Data from: Feeding the enemy: loss of nectar and nectaries to herbivores reduces tepal damage and increases pollinator attraction in Iris bulleyana

  • Ya-Ru Zhu (Contributor)
  • Min Yang (Contributor)
  • Jana C. Vamosi (Contributor)
  • Scott Armbruster (Contributor)
  • Tao Wan (Contributor)
  • Yan-Bing Gong (Contributor)



Floral nectar usually functions as a pollinator reward, yet it may also attract herbivores. However, the effects of herbivore consumption of nectar or nectaries on pollination have rarely been tested. We investigated Iris bulleyana, an alpine plant that has showy tepals and abundant nectar, in the Hengduan Mountains of SW China. In this region, flowers are visited mainly by pollen-collecting pollinators and nectarivorous herbivores. We tested the hypothesis that, in I. bulleyana, sacrificing nectar and nectaries to herbivores protects tepals and thus enhances pollinator attraction. We compared rates of pollination and herbivory on different floral tissues in plants with flowers protected from nectar and nectary consumption with rates in unprotected control plants. We found that nectar and nectaries suffered more herbivore damage than did tepals in natural conditions. However, the amount of tepal damage was significantly greater in the flowers with protected nectaries than in the controls; this resulted in significant differences in pollinator visitation rates. These results provide the first evidence that floral nectar and nectaries may be ‘sacrificed’ to herbivores, leading to reduced damage to other floral tissues that are more important for reproduction.,Herbivory and pollination ratesHerbivory rate of Iris bulleyana in relation to treatment (unmanipulated vs. protected) and tissue type (nectar and nectaries vs. tepals); and pollination rate of Iris bulleyana in relation to treatment. Treatment: 0 = unmanipulated, 1 = protected. Tissue type: 0= nectar and nectaries, 1= tepals.Iris.R-codeR-code for analysing pollinator-herbivore interaction in Iris bulleyana. Both analyses were conducted in the R packages "lme4", and the p-values were obtained using negative binomial regressions with a log link function, because the response variables were over-dispersed counts.,
Date made available1 Jan 2017

Cite this