A new study published in the journal New Phytologist has used artificial intelligence (AI) to predict the extinction risk of all 328,565 known species of flowering plants. The researchers from the Royal Botanic Gardens, Kew, in Richmond, U.K., hope that their findings will help accelerate conservation efforts and protect biodiversity.
How did they do it?
The researchers used a Bayesian Additive Regression Trees (BART) model, a type of machine learning algorithm, to estimate the extinction risk of plants based on their characteristics and distribution. They trained the model on a dataset of more than 53,000 plants that have already been assessed on the International Union for Conservation of Nature (IUCN) Red List, the world’s most comprehensive source of information on the conservation status of animal, fungi and plant species. They then applied the model to the remaining 275,004 unassessed species, and assigned them a provisional status of threatened or not threatened, along with a level of confidence in the prediction. The BART model is able to capture complex and nonlinear relationships between variables, and account for uncertainty and missing data.
The researchers used various sources of data to train and test their model, such as geographic range maps, climate data, elevation data, phylogenetic data and human population density data. They also used expert knowledge to validate their predictions and correct any errors or biases. They compared their results with previous studies that used different methods or datasets to estimate plant extinction risks.
What did they find?
The study revealed that 45% of all flowering plants are threatened with extinction, which is higher than the previous estimate of 40% based on a smaller sample of assessed species. The study also identified 100 species with the least certain predictions, which could be prioritized for further assessment and conservation action. The researchers made their predictions available online via Kew’s Plants of the World Online portal, where anyone can look up any plant species and see its probable status and confidence level. The portal also provides information on the distribution, morphology, ecology and uses of each plant species.
The study found that the main drivers of plant extinction risk are habitat loss due to agriculture, urbanization and deforestation; climate change; invasive species; overexploitation; and disease. The study also found that plant extinction risk varies by region, with the highest proportion of threatened species in tropical and subtropical areas, especially islands and mountains. The study also found that plant extinction risk varies by taxonomic group, with some families having more threatened species than others.
Why does it matter?
Plants are essential for life on Earth, as they provide oxygen, food, medicine, materials and ecosystem services. However, they are also facing unprecedented threats from human activities that degrade their habitats and reduce their populations. Assessing the extinction risk of plants is a crucial step for informing conservation decisions and actions, but it is also a time-consuming and resource-intensive process that requires expert knowledge and data. By using AI to predict the extinction risk of plants, the researchers hope to speed up the assessment process and fill in the knowledge gaps for unassessed species. They also hope to raise awareness and engagement among the public and policymakers about the importance and urgency of plant conservation.
Dr Steven Bachman, Research Leader in RBG Kew’s Conservation Assessment and Analysis team and author of the study, says: “We hope that these predictions can be used for people to apply to their own local biodiversity to find out if they’ve got a threatened species in their house, garden or local park that needs protecting. At a larger scale, our findings can be used by scientists to prioritise and accelerate extinction assessments for the plants we’ve identified as probably threatened but haven’t been officially assessed by the International Union for Conservation of Nature (IUCN) Red List yet. We hope that a commitment can be made to assess these species or we can encourage other people to carry out these assessments.”
Dr Richard Olmstead, Professor of Biology at the University of Washington and Curator of the UW’s Burke Museum Herbarium who collaborated on this study says: “Plants are under threat from many factors but we lack formal assessments for most plant species. This study provides a way to identify those species most in need of attention so that we can target our conservation efforts more effectively.”
Recent Blog : Could Aliens Identify Earth as Habitable Using Our Tech?