Deep learning of fossil pollen morphology reveals 25,000 y of ecological change in eastern African grasslands

A computer program learned to spot tiny shape differences in grass pollen that look almost the same under a normal microscope. It studied super sharp images of pollen from a lake near Mount Kenya, covering 25,000 years of history. The program tracked how many grass species lived there over time and whether they used a heat tolerant or cooler weather strategy to grow. Grass diversity dropped a lot between 21,500 and 16,000 years ago, during the coldest part of the last ice age. After the ice age ended, heat loving grasses slowly became less common as the climate warmed and fires happened less often.

One lake site in Kenya studied, lab based image analysis only.

Integrity screen: passed (3 checks) Checked 4 October 2026. Retraction record: none. DOI resolves at doi.org. Metadata record found (Proceedings of the National Academy of Sciences). Read the source

This screen checks process signals. It does not tell you the findings are true.