Pollinator-plant interactions in Phormium spp: the effects of an introduced pollinator

Abstract

Phormium is one of the few specialised bird-syndrome plant genera recognised in the New Zealand flora. This genus has historically been visited and pollinated predominantly by native nectar-feeding birds, in particular tui (Prosthemadera novaeseelandiae); however, introduced bees, the honey bee (Apis mellifera) and bumble bee (Bombus spp.), have been observed foraging at Phormium for pollen and nectar. I investigated the impact of introduced bees on the pollination ecology and evolution in Phormium. Changes to the pollinating fauna may exert unique selection pressures on floral morphology of New Zealand flowers, and many of the conditions for pollinator shifts (e.g., co-occurrence with another class of pollinator, and pollination by the 'wrong' pollinator) appear to occur in the Phormium plant-pollinator system. For a shift to occur introduced bees need to be abundant and effective pollinators and select for different morphological traits. While I found introduced bees were abundant visitors they were not effective pollinators in terms of mean number of pollen grains deposited on the stigma of flowers in a single visit. However, pollinator-mediated selection on floral traits did differ between tui and bees suggesting that there is potential for morphological changes and pollinator shifts in populations that receive little bird visitation. Moreover, on average flowers visited by tuis were longer, thinner, had less exserted stigmas, and greater anther-stigma separation than flowers visited by bees, and flowers with greater stigma exsertion received more pollen from birds, while bees deposited more pollen on flowers with less anther-stigma separation (between sites). On average, tuis and bees also displayed significantly different behaviours-compared to tuis, bees visited fewer flowers per plant and fewer plants within a population. To study the long-term implications of introduced bee visitation on the population ecology of Phormium I compared two island sites (bees absent) to two mainland sites (bees present). Outcrossing rates and inbreeding depression were high at all sites, and while estimates of inbreeding depression using allozyme markers did not differ between island sites and mainland sites, seed germination was substantially lower at both island sites in comparison to the mainland, potentially due to a greater relatedness between island plants leading to higher levels of early-acting inbreeding depression. Fruit set was positively influenced by the level of bird visitation, and bird visitation was high at island sites, however there were no consistent differences between fruit set or seed set in island and mainland populations. These results suggest that patterns of reproductive success on island versus mainland populations do not reflect differences in pollinator assemblage. Instead, individual site effects may be a more important determinant of population processes.