The institution behind this mobility was the nabo system, the circuit of seasonal imperial camps through which the Liao empire (907–1125) and, in modified form, the Jurchen-led Jin dynasty (1115–1234) conducted much of their politics. Instead of fixing the state in one capital, the emperor moved with his entire government through an annual rhythm of encampments, each staged in a landscape suited to the season. The spring nabo was positioned around the shallow lakes and wetlands of the Songnen Plain in what is now western Jilin Province, where the imperial party fished, hunted waterfowl, received envoys and held feasts. Court chronicles such as the Liaoshi and the Jinshi record both the spectacle and the machinery that sustained it: animals flowed in from state-managed imperial herds, from tribute levied on subordinate tribes and neighboring polities, and through procurement organized by officials who traveled with the court. Historians have long read these passages as evidence of a flexible, multi-channel provisioning economy. What no text could demonstrate is whether that system operated on the ground—inside one camp, in the bones and chemistry of the animals themselves.

Houmingzi, a locality within the spring nabo site complex on the lake margin near Chagan Lake in Qian’an County, supplies that ground truth. Excavated in 2016 by archaeologists from Jilin University and the Jilin Provincial Institute of Cultural Relics and Archaeology, the site sits in one of Northeast China’s most environmentally constrained landscapes: the Songnen Plain is a mosaic of saline-alkaline soils, marshes and grassland whose grazing capacity swings sharply between seasons and years. Radiocarbon dating of the deposit anchors the assemblage within the Liao–Jin era, matching the historical lifetime of the royal spring encampments. The setting drives the study’s central logic. A locality used only seasonally, on a shoreline with limited carrying capacity, could not plausibly have reared the volume of domestic animals represented in its refuse; any such accumulation must represent animals moved in from elsewhere, whether herded alongside the traveling court or dispatched to meet it. The bone deposit therefore reads as a ledger of provisioning rather than a portrait of local herding, and that distinction is what makes it archaeologically powerful.

The first line of evidence was classical zooarchaeology. Every fragment was identified to taxon, abundance was quantified with standard measures such as the number of identified specimens and the minimum number of individuals, and the survival of particular skeletal elements was recorded. Kill-off patterns—the ages at which animals died—were reconstructed from tooth eruption and wear stages and from the fusion timing of limb-bone epiphyses, the standard tools for distinguishing herds managed for meat, milk, traction or breeding. Cut marks, chop marks and other carcass-processing traces were mapped across the bones and screened against taphonomic damage from weathering, carnivore activity and recovery bias. The outcome was unambiguous: mortality profiles and butchery patterns were taxon-specific. Cattle, sheep, horses and pigs were not treated as interchangeable meat units; each species was slaughtered at different ages, processed in different ways and consumed under different conventions, implying that each taxon entered the camp through a distinct logic of supply and use.

The second line of evidence came from the animals’ own chemistry. The team extracted bone collagen and measured its stable carbon and nitrogen isotope ratios, expressed as δ¹³C and δ¹⁵N. Because collagen turns over slowly, bone integrates diet over roughly the final years of an animal’s life, so every measurement preserves a pre-death dietary archive. Carbon isotopes separate two photosynthetic worlds: C₃ plants, the trees, shrubs and cool-season grasses that dominate the region’s natural vegetation, and C₄ plants such as millet, a staple of the Liao–Jin world. Nitrogen isotopes climb with trophic level and can flag foddering on animal by-products or other distinctive feeds. Only collagen that passed strict quality controls—adequate yield and atomic carbon-to-nitrogen ratios within accepted windows—entered the analysis, a safeguard against the diagenetic alteration that can silently corrupt ancient samples. AMS radiocarbon dates, calibrated against the IntCal20 curve, fixed the assemblage chronologically within the era of the royal camps, tying the chemistry to the Liao–Jin period.

The isotope data split the assemblage into two very different groups. The wild animals, few in number but essential as a local reference, broadly reflect the C₃-based environment of the surrounding region, sketching what locally raised fauna should look like. Sheep occupy a comparatively narrow δ¹³C range, a pattern consistent with flocks raised under relatively uniform grazing regimes before they reached the camp. Cattle, horses and pigs behave altogether differently: each of these species shows marked intraspecific variation, with individual animals scattered across the isotopic spectrum. That spread cannot be produced by a single herd on a single pasture under a single feeding strategy. It means individual animals arrived carrying distinct pre-death feeding histories—different pastures, different fodder, plausibly different herder communities. Combined with the species-specific kill-off and processing patterns, the chemistry indicates that different animals were funneled into the camp through different channels and consumed according to different conventions.

