A week after it’s initial publication, I am finally getting round to posting a link to my new article on the cost-effective assessment of extinction risk with limited information. I wrote this paper during my first visit to CEED in 2013, working with Emily Nicholson and Michael McCarthy.
The setting: one in six species on the IUCN Red List are assessed as Data Deficient, seriously limiting our capacity to monitor biodiversity and track our progress towards international targets. Given the costs of field surveys and Red List assessments, re-assessing all 12,000+ species would cost USD320 million.
So how can we make the most of limited conservation dollars when assessing the status of Data Deficient species? The answer is double sampling, a simple technique borrowed from a statistics paper from the 1970s by Tenenbein, and described more intuitively in Mick McCarthy’s blog.
To estimate the proportion of Data Deficient species at risk of extinction, we can use:
– field surveys and IUCN Red List assessments, which are a perfect ‘gold standard’ but expensive
– models of extinction risk, which are cheaper but can make mistakes when predicting the status of species
I built predictive models of extinction risk for mammals, reptiles, amphibians and crayfish. I correlated species information (life-history traits, environmental niche and threat exposure) with true IUCN Red List status for well-known species. These simple data are often available for Data Deficient species, and can be collected from museum specimens, museum records and freely-available GIS layers. Good news, the models based on simple data were able to predict true Red List status very well, and for a fraction of the cost! The cost ratio between Red List assessments and my own work on predicting species extinction risk turned out to be 1,500 – a bold statement on how cheap PhD students are in the UK! This means that we could rapidly and cheaply predict the status of Data Deficient species with models, rather than wait for Red List assessments.
The paper demonstrates that by using both Red List assessments and models of risk (hence double sampling), we can estimate the levels of extinction risk in Data Deficient species for a third of the original costs of Red List assessments. This is because the status of all Data Deficient species are predicted with cheap models, whilst 1-6% of the species are assessed with Red List assessments to reduce errors in classification. Double sampling has considerable potential for estimating the proportion of Data Deficient at risk of extinction, and reducing the uncertainty in the Red List and Red List Index.
We also published an easy-to-use Excel spreadsheet to calculate the possible reductions in cost with double sampling (Tip: it works for any situation comparing two sampling methods differing in cost and reliability).