Reinforcing the antiquity of hominins in China

6 minute read

In Nature yesterday, Zhaoyu Zhu and collaborators published a paper describing the paleomag chronology of Shangchen, a site in Lantian county, China. The oldest layers bearing stone artifacts at this site appear to be approximately 2.1 million years old, making these stone tools the oldest yet documented within China.

I was really interested in seeing this new archaeological study published. China is taking on a much more important role in human evolution research. The science of human evolution has advanced markedly within China during the last 20 years, just as it has in many other parts of the world. I had the great privilege of visiting the Institute for Vertebrate Paleontology and Paleoanthropology in Beijing earlier this year, and I’ll be sharing more from that trip soon.

The Shangchen site shows two things clearly.

First, it confirms that there were hominins making stone tools in this part of China by 2.1 million years ago. That’s not the earliest date ever found in China — another cave site called Longgupo has stone tools that archaeologists think are 2.2 million years old or older. But other archaeologists have questioned the very early dates from Longgupo, so this new Shangchen chronology helps to put the early evidence on a more secure footing.

Second, the Shangchen site is not merely one ancient discovery but a long series of archaeological layers. With seventeen archaeological layers across the period from roughly 1.4 million to 2.1 million years ago, Shangchen presents a more impressive record of habitation of the Chinese Loess Plateau than any other part of Eurasia.

Loess stratigraphy showing position of artifacts from Zhu et al. 2018
Figure 3 from Zhu et al. (2018), showing the loess and paleosol stratigraphy at Shangchen. The position of artifact-bearing layers in the deposit is shown at left, while the global paleomag chronology is at far right. There is an impressive correlation between the paleomag signatures at the site and the global paleomag chronology.

Still, seventeen artifact-bearing layers in 700,000 years leaves a lot of room for significant changes and hiatuses. At some times, the climate in this region was very arid, the deserts of western and northern China became larger, and huge amounts of dust blew from those deserts and dropped onto this central part of China. At those times, there are some archaeological layers but not too many. Other times, the climate was moist, less dust blew from the deserts, and stable layers of soils formed on the surface at Shangchen. Those ancient paleosols are richer and denser in stone tool assemblages.

So this site seems to show evidence of a long habitation of China by human relatives, but whether it was the same hominin species or population across all that time is not clear. Whoever was living there, they were responding to the local climate in the way they behaved in this region.

Large areas of central and western China are covered in loess from the early Miocene period—as early as 22 million years ago—on up to recent times. During relatively arid periods, desert expanded in western and northern China and Mongolia, generating large amounts of windblown dust. Some of this dust settled in western and central China, creating layers of loess. Today, a thick succession of loess deposits occur in the Chinese Loess Plateau. The sediment is composed of fine particles that were carried by the wind. During wetter periods, the loess deposition greatly slowed, and stable soils formed on the ground surface, so today’s stratigraphy is an alternating series of loess layers and paleosols.

Water easily erodes these layers, so gullies and streams create deep cuts across the stratigraphy. Of course, hominins didn’t enter China as early as the Miocene, as far as we know. Archaeologists have focused on the Pleistocene stratigraphy of the Loess Plateau, surveying these layers where stone tools contrast markedly with the fine-grained loess. A huge region of China is covered by these ancient windblown sediments, and there have been many, many important archaeological discoveries, from Upper Paleolithic all the way back to the dawn of hominin occupation of China.

The paleomagnetic chronology presented in this paper is only one aspect of a much larger program of dating loess events across the Pleistocene. The systematic approach to identifying these dry time periods with large loess deposits is really important, because it is allowing scientists to tie together ancient landscapes and evaluate tools and fossils from many different sites.

For example, the same approach gave rise to the redating of the Gongwangling fossil material by the same research group in 2015. (They argued that the fossil is around 1.65 million years old, far more ancient than thought previously.) This program of chronology across sites has been underway for several years, and I think it is going to lead to many more important discoveries.

I’ll be writing a bit more in the next few days about what scientists know about the earliest movements of hominins out of Africa. All the headlines I’ve seen in the last day are wrong.

Every story I’ve seen has accentuated that Shangchen is 300,000 years older than Dmanisi, which the stories claim was previously the oldest evidence of human habitation of Eurasia. This is wrong.

Now, Dmanisi is very important, but it’s simply not accurate that it’s the oldest evidence outside Africa. There are older sites. Even within China, Longgupo has an earlier published chronology, and Longgudong is estimated to be as old as the oldest Shangchen layer. In India, there are sites with earlier published evidence, one as old as 2.6 million years.

That doesn’t mean scientists should blindly accept every “earliest” date for every archaeological occurrence. Any find that demands that we revise our models for human evolution should be supported by multiple lines of evidence. So when only one or two methods point to the geological age, I want to see more. Early sites often have only a few observations suggesting a minimum age, and history shows that those “earliest” dates are often revised. Even for the new Shangchen chronology, where the paleomagnetic chronology is impressive, we should seek more cross-validation.

Still, even with my mistrust of geological age estimates, I have to say that the evidence points against the old idea that the first out-of-Africa dispersal was by stone tool-making Homo erectus within the last 2 million years.

We now have five or six sites earlier than Dmanisi, and Dmanisi itself is the first occurrence of “H. erectus” anywhere in the world. With the Gongwangling cranium and the earliest Sangiran material older than 1.5 million years, H. erectus has a better record and greater diversity in Eurasia than in Africa in the same time period.

At least, what we’ve been calling H. erectus. If there was a single, early colonization of Eurasia by hominins, it must have been much earlier than the first occurrence of H. erectus. I would suggest instead that there were many movements and dispersals from Africa and back into Africa, starting much earlier than 2 million years ago and extending up to the most recent. We already know this is true of the Middle and Late Pleistocene—it is not a single dispersal and stasis, there were many partial replacements with interaction of earlier people and new migrants.

The old H. erectus single migration model just can’t account for the data of early sites in Eurasia, it doesn’t deal well with the morphology of the Dmanisi and early African H. erectus fossil material, and it doesn’t add any explanatory power.

There’s a big story brewing here.


Zhu Z, Dennell R, Huang W, Wu Y, Qiu S, Yang S, Rao Z, Hou Y, Xie J, Han J, Ouyang T. 2018. Hominin occupation of the Chinese Loess Plateau since about 2.1 million years ago. Nature (online). doi:10.1038/s41586-018-0299-4

Zhu, Z. Y., Dennell, R., Huang, W. W., Wu, Y., Rao, Z. G., Qiu, S. F., ... & Zhou, H. Y. (2015). New dating of the Homo erectus cranium from Lantian (Gongwangling), China. Journal of Human Evolution, 78, 144-157. doi:10.1016/j.jhevol.2014.10.001