Southeast Asia is of growing importance in the study of hominin evolution and migration. It was home to at least four hominin species identified from fossil material (Homo erectus, H. floresiensis, H. luzonensis, H. sapiens) as well as several hominid genera including Pongo, Meganthropus, and Gigantopithecus (Kaifu, 2017; Détroit et al. 2019; Welker et al. 2019; Zanolli et al. 2019). Major climatic and geographical changes over the Middle to Late Pleistocene saw the subsidence of the Sunda shelf as well as important shifts in the location and intensity of the Asian monsoon system (Meckler et al. 2012; Cheng et al. 2016; Sarr et al. 2019). These changes are thought to have precipitated the loss of savannahs and the increased dominance of rainforest environments, with deleterious effects on some hominids and megafauna (Louys et al. 2007; Louys, 2008; Louys and Roberts, 2020).
However, these environmental changes are likely to have been heterogeneous across time and space, with rainforest persisting throughout the Pleistocene, either in refugia or in major corridors, allowing the persistence of gene flow among rainforest species (e.g. Mason et al. 2019). Sumatra is a key region for identifying Pleistocene rainforests, as most of the island is thought to have remained humid throughout the Pleistocene (Laumier 1997). A key source of such palaeoenvironmental information, particularly as it pertains to hominins, are mammal fossils. Non-hominin mammal remains are ubiquitous in fossil deposits pertinent to human evolution and are commonly used to extract key palaeoecological and palaeohabitat information (e.g. Le Fur et al. 2009; Bedaso et al. 2013; Louys et al. 2015; López-García et al. 2019). The environmental reconstruction of sites dated to before the arrival of hominins provides important contextual information regarding potential migration and colonisation routes, as well as the impacts of hominins on ecosystems (Price and Webb, 2006; Louys and Turner, 2012; Potts and Teague, 2010; Martinez-Navarro 2010; Petraglia, 2017). However, the use of mammals as palaeoenvironmental indicators requires good chronological information and resolution of the sites in which they are found.
The palaeontological history of Sumatra has been the subject of renewed interest in recent years, in part driven by re-examination of fossil material originally collected by M. Eugène F.T. Dubois in the late 1800s from caves in the Padang Highlands, western Sumatra (e.g. Westaway et al. 2017, Volmer et al. 2017, Wirkner and Hertler, 2019; Louys et al., 2021a). Most attention has been paid to three caves in the region: Lida Ajer, Simbrambang, and Djamboe, as the bulk of the Dubois Sumatran fossils are derived from these sites (Hooijer, 1946, 1947, Drawhorn 1995). However, in addition to these sites, mention in the Dubois reports is also made of several other caves in the region, including Ngalau Pandjang, Ngalau Mansioe, Ngalau Batang Pagian, Ngalau Moeka Moeka, and Ngalau Sampit, to name a few (Anonymus, 1889–1890 cited in Hooijer, 1946).
Here, we examine the fossil-bearing breccias of Ngalau Sampit, not only one of Dubois’s sites mentioned in his field notes, but also one of the few Sumatran caves with a preserved fossil assemblage (Louys et al., 2017). A small amount of fossil material was initially collected from this site by Louys et al. (2017), who also provided brief notes on the age of the deposit and its formation history. Following this preliminary evaluation, another fieldwork campaign was subsequently organised in 2018 to get a better understanding of the site formation processes and complement the existing faunal list. In the present work, we date newly collected fossil material and associated sediments using three independent numerical methods, Electron Spin Resonance (ESR), U-series and Luminescence, in order to refine the age of the deposit as well as provide further inferences regarding its depositional history. Ngalau Sampit represents only the third site from the Pandag Highlands to be radiometrically dated, after Lida Ajer (Westaway et al., 2017) and Ngalau Gupin (Smith et al., 2021a), and only the second site explored and recorded by Dubois to have associated dates.
Ngalau Sampit is a relatively large cave (GPS Coordinates: S00°15’38.0” E100°36’24.3”) situated on the outskirts of Payakumbuh (Figure 1), about 70 km from the western coast of Sumatra. Included within the list of cave sites originally visited by Dubois, Ngalau Sampit was more recently partially surveyed and mapped by Louys et al (2017) (Figure 2). It is a multichambered cave system with at least three entrances observed. The most northern entrance can be found extending from a rock shelter which was in use by local villagers at the time of initial survey (Figure 3A). A small tunnel measuring approximately 0.5 × 0.5 m and extending in a northern direction for 5 m opens into a small chamber measuring 1.0 × 1.3 m. Two phreatic tunnels extend out of this chamber, one blind extending southwest, the second north for a short distance and connecting the first chamber to a larger chamber measuring 4.5 m × 2.8 m. A narrow tunnel extends at 200° from this second chamber, doglegging to 320° and expanding in height from 1 m to 8.5 m at its highest extent, before a 0.5 m restriction 12.7 m along the passage. The tunnel continues for a further 6 m before opening into a very large chamber. This is bounded in the east by partially submerged phreatic chambers and tunnels. These were only partly explored; besides further chambers, they also revealed a more southerly entrance with concrete stairs now underwater, but no breccia deposits were identified. A 2 m wide passage northwest of the submerged chambers extends for 2.5 m before opening into the main chamber of the cave, marked A in Figure 2. This is bounded to the north by the fossil-bearing passages of the cave. This was divided into two chambers during survey, Chamber 1 the eastern passage, that tapers quickly into a restriction (B-C in Figure 2), and Chamber 2, the north-western passage, which opens up into a 3.0 × 1.7 m chamber, 2 m high at its maximum extent (B-D-E in Figure 2). Sedimentary infilling of these passages is indicated by remnants of lithified and semi-lithified breccia deposits made of high-density speleothem matrix with angular allogenic clasts, positioned against the walls and ceilings of the NW and E passages (Figures 2 and 4). Capping and interstitial flowstones are observed at both ends of the passages, with Chamber 2 also preserving large cascade stalactites draping over most of the breccia. The breccia deposits host fossil remains (mostly isolated teeth), from large mammals. Some of these have visibly eroded from the cave walls and roof and been redeposited on the cave floor. A further sinkhole opening was recorded at the south-eastern end of the cave complex, but no additional breccia deposits were observed.
Ngalau Sampit (“scary cave”) was one of the first caves visited by Dubois in June 1888, but he only started to excavate there on 25 February 1889 after Ngalau Lida Ajer had largely been exhausted (Dubois 1889–1890; 40–447; 40–457) There are some fieldnotes (see Figure 3), but the most detailed report Dubois gives is in his hitherto unpublished monthly report (written in Dutch) of April 1889 (Dubois 1889–1890 [submitted 6 May 1889]; 50–011), which we present here, translated into English, in full (when needed, some clarifications are given between [ ]):
“Short overview of the results that were obtained during the month of April for the palaeontological research at Sumatra’s west coast.
At a distance of about 3 paal from Pajakombo, south of the road to Fort de Kock opposite the well-known cave of Pajakombo an opening can be seen in a picturesquely shaped and draped rock, which gives access to the ngalau sampit [sic] (scary cave). The name has probably to do with the narrowness of its entrance, while the rest is no more narrow than any other cave. The entrances turn to north-west, whilst it is only about 30 m above the bottom of the valley and easily accessible along a not very steep slope overgrown with Alang Alang.
From an earlier examination I did it had already become clear, that only the part close to the entrance and a small terrace were suitable for excavation. Within the cave there were such thick stone layers that were baked to very hard conglomerates by the dripping chalk [= speleothem], that only after very long or very hard work and the sacrifice of a lot of dynamite any fruitful work could be done. Near the entrance however, both inside and outside the cave the stony masses were not present – most likely flushed away earlier, so the floor here was about 2 m lower.
