Based on faunal similarities and the teeth of terrestrial mammals, specifically the crown height of their upper molars, the late Miocene (11.6-5.3 million years ago (Ma)) vegetation from China to the Eastern Mediterranean and East Africa has been reconstructed as a single cohesive biome, with modern African savannahs the surviving remnant1. I decided to test this reconstruction with plant fossils. 

To that end, I joined forces with Thomas Denk (Swedish Museum of Natural History, Stockholm), a paleobotanist with expertise in Miocene seed plants, Constantin Zohner (ETH Zurich), an ecologist with statistical expertise, and the paleobotanist Guido Grimm (Orléans, France) to review the relevant plant fossil record. Our data span sites in Greece, Bulgaria, Turkey, the Tian Shan Mountains and Baode County in China, and East Africa from 14–4 Ma. We found evidence that the Western Eurasian sites were covered by forests of both evergreen and deciduous flowering plants and needle trees, with 15% of 1602 fossil occurrences representing conifers, which are present at all but one of the sites. None of the genera (or sometimes species) of conifers and flowering plants that were present are typical of African savannas. The late Miocene Western Eurasia was covered by evergreen needleleaf forests and mixed forests had already been inferred by paleobiologists in the 1980s – as seen in a wall painting showing Miocene horses in a forest near Vienna, photographed in the Vienna Natural History Museum in February 2018.  

So why do plant-based reconstructions imply tree- and shrub-dominated vegetation, not grass-dominated savannas, in late Miocene Bulgaria, Turkey, and Greece, while average tooth crown height, the so-called hyposonty index, implies increasingly arid steppe climates in western Asia (ref. 1)? My colleagues and I suspect the crux may be the way in which the hypsodonty index averages over the height-to-length ratios of the second (upper or lower) molar of many species of animals, with just three categories into which all teeth found at a particular site are binned. This may be insufficient to infer past precipitation. Surely, the mammals that roamed late Miocene Eurasia, including antelopes, giraffes, horses (extinct genus Hipparion and other Equinae), rhinoceroses, elephants and mastodons, pigs, hyenas, cats, and many rodents, fed in different ways, and the crown height index may therefore be a problematic proxy for rainfall --and open or closed vegetation-- in the Miocene. 

Organismal communities are constantly changing in their species composition, and one of the reviewers of our paper suggested that the problem with the different reconstructions might lie in the biome concept itself, which places a huge trust in the existence of co-adapted local communities. What if the plants and mammals living together are not perfectly adapted to each other and to the local climate? It’s worth re-reading Gleason’s (ref. 2) individualistic concept of plant associations.

The paper in Nature Ecology and Evolution is here: go.nature.com/2PKn8o2

References
1. Kaya, F., Bibi, F., Žliobaitė, I., Eronen, J. T., Hui, T. & Fortelius, M. The rise and fall of the Old World savannah fauna and the origins of the African savannah biome. Nature Ecology & Evolution 2, 241–246 (2018).
2. Gleason, H. A. The individualistic concept of the plant association. Torrey Botanical Club Bulletin 53:7–26. 1926 (1926).