| Summary: | Habitat destruction and modification is the greatest cause of biodiversity loss on the planet. Biodiversity loss has and will continue to contribute to the disruption of crucial ecological functions such as seed dispersal; the movement of a plant's genetic material by abiotic vectors such as wind or biotic vectors such as birds. Severe avian population declines are well documented in New Zealand, where avian dispersers are vital for seed dispersal for many plants. Mutualistic interactions between seed dispersing birds and fruiting plants can form large complex webs that, until recently, have impeded community level analysis. The application of network theory to these complex webs of interactions provides the necessary tools to visualise and describe their structural properties and predict the ecological consequences of network dynamics on species. In this thesis, I applied a network theory approach to describe frugivore-plant interactions across different habitats within Tāwharanui Regional park (TRP), New Zealand's first open sanctuary, 90 km north of Auckland City. I achieved this by conducting bird and fruit counts within habitat types throughout TRP. Bush interior points and bush edge points had significantly higher frugivore species richness than pasture interior points, while bush interior points supported a significantly higher number of large frugivores per point than mānuka edge points. Network analysis showed a highly modular network structure of the long-distance and short-distance potential networks, indicative of a network that is resilient to disturbance. Extinction models indicate that the extirpation of tūi and kererū, the two most connected species in the short-distance potential network, would lead to 42% of plant species losing their dispersers. To my knowledge, this is the first implementation of a predictive network model for plant-frugivore interactions. This research underscores the benefits of applying network theory as a tool for conservation managers to identify and set conservation priorities. For example, management should ensure local populations of kererū are preserved by maintaining and replanting remnant bush habitats which contain several fruiting species favoured by kererū. Their high mobility could drive the recolonization of fruiting plant species to regenerating areas which would assist restoration efforts and reduce management costs. Furthermore, this research demonstrates the efficacy of predictive networks through the novel use of co-occurrence data from field observations in combination with literature of plant-frugivore interactions.
|
|---|