Within the cave it consisted of chalktuff and breccia which almost immediately delivered fossils. Some fragments of bone were found, but there were mainly molars of Elephas, Sus, Bos or Bubalus, Cervus, Simia satyrus [= Pongo]. These fossils were not very abundant, but a little bit further into the cave, where there used to be a [?unreadable?]– the opening, though currently closed by drip stone, the rain water from above had flushed away these tuffs and breccias and filled a deep hole with black soil that contained many much younger remains, mostly of Sus vittatus [= scrofa], Cervus [Muntiacus] muntjac and Antilope sumatrensis [= Capricornis sumatraensis] and a Canis species. These do not have any direct meaning for the knowledge of the diluvial fauna of these areas, but these finds are not unimportant as an illustration of the way in which bones of these animals are ending up in caves and accumulate there.
I have not been able to establish how the older bones had entered the cave, it does however seem most likely to me, given the length of the place where they were lying that these must have been dragged in by predators being already torn apart.
Both inside the cave and outside the soil, breccia and fallen limestone blocks to a maximum depth of 7 m were removed reaching a yellow loam layer at 1.50 m from which no more fossils were harvested, which was reason to ultimo [by the end of] April cease the work.”
A few lines were published in a quarterly report which in those days were also published as a supplement to a newspaper (Dubois 1889, p. 8). Dubois also briefly summarizes the results in a few lines in his annual report (Dubois 1889–1890 [submitted 23-1-1890]; 50–042) stating “The work done during the month of April in the Ngalau sampit [sic] near Pajakombo, yielded a not very large amount of teeth and molars, originating from the same species as had earlier been so numerously represented in the Ngalau lida ajer [sic] and are likely to be from the same time.”
To the best of our knowledge, no fossil material collected by Dubois in the Naturalis collection has been specifically attributed to this site. However, thousands of cave samples collected by Dubois bear no specific site name but are simply referred to as unknown or ‘Padang’ caves. At least some of this material will have derived from Ngalau Sampit and may yet be recognized as such in future. Finding Homo erectus on Java later completely overshadowed the Sumatran part of the Dubois collection.
When revisiting Ngalau Sampit, Louys et al. (2017) described a small but easily visible fossil assemblage including teeth assigned to Bovinae (cf. Bos or Bubalus), Cervidae, Pongo sp., Sus sp., and Hystrix sp., in breccia outcrops visible at the ends of Chamber 1 and 2 and the West Entrance to the fossil-bearing passages. The partial cranium of an unidentified ungulate was also observed in northern wall of Chamber 1 (Figure 2) by the authors (Figure 8b in Louys et al., 2017).
Like in many caves in SE Asia (O’Connor et al., 2017; Smith et al., 2020), the sedimentary infilling at Ngalau Sampit is made of breccia deposits (Figure 4), which are cemented to the limestone cave walls and floors (Smith et al., 2021a), and unconsolidated deposits lying on the floor (Figure 5). This loose sediment most likely result from the erosion of the breccia (e.g. Smith et al., 2021a), and some more recent sedimentary inputs into the cave system. Field observations at several locations within the cave suggest that the breccia may be divided into fossiliferous and sterile units (Figures 4 and 5). This visual distinction, though, might be simply result from the lateral variability of fossil density, and not corresponding to distinct phases of breccia formation. The breccia has been the subject of a specific study by Smith et al. (2021b), who analysed a series of blocks collected from various parts of the cave by neutron tomography (NT). Results show that these blocs do not represent a spatially homogeneous sedimentary unit, the nature and composition of the breccia showing significant lateral heterogeneity. The authors conclude that the primary agent in the formation of the breccia is likely to be sediment gravity flow, with clasts and inclusions incorporated within the breccia resulting from a short-distance sediment and water transport. The breccia has also been significantly impacted by post-depositional processes according to Louys et al. (2017)’s field observations, who reported not only a subsequent erosion of the deposits, but also the formation of flowstones or other carbonate formations (Figure 5) as the result of water movement in the cave following the brecciation event. Finally, field observations and laboratory analyses of the breccia did not lead to the identification of any stratigraphic order within the deposits, suggesting that the whole sedimentary infilling might roughly correspond to a single depositional event, although its synchronicity in the various areas of the cave remains unknown.
A preliminary indirect age constraint for the deposits and associated fossil assemblage has been previously obtained by Louys et al. (2017) with a couple of U-series dates from carbonate samples collected in Chamber 2 (Figures 2 and 5, Table 1). GS-4 and GS-5 samples provided ages of 91.2 ± 0.4 ka and 83 ± 5 ka (2σ), respectively. GS-4 comes from a flowstone that is stratigraphically capping the fossil-bearing deposits (see Figure 8e of Louys et al., 2017), thus providing a minimum age constraint for the breccia in that chamber. GS-5 corresponds to a calcite-filled vugh within that breccia, which is most likely a secondary deposit whose formation postdates the deposition of the breccia. Therefore, it also provides a minimum age constraint for the deposit.
|SAMPLE ID||TYPE||LOCATION||DATING METHOD||NT SAMPLE ID|
|SUM1818||?Sus sp. molar||NS18-4 Chamber 2, roof||combined U-series/ESR dating||SUM1872|
|SUM1819||cervid molar||NS18-3 Chamber 2, roof||combined U-series/ESR dating||SUM1872|
|SUM1820||bovid (c.f. Bos or Bubalus) upper molar||NS18-2 Passage to Chamber 2, southern wall||combined U-series/ESR dating||SUM1848|
|SUM1857||sediment||Passage to Chamber 1, southern wall||pIR-IRSL||SUM1871|
|SUM1860||sediment||Chamber 1, roof||pIR-IRSL||SUM1851|
|SUM1862||sediment||Passage to Chamber 2, southern wall||pIR-IRSL||SUM1848|
|GS-4||Speleothem||Chamber 2, roof||U-series|
|GS-5||Calcite-filled vugh||Chamber 2, roof||U-series|
Several teeth and breccia samples were collected in June 2018 for ESR and Luminescence (pIR-IRSL) dating purposes (Table 1), in order to (i) complement the existing preliminary chronological framework, and (ii) evaluate the potential spatial diachronicity of the deposits and associated fossil assemblage within the cave. Although samples were tentatively taken from various spots within the cave, the spatial resolution of the sampling was mostly driven by the presence or absence of datable material (fossils and/or breccia) at a given location.
All pIR-IRSL samples were collected from breccia deposits. Samples SUM1857 and SUM1860 were taken from the entrance of the eastern passage and end of Chamber 1 (Figures 2 and 4: East Entrance and Chamber 1), respectively. pIR-IRSL sample SUM1862 and ESR sample SUM1820 are both from the same area at the entrance of the north-western tunnel. Finally, the teeth SUM1818 and SUM1819 were taken from Chamber 2, at most a meter from the U-series samples previously dated by Louys et al. (2017) (Table 1).
Luminescence sampling consisted of extracting blocks of the cemented fossil-bearing breccia or speleothem from the section in-situ. Due to the hardness of the deposits, and difficulty in accessing the site, this was largely done with geological hammer, with breccia sections selected based on ease of sampling. Efforts were made to sample in all major areas of the deposit (Figure 2). Where the breccia was not completely cemented, breccia blocks were wrapped in plaster prior to extraction. No in situ gamma dose rates were performed for any of the dating methods. The sediment directly attached to each tooth was also collected for dose rate evaluation in the laboratory.
Figure 5 provides a stratigraphic description of the surrounding deposits and formations locally outcropping at each sampling point. Moreover, several blocks of breccia were also collected at various locations in the cave as part of an independent taphonomic study involving NT analyses (Figure 6; Smith et al. 2021b). Since these blocks of sediment were taken in close association with the dating samples (see correspondence in Table 1), they can provide useful information about the local nature, composition, and heterogeneity of the breccia at each sampling point.
