Authors: Lucie M. Bland, C. David L. Orme, Jon Bielby, Ben Collen, Emily Nicholson and Michael A. McCarthy
Source: Journal of Applied Ecology (Accepted Article 2015)
Brief summary of the paper: Cost-effective reduction of uncertainty in global biodiversity indicators is a central goal of conservation. Comprising a sixth of the 74,000+ species currently on the IUCN Red List, Data Deficient species contribute to considerable uncertainty in estimates of extinction risk. Estimating levels of risk in Data Deficient species will require large resources given the costs of surveys and Red List assessments. Predicting extinction risk from species traits and geographical information could provide a cheaper approach for determining the proportion of Data Deficient species at risk of extinction.
We use double sampling theory to compare the cost-effectiveness of predictive models and IUCN Red List assessments for estimating risk levels in Data Deficient terrestrial mammals, amphibians, reptiles and crayfish. For each group, we calibrate Machine Learning models of extinction risk on species of known conservation status, and assess their cost and reliability relative to field surveys followed by Red List assessments.
We show that regardless of model type used or species group examined, it is always more cost-effective to determine the conservation status of all species with models and assess a small proportion of species with IUCN criteria (double sampling), rather than spend the same resources on field surveys and Red List assessments alone (single sampling).
We estimate that surveying and re-assessing all Data Deficient species currently listed on the IUCN Red List (12,206 species) with IUCN criteria would cost a minimum of US $323 million. Double sampling reduces the cost of determining the proportion of Data Deficient species at risk of extinction by up to 68%, because less than 6% of Data Deficient species would need to be surveyed and assessed with IUCN criteria.
Synthesis and applications. Double sampling with models cost-effectively estimates extinction risk levels in poorly-known species, and can be used to reduce the impact of uncertainty in the Red List and Red List Index. We provide recommendations for uptake by managers and a sampling planner spreadsheet. Double sampling could be applied more widely in ecology and conservation to formally compare the cost-effectiveness of sampling methods differing in cost and reliability.