Inferential and forward projection modeling to evaluate options for controlling invasive mammals on islands.

Successful pest-mammal eradications from remote islands have resulted in important biodiversity benefits. Near-shore islands can also serve as refuges for native biota but require ongoing effort to maintain low-pest or pest-free status. Three management options are available in the presence of reinvasion risk: (1) control-to-zero density, in which immigration may occur but reinvaders are removed; (2) sustained population suppression (to relatively low numbers); or (3) no action. Biodiversity benefits can result from options one and two. The management challenge is to make evidence-based decisions on the selection of an appropriate objective and to identify a financially feasible control strategy that has a high probability of success. This requires understanding the pest species population dynamics and how it will respond to a range of potential management strategies, each with an associated financial cost. We developed a two-stage modeling approach that consisted of (1) Bayesian inferential modeling to estimate parameters for a model of pest population dynamics and control, and (2) a forward projection model to simulate a range of plausible management scenarios and quantify the probability of obtaining zero density within four years. We applied the model to an ongoing, six-year trapping program to control stoats (Mustela erminea) on Resolution Island, New Zealand. Zero density has not yet been achieved. Results demonstrate that management objectives were impeded by a combination of a highly fecund population, insufficient trap attractiveness, and a substantial proportion of the population that did not enter traps. Immigration is known to occur because the founding population arrived on the island by swimming from the mainland. However, immigration rate during this study was indistinguishable from zero. The forward projection modeling showed that control-to-zero density was feasible but required greater than a two-fold budget increase to intensify the trapping rate relative to population growth. The two-stage modeling provides the foundation for a management program in which broad-scale trials of additional trapping effort or improved trap lures would test model predictions and increase our understanding of system dynamics.

Using genetic techniques to quantify reinvasion, survival and in situ breeding rates during control operations.

Determining the origin of individuals caught during a control/eradication programme enables conservation managers to assess the reinvasion rates of their target species and evaluate the level of success of their control methods. We examine how genetic techniques can focus management by distinguishing between hypotheses of 'reinvasion' and 'survivor', and defining kin groups for invasive stoats (Mustela erminea) on Secretary Island, New Zealand. 205 stoats caught on the island were genotyped at 16 microsatellite loci, along with 40 stoats from the opposing mainland coast, and the age and sex were determined for each individual. Using these data, we compare and combine a variety of genetic techniques including genetic clustering, population assignment and kinship-based techniques to assess the origin of each stoat. The population history and individual movement could be described in fine detail, with results indicating that both in-situ survival and breeding, and reinvasion are occurring. Immigration to the island was found to be generally low, apart from in 1 year where around 8 stoats emigrated from the mainland. This increased immigration was probably linked to a stoat population spike on the mainland in that year, caused by a masting event of southern beech forest (Nothofagus sp.) and the subsequent rodent irruption. Our study provides an example of some of the ways genetic analyses can feed directly into informing management practices for invasive species.