Age control of the first appearance datum for Javanese Homo erectus in the Sangiran area

See allHide authors and affiliations

Science  10 Jan 2020:
Vol. 367, Issue 6474, pp. 210-214
DOI: 10.1126/science.aau8556

Dating the arrival of the first hominins in Java

The World Heritage archaeological site at Sangiran on the island of Java in Indonesia has major importance for the understanding of human arrival and evolution in Asia. However, the timing of the first appearance of Homo erectus at the site has been controversial. Using a combination of dating techniques for hominin-bearing sediments, Matsu'ura et al. resolved the arrival of H. erectus at ∼1.3 million years ago (see the Perspective by Brasseur). This dating suggests that the earliest hominins in Sangiran are at least 200,000 years younger than has been thought and may represent an important step to the resolution of the controversy.

Science, this issue p. 210; see also p. 147


The chronology of the World Heritage Site of Sangiran in Indonesia is crucial for the understanding of human dispersals and settlement in Asia in the Early Pleistocene (before 780,000 years ago). It has been controversial, however, especially regarding the timing of the earliest hominin migration into the Sangiran region. We use a method of combining fission-track and uranium-lead dating and present key ages to calibrate the lower (older) Sangiran hominin-bearing horizons. We conclude that the first appearance datum for the Sangiran hominins is most likely ~1.3 million years ago and less than 1.5 million years ago, which is markedly later than the dates that have been widely accepted for the past two decades.

The first appearance datum of Homo erectus in the Far East, a hominin species that originated in equatorial Africa or perhaps the Caucasus region of Eurasia (13), is important for modeling early human dispersals across Asia after the first out-of-Africa migration. Since the mid-1990s, researchers have believed that eastern Asia’s oldest hominin remains date back to 1.7 to 1.8 million years ago (Ma) (47), about half a million years earlier than previously thought (811). Despite uncertainties (12), this chronology and its consequences have been widely recognized as one of paleoanthropology's basic paradigms: H. erectus expanded rapidly to eastern Asia (1, 13) after its first appearance at ~1.85 Ma. Furthermore, combined with chronological revisions of H. erectus sites in China, different dispersal dynamics have been suggested for southeastern and northeastern Asia (14, 15): an earlier (>1.5 Ma) H. erectus dispersal along a southern route to equatorial southeast Asia and a later (after ~1.3 Ma) dispersing population along a northerly route to northeast Asia. Although the hypothesis of a later dispersal to northeastern Asia disagrees with current Chinese hominin fossil chronologies (6, 7), an early dispersal into southeast Asia has been based on evidence from the Sangiran area (Indonesia), where, on the basis of 40Ar/39Ar geochronology, it has been proposed that H. erectus dates to ~1.6 Ma (5). However, other 40Ar/39Ar ages (16, 17) and a newly refined magnetic polarity stratigraphy of the Sangiran area (18) have failed to support this chronology. This calls for reevaluation of the 1.6 Ma Sangiran H. erectus age. We report U-Pb and fission-track dates of volcaniclastic layers in and just above the Grenzbank zone, a key bed of the lower Sangiran hominin-bearing horizons. We also present dates of a marker tuff lying below the hominin-bearing horizons. Our results provide reliable age control points to infer the first appearance datum for Javanese H. erectus in Sangiran.

Sangiran in Central Java (Fig. 1A) is one of the most productive sites in paleoanthropology. It has produced a steady stream of H. erectus finds since 1936, now totaling >100 hominin specimens. However, it lacks an accepted chronostratigraphy. Geologically, the Sangiran area is a domelike structure, extending 8 km north–south and 4 km east–west (Fig. 1B). The dome is truncated by erosion, exposing a concentric pattern of strata, with older strata surrounded by younger strata (19) (Fig. 1B). The exposed sediments, >300 m thick, are divided into four units (19). The marine Puren Formation lies at the base (Fig. 1C). The overlying Sangiran Formation consists of the Lower Lahar (volcanic breccia) at its lowermost part and the black clay. The latter comprises shallow marine to lagoonal sediments and overlying lacustrine sediments that are partly pedogenized. The Sangiran Formation is overlain by the Bapang Formation, the base of which is known as the Grenzbank zone—a fossiliferous unit of carbonate-cemented gravelly sands. The Bapang Formation is further overlain by the Pohjajar Formation (Fig. 1C). The upper two formations are primarily fluviatile sediments intercalating many layers of pumice, volcanic ash, and lahar.

Fig. 1 Sangiran location and stratigraphy.

