Figure 1. Yuccas and yucca moths. (a) A female moth collects pollen actively with her mouthparts, after which she flies to a different plant (this is critical as yuccas are completely or partially self-incompatible). (b) The female then oviposits into the ovary of a new flower, after which she actively places pollen on the stigma. (c) Larvae develop within a single fruit and consume some, but not all, of its seeds. The plants abort fruits into which too many eggs are laid, thus exerting some control over ‘cheaters’ that devour most or all the seeds. From Riley CV (1892) The yucca moth and yucca pollination. Annual Report of the Missouri Botanical Garden 1892: 99–158.

Figure 2. A pollination web. Sampling of flower visitors over one fortnight in upland vegetation on the island of Mauritius, in the tropical Indian Ocean, yielded this bipartite graph of interactions among pollinators (numbered along the top, with width of each colored rectangle proportional to relative abundance of that species) and plants (along the bottom). Width of the wedge connecting animals to plants indicates the relative frequency of their interaction. The pollinators are colour coded as follows: red = Hymenoptera other than ants, magenta = Gekkonidae (only one species, Phelsuma cepediana, blue-tailed day gecko), blue = Coleoptera, dark green = Aves (one species, Zosterops mauritianus, grey white-eye), light green = Diptera, orange = Formicidae, yellow = Lepidoptera. The 32 plant species in this web were visited by a mean of 3.1 pollinator species each, whereas the 28 pollinator species visited a mean of 2.6 plants each. This web is a representative of pollination webs studied to date in that it contains a range of species from specialists to generalists, with specialists tending to interact with generalists rather than with other specialists. With permission from Kaiser (2006).

Figure 3. Disruptive selection on width of the flower tube (corolla) in Ipomopsis. In rare years, hawkmoths join hummingbirds as pollinators, and selection is disruptive, favouring the narrowest and widest tubes rather than intermediates. This situation recalls a scenario for pollinator-mediated speciation without geographical isolation. There is no significant selection on corolla length.