Animal teeth from two caves in Guangxi carry a dense-forest carbon signal, and three Homo erectus teeth found with them plot at the forest edge, but only four hominin teeth were tested for chemistry.

Peer reviewed · Fossil and tooth-chemistry study · Small hominin sample

Homo erectus, an extinct human relative, was living among rainforest animals in southern China sometime between 324,000 and 184,000 years ago. That is the conclusion of a study published in Nature on 18 November 2026 by Kira Westaway of Macquarie University, Renaud Joannes-Boyau of Southern Cross University, Wei Liao of Shandong University and colleagues. If it holds, it makes these the youngest securely dated H. erectus fossils on mainland Asia, and it weakens the idea that only our own species could live in rainforest. The catch: only three hominin teeth were tested for diet, and four for movement.

What you need to know

  • Finding: Eight H. erectus teeth from Baxian and Naxian caves in Guangxi date to a modelled range of 324,000 to 184,000 years ago, and the animals buried with them ate only plants from forest, mostly dense canopy.
  • Method: The team modelled 43 dates from five dating techniques, measured carbon and oxygen isotopes in animal and human enamel, and traced strontium, an element that records local rock, across four hominin teeth.
  • Source: Westaway et al., “Rainforest adaptation in late Homo erectus from southern China”, Nature, 18 November 2026 (open access), with a companion paper by Liao et al. identifying the teeth.
  • Why it matters: Rainforest spread in Southeast Asia has been linked to the disappearance of other human species, so evidence that H. erectus lived there changes the story of how it died out.
  • Limitation: The rainforest case rests mainly on the animals; direct diet data come from three H. erectus teeth from one cave, and the long-distance travel idea relies on a predicted rock-chemistry map with wide error.

Evidence at a glance

Evidence summary for the late Homo erectus rainforest study from Guangxi, southern China
ItemAssessment
Evidence statusPeer reviewed. Fossil study with radiometric dating and tooth chemistry. Small hominin sample: 3 teeth for carbon isotopes, 4 for strontium.
Primary sourceK. E. Westaway, R. Joannes-Boyau, W. Liao and colleagues, “Rainforest adaptation in late Homo erectus from southern China”, Nature (2026). Received 3 January 2026, accepted 11 September 2026, published 18 November 2026. CC BY-NC-ND 4.0. Companion: Liao et al., “Evidence for late Middle Pleistocene Homo erectus in mainland East Asia”, Nature (2026).
Evidence strengthModerate for the dates, because five methods and two statistical models give overlapping ranges. Moderate for a rainforest setting, because dozens of animal teeth agree. Limited for H. erectus eating rainforest food, because only three teeth from one cave were tested. Preliminary for long-distance travel, which rests on four teeth and a modelled map.
Principal limitationThe hominin chemistry is thin: three diet measurements from Naxian only, and strontium shifts that could mean childhood travel or simply food and water from several places.
Coverage at a glance5 distinct credible publishers found as of 19 November 2026, 09:40 UTC (Nature News, Reuters, National Geographic, ABC News Australia, Sci.News), from SciQuest web searches for the paper title, cave names and topic. Reuters reprints (Yahoo, bdnews24) were counted once; the Chinese Academy of Sciences release on EurekAlert was read but not counted. Coverage is not evidence that a finding is correct.

Sources: the two Nature papers (full texts read at nature.com), the Chinese Academy of Sciences release, and SciQuest’s coverage search at the time stated.

Is this a big deal?

For the timeline, yes. The dating is careful, and it fills a gap of more than 200,000 years in the mainland record, at least at the young end of the range. For the rainforest claim, it is a solid first hint rather than a settled fact. The animals around these hominins clearly lived in closed forest, but the hominins’ own diet signal comes from three teeth and sits at the forest edge. The paper’s authors call the evidence unequivocal; on a sample this small, we would say strongly suggestive.

What scientists found

Baxian and Naxian are two small limestone caves near the city of Chongzuo in Guangxi Zhuang Autonomous Region, 15 km apart and joined by a straight valley through karst hills. Tucked in a chamber at the back of each is a hard, cemented rubble of bone. Together the two chambers have yielded more than 18,000 mammal fossils, first gathered by porcupines, which drag bones into caves to gnaw. Among them were nine hominin teeth.