(A) Location of Sangiran. (B) Geological map of the Sangiran area (after 19). (C) Generalized stratigraphy of the Sangiran area (after 19, 20, 25, 32). In (B), a thick red arrow (S48) and a short red line (V-1) show the columnar section localities involved with samples for dating. Crosses show H. erectus fossil sites, some of which are indicated with registered designations: mandible Pithecanthropus B (Pb), mandible Pithecanthropus C (Pc), mandible Pithecanthropus E (Pe), mandible Pithecanthropus F (Pf), mandible Meganthropus B (Mb), calotte Pithecanthropus II (P II), calotte Pithecanthropus III (P III), cranium Pithecanthropus IV (P IV), calotte Pithecanthropus VI (P VI), calotte Pithecanthropus VII (P VII), and cranium Pithecanthropus VIII (P VIII). F., Formation.

Previous lithostratigraphic and geochemical investigations have revealed that the hominin-bearing horizons of the Sangiran area could range from the Upper Tuff of the Bapang Formation down to Tuff 11 of the uppermost Sangiran Formation (19) or plausibly to the lower levels in the Upper Sangiran Formation (8, 20) (Fig. 1C). A systematic investigation of the chronostratigraphy of this sequence was first conducted in the late 1970s to early 1990s, including fission-track dating and magnetostratigraphic analysis, which suggested a hominin time span of ~0.8 to 1.1 Ma (10, 19) or possibly to ~1.3 Ma (8). However, in 1994, a substantially older 40Ar/39Ar date of 1.66 Ma was reported for pumices presumed to overlie the two allegedly oldest hominin remains (4). Although this date has been questioned because of the uncertain stratigraphic relationship between the dated pumice and the hominin specimens (5, 21, 22), Larick et al. (5) subsequently reported a series of hornblende 40Ar/39Ar dates that supported the older chronology and placed the Sangiran hominin-bearing sequence from >1.5 to ~1.0 Ma. The widely cited age of >1.5 Ma derives from a date of 1.51 ± 0.08 Ma (5) for a pumice lens lying a few meters above the Grenzbank zone at the base of the Bapang Formation (Fig. 1C and Fig. 2). However, other pumice and tuff samples collected from the lowermost part of the Bapang Formation have produced a group of notably younger hornblende 40Ar/39Ar dates of 0.8 to 0.9 Ma (16, 17). One of the samples yielded a date of ~1.5 Ma; however, this has been provisionally considered to relate to natural reworking (17). These 0.8 to 0.9 Ma dates—centering around 0.88 Ma—are, in turn, consistent with the newly refined magnetostratigraphy (18), which securely establishes the Matuyama-Brunhes polarity transition within the Upper Tuff complex of the Bapang Formation and constrains the age of the uppermost hominin-bearing sediments to ~0.79 Ma. Thus, the controversy over the long (older) and short (younger) chronologies for the Sangiran hominins is ongoing (15, 18, 23). Further studies are needed to resolve this issue and to better understand the biogeographic origins and evolution of Javanese H. erectus.

Fig. 2 Geological columnar sections showing the stratigraphic levels of the tephra samples treated in this study (BP-LMTCL-1, BP-GB-BC-P1, and Pb-T8).

For the locations and photographs of the sampling sites, see fig. S1. Note that the U-Pb age represents an average of timing of zircon crystallization, which commonly occurs substantially before eruption, and that the deposition time of the unit layer is younger than the U-Pb age (24) (see Fig. 3). FT, fission-track.

To solve this long-standing chronological controversy, we applied a method of combining fission-track and U-Pb dating on zircon grain populations taken from the lowermost Bapang section—which directly contains the former sampling location of the 40Ar/39Ar 1.51 ± 0.08 Ma (5) sample—and also on zircon grains taken from a tuff that could potentially provide a maximum age for the Sangiran hominins. Zircon is commonly observed in volcanic rocks and is a markedly robust mineral. The U-Pb zircon method has been recently improved in dating young Quaternary samples (24). The U-Pb and fission-track methods are complementary because they can be applied to the same zircon grain population. Note that closure temperatures for fission-track and U-Pb chronometry on zircons are ∼240°C and >900°C, respectively (24). Because zircon variably crystallizes before eruption (24), U-Pb ages indicate the timing of crystallization, whereas the fission-track age closely reflects the timing of eruption.