Sediment block SUM1871 was collected with the Luminescence sample SUM1857 in the East Entrance (Figure 2 and Table 1), in the passage leading to Chamber 1. It shows an overall homogeneous character with no apparent structure or bedding. NT analyses led to the identification of two main sedimentary fabrics (Smith et al. 2021b): a muddy sand matrix with scarce (about 20% of the volume), poorly-sorted and sub-rounded clasts (1-mm to 1-cm size), and a matrix dominated by fine sands and with numerous sub-angular and high-density clasts (~40%). In comparison, the breccia sediment from Chamber 1 (NT sample ID SUM1851, corresponding to pIR-IRSL sample SUM1860; see Figure 2 and Table 1) shows a silty matrix with a large variety of rounded and angular clasts of granule to pebble sizes and made of different colors indicating variable nature and origin (Smith et al. 2021b). Unlike at the East entrance, fossils were identified within the sediment breccia during the NT analyses.
The sediment sample collected from the passage to Chamber 2 (corresponding to pIR-IRSL sample SUM1862 and ESR sample SUM1820), shows that the breccia deposits are dominated by a single sandy facies with a limited proportion of mostly rounded clast (about 10%), i.e., lower than that observed in the East entrance (Smith et al. 2021b). NT analyses also reveal a relatively high (but unquantified) water content in this sediment.
Finally, sediment block SUM1872 originates from Chamber 2, where samples SUM1818 and SUM1819 were collected. The breccia deposits appear to be very heterogeneous, dominated by three main sedimentary facies, a fine-to-medium sandy matrix with rounded clasts of millimetric size, a flowstone that appears massive, and a coarse muddy sand with clusters of brittle angular carbonate clasts of length ranging from 5 to 30 mm (Smith et al. 2021b).
ESR dating is one of the very few methods that can provide a direct and finite age constraint to fossils that are older than 50 ka (e.g., Grün et al., 2010). The optimum time range for the application of the ESR method to fossil teeth is usually considered to be between ~50 and ~600 ka, although it can potentially cover the whole Quaternary if conditions and samples are suitable (Duval, 2015). Since dental tissues behave as open systems for U-series elements, uranium uptake over time needs to be properly modelled for the dose rate evaluation. This is typically done by combining U-series and ESR data, as per first proposed by Grün et al. (1988). Several uranium uptake models have been proposed over the last decades (e.g., Grün et al., 1988; Grün, 2000b; Hoffmann et al., 2003; Shao et al., 2012). Further details about the method can be found in the overview paper by Duval (2015).
The teeth were prepared as in Smith et al. (2021a): the enamel layer was mechanically separated from the other dental tissues and both inner and outer surfaces were removed with a dentist drill to eliminate the volume that received an external alpha dose. The initial and removed thicknesses of the enamel layer were measured with a digital calliper. The clean enamel and dentine samples were ground and sieved <200 µm to obtain homogenous powders.
The ESR dosimetry of the enamel samples was performed at the Centro Nacional de Investigación sobre la Evolución Humana (CENIEH), Spain, following the multiple aliquot additive dose (MAAD) method. The enamel powder of the three teeth was split into ten aliquots and gamma irradiated with a Gammacell 1000 Cs-137 gamma source. SUM1818 and SUM1819 were irradiated to the following doses (dose rate = 6.39 ± 0.15 Gy/min): 0, 49.9, 99.8, 149.6, 249.5, 349.2, 498.9, 698.5, 898.1 and 1496.8 Gy. SUM1820 was irradiated as follows (dose rate = 6.27 ± 0.14 Gy/min): 0, 49.0, 98.1, 147.0, 196.0, 294.1, 392.1, 588.0, 784.0 and 980.0 Gy.
Room temperature ESR measurements were carried out with an EMXmicro 6/1 Bruker ESR spectrometer coupled to a standard rectangular ER 4102ST cavity. The following procedure was used to minimise the analytical uncertainties: (i) all aliquots of a given sample were carefully weighted into their corresponding tubes and a variation of <1 mg was tolerated between aliquots; (ii) ESR measurements were performed using a Teflon sample tube holder inserted from the bottom of the cavity to ensure that the vertical position of the tubes remains exactly the same for all aliquots. The following acquisition parameters were used: 1–25 scans depending on the aliquot and sample considered, 1 mW microwave power, 1024 points resolution, 15 mT sweep width, 100 kHz modulation frequency, 0.1 mT modulation amplitude, 20 ms conversion time and 5 ms time constant. All aliquots of a given sample were measured within a short time interval (<1 h). This procedure was repeated two times over successive days without removing the enamel from the ESR tubes between measurements in order to evaluate measurement and equivalent dose (DE) precisions (Table 2).
|Average weight per aliquot (mg)||8.1 ± 0.3||12.3 ± 0.2||12.2 ± 0.2|
|Number of repeated measurements||2||2||2|
|Measurement precision (%)||1.1||0.6||0.7|
|SSE fitting (data weighting by 1/I2)|
|Non-corrected ESR intensities|
|DE precision (%)||4.8||1.2||4.5|
|DE1 (Gy)||90.6 ± 2.5||96.8 ± 1.5||135.1 ± 3.2|
|DE2 (Gy)||90.4 ± 3.9||95.6 ± 1.7||–|
|Baseline-corrected ESR intensities|
|DE3 (Gy)||88.1 ± 3.6||95.5 ± 2.0||132.9 ± 3.4|
|Noise-subtracted ESR intensities|
|DE4 (Gy)||81.0 ± 3.3||94.2 ± 1.7||131.4 ± 3.2|
|SSE fitting (data weighting by 1/s2)|
|Noise-subtracted ESR intensities|
|DE5 (Gy)||85.3 ± 8.5||106.7 ± 7.5||129.6 ± 4.6|
|DE3/DE2 ratio||0.97||0.99||0.98 (1)|
|DE4/DE2 ratio||0.90||0.95||0.97 (1)|
|DE5/DE2 ratio||0.94||1.12||0.96 (1)|
The ESR intensities were extracted from T1-B2 peak-to-peak amplitudes of the ESR signal (Grün, 2000a), and then normalised to the corresponding number of scans and aliquot mass. DE values were obtained by fitting a single saturating exponential (SSE) through the mean ESR intensities and associate errors (1 s.d.) derived from the repeated measurements. Fitting was performed with Microcal OriginPro 9.1 software, which is based on a Levenberg-Marquardt algorithm by chi-square minimisation. Data were weighted by the inverse of the squared ESR intensities (1/I2) and inverse of the squared experimental errors. ESR dose response curves (DRCs) are shown in Figure 7. Final 1-σ DE error used for age calculation (Table 3) include both the fitting and gamma source dose rate errors.
