Primary pollinator exclusion has divergent consequences for pollen dispersal and mating in different populations of a bird-pollinated tree.

Pollination by nectarivorous birds is predicted to result in different patterns of pollen dispersal and plant mating compared to pollination by insects. We tested the prediction that paternal genetic diversity, outcrossing rate and realized pollen dispersal will be reduced when the primary pollinator group is excluded from bird-pollinated plants. Pollinator exclusion experiments in conjunction with paternity analysis of progeny were applied to Eucalyptus caesia Benth. (Myrtaceae), a predominantly honeyeater-pollinated tree that is visited by native insects and the introduced Apis mellifera (Apidae). Microsatellite genotyping at 14 loci of all adult E. caesia at two populations (n = 580 and 315), followed by paternity analysis of 705 progeny, revealed contrasting results between populations. Honeyeater exclusion did not significantly impact pollen dispersal or plant mating at Mount Caroline. In contrast, at the Chiddarcooping site, the exclusion of honeyeaters led to lower outcrossing rates, a threefold reduction in the average number of sires per fruit, a decrease in intermediate-distance mating and an increase in near-neighbour mating. The results from Chiddarcooping suggest that bird pollination may increase paternal genetic diversity, potentially leading to higher fitness of progeny and favouring the evolution of this strategy. However, further experimentation involving additional trees and study sites is required to test this hypothesis. Alternatively, insects may be effective pollinators in some populations of bird-adapted plants, but ineffective in others.

Conservation of old individual trees and small populations is integral to maintain species' genetic diversity of a historically fragmented woody perennial.

Historically fragmented and specialized habitats such as granite outcrops are understudied globally unique hot spots of plant evolution. In contrast to predictions based on mainstream population genetic theory, some granite outcrop plants appear to have persisted as very small populations despite prolonged geographic and genetic isolation. Eucalyptus caesia Benth. is a long-lived lignotuberous tree endemic with a naturally fragmented distribution on granite outcrops in south-western Australia. To quantify population to landscape-level genetic structure, we employed microsatellite genotyping at 14 loci of all plants in 18 stands of E. caesia. Sampled stands were characterized by low levels of genetic diversity, small absolute population sizes, localized clonality and strong fine-scale genetic subdivision. There was no significant relationship between population size and levels of heterozygosity. At the landscape scale, high levels of population genetic differentiation were most pronounced among representatives of the two subspecies in E. caesia as originally circumscribed. Past genetic interconnection was evident between some geographic neighbours separated by up to 20 km. Paradoxically, other pairs of neighbouring stands as little as 7 km apart were genetically distinct. There was no consistent pattern of isolation by distance across the 280 km range of E. caesia. Low levels of gene flow, together with strong drift within stands, provide some explanation of the patterns of genetic differentiation we observed. Individual genet longevity via the ability to repeatedly resprout and expand from a lignotuber may enhance the persistence of some woody perennial endemic plants despite small population size, minimal genetic interconnection and low heterozygosity.