Through space and time: novel methods to prioritise management actions for invasive species

Abstract

Systematic eradication techniques were developed during the twentieth century to stem the negative impacts of invasive mammals and to facilitate restoration of island communities. Eradication projects have increased demonstrably in size, scope, and complexity over the last forty years and are now a principal conservation intervention on islands globally. Conservation practitioners are now setting their sights on projects with unprecedented levels of biogeographical and social complexity due to their potential for substantial conservation gains. However, the successful outcome of these projects is not guaranteed due to the high risk of reinvasion and often complex social structure. In this thesis, I investigate novel analytical techniques to prioritise the implementation of management actions for invasive species so that conservation practitioners can maximise the probability of restoring threatened island communities. I use New Zealand’s Predator Free 2050 programme as the context for this research. Following a brief introduction (Chapter 1), I solve a strategic issue associated with where invasive species eradications should be conducted given limited time resources. I first (Chapter 2) quantify the underlying structure of insular isolation measures to understand how they mechanistically drive island biogeographical patterns. The measures synthesised in this chapter function as proxies for reinvasion, which represents one of the most-limiting factors behind eradication success. I then (Chapter 3) take these findings, along with metrics describing social complexity, to temporally prioritise eradications in New Zealand using statistical tools developed for survival analysis. These results can be used to determine whether a conservation priority should be selected for intervention immediately or in the future. I then (Chapter 4) solve a tactical issue associated with determining appropriate management actions for the New Zealand mainland given a suite of potentially suitable options. This prioritising model employs a machine learning technique to predict where, within the landscape, different management methods have the highest probability of being implemented. These results can be used to create a contiguous management network across large and heterogeneous areas. The next-step in invasive species management requires that conservation action takes place in large and complex areas. This research gives practitioners tools to help them determine the feasibility of proposed management plans at unprecedented spatiotemporal scales.