New Zealand Journal of Ecology (2015) 39(2): 286- 290

Using home-range data to optimise the control of invasive animals

Research Article
Des H. V. Smith 1*
Richard Clayton 2
Dean Anderson 2
Bruce Warburton 2
  1. Wildland Consultants, Level 1, Unit B, 238 Barrington Street, Christchurch 8244, New Zealand
  2. Landcare Research, PO Box 69040, Lincoln 7640, New Zealand
*  Corresponding author

Invasive species have been identified by the Convention on Biological Diversity as a significant threat to biodiversity. Conservation managers often lack tools for addressing uncertainty about the control intensity required to achieve cost-effective management of invasive species. We describe a modelling approach for informing the spacing of control-device lines given the availability of home-range data. To demonstrate its utility, we used data on stoats (Mustela erminea), an introduced mammalian predator responsible for the decline of endemic birds in New Zealand. We calculated home-range widths using three methods: kernels, circles and the narrowest distance across the raw point data. Using the widths from each method, we then permuted iteratively the relative location and orientation of home ranges between control-device lines, and calculated the probability of encounter with varying distances between lines. Widths across raw points gave lower estimates of the probability of encounter of device lines than kernels, while circles gave estimates that were intermediate between the two. For stoats, the simulation on point-data widths indicates that to ensure control-device lines will intersect 100% of female stoat home ranges they need to be ≤ 400 m apart, while the simulation on kernels and circles allowed ≤ 700 m. When needing to address uncertainty about the intensity of control to apply, managers should give priority to the collection of home-range data so that control-line spacing can be determined using the simulation described. If sufficient home-range data are available then simulating kernels provides better predictions, otherwise simulating the width across point data provides a conservative option when such data are insufficient.