Asynchronous crop-record expansions reveal spatial opportunity structures in Holocene China

For thousands of years, the story of how humans learned to farm has often been told as a simple march outward. The image is of a single agricultural frontier pushing steadily across the landscape, like a wave rolling onto a shore, carrying seeds and tools from a central point of origin to new lands. This view suggests that as people moved, they simply planted the same crops in every place they found suitable, driven by a uniform expansion of a single farming package. However, the reality of how food systems spread across ancient China was far more complex. It was not a single wave, but a collection of different stories unfolding at different times and in different ways. To understand this, researchers must look at the physical remains of ancient seeds and grains, known as archaeological crop records, and ask why some crops appeared in certain places while others stayed put, or why they arrived in some regions much later than others. The question is not just where crops grew, but how the landscape itself, the history of previous farming, and the connections between different groups of people shaped the path of these plants.

A team of researchers led by Yu Li and Mingjun Gao at Lanzhou University has tackled this question by building a new way to map the movement of three major crop systems in China over a period of four thousand years, from 6000 to 2000 years before the present. They focused on three distinct groups of plants: millets, which formed the backbone of dry farming in the north; rice, which thrived in the wet, low-lying south; and wheat and barley, which were introduced from the west. Instead of trying to trace a single, straight line of travel for each plant, the team created a computer model that acted like a probability map. This model did not predict exactly where a farmer planted a seed in a specific year. Instead, it estimated the likelihood that a new piece of archaeological evidence for a crop would appear in a specific location, based on what was already known about the environment, where that crop had been found before, and how easily people could travel between regions.

The researchers fed their model with thousands of data points from ancient sites, including the dates of carbon-dated seeds and the specific types of grains found. They combined this with detailed maps of the ancient climate, the shape of the land, and the network of rivers. By running simulations, they could see how the probability of finding a new record for each crop changed over time. The results revealed that these three crop systems did not follow the same rules. The spread of millet records across northern China looked like a steady, continuous expansion. The model showed that new finds of millet were most likely to appear in areas where millet had already been grown, suggesting that once a farming tradition was established in the dry north, it tended to persist and slowly spread into neighboring dry and semi-dry lands. This pattern pointed to a strong sense of regional continuity, where the history of farming in a place made it more likely that farming would continue there.

Rice told a different story. Its records remained tightly clustered in the humid, low-lying areas of the Yangtze River basin and the southern wetlands. The model showed that for rice, the environment acted as a strict filter. Even if there was a strong pressure for the crop to spread from nearby areas, new records were unlikely to appear unless the local climate was wet and the land was flat enough to hold water. The expansion of rice was not just about moving outward; it was about finding the right kind of wet environment. This suggests that the spread of rice was tightly coupled with the specific conditions of the landscape, and it did not easily cross into drier or more rugged terrain, even when it was close to existing rice-farming communities.

Wheat and barley presented a third, distinct pattern. As crops introduced from Central Asia, their records appeared later in the timeline, first showing up in the northwest and then moving toward the central plains. The model indicated that these crops did not rely on the same kind of local, step-by-step spread as millet. Instead, their appearance in new places was strongly linked to "corridor-mediated accessibility." This means that new records of wheat and barley were most likely to appear in places that were easily reachable from existing agricultural hubs through established travel routes, such as the Hexi Corridor. The spread of these introduced crops seemed to happen in leaps, appearing in distant but accessible locations rather than slowly filling in the space between them. This pattern suggests that the movement of wheat and barley was driven by long-distance exchange networks and the ability to reach new areas quickly, rather than by a slow, local expansion of farming communities.

The study explicitly challenges the idea that a single climatic event or a uniform expansion front drove all agricultural change in ancient China. The researchers found that while climate acted as a boundary condition—determining where a crop could physically survive—it did not dictate the exact path of expansion on its own. For rice, the climate was a gatekeeper, allowing expansion only where the humidity was right. For millet, the climate was less of a barrier, allowing the crop to follow the path of existing farming traditions. For wheat and barley, the ability to travel along trade and exchange routes was more important than the immediate local climate. The researchers also ruled out the possibility that these patterns were simply the result of where modern archaeologists have dug the most. Their model showed that the specific ways these crops spread could not be explained by random chance or by the simple fact that some areas have been studied more than others.

By treating the spread of crops as a process of "spatial reorganization" rather than a simple outward march, the study offers a clearer picture of how ancient people adapted to their world. It suggests that the landscape of ancient China was not a blank canvas waiting to be filled, but a complex web of opportunities. Some crops, like millet, followed the path of least resistance, building on what was already there. Others, like rice, waited for the perfect environmental match. And still others, like wheat and barley, rode the waves of human connection, appearing in new places because the roads and rivers made them accessible. The work does not claim to have reconstructed the exact routes every farmer took, but it does provide a quantitative way to see how different crops used the landscape differently. It reveals that the history of agriculture in China was not a single story of expansion, but a multicentered process where different crops found their own unique opportunities in the same ancient world.