Beyond the Savanna: Homo erectus and the Rainforests of Southern China
Eight fossil teeth from two limestone caves in southern China have provided unusually direct evidence of an ancient human population living in tropical rainforest environments. The teeth, attributed to Homo erectus, belong to deposits dated within approximately 324,000–184,000 years ago. They come from Baxian and Naxian caves in Guangxi, a region of steep karst hills, enclosed valleys and extensive tropical vegetation near the modern border with Vietnam.
The fossils are modest objects on which to reconstruct an extinct human population. They do not reveal the appearance of the individuals who possessed them, the tools they carried or the precise foods they collected. Yet dental tissues preserve information unavailable from many other archaeological remains. Their shape and microscopic structure can help identify a hominin lineage; their surviving proteins can provide molecular evidence of evolutionary relationships; and their chemical composition can record aspects of diet, environment and movement during life.
Two complementary studies published in Nature on 18 November 2026 have brought these approaches together. One establishes the taxonomic identity of the fossils, while the other reconstructs their chronology and environmental context. The resulting picture places Homo erectus in southern China considerably later than previously documented there and indicates that these hominins exploited dense tropical forests, habitats often regarded as especially difficult for earlier members of the human genus.
Two caves and an unexpected population
Baxian and Naxian lie approximately 15 kilometres apart in the Chongzuo region of Guangxi Zhuang Autonomous Region. Their limestone setting belongs to the wider karst landscapes of southern China and northern mainland Southeast Asia, where caves have preserved important assemblages of Middle and Late Pleistocene mammals and hominins.
These deposits are valuable because much of the archaeological record outside caves has been destroyed, buried or rendered difficult to identify by tropical weathering and landscape change. Cave sediments can protect fossils, although their accumulation is rarely straightforward. Bones and teeth may enter through animal activity, water transport, sediment movement or other processes. The presence of a fossil in a cave does not establish that its owner lived inside it.
The researchers examined nine hominin teeth from the two sites, alongside thousands of associated animal fossils. Eight teeth were assigned to Homo erectus through morphological and molecular evidence. The ninth is important because it indicates that the assemblage cannot simply be treated as representing one uniform hominin population. Its taxonomic position requires separate consideration.
The identification matters particularly in southern China, where isolated fossil teeth have sometimes been difficult to distinguish from those of other hominins or even large apes. Similarities in dental dimensions and morphology can complicate attribution, especially when the rest of the skeleton is absent. In tropical Asia, fossil orangutans add another potential source of confusion.
The accompanying taxonomic study, led by W. Liao and colleagues, uses dental morphology together with palaeoproteomics, the analysis of ancient proteins preserved in fossil tissues. Proteins can survive in circumstances where recoverable ancient DNA is absent or severely degraded. Their molecular sequences provide another line of evidence for evolutionary relationships, although their resolution depends on which proteins survive and how informative those sequences are.
This combination is more persuasive than assigning isolated teeth to a species on size or shape alone. It also illustrates a growing methodological shift in palaeoanthropology: fragmentary fossils can sometimes be investigated through multiple independent biological records rather than relying exclusively on comparative anatomy.
Establishing the chronology
Before the Guangxi discoveries, securely identified Homo erectus fossils in China were concentrated largely in northern and central regions. Important occurrences include Yuanmou, Gongwangling, Zhoukoudian, Nanjing and Hexian. Together they document a long history of hominin occupation, extending from the Early Pleistocene into the Middle Pleistocene. However, reliably identified Homo erectus fossils younger than approximately 400,000 years were notably scarce on the Chinese mainland.
This gap was difficult to interpret. It could indicate that Homo erectus populations disappeared from substantial parts of mainland East Asia, perhaps in response to environmental change or competition. Alternatively, it could reflect the uneven distribution of excavation, poor preservation, uncertain fossil identification or insufficient chronological control.
The new studies address the dating problem through several methods rather than depending on a single measurement. Westaway and colleagues integrated 43 radiometric age determinations obtained through five dating techniques. These included optically stimulated luminescence dating of quartz, post-infrared infrared-stimulated luminescence dating of feldspar, uranium-series dating, combined uranium-series and electron spin resonance dating of teeth, and uranium-series dating of carbonate formations.