We collected samples for dating from two tephra horizons at the type site of the middle to lower levels of the Bapang Formation near the Bapang Village [columnar section site S48 (Fig. 1B); fig. S1, A and G]. Tuff sample BP-LMTCL-1 was collected from the yellowish white tuffaceous sand layer immediately above the Grenzbank zone, whereas pumice sample BP-GB-BC-P1 was taken from the bottom part of the bluish gray clay layer intercalated in the Grenzbank zone (Fig. 2 and fig. S1, A to D). We also collected sample Pb-T8 from the yellowish white tuff (Tuff 8) at the V-1 section site (Fig. 1B and fig. S1, E to G). Tuff 8 is a marked tephra that lies slightly below the base of the Upper Sangiran Formation (Fig. 1C and Fig. 2), i.e., stratigraphically underlying the hominin-bearing sediments.

Analytical results are summarized in Fig. 3. Tuff sample BP-LMTCL-1 yielded numerous colorless and euhedral zircon grains that suggest proximal magmatic sources. To obtain a robust fission-track date from younger zircons with low spontaneous track density (table S1), we measured >1600 grains and calculated cumulative grain plateau ages (fig. S2). The weighted mean of two plateau ages was 0.884 ± 0.031 Ma (table S1). As mentioned earlier, it is necessary to consider the time difference of crystallization and eruption. Zircons usually crystallize over an extended period of time in a magma reservoir, and individual crystals potentially have various, prolonged durations of growth (24). The U-Pb age distribution of crystal faces of 75 grains from sample BP-LMTCL-1 is shown in Fig. 3, corresponding to crystallization before and around eruption, with a weighted average age of 0.991 ± 0.005 Ma. The lower tail (younger ages) of the U-Pb age distribution is consistent with the fission-track date of 0.884 ± 0.031 Ma, supporting the interpretation that the lower end of the U-Pb age range approximates the eruption age. These results for the tuff that lies just above the Grenzbank zone demonstrate that the Grenzbank zone is close to 0.9 Ma.

Fig. 3 Zircon fission-track age for BP-LMTCL-1 and Pb-T8 and the U-Pb age results of zircon grains for BP-LMTCL-1, BP-GB-BC-P1, and Pb-T8.

Data for the fission-track age results are available in table S1, and data for the U-Pb age results are in table S5. The closure temperature (~240°C) of zircon fission-track chronometry is much lower than that (>900°C) of U-Pb systematics. The weighted mean of grain U-Pb ages signifies an average of timing of crystallization which variably occurs before eruption (24), where the fission-track date should represent the eruption age. Note that the 2σ ranges of the weighted mean U-Pb ages of BP-LMTCL-1 and BP-GB-BC-P1 overlap considerably, and the two groups of grain U-Pb ages show statistically comparable overall crystallization distributions (P = 0.483, two-sample, two-sided Kolmogorov-Smirnov test). Wtd., weighted; n, number of grains; MSWD, mean square weighted deviation.

As for the pumice sample BP-GB-BC-P1, taken from within the Grenzbank zone, the number of zircon grains was comparatively fewer and reliable fission-track dating was not possible. We note that the zircon grains included some reddish detrital ones that dated from ~4 to ~90 Ma (24). After excluding xenocrystals, weighted averaging of the U-Pb ages from the zircon crystal faces provided a 0.971 ± 0.009 Ma date (Fig. 3). Notably, the U-Pb age distributions of the two tephra units (BP-GB-BC-P1 and BP-LMTCL-1) correspond well with each other (Fig. 3), suggesting similar crystallization and eruption times, which demonstrates that the Grenzbank zone is close to 0.9 Ma. This is consistent with the recently confirmed absence of the Jaramillo subchronozone in the Bapang Formation (18), indicating an age younger than 0.99 Ma.

Sample Pb-T8 from Tuff 8, located several meters stratigraphically below the base of the Upper Sangiran Formation (which plausibly corresponds to the first appearance datum for the Sangiran hominins), yielded zircons suggesting proximal magmatic sources (24). Although the number of grains was comparatively fewer, higher spontaneous track density (higher uranium content) allowed reliable fission-track dating, producing a plateau age of 1.345 ± 0.108 Ma (table S1). We consider this date to represent the eruption age of Tuff 8. In our U-Pb dating of the same zircon assemblage, we aimed to estimate a maximum limit of the eruption age and hence analyzed the inner crystal zones, which likely represent pre-eruptive crystallization regions (24). The weighted mean of 1.688 ± 0.010 Ma for the grain U-Pb ages reflects an average timing of crystallization before eruption. The lowermost (youngest) tail of the U-Pb age distribution overlaps with the +2σ range (1.56 Ma) of the Tuff 8 fission-track date (Fig. 3), implying that the maximum possible age of Tuff 8 is ~1.55 Ma.