|U (ppm)||0.69 ± 0.00||1.05 ± 0.00||1.39 ± 0.00|
|234U/238U||1.304 ± 0.001||1.323 ± 0.001||1.245 ± 0.001|
|230Th/234U||0.502 ± 0.001||0.417 ± 0.002||0.626 ± 0.002|
|230Th/232Th||321.8 ± 0.929||136.3 ± 0.553||94.02 ± 0.397|
|Age (ka)||73.4 ± 0.26||57.20 ± 0.28||102.03 ± 0.53|
|Alpha Efficiency||0.13 ± 0.02||0.13 ± 0.02||0.13 ± 0.02|
|Water content (%)||0||0||0|
|Initial enamel thickness (μm) (1)||928 ± 93||981 ± 98||1367 ± 137|
|U (ppm)||25.40 ± 0.01||27.64 ± 0.01||21.56 ± 0.01|
|234U/238U||1.319 ± 0.001||1.336 ± 0.000||1.275 ± 0.001|
|230Th/234U||0.605 ± 0.002||0.599 ± 0.002||0.526 ± 0.002|
|230Th/232Th||116794 ± 1765||4258 ± 18.33||6004 ± 28.85|
|Age (ka)||96.4 ± 0.38||94.8 ± 0.39||78.8 ± 0.42|
|Water (%)||5 ± 3||5 ± 3||5 ± 3|
|Removed enamel thickness (μm) (1)||18 ± 2||112 ± 11||12 ± 1|
|U (ppm)||1.360 ± 0.084||1.890 ± 0.095||2.700 ± 0.117|
|Th (ppm)||1.370 ± 0.109||3.580 ± 0.157||4.220 ± 0.183|
|K (%)||0.133 ± 0.005||0.196 ± 0.008||0.201 ± 0.008|
|Water (%)||15 ± 5||15 ± 5||15 ± 5|
|Removed thickness (μm) (1)||65 ± 7||123 ± 12||273 ± 27|
|US/AU-ESR age calculations|
|Internal dose rate (μGy/a)||144 ± 58||171 ± 73||388 ± 103|
|Beta dose rate, dentine (μGy/a)||289 ± 116||262 ± 111||178 ± 47|
|Beta dose rate, sediment (μGy/a)||41 ± 3||54 ± 4||35 ± 3|
|Gamma + cosmic dose rate (μGy/a)||320 ± 14||437 ± 19||542 ± 24|
|Total dose rate (μGy/a)||794 ± 130||924 ± 134||1143 ± 116|
|p dentine or n dentine||–0.01||–0.01||–0.65|
|US-ESR age (ka)||102 ± 16||98 ± 14||115 ± 11|
Powdered enamel and dentine samples were weighed then spiked using a 229Th-233U tracer before being digested in concentrated HNO3. The solutions were then treated with H2O2 to remove trace organics, with U and Th then separated using conventional column chemistry techniques described in Clark et al. (2014). Both U and Th were collected into the same pre-cleaned test tube using 3 ml of 2% HNO3 mixed with a trace amount of HF. U-Th isotopic ratios were then measured using a Nu Plasma multi collector inductively-coupled plasma mass spectrometer (MC-ICPMS) in the Radiogenic Isotope Facility at The University of Queensland, Brisbane, Australia, following analytical protocols established by Clark et al. (2014) and Zhao et al. (2009). U-series results are displayed in Table 3.
No in situ evaluation of the gamma dose rate associated with the teeth was performed. Consequently, both the beta and gamma dose rates were derived from the laboratory analysis of the sediment attached to each tooth. The sediment samples were crushed and homogenized prior to ICPMS analyses by Genalysis Laboratory Services, following a four-acid digest preparation procedure.
The following parameters were used for the dose rate calculations: an alpha efficiency of 0.13 ± 0.02 (Grün and Katzenberger-Apel, 1994), Monte-Carlo beta attenuation factors from Marsh (1999), dose-rate conversion factors from Guérin et al (2011), an assumed water content 5 ± 3 wt.% in dentine. In order to ensure consistent dose rate calculations across ESR and Luminescence methods, we employed similar long-term water content and cosmic dose rate estimates. A water content of 12 ± 5% (% wet weight) in sediment was used for the three teeth. This value is equivalent to the average value of 13% (% dry weight) measured for the Luminescence samples SUM1857 and SUM1862. The measured water content is estimated to be the most accurate estimate of the long-term water content, as fluctuations of the humidity within the cave are assumed to remain somewhat limited in this closed environment. Nevertheless, the 5% 1-σ associated error is meant to cover potential random fluctuations over time of between 5% to 22% at a 2-σ confidence level. Similar to Luminescence dating, a cosmic dose rate of 63 ± 2 µGy/a was used for the three teeth.
Age calculations were performed with USESR, a Matlab-based program (Shao et al., 2014) using the US and AU models defined by Grün et al. (1988) and Shao et al. (2012), respectively. The US model was preferentially employed, unless for dental tissues showing apparent uranium leaching. In this case, the AU model was used instead. Additional CSUS-ESR age calculations were also carried out using DATA, a DOS-based program (Grün 2009). The CSUS-ESR model defined by Grün (2000b) is based on the assumption that all of the uranium migrated into the sample at a time given by the closed system U-series age. The CSUS-ESR age is the maximum age that can be derived from a given U-series and ESR data set. Age calculations using the US or AU and CSUS models encompass all possible uptake scenarios. Data inputs and outputs are given in Table 3. Age results are given at 1-σ confidence level.
Luminescence dating is one of the only dating techniques that can constrain the sedimentary context at the site. Luminescence measures the time since minerals, such as quartz or feldspar were last exposed to sunlight, which resets this light sensitive signal to zero (Aitken, 1998). After burial, this signal then builds up as charge from naturally occurring radiation, which is stored in the crystal lattice. As this radiation (or environmental dose rate) is high in these cave sediments (e.g., Westaway et al., 2017; Louys et al., 2021a) we employed potassium feldspars (KF) rather than quartz minerals for luminescence analysis. This choice is based on the much higher dose saturation characteristics of KF compared to quartz, which is crucial for reliable De estimation in these high dose rate environments. In addition, we used a single-grain rather than multiple grain feldspar technique that is designed to isolate the most stable part of the feldspar infra-red stimulated signal (pIR-IRSL) (Rhodes, 2015). Volcanic feldspars from this region are known to fade over geological timescales (Morwood et al., 2004): the use of the post-infra-red protocol overcomes this problem, so the fading is often reduced to negligible levels (Thomsen et al., 2008; Buylaert et al., 2009; Thiel et al., 2011).
Within subdued red light conditions, the light exposed outer layer was removed using a chisel and hammer and was retained as the dosimetry sample. The unexposed inner core was gently broken up using a pestle and mortar and was processed using the standard sample purification procedures for feldspar separation (Aitken, 1998) including a 10% wash in hydrofluoric acid for 10 mins to remove the external alpha-dosed rinds.
Previous luminescence analysis of a deposit in the nearby cave site of Lida Ajer (Westaway et al., 2017) revealed a non-existent blue quartz signal and a feldspar signal with a high dose response and a stable component that could be isolated from the unstable fading component (Westaway et al., 2017). We adopted the same standard post-IR-IRSL protocol to overcome the problems of anomalous fading (Thompsen et al., 2008; Murray et al., 2009) using a 300°C preheat and 270°C pIR-IRSL stimulation combination following a standard 50°C IR stimulation (Westaway et al., 2017; Shackelford et al., 2018). This combination plotted within the flattest part of the preheat plateau, provided the best recovery of the surrogate dose, with the least fading of all the pIR-IRSL signals (g value = 1.0% per decade) and lowest residual value (<10 Gy). Single aliquot protocols tested in the neighbouring site of Lida Ajer (Westaway et al., 2017) produced only maximum ages for sediment deposition due to the number of feldspar grains used on each disc, which had an averaging effect on the final De. Therefore, due to this issue combined with an extremely low yield in potassium feldspar grains, we decided to only apply a single-grain pIR–IRSL technique to this site. The procedural tests employed by Shackleford et al. (2018) were also applied to these samples with some modifications for the use of single-grains as outlined in Rhodes (2015). All luminescence analysis was conducted at the ‘Traps’ luminescence dating facility at Macquarie University in Sydney, Australia.