Each method measures a different physical process. Luminescence techniques estimate when mineral grains were last exposed to sufficient light before burial. Uranium-series measurements exploit radioactive decay within suitable materials, while electron spin resonance records radiation-related changes in fossil tissues. Carbonate formations can constrain the timing of sediment accumulation when their stratigraphic relationships are secure.
The results were integrated using Bayesian chronological models, allowing the researchers to assess the relationships between individual dates, archaeological layers and their uncertainties.
For Baxian, the fossil-bearing deposits were modelled within approximately 324,000–184,000 years ago. The corresponding interval at Naxian was approximately 319,000–241,000 years ago. These ranges overlap substantially, although overlap does not establish that the two caves contained hominins at precisely the same moment.
Nor should the youngest boundary of the Baxian interval be interpreted as the date when Homo erectus became extinct in southern China. The dates constrain the deposits containing the fossils. They do not establish continuous occupation throughout the entire interval, and they cannot identify the last surviving individual.
Nevertheless, the chronological result is substantial. It places securely identified Homo erectus in mainland China much later than the previously established regional record, and it does so with multiple dating methods that are broadly consistent within their uncertainties.
Reading a rainforest from fossil teeth
The ecological argument rests on more than the modern appearance of Guangxi. Tropical landscapes have changed repeatedly during the Pleistocene, responding to climatic oscillations, rainfall patterns and vegetation shifts. A rainforest existing in a region today cannot be assumed to have occupied the same ground hundreds of thousands of years ago.
Westaway and colleagues therefore reconstructed the ancient environment through animal fossils, stable isotopes and other geochemical evidence. The faunal assemblages include forest-associated mammals and extinct or locally vanished taxa such as Stegodon, alongside elephants, rhinoceroses and orangutans. Their ecological associations provide an initial indication of the habitats represented in the deposits.
More direct evidence comes from carbon isotopes preserved in tooth enamel. Plants using the C3 photosynthetic pathway, including most trees and forest vegetation, have different carbon isotope signatures from many tropical grasses using the C4 pathway. Under dense forest canopies, additional isotopic effects can produce especially depleted carbon values.
By measuring these signatures in herbivore teeth, researchers can investigate the vegetation supporting ancient food webs. At Baxian and Naxian, the faunal measurements indicate diets dominated by C3 vegetation, with numerous values consistent with dense-canopy conditions. The researchers found no evidence in the analysed fauna for substantial consumption of C4 resources.
Oxygen isotope and clumped isotope analyses provide complementary environmental information. Together, the results support humid tropical conditions and extensive forest cover during the relevant periods.
The hominin teeth offer an especially important comparison. Three Homo erectus teeth from Naxian yielded carbon isotope values consistent with consumption of resources from a closed-canopy C3 ecosystem. These results align with the broader faunal evidence.
The limitation is the sample size. Three hominin teeth cannot establish the dietary behaviour of every Homo erectus population in southern China, nor do their isotope values identify particular foods. They cannot tell us whether the individuals ate forest fruits, underground plant resources, insects, small vertebrates or larger hunted animals in any particular proportion.
What they do establish, in conjunction with the faunal assemblages and environmental reconstruction, is that these hominins obtained resources within a strongly forested ecological system. The possibility that Homo erectus was categorically excluded from dense tropical forests is difficult to maintain in the face of this evidence.
How far did they travel?
The research also examines movement through the landscape. Strontium isotope ratios in tooth enamel reflect, indirectly, the geology of the places where food and water were obtained while the enamel was forming. Different geological formations may possess distinguishable strontium signatures, allowing researchers to compare individual enamel measurements with regional environmental baselines.
The team analysed strontium and trace element distributions in several hominin teeth and developed a regional isotopic reference model. Some measurements appear inconsistent with a lifetime restricted to the immediate surroundings of the caves. The researchers propose that individuals may have travelled across substantial areas, potentially extending towards northern Vietnam and other parts of mainland Southeast Asia.
