Ancient ocean deposits reveal that Africa’s open grassy plains began to form millions of years earlier than previously believed—long before modern grasses took over. The study’s findings were published in the Proceedings of the National Academy of Sciences.
The African savanna is one of the planet’s most recognizable landscapes: open grasslands, scattered flat-topped trees, and herds of wild animals roaming the terrain. It is generally believed that this landscape emerged around 10 million years ago, when grasses utilizing a drought-resistant form of photosynthesis spread across the tropics.
However, this open landscape existed long before the arrival of those grasses. By studying silt deposits off the northwest coast of Africa, a research team dated the formation of this landscape to approximately 15 million years ago. The grasses growing there at the time were of a completely different type—nothing like them exists in subtropical Africa today.
Anna Schartman of the University of California, Santa Cruz, conducted the study alongside her advisor, Pratigya Polissar, and Caroline Strömberg of the Burke Museum in Seattle. The researchers reconstructed Africa’s ancient landscape using plant material carried to the sea by the wind.
Every leaf is coated in a waxy layer that slows moisture loss. Wind lifts dust from northwest Africa and carries this wax out over the Atlantic Ocean, where it settles into the silt and remains preserved. Mud deposits accumulated continuously from the late Oligocene through the late Miocene on the Cape Verde Plateau and Rise, located approximately 500 kilometers offshore. Shartman worked with two core samples taken from the area, sampling at 100,000-year intervals during the middle of this period and taking about 40 additional samples from the second core for cross-verification. All data were obtained through the International Ocean Discovery Program. Of the measurements she performed, 158 samples were clean enough to be identified as terrestrial plants, representing about 96% of the total sample set.
Wax molecules vary in length. Trees and shrubs primarily produce short molecules, whereas grasses and low-growing leafy plants produce longer ones. The team verified this by comparing the results with 967 wax measurements obtained from modern African plants.
Between approximately 16 and 14 million years ago, the chemical composition shifted. Long molecules became far more common than short ones—a pattern that occurs when grasses spread across territory previously covered by trees and shrubs.
Around 15 million years ago, each batch of samples differed markedly from the older ones. The landscape never reverted to its previous state. Between 16.9 and 14.7 million years ago, Earth experienced its warmest Miocene interval. This was followed by a sharp cooling and drying trend, and it was during this period that grasses spread most rapidly.
Previous research teams working along this coast had dated the region’s grassy landscapes to approximately 10 million years ago. The current findings push the date for these grassy landscapes back by five million years.
A second measurement of the same molecules makes it possible to distinguish between two types of plants. Grasses in hot, arid environments often employ a form of photosynthesis known as C4, which concentrates carbon dioxide before fixing it. Virtually all other methods employ an older technique, and both methods leave carbon signatures distinct enough to determine the mixture’s composition in any given silt sample.
Throughout the Middle Miocene, this characteristic feature remained associated with the older type. The grasses that appeared in northwest Africa were not drought-tolerant at all, and such a combination does not grow in any subtropical lowland area today. This alters the sequence of events: forests and groves gave way to grasslands, which were later replaced by other types of grasslands.
The new type of photosynthesis did not pioneer the colonization of Africa; it emerged in areas that were already inhabited by vegetation. Other regions fit this pattern as well. In East Africa, grass pollen accounted for up to 40% of the total count around 11 million years ago, yet the wax found there still resembled the older type.
A 2023 study revealed that grasses—which had been present in the area as early as 18 million years ago—grew within a fragmented landscape where forest cover still persisted.
The newer types of grasses began to spread after 11 million years ago. A 2019 study of this coastline lacked sufficient samples to show what happened next, whereas the current study provides enough data.
They spread rapidly between approximately 9.3 and 8.4 million years ago, but then showed virtually no further expansion during…
…over the course of the next million years. Growth resumed through the end of the Miocene, with the most rapid expansion occurring between 7.4 and 6.4 million years ago.
Grasslands in other regions underwent changes during the same period. In Nepal, C4 grasses displaced other plants approximately 7.5 million years ago, followed by the plains of the Indus and Bengal regions 7.4 and 6.9 million years ago, respectively. Next came Argentina (6.7 million years ago), the Great Plains of North America (6.4), and China’s Loess Plateau (6.1).
It is difficult to attribute such simultaneous climatic shifts across so many continents to regional weather patterns alone. Australia is an exception; its grasses emerged roughly 3.5 million years ago. These developments coincide with a period of global cooling between 7 and 5.3 million years ago, during which sea-surface temperatures dropped worldwide.
Climate change alone is not enough to maintain a savanna in its natural state. Across much of Africa, forests would grow where savanna now exists were it not for the impact of fires and large herbivores.
Shartman’s team found evidence that both of these factors were at play long before the new grass species appeared. Fragments of burnt grass have been discovered in Middle Miocene deposits in the Niger Delta, alongside abundant grass pollen. Thorny trees and shrubs diversified during this same period, coinciding with the rise of African bovids—the antelope family.
Later, underground woody structures evolved, enabling shrubs to regenerate after fires. The same pattern applied to grazing: starting 10 million years ago, an increasing number of large East African mammals—including relatives of horses, rhinoceroses, and bovids—switched to a diet of C4 grasses. Alongside the initial opening of the landscape, leaf- and twig-browsing plant species diversified; similarly, fire-adapted plants diversified alongside the second phase of landscape opening.
The authors cannot explain why the landscape opened up or why new grasses subsequently took over. Reconstructions of past carbon dioxide levels are so contradictory that the gas’s role remains unclear.
Precipitation levels increased. While climate models generally show a steady drying trend in North Africa, some sedimentary data point to increased rainfall. A separate analysis based on hydrogen content in the same wax molecules revealed no long-term changes in precipitation.
Shartman and her colleagues describe their drought-based explanation as speculative. The question of whether rainfall became more seasonal has never been directly tested in this region.
Fire and grazing continue to shape the African savanna, and Miocene-era data suggest that relatively minor changes can reorganize the entire system.
Future changes in carbon dioxide concentrations and temperatures are projected to match or exceed those observed during the Miocene. These samples capture dust from a specific latitudinal band of the continent—roughly 10 to 25 degrees north. Data from a single coast cannot determine whether the rest of Africa opened up simultaneously.
The research team has called for an examination of data from other latitudes to trace this pattern across the African continent.
About the Author