The companion paper, led by Wei Liao with Frido Welker of the University of Copenhagen as a senior author, identifies eight of those teeth as H. erectus from their shape. For four of them, ancient proteins in the enamel carry an amino-acid change so far seen only in H. erectus from China. The ninth tooth, PA1796 from Baxian, matches neither H. erectus nor Neanderthals nor modern humans. The authors say it could belong to a Denisovan or to another, unknown group, and they do not settle it.

Westaway’s paper then asks when these hominins lived and in what kind of country. The answer it gives: between 324,000 and 184,000 years ago at Baxian and between 319,000 and 241,000 years ago at Naxian, in a humid tropical forest with no sign of grassland.

How they found it

Dating cave rubble is hard, so the team used five methods and checked them against each other. They dated sand grains in the rubble by the light they had stored since burial (two luminescence methods), dated fossil teeth by the decay of uranium and by a combined uranium and electron-spin method, and dated the crust of flowstone that sealed each deposit from above. Of 45 ages, 43 went into two statistical models. One, ChronoModel, gave 324,000 to 184,000 years at 95% probability; the other, OxCal, gave 312,000 to 161,000 years. The authors chose ChronoModel because it handles outlying dates better.

For the setting, they turned to carbon in tooth enamel. Trees and shrubs under a dense canopy make plant tissue with very low levels of the heavier carbon isotope, carbon-13, and grazers on tropical grass carry much higher levels. After correcting for body size, the animal teeth from Baxian averaged −27.6‰ (range −31.5 to −23.0‰) and those from Naxian −29.4‰ (range −34.0 to −25.1‰). Both averages fall at or below the −27.5‰ cut-off the authors use for dense-canopy forest, and no animal showed any sign of eating tropical grass. The three H. erectus teeth tested, all from Naxian, sit at the boundary between dense forest and more open forest. Oxygen isotopes and a temperature method based on rare isotope pairs in cave sediment point to wetter and cooler conditions than the region has today.

What the two Guangxi caves hold, and what was measured
MeasureBaxian caveNaxian cave
Modelled age of fossil layer (95% range)324,000 to 184,000 years ago319,000 to 241,000 years ago
Height above sea level218 m226 m
Animal carbon, plant-equivalent, mean (range)−27.6‰ (−31.5 to −23.0‰)−29.4‰ (−34.0 to −25.1‰)
H. erectus teeth tested for carbon03 (forest-edge values)
H. erectus teeth tested for strontium1 (0.7110 to 0.7220)3 (0.7095 to 0.7165 combined)

Source: Westaway et al., Nature (2026), main text, Methods and Extended Data Fig. 10. Carbon values in per mil (‰) relative to the VPDB standard; lower values mean denser forest, with −27.5‰ used as the dense-canopy cut-off. The cave heights are given in the paper in the order Naxian, Baxian.

Movement came from strontium. Rock releases strontium into soil, water and plants, and the ratio of two of its isotopes, strontium-87 to strontium-86, varies with the type and age of the rock. A child’s enamel records the ratio of what it ate and drank while the tooth formed, here between about 2 and 6 years of age. The team scanned along four H. erectus teeth with a laser. The Baxian tooth ran from 0.7110 to 0.7220, and the three Naxian teeth from 0.7095 to 0.7165.

Late Homo erectus in southern China: when, and how far it may have roamed Left: dated H. erectus records (thousands of years ago, ka). Right: strontium ratios in four fossil teeth compared with modelled rock signatures 1. A younger date for mainland Asia Zhoukoudian, north China: 780–400 ka Nanjing, east China: about 650 ka Hexian, east China: 410 ka Baxian cave, Guangxi (this study): 324–184 ka Naxian cave, Guangxi (this study): 319–241 ka Ngandong, Java (island): 117–108 ka 800 ka 600 400 200 today Dashed line: about 400 ka, the age of the previous youngest unequivocal mainland H. erectus fossils. Guangxi ranges are modelled 95% ranges. 2. Childhood chemistry in four teeth 0.710 0.715 0.720 0.725 Strontium ratio (87Sr/86Sr) 0.722 Baxian BAX-1 Naxian NX-A1 Naxian NX-B1 Naxian NX-B2 Measured in enamel Modelled rock signatures Caves’ area (limestone) South, incl. Red River valley West and northwest mountains Shaded band: modelled local limestone range. Values above it point farther afield, but the map’s own prediction uncertainty (0.004 to 0.016) is wide. Sources: Westaway et al., Nature (2026), doi:10.1038/s41586-026-11155-6 (main text, Extended Data Fig. 10); earlier site ages as cited in that paper. Bars show ranges, not single values. SciQuest original graphic.
Left: the Guangxi caves’ modelled age ranges sit well after the roughly 400,000-year-old fossils that were the previous youngest unequivocal H. erectus on the Asian mainland, though still older than Ngandong on the island of Java. Right: strontium ratios measured along four H. erectus teeth, against the values a computer-predicted map gives for the caves’ limestone and for more distant rocks. Only the Baxian tooth climbs well above the local range. Data from Westaway et al., Nature (2026), main text and Extended Data Fig. 10. SciQuest original graphic.