Individual 180–212 µm feldspar grains were mounted onto coated single grain discs in a 10 by 10 grid. The disc were loaded onto a carousel and processed in a Riso TL-DA-20 containing an automated Detection and Stimulation Head (DASH) set up with a Dual laser single grain attachment with a Blue/UV sensitive Electron Tube PMT (PDM9107Q-AP-TTL-03), with maximum detection efficiency between 200 and 400 nm. The filters in the automated detection changer were set on the blue filter pack (Schott BG-39 and Corning 7–59 filters to transmit wavelengths of 320–480 nm (Hutt, 1988). The grains were stimulated using an IR (830 nm) 140 mW TTL modulated laser with a 3 mm RG-780 long-pass filter (mounted directly in front of the IR laser) and the emissions were detected using the blue filter pack combination describe above. The laser stimulated the grains for 2.5 s first at 50 °C and secondly at 270 °C according to the procedures of the pIR-IRSL protocol (Roberts, 2012). The uncorrected single-grain equivalent doses were plotted on a radial plot centred on the central age, then were run through a minimum age model (MAM) (Galbraith et al., 1999) to identify the population that had the most bleaching prior to burial. Based on the cave location and high potential for partial bleaching in this sedimentary environment (see the wide range of De values in Figure 8c) it was decided that the MAM offered the best opportunity for isolating the most bleached population of grains, and to provide the closest estimation of the burial age. As suggested by Galbraith et al., (1999), an estimation of the natural overdispersion value of 15% (obtained by measuring a modern analogue sample) was added to the errors prior to running the model. The resulting MAM De was then corrected according to the results of the anomalous fading tests using a weighted mean fading rate of 1.0 ± 0.2% per decade. Residual values (in Gy) were observed, but they were minimal in comparison to the De values so no residual corrections were undertaken (Figure 8). Thus, this value and associated error were used as the fading corrected De.
To obtain an estimate of the environmental dose rate for each of the samples, we first measured beta dose rates using a Geiger-Muller multi-counter beta counting of dried and powdered sediment samples (Botter-Jensen and Mejdahl, 1988) in the laboratory. Allowance was made for the effect of sample moisture content (Aitken, 1998), different grain sizes (Brennan, 2003) and HF etching (Bell and Zimmerman, 1978) on attenuation of the beta dose and the total beta dose-rate contribution was calculated by comparing the beta count rate to a standard beta source (Shap with a dose rate of 5.99 Gy/ka) and magnesium oxide as a non-beta emitting background material. Secondly, thick source alpha counting using a Daybreak 583 intelligent alpha counter was used to obtain estimates of Uranium and Thorium (Wang and Xia, 1991) to estimate the gamma dose rate, and thirdly the difference between beta and alpha counting was used to estimate potassium values. These estimates were then converted to gamma dose rates using the conversion factors of Guérin et al. (2011). An effective internal beta dose rate of 0.84 Gy/ka (Huntley and Baril, 1997; Huntley and Hancock, 2001) was used for the 180–212 µm feldspar samples (due to the radioactive decay of 40K and 87Rb), which was made assuming K (12.5 ± 0.5 %) and 87Rb (400 ± 100 μg.g–1) concentrations, and was included in the total dose rate. Cosmic-ray dose rates were estimated from published relationships (Prescott and Hutton, 1994), making allowance for the limestone overburden at the sample locality (~15 m), the altitude (~522 m above sea level) and geomagnetic latitude and longitude (00° and 100°) of the sampling site. The total dose rate was calculated using a long-term water content of 12 ± 2% (% dry weight), which is similar to the measured (field) water content of 11–15%. Sample SUM1860 was significantly drier than this amount and the dose rate was obtained using a lower water content of 3 ± 0.2%. Detailed age results (1-σ confidence level) are given in Table 4.
|SAMPLE CODEa||SAMPLE DEPTH (M)||GAMMA DOSE RATE (GY.KA–1)b||BETA DOSE RATE (GY.KA–1)b||COSMIC-RAY DOSE RATE (GY.KA–1)c||INTERNAL DOSE RATE (GY.KA–1)||WATER CONTENT (%)d||TOTAL DOSE RATE (GY.KA–1)e||EQUIVALENT DOSE (GY)g,h||PIR-IRSL AGE (KA)h|
|SUM1862||0.2||0.848 ± 0.047||0.876 ± 0.143||0.061 ± 0.006||0.84 ± 0.20||15.0/12 ± 2||2.626 ± 0.278||228 ± 27||87 ± 14|
|SUM1857||0.2||0.799 ± 0.054||0.822 ± 0.143||0.061 ± 0.006||0.84 ± 0.20||11.0/12 ± 2||2.522 ± 0.283||237 ± 44||94 ± 22|
|SUM1860||0||0.364 ± 0.031||0.360 ± 0.140||0.068 ± 0.006||0.84 ± 0.20||3.0/3 ± 0.2||1.631 ± 0.263||160 ± 41||98 ± 30|
Due to the limited amount of enamel material that could be extracted from the teeth, ESR measurements were performed with aliquot weights ranging from ~8 to ~12 mg (Table 2) depending on the sample considered. This resulted in somewhat noisy ESR spectra and relatively weak ESR intensities. For this reason, up to 25 scans were accumulated for the least irradiated aliquots in order to achieve a sufficient signal-to-noise ratio. Despite these issues, ESR measurement are highly repeatable, with the ESR intensities varying by 0.7–1.1% among the three samples, resulting in a very small DE variability of <5% over repeated measurements (Table 2). Fitting performed on the non-corrected ESR intensities over the full dose range yield DE values ranging from 91 to 135 Gy depending on the sample considered (Figure 7). Fitting results of SUM1818 and SUM1819 are above the Dmax/DE ratio of 5–10 recommended by Duval and Grün (2016) to avoid DE overestimations, while SUM1820 meets this specific requirement (Dmax/DE = 7.4). New fittings using the appropriate maximum irradiation dose (Dmax) result in slightly lower DE values by <1% for SUM1818 and SUM1819.
Given the weak ESR signals measured for the three samples, the ESR intensities were successively corrected for the non-horizontal baseline and significant noise intensity observed in the spectra. Practically, a baseline correction based on a cubic function was employed, and the mean ESR intensity of the noise measured in all aliquots of a given sample was subtracted to each signal intensity. The same corrections were employed for all aliquots of a given sample. DE values derived from baseline-corrected ESR intensities are slightly lower by <3%, while those obtained from the baseline-corrected and noise-subtracted ER intensities are lower by between 3 to 10%. In both situations, the resulting DE values agree within 2σ error with the previous dose estimates, indicating the limited impact of such corrections. Finally, SSE fitting with data weighted by 1/s2 was also performed for comparison, resulting in DE results varying by –6% to +12% with the previous estimates. This outcome indicates that the fitting options have a non-significant impact on the DE results for two samples. In contrast, SUM1819 seems more impacted, but the fitting results nevertheless agree at 2σ (Table 2).
In summary, the excellent repeatability of the ESR measurements, combined with the excellent goodness-of-fit (adjusted r2 values systematically >0.998) and the limited impact of the different fitting options tested indicate the good reliability and robustness of the ESR data collected for these three samples.
Apparent U-series range between 57 to 102 ka (Table 3). They should be regarded as minimum age constraints for the fossils. The dentine tissues vary within little age range (79–96 ka), while the enamel shows more scatter (57–102 ka). Teeth SUM1818 and SUM1819 show delayed uranium uptake in the enamel compared with the dentine, while this is the opposite for SUM1820.