Distances approaching 250 kilometres are discussed as possibilities within the geographical modelling. Such estimates require caution. Strontium isotopes do not provide a travel itinerary, and similar geological signatures may occur in different places. The model identifies plausible geographical origins rather than proving a particular route.
Furthermore, enamel records the period of tooth formation, not an entire lifetime. Differences within or between teeth can suggest changing geographical exposure during childhood, but they do not establish whether movements were seasonal, repeated, associated with migration or undertaken for specific resources.
Even with these qualifications, the evidence invites a more expansive view of landscape use. Karst environments are not uniform barriers or isolated pockets of habitation. They contain valleys, forest corridors, water sources and ecological transitions that may have enabled hominins to exploit resources across broad territories.
The archaeological record does not yet reveal how these movements were organised. There is no basis for reconstructing particular social groups, exchange systems or planned migration routes from the isotope measurements alone.
Was Homo erectus really a savanna specialist?
Homo erectus has long occupied a central place in accounts of human evolution. Appearing in the fossil record around 1.9 million years ago, the species or closely related forms dispersed widely across Africa and Eurasia. Its archaeological associations include stone-tool technologies, substantial geographical expansion and occupation of varied environments over an exceptionally long period.
Yet popular reconstructions have often portrayed Homo erectus as particularly associated with open grasslands and woodland mosaics. Some of this emphasis derives from major Javanese fossil localities, including Trinil, Sangiran and Ngandong, where environmental reconstructions have highlighted relatively open habitats.
These associations were never sufficient to demonstrate that the species was biologically incapable of using dense forest. A fossil population occupying one type of environment does not establish the ecological limits of the entire species. Nevertheless, the scarcity of securely identified Homo erectus remains from tropical rainforest settings encouraged the suggestion that closed-canopy forests formed a significant ecological barrier.
Such habitats can present genuine challenges to terrestrial foragers. Dense vegetation restricts visibility and movement, while edible resources may be unevenly distributed in space and time. Large terrestrial herbivores may be less accessible than in open environments, and forest food webs require knowledge of resources that are not necessarily obvious from the ground.
These difficulties should not be exaggerated into universal rules. Rainforests are heterogeneous landscapes, with rivers, gaps, slopes, edges and seasonal variation. Their resources are diverse, and the strategies required to exploit them may differ substantially between regions.
Evidence for Homo sapiens exploiting tropical forests has played an important role in discussions of modern human ecological flexibility. The Guangxi results complicate any interpretation that treats rainforest use as an exclusive behavioural achievement of our own species.
They do not, however, demonstrate that Homo erectus possessed the same technologies, social organisation or subsistence strategies as later Homo sapiens. Ecological occupation and behavioural equivalence are separate propositions. The fossils establish participation in rainforest food webs, not the particular cultural mechanisms that made that participation possible.
Survival in a crowded human landscape
The chronology of the Guangxi fossils places them within a period of considerable hominin diversity in Asia. By the later Middle Pleistocene, populations representing different evolutionary lineages occupied parts of eastern Eurasia. Fossil and molecular discoveries associated with Denisovans and other archaic humans have made the regional evolutionary picture increasingly complicated.
Potential chronological overlap is particularly relevant because Homo erectus is often discussed as though its history in Asia were a relatively straightforward sequence ending before the appearance of other populations. The new evidence indicates that the situation on the mainland was less easily reduced to such a succession.
However, contemporaneity at a broad regional scale does not demonstrate interaction. The Guangxi fossils provide no direct evidence of encounters, interbreeding, competition or cultural exchange between Homo erectus and other hominin lineages.
The taxonomic uncertainty surrounding the ninth hominin tooth also illustrates the need for restraint. A specimen that cannot be confidently assigned to Homo erectus should not automatically be attributed to Denisovans. Establishing such relationships requires diagnostic anatomical or molecular evidence, and isolated teeth can remain ambiguous even after detailed analysis.
Further south, the Ngandong fossils from Java indicate that Homo erectus survived until approximately 117,000–108,000 years ago. These remains are substantially younger than the dated Guangxi deposits. Consequently, the new Chinese fossils do not represent the last known occurrence of the species globally.