The Grenzbank zone, located at the base of the Bapang Formation, is a key bed of the Sangiran hominin-bearing horizons. The above results for the Grenzbank zone (~0.9 Ma) contradict the widely cited bulk hornblende sample 40Ar/39Ar date of 1.51 Ma (5) for the pumice-rich layer lying a few meters above the Grenzbank zone (Fig. 2). They are consistent with previous 40Ar/39Ar determinations, centering around 0.88 Ma, on single hornblende grains (16, 17) of pumice and tuff sampled at another locality of the lowermost Bapang Formation. The Larick et al.’s (5) notably older 40Ar/39Ar date may be provisionally considered to relate to inherited hornblende grains or possible reworking of epiclastic pumice balls.

The majority of the Sangiran hominin fossils were found by chance by local inhabitants, sometimes recovered without precise provenance. Despite this problem, geological and geochemical surveys (8, 19, 20), as mentioned earlier, have shown that the hominin specimens could derive from sediments between the Upper Tuff of the Bapang Formation and the upper part of the Sangiran Formation (Fig. 1C). The Sangiran H. erectus materials, hitherto variously referred to as Pithecanthropus and Meganthropus, are tentatively divided into chronologically older and younger groups in light of paleontologic and stratigraphic contexts (8, 2527).

The chronologically younger group includes crania Pithecanthropus III (Sangiran 3), VI (Sangiran 10), VII (Sangiran 12), and VIII (Sangiran 17); Skull IX; and mandibles Sb 8103 and Ng 8503. This group is derived from the Bapang Formation above the Grenzbank zone, mainly from sediments near the Middle Tuff (Fig. 1C) (8, 19), and it is associated with the Kedung Brubus fauna (25). Currently, the uppermost datum of the Sangiran hominin fossils has been stratigraphically placed at just below the Upper Middle Tuff (18) (Fig. 2); this stratigraphic level lies a few meters below the Matuyama-Brunhes transition and is dated to ~0.79 Ma (18).

The chronologically older group includes crania Pithecanthropus II (Sangiran 2) and IV (Sangiran 4); crania Sangiran 27 and 31; frontal Bp 9408; maxilla Bpg 2001.04; mandibles Pithecanthropus B (Sangiran 1b), C (Sangiran 9), and F (Sangiran 22); mandibles Bk7905 and Bk8606; and mandible Meganthropus B (Sangiran 8). The provenience of this group, based on hominin remains with known stratigraphic positions, ranges from the Grenzbank zone (the basal layer of the Bapang Formation, now dated to ~0.9 Ma) to Tuff 11 of the uppermost Sangiran Formation, but it plausibly goes down to lower levels in the Upper Sangiran Formation (8, 19, 20). This group is associated with the Trinil H.K. (Trinil Haupt Knochenschicht) fauna and the Ci Saat fauna (25) (Fig. 1C). Magnetostratigraphy (10, 18) has suggested that the Jaramillo subchronozone lies in the uppermost Sangiran Formation with the lower boundary of the zone (1.07 Ma) near Tuff 11. We estimate the base of the Upper Sangiran Formation—i.e., the plausible first appearance datum (FAD) for the Sangiran hominins (20)—as follows. Using the chronological brackets of 0.884 Ma (the tuff immediately overlying the Grenzbank zone) and 1.345 Ma (Tuff 8), the base of the Upper Sangiran Formation (Fig. 1C and Fig. 2) is estimated to be 1.27 Ma, assuming a constant depositional rate during this stratigraphic interval. This age estimate is also concordant with the stratigraphic position of Tuff 8 relative to the age of the Lower Lahar lying at the basal Sangiran Formation (Fig. 1C). The latter has been dated at ~1.7 Ma by 40Ar/39Ar dating of single hornblende grains (22) or ~1.9 Ma by 40Ar/39Ar dating of bulk sample hornblendes (28). Considering that the maximum possible age of Tuff 8 is ~1.55 Ma, the FAD for the Sangiran hominins is best considered to be either ~1.27 or <1.45 Ma.

Some consequences of the short chronology supported by this work include paleoenvironmental context that relates to adaptation and evolution of Javanese H. erectus. The Sangiran hominins have been shown to comprise two morphologically distinguishable groups—in mandibular, dental, and cranial morphology (23, 26, 27)—with the top of the Grenzbank zone being the stratigraphic and temporal boundary. The older hominin group, although highly variable, displays relatively primitive features that are comparable in morphology to the 1.4 to 1.7 Ma African H. erectus (Homo ergaster) (27). The younger hominin group is comparatively advanced, showing a larger neurocranial size and a degree of dentognathic reduction comparable to Middle Pleistocene Chinese H. erectus (23).