The combination of U-series and ESR data using the US model (Grün et al., 1988) do not return any finite age result for teeth SUM1818 and SUM1819: U-series data collected from the dentine samples of both teeth preclude US-ESR age calculations, suggesting thus that these tissues have experienced uranium leaching. This is one of the main limits of the US model initially defined by Grün et al. (1988). Consequently, the AU model (Shao et al., 2012), which can take into account uranium leaching, was specifically employed for these two tissues, while the US model was used for the others. Combined U-series/ESR age calculations return highly consistent results of 102 ± 16 ka, 98 ± 14 and 115 ± 11 ka for the three teeth. The dose rate is mostly dominated by the sediment (beta and gamma) and cosmic dose rate components, which represent 45–53% of the total dose rate, while the weight of the internal rate is 18–34%. The use of the AU model resulted in the modelling of a somewhat limited uranium leaching for the dentine of SUM1818 and SUM1819 (about 20% for both). Interestingly, it is probably not a coincidence that the only sample showing no apparent uranium leaching (SUM1820) yields the oldest age, but is still consistent with the other results. It is possible that the uranium leaching may induce a limited dose rate overestimation, resulting in slight age underestimation. However, since all samples yield 1σ consistent ages, the mean age of 105 ± 9 ka (1 s.d.) is probably in the first instance the most accurate chronological constraint for the fossil assemblage at Ngalau Sampit.
The impact of uranium uptake modelling on the calculated age results may be evaluated by performing additional age simulations using other models or programs. For example, unlike DATA program (Grun, 2009), USESR allows the p parameter to be lower than –1, as first suggested by Bahain et al. (1992). Such calculation would be based on the assumption of an early uptake followed by a continuous uranium leaching over time. The resulting US-ESR age results of 93 ± 13 ka and 91 ± 10 for SUM1818 and SUM1819, respectively, are younger by 7–9% than the AU-ESR estimates initially calculated. This slight age difference is explained by the modelled p values, which remain very close to –1 (–1.02 and –1.03). When using the CSUS model for SUM1820, an age of 122 ± 11 ka may be obtained, i.e. only 6% older than the previous result for this sample. Finally, additional calculations were carried out for SUM1820 with water content in the sediment varying between 5 and 25% (Table 5). Resulting age estimates vary by –5% to +12%, but remain systematically within error with the initial age result.
|WATER CONTENT (%WET WEIGHT)||US-ESR AGE (KA)||AGE RATIO|
|5%||109 ± 10||0.95|
|10%||114 ± 10||0.99|
|12%||115 ± 11||1.00|
|15%||118 ± 11||1.03|
|20%||123 ± 12||1.07|
|25%||129 ± 13||1.12|
To summarize, these various simulations illustrate the overall limited impact of uranium uptake modelling and water content assumptions on the calculated ESR ages, which remain overall within error. They consistently point to the robustness of the ESR results obtained for the three samples from Ngalau Sampit.
The feldspars had low sensitivities yielding very low acceptance rates; over 4000 grains were processed with only 40 accepted grains (0.01% acceptance). Rejections were based on the protocol of Jacobs et al. (2006). Most of the grains had no decay whatsoever. These results were challenging considering the small amounts of sample yield meant that only between 8–17 discs could be made for each sample and we were unable to get the accepted grains to a statistically meaningful number of ~100. However, the grains that did decay produced consistent shine-downs and good growth in the dose response curves with 2 × D0 values that were well within the acceptable saturation limits (Figure 8b). The accepted grains for sample SUM1862 (27 grains), SUM1857 (10 grains) and SUM1860 (3 grains) produced consistent age estimates of 87 ± 14 ka, 94 ± 22 ka and 98 ± 30 ka, respectively (Table 4).
The minimum age model (MAM) was chosen to analyse the single-grain feldspar data set because of the sedimentary characteristics of the cave environment. The depositional processes that transport sediments into the caves sometimes most likely prevent full bleaching of the sediments before burial. Thus, the MAM allows the isolation of the grains that have received the most bleaching. Therefore, although the number of accepted grains is low, the resulting age estimates do provide an estimation of the burial age that is not restricted by the averaging effect of the single-aliquot procedure, which tends to produce maximum ages. This results in age estimates that are close to the true burial age of the sediments.
The corrected De values for samples SUM1862 and SUM1857 agree within errors and the estimated dose rates are also in agreement, suggesting internal consistency in the methods and timing of deposition. In comparison, while SUM1860 has a much lower water content and dose rate resulting in a lower De value, the age estimate is nevertheless in close agreement with the others (Table 4). This sample recovered the lowest amount of feldspars from all three samples. Only 8 single-grain discs could be processed, and out of the 800 grains only 3 produced any PIR-IRSL decays. This resulted in the least reliable age estimate, but it has been included here to provide a rough age estimate of the breccia from this section of the cave. Additionally, this sample shows a significantly lower dose rate, which is less than half of the values calculated for the other two samples. NT analyses suggest that the breccia located in the vicinity of SUM1860 is somewhat locally different in nature and composition to that observed in other parts of the cave: (i) it includes many pebble-sized limestone clasts and is overall more heterogeneous than the other blocks analysed (Figure 6), and (ii) it seems to result from a more complex pre-depositional history, as it incorporates elements of a previously eroded primary breccia deposit. This would explain the large difference in dose rate among the samples, although final age results vary within very narrow range.
Like many caves in SE Asia (e.g., Louys et al. 2017; Smith et al., 2020; O’Connor et al., 2017), Ngalau Smapit is characterised by the presence of remnant breccia deposits hanging on cave walls. These cemented deposits have traditionally received much less attention from archaeologists and palaeontologists in comparison with well-stratified unconsolidated sediment of cave floors (Smith et al. 2020). However, breccia can document a phase of sedimentary input, mammal and human occupation that is not visible in the cave floor. The integrity of the latter can also be impacted by incorporation of material resulting from the erosion/alteration of the breccia, strongly limiting their paleoenvironmental, paleontological or chronological significance. Conversely, remnants of breccia deposits may be considered as useful ‘time capsules’ (Smith et al. 2020; O’Connor et al., 2021).
Dating breccia deposits using trapped charge dating methods remains, however, quite challenging. In addition to their hardness, these deposits are by definition heterogeneous, which strongly complicates the dose rate evaluation in absence of in situ dosimetry. In particular, gamma dose rate calculations performed for the teeth in the present study are based on a series of assumptions or considerations that require a proper evaluation of the associated uncertainty: (i) the sediment surrounding the teeth is relatively homogenous, (ii) the teeth are totally surrounded by sediment, (iii) the limestone bedrock is located >30 cm away from the teeth and does not contribute to the gamma dose rate. Various sensitivity tests were performed to roughly evaluate the potential impact of these assumptions/considerations on the resulting ESR age estimates (Figure 9).
First, dose rate evaluation was performed using radioelement concentrations from homogenized sediment samples that were collected directly from each tooth. The spatial heterogeneity of the breccia may be indirectly assessed by evaluating the variability of the radioelement concentrations in the breccia across the cave. U, Th and K concentration vary by about 34%, 28 and 22%, respectively, resulting in a gamma dose rate variability of about 30% (1.s.d.) (Table 3). A variation on the gamma dose rate of such magnitude would have a relatively limited impact of 2 to 8% on the calculated ages depending on the tooth considered (Figure 9A). Such uncertainty due to the heterogeneity of the surrounding sediment would impact the pIR-IRSL age estimates to a similar extent.
Second, the teeth were found outcropping from the breccia deposits, which means that the infinite matrix conditions (= 4π geometry) are not met. While it is virtually impossible to determine when the missing part of the breccia eroded, and thus for how long the tooth has been exposed, age simulations may nevertheless be performed assuming that the current geometry (= 2π geometry) has prevailed throughout the burial history of the tooth. Considering that the tooth was on surface and therefore the sediment contribution would only of 50% of the infinite matrix conditions (Aitken, 1985), and assuming a negligible contribution from the air space, new calculations return older ages by 9 to 25% and reaching ~111 to ~136 ka, depending on the tooth considered (Figure 9B). These results are based on the (unlikely) assumption that the teeth have spent their whole time exposed on the surface of the breccia. Since this geometry leads to the calculation of a minimum gamma dose rate value, the age estimates should be regarded as maximum constraints for the fossils in that regard. Therefore, the true age lies somewhere between 95–115 ka and 111–136 ka. While these calculations enable us to roughly quantify the impact of the geometry of the surrounding sediment, we nevertheless do acknowledge their oversimplification, since both the Rn content and the reflection of gamma rays originating from the sediment against the cave walls most likely build up a non-null contribution of the air to the gamma dose rate. This contribution can, however, hardly be properly quantified.
