform management activity surrounding both rare and invasive species.
We characterized pollination networks at 8 sites within Badlands National Park (South Dakota, USA) that included the rare species, Eriogonum visheri (4 sites) or Chrysothamnus parryi (4 sites). Eriogonum sites included the invasive Melilotus officinalis. We quantified floral abundance by species and sampled insect visitors in contact with anthers or stigmas on 1-ha plots while Eriogonum and Chrysothamnus were in flower (approximately one month each). Observed interaction matrices were compared to null models that constrained both marginal totals of interactions and network connectance (proportion of nonzero cells in the matrix).
Results/Conclusions
Eriogonum plots did not differ from null-model predictions with respect to diversity or evenness of plant-insect interactions, nor nestedness of the interaction matrix. Interaction diversity and evenness were lower than expected on Chrysothamnus plots, and the interaction matrix was significantly less nested than the null model prediction. Analysis of a select subset of individual species from our plots revealed that Chrysothamnus parryi on Chrysothamnus plots interacted with a more diverse suite of insects than expected by chance, and Eriogonum pauciflorum on the same plots interacted with fewer insect species but had a higher interaction diversity index than expected by chance; these two results together suggest even usage of E. pauciflorum by those insects that visited it. The network roles of all of the insects we analyzed were not significantly different from null-model predictions.
Our results indicate little specialization within the pollination networks on Eriogonum plots where interaction patterns could be predicted by abundances of interactions. Deviations from null-model predictions for the Chrysothamnus plots suggest some specialization there, possibly because Chrysothamnus is a late-season flower with few co-flowering species (4 species, compared to 18 co-flowering species on Eriogonum plots). Management activities that affect relative abundances of plant floral resources may have little impact on pollination networks at Badlands National Park except when floral resources are particularly scarce, such as early and late during the season, or during drought years.
