Abstract
Species translocations, including reintroductions, are an increasingly important tool for the management of many threatened species. However, globally the success rate of reintroductions has been low. This is due in large part to gaps between ecological theory and conservation practice inhibiting effective planning and execution of reintroductions. Better integration between ecological theory and practice can be achieved through the application of clear strategic frameworks. Predictive ecological models are also increasingly being used to inform conservation practice, and these models have potential to bridge gaps between theory and practice in translocation management.
South Island takahē (Porphyrio hochstetteri), like many species subject to reintroduction efforts, persist in relict and small, restored, populations in often marginally suitable environments. Generating useful models and predictions of habitat suitability in reintroduction landscapes based on data collected within a relict range is therefore difficult, requiring careful treatment and interpretation. I illustrate this for the takahē by modelling habitat suitability in a reintroduction landscape based on location records from a relict distribution. Through careful interpretation of relict-population data potentially useful predictions are generated for informing management of the translocated population, including supporting the view that there is suitable habitat around release sites in the Gouland Downs. However, these predictions are sensitive to model settings relating to the interpretation of input data, as there was an 84% greater area of suitable habitat predicted by models that were fitted to allow more extrapolation, when compared with conservatively fitted models.
Post-release monitoring is crucial for the adaptive management of translocated populations. Satellite and radio-tag tracking methods were used to monitor the reintroduced takahē population. I first use data collected from this monitoring to provide an early evaluation of reintroduction success, finding that following initial breeding success and high survival, high rates of post-release dispersal appeared to be associated with mortality events two years after initial releases. Following a preliminary evaluation of reintroduction success, I develop models of space use based on the radio tracking dataset, and models of resource selection based on the satellite tracking dataset. Integration of these two models in a hybrid model framework enabled me to combine insights from these individually limited, but complementary, datasets, producing a high resolution estimated distribution for the reintroduced population.
Model validation is a crucial step in the development of ecological models, as this can highlight the strengths and limitations of predictive models and ensure that they are applied to management with a suitable weight and confidence. I use independently observed and modelled post-release data to evaluate the habitat predictions based on observations fromthe takahē’s relict distribution. These evaluations suggest that predictive models are reasonably effective, with the best performing models of post-release movements, resource selection, and population distribution having an estimated accuracy of 74.6%, 59.4%, and 72.2%. Routes for further refinement of the predictive models as well as estimating their overall predictive accuracy are also indicated by these evaluations.
I expect that the developments made in this thesis will be of value, providing a strong evidence-base that enables a rigorous application of ecological data and theory to drive improvements in conservation practice. These benefits should apply both to takahē and to other species management programmes to which the developed ideas and methods can be readily applied.