Assessing extinction risk in the absence of species-level data: quantitative criteria for terrestrial ecosystems

The conservation of individual plant and animal species has been advanced greatly by the World Conservation Union’s (IUCN) development of objective, repeatable, and transparent criteria for assessing extinction risk, which explicitly separate the process of risk assessment from priority-setting. Here we present an analogous procedure for assessing the extinction risk of terrestrial ecosystems, which may complement traditional species-specific risk assessments, or may provide an alternative when only landscape-level data are available. We developed four quantitative risk criteria, derived primarily from remotely sensed spatial data, information on one of which must be available to permit classification. Using a naming system analogous to the present IUCN species-specific system, our four criteria were: (A) reduction of land cover and continuing threat, (B) rapid rate of land cover change, (C) increased fragmentation, and (D) highly restricted geographical distribution. We applied these criteria to five ecosystems covering a range of spatial and temporal scales, regions of the world, and ecosystem types, and found that Indonesian Borneo’s lowland tropical forests and the Brazilian Atlantic rainforest were Critically Endangered, while South Africa’s grasslands and Brazil’s Mato Grosso were Vulnerable. Furthermore, at a finer grain of analysis, one region of Venezuela’s coastal dry forests (Margarita Island) qualified as Vulnerable, while another (the Guasare River watershed) was Critically Endangered. In northern Venezuela, deciduous forests were classified as Endangered, semi-deciduous forests Vulnerable, and evergreen forests of Least Concern. We conclude that adoption of such a standardized system will facilitate globally comparable, repeatable geographic analyses that clearly separate risk assessment (a fundamentally scientific process), from the definition of conservation priorities, which should take into account additional factors, such as ecological distinctiveness, costs, logistics, likelihood of success, and societal preferences. Jon Paul Rodríguez and Jennifer K. Balch are contributed equally to this work