Finally, the limestone cave wall may also itself contribute to the gamma dose rate, depending on its distance from the sample. Given the significant difference between the radioactivity of the sediment and the cave wall, the latter may be considered in as radioactively inert. Gamma dose rates have been recalculated using Aitken (1985) and considering distances of 5, 10, and 20 cm from the limestone, corresponding to ca. 78%, 88% and 96% of the infinite matrix gamma dose rate from the sediment. Resulting ages get older by up to 10% when considering a 5-cm distance and remain in agreement within 1σ error (Figure 9C). While the exact distance of the teeth to the limestone can hardly be properly evaluated in all directions, field observations nevertheless indicate that bedrock is >10 cm from each tooth. In this case, the age estimates are older by only 4–5%, demonstrating the relatively limited impact of this source of uncertainty on the calculated ages. It is worth mentioning that the pIR-IRSL age estimates initially calculated considering no influence from the cave walls would be affected in a similar way. Finally, we have considered here an extreme scenario with a radioactively inert cave wall, but the magnitude of this potential dose rate bias would be even more reduced if the cave walls display a non-negligible radioactivity, i.e. closer to that of the sediment.
In summary, although we acknowledge the existing uncertainty around the gamma dose rate evaluation, the calculated age results are overall little impacted and remain systematically within 1-σ error. Interestingly, all calculated ages consistently point towards an early MIS 5 chronology (Figure 9). Finally, it should be noted here that in situ evaluation of the gamma dose rate would not have provided either a fully accurate estimate of the true gamma dose rate, at least for the teeth. These were found outcropping, meaning that a large part of the sedimentary environment that contributed to the gamma dose rate is simply missing. Consequently, since 2π or 4π in situ measurements would simply not help to reconstruct this missing part, in situ evaluations would also be affected by a significant associated uncertainty.
The numerical age results obtained at Ngalau Sampit may be summarized as follows:
Two different interpretations can be made from the existing chronological data set (Figure 10):
Interpretation 1. There is an apparent systematic and non-neglegible deviation between the pIR-IRSL and US-ESR age results (12 ka between the mean age estimates). Consequently, it is possible that the fossil assemblage is (partly or entirely) older than the formation of the breccia, which is not uncommon in cave environments: the fossils may have been moved around the landscape for long periods before being buried in the cave. The pIR-IRSL technique provides an age for the burial of sediment in the cave, not the death of the organisms, and rarely do these two events occur simultaneously. In this case, it is virtually impossible to estimate how much older the fossil assemblage would be. Indeed, the combined U-series/ESR age results would then be treated with extreme caution, as the dose rate can only be evaluated from the present-day conditions. Past surrounding sedimentary environment, prior to the breccia formation, is simply unknown, leaving a significant uncertainty in the dose rate estimation. Consequently, the apparent U-series age results obtained from the dental tissues would become the only reliable and meaningful direct chronological constraints for the fossil teeth, and would be regarded as minimum ages (>96 ka for SUM1818; >95 ka for SUM1819; >102 ka for SUM1820).
Interpretation 2. The apparent systematic deviation between the two independent data sets is considered to be not significant, as all ESR and pIR-IR age results are consistent at either 1σ or 2σ confidence level, depending on the sample considered (Figure 10). This consistency can be directly evaluated through the comparison of pIR-IRSL sample SUM1862 and ESR sample SUM1820, which were collected in close proximity (Figure 2). Although age estimates differ by about 28 ka (87 ± 14 vs 115 ± 11 ka), they nevertheless agree at 2σ. We suspect this age difference most likely comes from the evaluation of the gamma dose rate, which varies by a factor of ~1.8 (SUM1862: 848 µGy/a; SUM1820: 480 µGy/a). While we cannot exclude that part of this difference may originate from the techniques employed (ICPMS for ESR dating vs. alpha and beta counting for pIR-IRSL), spatial heterogeneity of the radioactivity within the breccia does undoubtedly play a major role, as highlighted by the NT analyses (Smith et al., 2021b; Figure 6). pIR-IRSL age simulation performed with a gamma dose rate of 480 µGy/a returns an estimate of ~98 ka for SUM1862, i.e. 13% older than the age initially calculated (Table 4). This result is in excellent agreement with the US-ESR result obtained for SUM1820 (Table 3). Consequently, the age difference observed between the two samples most likely results from the existing uncertainty around the evaluation of the gamma dose rate, which naturally arises from the absence of in situ dosimetry. However, despite this uncertainty, all the numerical ages consistently and systematically correlate the breccia and associated fossil assemblage to the MIS 5 (130–71 ka; Lisiecki and Raymo, 2005). This is probably the safest interpretation of the chronological data obtained.
In summary, our data indicate that the fossil assemblage and the sediment deposition are coeval at the timescale discernable by the two dating techniques (Figure 10). Furthermore, combining these data with field observations suggest that the breccia deposit and its fossil inclusions represent an overall single and synchronous event. However, given the number of samples dated, and since these samples do not cover the entire extension of the breccia deposit in this section of the cave, nor of the original deposit, we acknowledge this represents the simplest explanation for our data. We cannot exclude a more complex depositional history that included multiple phases of faunal deposition and/or multiple sedimentary events that are no longer recorded in the preserved breccia and cannot be chronologically constrained due to the existing age uncertainty of the dating techniques used and the limited sampling.
Determining precise and accurate ages of fossils is fundamental for resolving faunal responses to climate change, the timing of major biogeographical events, and causes of population and species extinctions. Such data are critical for the development of useful biochronological schemes and understanding the environmental backdrop of hominin arrival in a region. The Ngalau Sampit breccias are most parsimoniously understood to be the result of a single depositional event that occurred during MIS 5, with minimal time-averaging of the fossils incorporated therein. However, even in this cave system, where minimal incongruencies exist between geochronological techniques, their associated errors, and field observations, means that more complicated depositional histories could be accommodated by our data.
These results illustrate the interest of using a multi-technique dating approach at SE Asian sites, order to obtain direct chronological constraints on both the fossils and associated sedimentary matrix. On the other hand, they also highlight some of the difficulties in using geochronological approaches (e.g., Luminescence: absence of datable quartz and limited number of useful K-feldspars; ESR: possible occurrence of uranium leaching in dental tissues) to resolve in detail the geological history of sediment deposition, and thus fossil ages, in tropical caves that can’t be analysed using radiocarbon techniques. While the application of an even more vigorous dating program could test different depositional hypotheses, there remains a natural limit to how much resolution can be obtained using all dating techniques, given the existing age uncertainty provided (typically 10–15% at 1σ for palaeodosimetric methods such as ESR and Luminescence). In this context, additional analyses of other geological and taphonomic data can provide further resolution of the timing and order of events leading to fossil-bearing breccia (Smith et al, 2020, 2021a, 2021b).
This site represents only the third vertebrate locality in the Padang Highlands to be numerically dated. Thus, although it has not yielded a rich or diverse fossil mammal assemblage to date, it does provide a chronological baseline that suggests further sampling of this deposit would be useful. At this point, it is the only site in Sumatra that chronologically corresponds to MIS 5, and thus that correlates with the regionally important site of Punung in Java. Punung, dated to approximately 128–118 ka, is the oldest site thought to preserve rainforest environments in Java (Westaway et al., 2007), with all fauna present in Ngalau Sampit are also represented in the Punung deposits. The Late Pleistocene expansion of rainforests have been correlated with extinctions of megafauna and hominins throughout Southeast Asia (Louys and Roberts, 2020), and understanding their development through MIS 5 is important for efforts aimed at resolving these events at greater chronological and ecological resolutions.
