The South West Australian Floristic Region (SWAFR) is globally exceptional in terms of the number of plant species that appear to be predominantly pollinated by nectar-feeding birds. Many bird-pollinated plants in the SWAFR are also rare endemics that occur in anciently fragmented populations. Plants with anciently fragmented population structures may show distinct molecular ecological characteristics compared to recently fragmented populations, with important implications for conservation genetic management. Eucalyptus caesia is a long-lived, bird-pollinated mallee or tree with a natural distribution on scattered granite outcrops in south-western Australia. For my PhD research, I focused on two key questions. Firstly, what are some of the genetic and conservation implications of small population size in historically fragmented populations? To answer this question, spatial genetic structure at the population through to landscape level was assessed by comprehensive mapping and genotyping of individual trees at 18 locations across the 280 km distribution of E. caesia. Secondly, what evolutionary adaptations might explain the ability of E. caesia to persist on its granite rock habitat? Here, pollinator exclusion experiments were applied to test the hypothesis that paternal genetic diversity, outcrossing rate and mean pollen dispersal distance will be reduced when the primary pollinator group is excluded from the inflorescences of predominantly bird-pollinated plants. In addition, the growth and survivorship of seedlings resulting from self-pollination and outcrossing was measured over consecutive surveys to test whether purging of genetic load may reduce the negative impacts of early inbreeding depression in E. caesia.