To read those numbers, the researchers built a predicted strontium map of southern China and northern Southeast Asia, using a machine-learning method trained on a global database of measured samples. Around the caves, on Carboniferous to Permian limestone, roughly 360 to 250 million years old, the map predicts 0.710 to 0.715. Older metamorphic rock in mountains to the west and northwest predicts 0.717 to 0.723, and areas to the south, including the Red River valley of northern Vietnam, 0.715 to 0.720. The highest values in the Baxian tooth fit those distant areas best, so the team suggests that this child moved between the local hills and places perhaps as far away as northern Vietnam, a range they put at potentially around 250 km.

Think of strontium as a postmark stamped into a growing tooth: a run of local postmarks followed by a foreign one suggests a trip. The comparison has limits. Many places share the same postmark, so a match narrows the options without naming one town, and food carried in from elsewhere can leave the same mark as travel.

Show the maths: how much do the dates and the strontium map really pin down?

Ages. The two caves’ ranges overlap from 319,000 to 241,000 years ago: 319,000 − 241,000 = 78,000 years, the figure the paper gives. The full Baxian range spans 324,000 − 184,000 = 140,000 years, so the study fixes a window, not a moment.

How much younger? The previous youngest unequivocal mainland fossils, from Hexian, are about 410,000 years old. At the young end of the new range, 410,000 − 184,000 = 226,000 years later; at the old end, 410,000 − 324,000 = 86,000 years later. Headlines saying “200,000 years longer” use the young end.

Strontium. The Baxian tooth’s top value, 0.7220, is 0.007 above the 0.715 ceiling of the local limestone prediction. The paper’s predicted map carries a stated uncertainty of 0.004 to 0.016 in strontium ratio, so in the less certain parts of the map that gap is within the error. That is one reason SciQuest rates the travel finding as preliminary.

Distance. For scale, Chongzuo to Hanoi is about 216 km in a straight line (SciQuest calculation from city coordinates), close to the authors’ “potentially around 250 km” range.

Why it matters to you

Nothing here changes daily life. It speaks to the “where did we come from” question: which human species could cope with which places. Rainforests are hard places to live, with scattered food and dense cover, and for years the archaeological record suggested that Homo sapiens alone used them regularly, from about 150,000 years ago in West Africa. “This is the first clear, direct evidence we have for another hominin living within these habitats,” Patrick Roberts of the Max Planck Institute of Geoanthropology, a co-author, said in the release. If that holds up, a skill we have treated as ours alone was shared.

What the evidence does not show

The study does not show what H. erectus ate in the forest, how it got food or whether rainforest was its preferred home. The authors themselves are careful on that last point: “H. erectus was, at minimum, capable of persisting among rainforest fauna, even if such habitats did not represent its optimal ecological niche.”

The hominin data are thin. Carbon isotopes come from three teeth, all from Naxian, and the companion paper notes that two Naxian molars may come from one person. Strontium comes from four teeth. The paper notes that the shifting ratios could mean either travel between rock types or food from several places, and the predicted map has wide error in parts of the region.

The teeth also did not arrive in the caves with their owners. Porcupines collected the bones, so the fossils record the animals living within a porcupine’s foraging range, not a campsite. The paper reports no stone tools from these deposits. And the ages are ranges: the hominins lived sometime in a 140,000-year window, not necessarily at its young end.

How this fits what we already knew

H. erectus first appears in Africa about 1.9 million years ago and reached China early: the paper lists Yuanmou at 1.7 million years and Zhoukoudian, home of Peking Man, at 780,000 to 400,000 years. On the mainland, though, securely identified fossils younger than about 400,000 years were missing. The youngest known members of the species lived on Java at Ngandong, 117,000 to 108,000 years ago, in open woodland as rainforest spread around them, and some researchers have argued that the spreading forest helped finish them off. The companion paper notes that teeth from Changyang and Chaoxian caves in China, attributed to H. erectus, have ages of about 196,000 to 143,000 and 200,000 to 160,000 years, but these are minimum ages, so the fossils could be much older.

