No abstract

Scofield et al. discredited the utility of pest‐exclusion fences for restoring biodiversity partly on the grounds of unquantified costs and benefits. We estimated the discounted costs of mammal exclusion fences, semi‐permeable (‘leaky’) fences and trapping, over 50 years and adjusted costs by their observed effectiveness at reducing mammalian predator abundance. We modelled data from two large predator management programmes operated by the New Zealand Department of Conservation. Using typical baseline costs and predator control efficacies (scale 0 to 1), the model predicted that an exclusion fence (efficacy 1.0) is the cheapest and most cost‐effective option for areas below about 1 ha, a leaky fence (efficacy 0.9) is most cost‐effective for 1–219 ha, and trapping (efficacy 0.6, based on 0.2 traps per hectare and a 1500‐m buffer to reduce predator reinvasion) for areas above 219 ha. This ranking was insensitive to adjustments in efficacy, but reducing efficacy of leaky fences to 0.8 or increasing trapping efficacy to 0.7 reduced the cost‐effective range of leaky fences by about 90 ha. Reducing trap maintenance costs from $300 to $100 per trap per year (e.g. using long‐life lures), or reducing trap buffer widths to 500 m, significantly elevated trapping as the most cost‐effective method for areas greater than 11–15 ha. These results were largely consistent with an ecological measure of effectiveness based on observed rates of recovery of two indigenous skink species inside exclusion fences or with trapping. The results support criticisms that exclusion fences are generally not cost‐effective, but highlight the value of considering cheaper leaky designs for small‐ to medium‐sized areas. Because this study is based largely on reductions in predator abundance, it has general application to broader biodiversity protection interests, but not to indigenous species that are highly sensitive to predation and only ever adequately protected on the mainland by exclusion fences.

The characterization of individual animal life history is crucial for conservation efforts. In this paper, Sloop, an operational pattern retrieval engine for animal identification, is extended by coupling crowdsourcing with image retrieval. The coupled system delivers scalable performance by using aggregated computational inference to effectively deliver precision and by using human feedback to efficiently improve recall. To the best of our knowledge, this is the first coupled humanmachine animal biometrics system, and results on multiple species indicate that it is a promising approach for large-scale use.