Fossil pollen reveals 25,000 years of grassland change in eastern Africa, researchers say
New imaging and learning methods link pollen shape to grass diversity, with limits.
- Publication
- PNAS
- Stage
- Journal article
- What we read
- Summary of the abstract
- Authors
- Marc-Élie Adaimé, Shu Kong, Michael A. Urban, F. Alayne Street-Perrott, Dirk Verschuren, Surangi W. Punyasena
- Universities and research institutions
- Smithsonian Institution, University of Illinois Urbana-Champaign, University of Macau, University of New Brunswick, Swansea University, Ghent University
Institution metadata: OpenAlex record ↗
What they did and found
Researchers used high-resolution pollen imaging and a mixed data learning approach to tie pollen appearance to grass types and their photosynthesis. Across a 25,000-year lake record, they found a decline in grass diversity between roughly 21,500 and 16,000 years ago, and a gradual drop in the share of fast-photosynthesizing grasses as conditions cooled.
Why it matters
Understanding past grassland shifts helps land managers plan for biodiversity, fire risk, and habitat restoration, while recognizing limits tied to sites and time.
The study treats pollen images as data: take close-up photos, label what is known, and look for patterns that hint at grass type and photosynthesis. It shows shape can signal different grass groups.
In practice, map past grass changes to guide land-use decisions, but results depend on the studied region and time window, so keep expectations modest.
What remains uncertain
Results come from a single lake- sediment record; findings depend on site and time window and are not a global history.
Original sources · 1
- Deep learning of fossil pollen morphology reveals 25,000 y of ecological change in eastern African grasslands ↗PNAS · 2026-10-02
Check the original paper for its authors, methods, version and access terms.
What is your take?
Ask a question, add useful context or share a different perspective. Keep the conversation respectful and grounded.