The Guangxi animal signal looks less like the grassy settings of Trinil and Sangiran on Java and more like Lida Ajer on Sumatra, a dense-rainforest site where Homo sapiens lived 73,000 to 63,000 years ago. The new dates also put H. erectus in the same broad region and era as Denisovans, known from a molar in Cobra Cave in Laos and a jaw recovered from the seabed of the Penghu Channel off Taiwan. We covered how ancient proteins are being used to sort these fossils in our story on the Xujiayao Denisovan proteome.

What outside researchers said

SciQuest did not interview outside researchers for this article. Others did. Julien Louys, a palaeontologist at Griffith University who was not involved, told ABC News Australia that this is the first evidence of H. erectus in rainforest but stressed that the sample is small and the ecological flexibility needs to be substantiated. Jesse Martin of La Trobe University told ABC that the result is not surprising given how long the species lasted on Java. In National Geographic, Richard Potts of the Smithsonian’s Human Origins Program judged the identification of the teeth convincing, and Ben Marwick of the University of Washington said the idea that only Homo sapiens could live in rainforest may not hold. These views are reported second-hand; we have not checked them against transcripts.

Nature lists four peer reviewers for the paper, including Clement Bataille, who works on strontium mapping, and the authors built their map from a global strontium database he co-authored.

How other headlines framed it

Most coverage stressed survival and skill. Reuters said the species “mastered” the rainforest, and Nature‘s news team wrote that the two papers point to it having “thrived” there, which goes beyond what three diet measurements can show. National Geographic led on the timeline, saying the species may have survived 200,000 years longer than thought; that figure holds only at the young end of the date range. ABC News Australia framed it as evidence that more than one human species lived in rainforests and gave the most room to the small-sample caveat. Sci.News ran a “thrived … far later than thought” headline. The Chinese Academy of Sciences release says the teeth were identified partly by “ancient genetic signatures”; strictly, the evidence is protein sequences from enamel, which reflect genes but are not DNA.

What happens next

The authors call for stable-isotope work at other Middle Pleistocene sites to test whether rainforest living was special to these southern populations or common to the species. More hominin teeth from Baxian and Naxian, measured for both carbon and strontium, would test the thin parts of this study directly. The identity of the ninth tooth waits on more Denisovan reference proteins and genomes. The study’s data are public on Zenodo, so other groups can rerun the age models.

In one sentence

Fossil teeth from two Guangxi caves suggest Homo erectus lived among rainforest animals in southern China between about 324,000 and 184,000 years ago, much later than other well-dated mainland fossils, though the direct chemistry of the hominins themselves rests on a handful of teeth.

Quick check: did it click?
  1. How many hominin teeth were tested for carbon isotopes? Three, all from Naxian cave. Most of the rainforest evidence comes from dozens of animal teeth.
  2. Why does strontium in enamel hint at movement? It reflects the rock under the food and water a child consumed while the tooth grew; a shift in the ratio suggests a change of place, or food from several places.
  3. Was every one of the nine teeth H. erectus? No. Eight were assigned to H. erectus; one, PA1796, matches no known group and might be Denisovan.

Read next

Sources and disclosure

  1. Westaway, K. E., Joannes-Boyau, R., Liao, W. et al. Rainforest adaptation in late Homo erectus from southern China. Nature (2026). Data: Zenodo.
  2. Liao, W., Zanolli, C., Ásmundsdóttir, R. D. et al. Evidence for late Middle Pleistocene Homo erectus in mainland East Asia. Nature (2026).
  3. Chinese Academy of Sciences, “Hidden in the rainforest: Fossils reveal last chapter of homo erectus in China”, EurekAlert, 18 November 2026.
  4. Rizal, Y. et al. Last appearance of Homo erectus at Ngandong, Java, 117,000–108,000 years ago. Nature 577, 381–385 (2020).

Disclosure: This article was drafted with AI assistance and checked against the sources above. The age differences, the strontium gap and the Chongzuo to Hanoi distance are SciQuest calculations; the graphic is a SciQuest original based on values in the Westaway paper. Comments from outside researchers are reported from ABC News Australia and National Geographic. The study was funded by the Australian Research Council, the Chinese Academy of Sciences and other agencies, and the authors declare no competing interests. Spotted an error? See our corrections policy.


Discover more from SciQuest

Subscribe to get the latest posts sent to your email.