Using Machine Learning to Predict Patterns of Biodiversity

Jack Sullivan
Biological Sciences

Tuesday, February 3

12:30-1:30 p.m. PT
Library 1st floor near the main entrance

As the biodiversity crisis accelerates, the need arises to hasten the pace of biodiversity research. Traditionally, this has adopted a taxon-by-taxon or a single-ecosystem approach. We have been applying the power of supervised machine learning to large, publicly available databases in order to develop predictive models and apply them to two issues in biodiversity: the discovery of cryptic diversity and assessing conservation status. The former is typically assessed by genetic studies of single species, whereas our approach can be applied to species that lack genetic data. Here, we use Pacific Northwest rainforests as a model system. Similarly, assessing conservation status is labor intensive and the International Union for Conservation of Nature (IUCN) lacks data for 90% of known plant species. Our approach allows us to predict conservation status for the vast majority of plant species of the world, allowing limited resources to be concentrated to areas and to species likely to be under greatest threat.

Jack Sullivan received his B.S. and M.S. degrees from Vermont and his Ph.D. from Connecticut. He then did a postdoc at the Smithsonian and arrived at UI in 1997. He served 12 years as Associate Editor and then Editor-in-Chief of Systematic Biology and was elected President of the Society of Systematic Biologists. He was a founding member of IBEST at UI and has also served as its Director. He’s been listed among the top 2% of scientists worldwide. He has co-owned One World Café since 2005 and he coaches both men’s and women’s rugby teams at UI.