In 1998 two palaeontologists, Jin Meng and Malcolm McKenna, described a faunal change on the Mongolian Plateau in Nature. Around 34 million years ago, as the Earth cooled, perissodactyls, the group of horses, tapirs and rhinos, gave way to rodents and lagomorphs, the group of hares and rabbits. They called it the Mongolian Remodeling. It has been a textbook chapter ever since.
Gemma Benevento, of the Senckenberg centre in Frankfurt and the German biodiversity centre iDiv, put it to the test again with five colleagues from Frankfurt, Jena and Tübingen. The paper appeared in Nature Communications on 8 September 2026. It covers the Paleogene, from 66 to 23 million years ago, in Mongolia and in China north of the thirtieth parallel.
The data come from two public fossil archives, the Paleobiology Database and the NOW database, and from the literature. They add up to 1,887 finds of 393 genera at 426 sites. Time is split into eleven ages, the ones palaeontologists use for Asian mammals, from the Shanghuan to the Tabenbulakian. For body weight the team used the first lower molar, which preserves well and scales with size. Climate comes from a model that simulates atmosphere, ocean and vegetation together. It yields an aridity index, the ratio of rainfall to the water the ground could lose to evaporation.
The starting problem is simple. Where more fossils are collected, more genera are counted. In the Hsandagolian, the age after the cooling, there are 722 finds. In the Ergilian, the age before, there are 126. Counted that way, genera go from 61 to 129 and seem to double. The authors used rarefaction, a method that compares ages at equal sample completeness, as if every age had been dug with the same effort.

With the counts corrected, the record shows three turning points. The first is at the boundary between the Paleocene and the Eocene, about 56 million years ago, when a sudden heat struck the whole planet. The number of genera and species triples. The early Eocene is the richest moment of the whole period. Between the Bumbanian and the Arshantan, body size shifts sharply.
The second is regional. In the middle Eocene, around 40 million years ago, richness collapses almost to Paleocene levels. Archaic groups disappear above all, pantodonts, condylarths and creodonts. In the model Central Asia's climate turns from semi-arid to arid between the early and the middle Eocene. The authors link the two. They also write that the decline could reflect the slow cooling of the time or a brief warming known as the MECO. The link with drying is an inference.

The third coincides with the cooling at the Eocene–Oligocene boundary. Richness falls again and median body size drops. The Mongolian Remodeling, by contrast, shrinks. After correction the change of groups around the cooling is less abrupt. The rise of rodents and lagomorphs in the Oligocene is partly a sampling effect. Fossil-bearing Oligocene rocks are more numerous than late Eocene ones, and small animals get more sieving. Rodents and lagomorphs always outnumbered perissodactyls in genera, as expected for small animals. Sites with perissodactyls even turn out drier than sites with rodents and hares. That runs against the idea that the small animals coped better with drought. If a shift happened, the authors write, it was weak and perhaps already under way in the middle Eocene.
The limits are stated. The Irdinmanhan has so few fossils that its richness could not be estimated. The dates of Asian mammal ages remain uncertain against other continents. The climate model has a coarse grid of several degrees, and under high carbon dioxide plants use water more efficiently, so the aridity index does not translate directly into vegetation.

The study was funded by the Leibniz Association, the German Research Foundation and the state of Hesse. The authors declare no competing interests. The paper has no statement on the use of artificial intelligence. Data and code are in the supplementary materials. A German popular outlet wrote that the drying began six million years earlier than thought. That figure is not in the paper.

References
Benevento G.L., Meijer N., Brugger J., Mulch A., Hickler T., Fritz S.A., "Global and regional climate change impacted Paleogene mammal faunas in Central Asia", Nature Communications, 17, 9606, published online 8 September 2026, open access, CC BY 4.0 licence. doi: 10.1038/s41467-026-77374-7
Meng J., McKenna M.C., "Faunal turnovers of Palaeogene mammals from the Mongolian Plateau", Nature, 394, 364–367, 1998. doi: 10.1038/28603
The NOW Community, "NOW – New and Old Worlds: Database of fossil mammals", version 1, Zenodo, 11 November 2020, CC BY 4.0 licence. doi: 10.5281/zenodo.4268068