Context . Effective design of conservation management programs for long-term population control requires an accurate definition of the spatial extent of populations, along with a proper understanding of the ways that landscape patchiness influences demography and dispersal within these populations.

Aims . In the present study, genetic techniques are used to describe the population genetic structure and connectivity of invasive stoats (Mustela erminea) across the Auckland region, New Zealand, so as to assist planning for mainland stoat control, and define potential future eradication units.

Methods . A sample of stoats from across the region (n = 120), was genotyped at 17 microsatellite loci, and a combination of clustering, genetic population assignment and various migration estimation methods were applied to these data.

Key results . Moderate population structure was observed (F_ST = 0.03–0.21), with five geographic populations defined by genetic clustering. Almost all individuals were correctly assigned to the location of origin, and recent migration rates among forest patches were found to be low.

Conclusions . It is possible to define the origin of stoats at this regional scale using genetic measures. From this, we show that the stoat incursion on Rangitoto Island that occurred post-eradication in 2010 probably came from East Auckland (P < 0.0001), whereas the 2014 stoat incursion on Motutapu Island probably originated from a population linked to the Waitakeres. Also, the Waiheke Island stoat population has minimal connection to all other populations and it is therefore a potential eradication unit.

Implications . The low migration rates among forest patches indicated that if thorough control is imposed on a discrete forest patch, reinvasion from other forest patches will be relatively low. Importantly, for stoat control in the region, the isolation of the Waiheke Island stoat population means that eradication here is likely to be feasible with low reinvasion pressure.