Those converging channels are the heart of the finding. At a seasonally occupied lake-margin camp that could not have reared such quantities of livestock itself, animals with divergent carbon isotope signatures almost certainly originated in more than one ecological or management background. The researchers interpret Houmingzi as a point of concentration and consumption: a node where livestock raised elsewhere was gathered to sustain a mobile political center. The inference is structural as well as chemical, because a seasonal camp on the saline-alkaline shores of the Songnen Plain lacked the carrying capacity to supply the cattle, sheep, horses and pigs its deposits contain. The conclusion aligns closely with the historical record, which documents a multi-channel provisioning framework for the Liao court—state herds, tribute from subordinate groups and locally arranged procurement—administered by officials attached to the moving household. The refuse of a single encampment now shows that this framework left a measurable fingerprint in bone.

The study also extends the archaeology of Inner Asian empire into an underexamined setting. Because power in these polities resided in moving assemblies rather than fixed cities, provisioning was spatially separated from production, and the supply systems behind it are notoriously difficult to detect. Most isotopic work on elite provisioning has therefore focused on permanent capitals—among them a recent investigation at medieval Karabalgasun in Mongolia, which used zooarchaeological and isotopic evidence to examine the pastoral provisioning of Uyghur elites in an urban environment. Houmingzi offers the complementary and far rarer case: zooarchaeological and isotopic evidence from within a locality of a mobile royal camp, where the encampment itself, not a walled city, was the seat of power for the season. The results show that even a temporary court sat atop a logistics network capable of assembling and differentiating large volumes of livestock, moving the study of imperial supply from urban centers into the mobile political landscape of the steppe frontier.

The authors are careful about what the evidence cannot yet say. Carbon and nitrogen isotopes record feeding histories rather than birthplaces, so they cannot establish whether a given animal came from imperial herds, a tribute consignment or a market transaction, and the specific origins and supply channels behind the isotopic variation remain unresolved. The wild-fauna sample is also small, leaving the local isotopic baseline broad rather than fine-grained. The way forward is clearly defined: strontium isotope analysis of tooth enamel, which fingerprints the geology of an animal’s home range; sequential sampling of dental tissues, which tracks seasonal movement and dietary change across an individual’s life; and systematic comparison with faunal assemblages from other nabo localities and from the permanent settlements and capitals of the Liao and Jin worlds. Together, these approaches could convert the camp’s differential chemistry into an actual map of supply routes across Northeast China.

Even with those limits, the findings change how a chapter of imperial history can be read. The Khitan and Jurchen courts that moved through the Songnen Plain did not travel light: they carried with them, or summoned to meet them, a hinterland of herders and tribute payers whose animals converged on the lakeshore each spring. At Houmingzi, that hinterland is visibly non-uniform—sheep arrived on one dietary regime, cattle and horses on several, pigs on others—so that every animal carried the chemical record of the routes and management systems that delivered it. In butchered bone and collagen chemistry, archaeologists have recovered the ledger of a state in motion, and with it confirmation that the most sophisticated infrastructure of a nomadic empire may have been the provisioning system that traveled quietly beside it, assembled species by species long before the first imperial pot went on the fire.

The research, led by Dr. Wei Wang of Nanjing University of Information Science and Technology and Professor Chengbang An of Fudan University, centers on the Baiyanghe cemetery in the northern foothills of the Tianshan Mountains in Xinjiang. The team analyzed carbon and nitrogen isotopes preserved in human and animal bones excavated from the cemetery, whose burials span the Late Bronze Age through the Tang Dynasty. By combining isotope measurements with radiocarbon dating, the researchers established a long-term record of what people consumed and how their diets changed across successive historical periods.

The study then placed the Baiyanghe results within a much larger regional framework. Nearly 1,500 previously published isotopic records were assembled from the Minusinsk Basin, the Altai Mountains, central Kazakhstan, southeastern Central Asia and central Mongolia. This extensive comparison allowed the researchers to distinguish local dietary variation from changes occurring across the steppe as a whole. Despite major differences in ecology, mobility, political organization and cultural traditions, populations in the six regions displayed a remarkably synchronized dietary trajectory.

