Pre-assessments of plant conservation status in islands: the case of French Overseas Territories

Assessment methods have been developed to estimate a preliminary conservation status for species and subsequently to facilitate the building of Red Lists. Such pre-assessment methods could be particularly useful in the French Overseas Territories (FOTs) where Red Lists tend to be out-dated or absent and where a high number of endemic species face detrimental anthropogenic pressures. We first aimed to conduct a preliminary assessment (hereafter, pre-assessment) of the conservation status of endemic plants from Guadeloupe, Martinique, Reunion, Mayotte, French sub-Antarctic islands, New Caledonia, and Scattered Islands. We then compared the various methods used in conducting the pre-assessment and discussed ways to adapt these methods to small territories. We compiled occurrence data of endemic species identified thanks to a previous taxonomic work and pre-assessed their conservation status under Red List criteria A and B and the use of a Random Forest algorithm. We then measured the accuracy, specificity, and sensitivity of each method based on existing Red Lists. The Random Forest algorithm and a method based on range-size performed best at correctly attributing conservation status. Using these pre-assessment methods, we estimated that up to 60% of the endemic flora of the FOTs is potentially threatened. Range restriction but also anthropogenic pressures were key factors that explained these risks. Pre-assessment methods are useful tools to get a first measure of species conservation status. These methods should be adapted to the territories considered and their conservation issues in order to reach a good performance.

Automated summary

Assessment methods for estimating conservation status and building Red Lists are valuable tools, especially in French Overseas Territories where out-dated or absent Red Lists and anthropogenic pressures pose threats to endemic species. This study conducted a pre-assessment of conservation status for endemic plants in various territories, comparing different methods used and discussing adaptations for smaller territories. The Random Forest algorithm and range-size-based methods showed the highest accuracy in attributing conservation status, revealing that up to 60% of endemic flora in the FOTs is potentially threatened due to range restriction and anthropogenic pressures.