The recent Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services (IPBES) report highlighted the large scale of extinction risks to biodiversity (Díaz et al., 2019). Assessing species' extinction risk is vital for setting conservation priorities and the first step toward protecting particular areas or groups. A widely accepted approach to assess extinction risk, and a key source of data underpinning the IPBES report, is the IUCN Red List of Threatened Species (hereafter Red List). However, with only 9% of plants represented by assessments at the latest update (IUCN, 2019), slow progress in increasing Red List coverage of mega-diverse groups like plants has limited their inclusion in analyses of global conservation priorities (Venter et al., 2014; Betts et al., 2017; Di Marco et al., 2018). Responding to this problem, there is growing interest in speeding up the assessment process. Automation, particularly through machine learning, offers an attractive solution. However, we advocate caution before adopting it to help set global conservation priorities.
We draw on two recent examples from the literature (Pelletier et al., 2018; Stévart et al., 2019) that deserve attention as the largest studies to date that use machine learning or automation to predict the conservation status of plants. Each study recommends a protocol for rapidly generating preliminary conservation assessments, and both share the goal of using their preliminary assessments to highlight global or continent-wide conservation priorities. The potential impact of these studies merits careful scrutiny. Herein we highlight aspects of their design and reporting that can be improved so that future studies of this kind can have maximum impact.
This is a metadata only record.