Invasive species threaten the biodiversity of native communities, but most invasive species research has focused on direct, competitive interactions between invasive and native plants. An indirect and less-studied consequence of plant invasions is the disruption of native plant-pollinator interactions. We examined the effects of removal of the invader baby’s breath (Gypsophila paniculata) on pollinator visits to the threatened endemic Pitcher’s thistle (Cirsium pitcheri) on Lake Michigan sand dunes. We applied two experimental treatments to G. paniculata: elimination of flowers only (clipped) and removal of entire plants (removal). We compared visitation rates by putative pollinators to C. pitcheri among invaded, clipped, and removal plots, and also between removal and naturally uninvaded plots. In a second experiment, we observed pollinator visitation rates to C. pitcheri surrounded by temporary experimental arrays of cut native (Monarda punctata) or non-native (G. paniculata or Centaurea maculosa) flowers at low, medium, or high floral densities.

Results/Conclusions

Compared to invaded areas, the removal treatment had a positive net effect on total pollinator visits and on visits by the most common pollinator, Dianthidium simile (Megachilidae, Hymenoptera). Visitation rates were not significantly different between removal and naturally uninvaded plots, indicating that removal successfully restored visitation rates to natural levels. In addition, G. paniculata removal significantly affected the local plant community. When plot-level plant richness was added as a covariate to pollinator visitation analysis, treatment effects became non-significant, suggesting that invader-mediated changes in plant species richness may underlie the net effects of treatment on pollinator responses. In experimental arrays, the absolute pollinator abundance to C. pitcheri was not affected by density or species identity. However, the proportion of whole-system visits that were to C. pitcheri was higher for M. punctata (native) compared to C. maculosa (invasive) arrays, and the number of interspecific movements was highest for C. maculosa arrays. Although fitness measures were not possible for the threatened thistle, we found evidence that both G. paniculatai and C. maculosa have the potential to negatively impact the fitness of C. pitcheri via reduced pollinator visits and interspecific pollen transfer, respectively.