Year: 2014
Type: article
Abstract: Aim To apply mathematical models to the task of predicting extinction risk for species currently listed as ‘Data Deficient’ ( DD ) by the International Union for the Conservation of Nature ( IUCN ). We demonstrate this approach by applying it globally to amphibians, the vertebrate group recognized as being most extinction threatened and having the largest proportion of DD species. We combine model predictions with current extinction risk knowledge to highlight regions of greatest disparity between known and predicted risk, where potential species extinctions may be overlooked. Location Global. Methods Using global amphibian distribution data obtained from the IUCN and species trait data, we apply machine learning randomForest models to predict extinction risk of DD species from life history traits, environmental variables and habitat loss. These models are trained using data for species that have been assigned to an extinction risk category (other than DD) by the IUCN . We then combine predictions for DD species with IUCN assessment data in a GIS framework to highlight anomalies between current knowledge of amphibian extinction risk and our model predictions. Results We show that DD amphibian species are likely to be more threatened with extinction than their fully assessed counterparts. Regions in South America, central Africa and North Asia are particularly at risk due to lack of species knowledge and higher extinction risk than currently recognized. Main conclusions Application of predictive models ranking regions and species most in need of primary research allows prioritization of limited resources in an informed context, minimizing risk of unnoticed species' extinction.
Source: Diversity and Distributions
DOI: 10.1111/ddi.12218
Authors: Sam D. Howard, David P. Bickford
Institution: National University of Singapore
Language: English
Domain: Life Sciences
SDG: Life on land
Open Access status: bronze
Funder: National University of Singapore
Award: G5834026885
OpenAlex ID: W1977853062