Instead, they narrow an important geographical gap between better-known mainland East Asian populations and the late survivors of island Southeast Asia. Whether these populations were connected through dispersal, shared ancestry or long-term regional continuity remains unresolved.
The question is especially difficult because the fossil record is discontinuous. A long interval without identified fossils may reflect genuine population absence, but it may equally arise from preservation and sampling biases. Tropical soils and humid environments are often destructive to skeletal material, while caves preserve a selective and sometimes complicated record of the animals that entered them.
The new discoveries therefore provide evidence of survival without establishing uninterrupted regional occupation or a direct evolutionary relationship with later Javanese populations.
What the fossils cannot yet tell us
There is a temptation to move directly from evidence of rainforest occupation to a reconstruction of sophisticated forest survival strategies. That would exceed the present evidence.
The analysed teeth do not establish which stone tools were used by these individuals, whether they manufactured organic implements, how they acquired animal protein or whether fire played a regular role in their subsistence. Such questions require securely associated artefacts, activity areas, residues, cut-marked bones or other archaeological evidence.
Nor can the teeth establish population size or duration of settlement. Fossils deposited within a long chronological interval may represent occasional visits, repeated occupation or a more persistent regional population. Distinguishing these possibilities will require additional securely dated localities and better-resolved depositional histories.
The environmental reconstruction is stronger than any proposed account of specific behaviour. It rests on convergent evidence from faunal ecology, stable isotopes and geochemical measurements. The movement reconstruction is more tentative because geographical assignments depend on the resolution of the strontium reference model and the limited number of hominin specimens analysed.
There is also a distinction between exploiting rainforest resources and living exclusively within unbroken closed-canopy forest. The regional evidence supports extensive rainforest environments, but the degree to which individuals moved between forest interiors, openings, valleys and neighbouring habitats remains unknown.
These qualifications do not diminish the discovery. They identify the questions that the fossils are capable of answering and those for which the archaeological record remains incomplete.
A different account of ecological flexibility
The Baxian and Naxian discoveries establish three important points. First, Homo erectus persisted in southern China considerably later than previously documented through securely identified fossils. Second, the individuals represented by these remains exploited resources within a predominantly tropical rainforest ecosystem. Third, their dental chemistry suggests that at least some may have obtained food and water across geographically extensive areas during tooth formation.
None of these conclusions requires assuming that every Homo erectus population behaved alike. The species occupied a vast geographical range over an immense period, and its ecological capabilities probably varied across populations and landscapes.
The most productive next step is to examine comparable fossil assemblages elsewhere in southern China and Southeast Asia using similarly integrated chronological, molecular and environmental methods. Such work could establish whether the Guangxi populations were exceptional or part of a wider pattern of rainforest occupation that has remained largely invisible in the archaeological record.
For now, the evidence supports a more flexible interpretation of Homo erectus ecology than the familiar open-country model. The unresolved problem is how these populations acquired and organised the resources needed to survive in dense tropical forests, and whether that adaptability contributed to their persistence during a period of environmental and evolutionary change.
Selected Sources and Further Reading
- Westaway, K. E., Joannes-Boyau, R., Liao, W., et al. (2026). Rainforest adaptation in late Homo erectus from southern China. Nature. https://doi.org/10.1038/s41586-026-11155-6.
- Liao, W., Zanolli, C., Ásmundsdóttir, R. D., et al. (2026). Evidence for late Middle Pleistocene Homo erectus in mainland East Asia. Nature. https://doi.org/10.1038/s41586-026-11271-3.
- Chinese Academy of Sciences (2026). Hidden in the Rainforest: Fossils Reveal Last Chapter of Homo erectus in China. Chinese Academy of Sciences.
- Max Planck Institute of Geoanthropology (2026). Hidden in the rainforest: the surprising last chapter of Homo erectus in China. Max Planck Institute of Geoanthropology.
- Archaeology Magazine (2026). Homo Erectus Survived in Southern China's Tropical Rainforests. 19 November 2026. Archaeology Magazine.