Solving Paleobotany's 200-year problem with fluorescence and artificial intelligence

Abstract

Paleobotany has a 200-year problem: leaf fossils are often too poorly preserved to make proper taxonomic identification. Though they are the most abundant, and well-studied macroscopic fossil plant organ, many fossil leaves-- including specimens held in museum and university collections-- appear to lack the detailed features needed to take crucial next steps, like placement into phylogenetic context, generation of ecological interpretations, or reconstruction of paleoclimate. However, we have discovered that under UV and laser light [~]50% of leaves across a diverse survey of fossil deposits fluoresce in a way that reveals spectacularly detailed leaf venation and gross morphology that is invisible under white light. These 'Fossil Ciphers' bring new life into each step of the paleobotanical method, unlocking previously unattainable taxonomic classifications, quantitative morphometrics, and promising new avenues for the incorporation of AI into fossil leaf analyses. With the aid of fluorescence techniques, we can help solve paleobotany's 200-year problem.