It is well known that the global climate underwent a fundamental change between ~1.2 and 0.7 Ma, the mid-Pleistocene transition (MPT), when there was a shift from low-amplitude 41-kyr (thousand years) to high-amplitude 100-kyr climate cycles accompanied by more intensive global glaciation and exposure of continental shelf (29, 30). The MPT is characterized by the first major cooling phase at about 900 kyr ago, associated with a rapid increase in global ice volume in marine isotope stage (MIS) 22 (29, 30), initiating successive major glaciations. These climatic dynamics profoundly affected the biota and environment. Our estimated age of the Grenzbank zone is approximately at the onset of MIS 22 (~900 to 866 kyr ago). It is therefore possible that the morphological changes observed between the older and younger Sangiran hominin assemblages are associated with some drastic change induced by MIS 22. Although in situ dynamic microevolution within the same H. erectus population is conceivable (23), a new immigration event might have been the dominant source of the observed morphological changes. MIS 22 marks the first time in the Pleistocene that the sea level dropped ~120 m below the present level (30) and exposed the Sunda shelf around the Indonesian archipelago more widely than before, forming a large landmass. The Kedung Brubus fauna, associated with the younger group of Sangiran hominins, suggests massive interchange with the Asian continent (25). To the contrary, the Ci Saat and Trinil H.K. faunae associated with the older hominin group are more insular in character, albeit indicating some connection with the mainland.

Concerning the FAD of H. erectus in the Sangiran area, our results provide a probable FAD of ~1.3 Ma and a maximum possible FAD of ~1.45 Ma. Another hominin specimen that has been contended to be the earliest Javanese H. erectus is the Mojokerto skull from the Perning site in East Java. This skull is now concluded to be less than ~1.49 Ma on the basis of fission-track age determinations (31). Thus, the hominin dispersal into Java is resolved to be <1.5 Ma. The comparative primitive morphology of the Javanese H. erectus of the older chronological group may represent either primitive retentions (27) or derived features independently acquired in this hominin lineage.

In conclusion, our results provide important evidence that supports the short (younger) chronology. The Grenzbank zone is securely anchored at ~0.9 Ma, and our best estimate for the first hominin colonization into the Sangiran area is ~1.3 Ma (or <1.5 Ma), both much later than the estimate that has been widely accepted for more than two decades.

Supplementary Materials

Materials and Methods

Figs. S1 to S5

Tables S1 to S5

References (3388)

References and Notes

  1. Materials and methods are available as supplementary materials.
Acknowledgments: This chronological research was done under the auspices of the Centre for Geological Survey (CGS, formerly Geological Research and Development Centre) in Bandung, Indonesia. We thank the successive directors of CGS for continued support and courtesy lasting more than 25 years. This joint research project was conducted partly under the general agreement of cooperation between CGS and the National Museum of Nature and Science, Tokyo (NMNS). We thank H. Baba and Y. Kaifu of NMNS for consideration and help. Special thanks to the late Sudijono of CGS for long-term collaboration since the 1970s. We owe a very important debt to the late H. Kumai of Osaka City University, without whose enormous amount of work and instruction this project would not have been possible. We are also grateful to N. Watanabe (deceased) and D. Kadar, who made a beginning of Indonesia-Japan research cooperation in this field. Funding: This study was partially supported by the Japan Society for the Promotion of Science (grants 13440255, 15403015, 18200053, and 22320154) and by NMNS (fund no. 6503). Author contributions: S.M. and F.A. planned and coordinated the original chronological project with I.Ku., M.H., and M.K.; T.D., T.H., M.K., F.A., and S.M. planned dating analyses; F.A., I.Ku., M.H., M.K., Y.T., E.S., I.Ki., H.I., S.S., T.D., T.H., M.S., Y.D., and S.M. performed research; T.D., H.I., S.S., M.K., Y.D., T.H., and M.S. undertook dating experiments; and S.M. prepared the manuscript, with input from H.I., S.S., T.D., Y.T., and M.H., including tables and figures, with additional contributions from other authors. Competing interests: The authors declare no competing interests. Data and materials availability: The data that support the findings of this study are available in the paper and supplementary materials.
View Abstract

Stay Connected to Science

Navigate This Article