Abstract
Bat pollination of Dillenia in Fiji, a genus that was presumed to be pollinated by bees, posits that other Dillenia species may be bat-pollinated, with implications for conservation and the understanding of angiosperm evolution. Botanical descriptions of some corolla behaviours ('falling as a whole') suggest bat removal of permanently closed corollas, as in D. biflora. Considering the remoteness of species of interest, we reviewed some Dillenia floral traits to hypothesise what they may mean for bat pollination of the genus. We investigated D. biflora pollen grains apertures and reviewed Dillenia literature concerning corolla behaviour and colour, and pollen apertures and presentation, including pores and staminodes. Our samples had dramatically different ratios of tricolpate to tetracolpate pollen grains, a trait that does not exclude pollination by bees. Petal colour polymorphism occurs, with mixed colours proportionately less common in flowers with corollas that open. The proportion of species with staminodes did not differ between those presumed to be pollinated by bats and others, but anthers of the former were significantly more likely to have apical pores, and stamens all had similar length or were slightly longer in the middle, whereas stamens in two distinct groups occurred in 55% of bee-pollinated species. Pollen heteromorphy may facilitate pollination by different taxa in tropical environments. However, anther apical pores and stamen uniformity are more likely to be associated with bat-pollinated species than are other morphologies. Dillenia could be a useful model to examine evolutionary aspects of colour, heteranthery, staminodes and pollen heteromorphy. Only field work will verify bat pollination and the implications of bat dependence for Dillenia species.