The principal evidence came from carbon and nitrogen stable isotopes, chemical signatures that remain locked in bone collagen during an individual’s lifetime. Carbon isotope ratios, expressed as δ¹³C values, help identify the types of plants at the base of a food chain. C3 plants include wheat, barley and most temperate vegetation, while C4 plants include broomcorn millet and related crops. Because the isotopic signatures of these plants can pass through livestock and into human tissues, δ¹³C values can reveal whether people relied more heavily on C3 or C4 foods, even when ordinary food remains are scarce or poorly preserved.

Nitrogen isotopes, expressed as δ¹⁵N values, provide a different kind of information. They generally rise with each step up the food chain, meaning that consumers of animal protein often show higher values than the herbivores they eat. The researchers compared human δ¹⁵N measurements with those of sheep and goats, then used a typical trophic enrichment interval of approximately 3 to 6 parts per thousand to estimate the isotopic range associated with diets dominated by domesticated herbivores. This approach helped assess changes in animal-protein consumption while accounting for the ecological background of each region.

During the Bronze Age, roughly 5000 to 3000 years before the present, people across the central-eastern steppe appear to have relied predominantly on C3-based foods. These included wheat, barley and other C3 crops, together with meat and dairy products from domesticated herbivores such as sheep and goats. The picture changed dramatically between about 3000 and 2000 years before the present, corresponding broadly to the Early Iron Age through the Han Dynasty. Across several regions, δ¹³C values increased, indicating a substantial expansion of C4 foods, especially broomcorn millet. At the same time, elevated δ¹⁵N values suggest that animal protein also became more important in many communities.

The millet transition, however, was not equally intense everywhere. The Minusinsk Basin, southeastern Central Asia and the northern Tianshan Mountains emerged as high-intensity C4-use zones, where millet appears to have become a staple for much of the population. Central Kazakhstan, the Altai Mountains and central Mongolia showed more limited use. In these areas, millet may have been consumed primarily by elites or used in contexts where imported food carried social meaning. The evidence illustrates that a shared regional trend did not erase social distinctions: the same crop could function as everyday nourishment in one community and as a marker of prestige in another.

The Baiyanghe cemetery offers particularly suggestive evidence for the relationship between diet and status. Individuals buried with abundant grave goods or gold artifacts generally had higher δ¹⁵N values, consistent with greater access to animal protein. The pattern was not universal, and the researchers emphasize that social rank cannot be inferred from isotope values alone. Nevertheless, the association indicates that food inequality may have accompanied broader transformations in wealth, ritual display and political organization during the period when steppe societies were becoming more interconnected.

Between approximately 2000 and 1000 years before the present, during the Han-Tang era, millet consumption declined substantially and many populations shifted again toward C3-based staples. The researchers propose that the earlier move toward millet was supported by a warmer and more humid climate that encouraged movement and exchange among pastoralist and farming communities. The later reversal may have been linked to cooler, drier conditions as well as improved grain-processing technologies, which made C3 cereals more practical and attractive in changing economic environments. These explanations remain interpretations rather than direct proof of causation, but the alignment of climatic, archaeological and isotopic evidence provides a compelling framework for understanding the transition.

Published in Science China Earth Sciences, the study offers one of the most extensive reconstructions yet of dietary change across the central-eastern Eurasian Steppe. Its central message is that ancient food systems were neither static nor purely local. Climate fluctuations, crop dispersal, livestock economies, technological innovation and social hierarchy interacted across vast distances, producing dietary changes that were both regionally synchronized and locally distinctive. The findings show how the chemistry of human bones can transform scattered archaeological evidence into a continent-scale record of adaptation, exchange and inequality.

Subject of Research: Long-term dietary evolution and human-environment interaction across the central-eastern Eurasian Steppe, reconstructed through carbon and nitrogen stable isotopes.

Article Title: “Diachronic dietary evolution from the Bronze Age to the Tang Dynasty in the central-eastern Eurasian steppe revealed by stable isotopes of the Baiyanghe cemetery, Xinjiang”

Web References: https://doi.org/10.1007/s11430-025-1861-y

References: Wang W, Duan F, Yan X, An C B. 2026. “Diachronic dietary evolution from the Bronze Age to the Tang Dynasty in the central-eastern Eurasian steppe revealed by stable isotopes of the Baiyanghe cemetery, Xinjiang.” Science China Earth Sciences, 69(6): 2102–2117.

Image Credits: ©Science China Press

Keywords: Eurasian Steppe, ancient diets, stable isotopes, carbon isotopes, nitrogen isotopes, broomcorn millet, Bronze Age, Han Dynasty, Tang Dynasty, Xinjiang, archaeology, paleoclimate, human-environment interaction