The fossil taxa thus far observed at Ngalau Sampit are also characteristic of Sumatra, matching Lida Ajer as Dubois already noted, and indeed Southeast Asia in general, with stable isotope studies of these fossils ongoing. The recovery of an orangutan lower molar from Ngalau Sampit indicates some habitat continuity with the earlier Ngalau Gupin and the later Lida Ajer deposits (Smith et al. 2020, Westaway et al. 2017), as well as the Pongo-bearing faunas of Punung. Microwear analysis of this tooth indicates it has relatively high complexity and low anisotropy, although not outside the range of sampled modern and fossil orangutans (Louys et al. 2021b). Initial indications would thus suggest rainforest conditions for Ngalau Sampit during MIS 5, perhaps extending along the western edge of Sumatra and contiguous with rainforests in Java. Further sampling and analyses will be needed to confirm these hypotheses.
The present work provides new chronological constraints of the fossil material and associated deposits from Ngalau Sampit, one of the Sumatran sites initially reported by Dubois, and so far, only the third to be numerically dated from the island. Our results demonstrate the utility of integrating multiple geochronological techniques to archaeo- and palaeontological sites. Obtaining direct ages for both the fossil assemblage and the deposits offers the possibility to evaluate their contemporaneity, providing key insights about site formation processes. At Ngalau Sampit, combined U-series/ESR dating of three fossil teeth yields a mean age of 105 ± 9 ka, while three breccia samples give a mean pIR-IRSL age of 93 ± 6 ka. The 2σ consistency of the results derived from the two independent methods suggest that the breccia deposit and its fossil inclusions represent a single and synchronous event. We interpret the existing age difference between the two methods as being not significant, and likely to be partly resulting from the heterogeneity of the breccia, which makes the evaluation of the environmental dose rate quite complex. This illustrates the challenge of dating fossil-bearing breccias, which are commonly encountered in SE Asian caves, using palaeodosimetric methods (i.e., ESR and Luminescence), and especially in the absence of in situ measurements of radioactivity. Nevertheless, all sources of uncertainty considered, our numerical dating results systematically point towards a MIS 5 age for Ngalau Sampit. A finer correlation to sub-stages within MIS 5 would be too speculative based on the data available. Consequently, the fossil assemblage is chronologically positioned between Ngalau Gupin (Smith et al., 2021a) and Lida Ajer (Westaway et al., 2017; Louys et al. 2021a): these three localities now provide a sub-continuous Sumatran fossil record spanning from the late MIS 6 to the MIS 4. Sites dating to MIS 5 are rare in Southeast Asia but are critical for reconstructing faunal movements and evolution. Future work at Ngalau Sampit will focus on extending and describing the existing faunal remains to increase its biochronological and palaeoecological significance and enable cross-comparisons with other key Pleistocene fossil assemblages of SE Asia.
Permission for the research was granted by the Indonesian government – RISTEK Foreign Research Permit (Louys 2483/FRP/E5/Dit.KI/V/2018). The U-series and ESR dating studies have been funded by the Australian Research Council Future Fellowship Grant FT150100215 and the Spanish Ramón y Cajal Fellowship RYC2018-025221-I to M.D. The luminescence dating aspects were funded by an Australian Research Council Discovery grant DP170101597 to K.W. The fieldwork was supported by an Australian Research Council Future Fellowship to J.L. (FT160100450). We are grateful to María Jesús Alonso Escarza and Javier Iglesias Cibanal, CENIEH, for their support around the ESR dating analytical procedure. Finally, we thank Naturalis Biodiversity Center in Leiden, The Netherlands, for providing access to the Dubois’ archives.
The authors have no competing interests to declare.
Bedaso, ZK, Wynn, JG, Alemseged, Z and Geraads, D. 2013. Dietary and paleoenvironmental reconstruction using stable isotopes of herbivore tooth enamel from middle Pliocene Dikika, Ethiopia: Implication for Australopithecus afarensis habitat and food resources. Journal of human evolution, 64(1): 21–38. DOI: https://doi.org/10.1016/j.jhevol.2012.05.015
Bell, WT and Zimmerman, DW. 1978. The effect of HF acid etching on the morphology of quartz inclusions for thermoluminescence dating. Archaeometry, 20: 63–65. DOI: https://doi.org/10.1111/j.1475-4754.1978.tb00213.x
Bøtter-Jensen, L and Mejdahl, V. 1988. Assessment of beta dose-rate using a GM multicounter system. International Journal of Radiation Applications and Instrumentation. Part D. Nuclear Tracks and Radiation Measurements, 14: 187–191. DOI: https://doi.org/10.1016/1359-0189(88)90062-3
Buylaert, JP, Murray, AS, Thomsen, KJ and Jain, M. 2009. Testing the potential of an elevated temperature IRSL signal from K-feldspar. Radiat. Meas., 44: 560–565. DOI: https://doi.org/10.1016/j.radmeas.2009.02.007
Brennan, BJ. 2003. Beta doses to spherical grains. Radiat. Meas., 37: 299–303. DOI: https://doi.org/10.1016/S1350-4487(03)00011-8
Cheng, H., et al. 2016. The Asian monsoon over the past 640,000 years and ice age terminations. Nature, 534: 640–646. DOI: https://doi.org/10.1038/nature18591
Clark, TR, Roff, G, Zhao, J–X, Feng, Y–X, Done, TJ, Pandolfi JM. 2014. Testing the precision and accuracy of the U-Th chronometer for dating coral mortality events in the last 100 years. Quat. Geochronol. 23: 35–45. DOI: https://doi.org/10.1016/j.quageo.2014.05.002
Détroit, F., et al. 2019. A new species of Homo from the Late Pleistocene of the Philippines. Nature, 568: 181–186. DOI: https://doi.org/10.1038/s41586-019-1067-9
Dubois, MEFT. 1889–1890. unpublished notes and reports. In: The Dubois Aarchives, Naturalis Biodiversity Center, Leiden, the Netherlands, nos. MM774C-000040-447, MM774C-000040-457, MM774C-000050-011, MM774C-000050-042.
Duval, M. 2015. Electron Spin Resonance Dating of fossil tooth enamel. In Rink, WJ and Thompson, JW (eds.) Encyclopedia of Scientific Dating Methods, 239–246. Springer Netherlands. DOI: https://doi.org/10.1007/978-94-007-6304-3_71
Duval, M and Grün, R. 2016. Are published ESR dose assessments on fossil tooth enamel reliable? Quaternary Geochronology, 31: 19–27. DOI: https://doi.org/10.1016/j.quageo.2015.09.007
Galbraith, RF, Roberts, RG, Laslett, GM, Yoshida, H and Olley, JM. 1999. Optical dating of single and multiple grains of quartz from Jinmium rock shelter, northern Australia, part 1, Experimental design and statistical models. Archaeometry, 41: 339–364. DOI: https://doi.org/10.1111/j.1475-4754.1999.tb00987.x
Grün, R. 2000a. Methods of dose determination using ESR spectra of tooth enamel. Radiation Measurements, 32(5–6): 767–772. DOI: https://doi.org/10.1016/S1350-4487(99)00281-4
Grün, R, Aubert, M, Hellstrom, J and Duval, M. 2010. The challenge of direct dating old human fossils. Quaternary International, 223–224: 87–93. DOI: https://doi.org/10.1016/j.quaint.2009.10.005
Grün, R, Schwarcz, HP and Chadam, J. 1988. ESR dating of tooth enamel: Coupled correction for U-uptake and U-series disequilibrium. International Journal of Radiation Applications and Instrumentation. Part D. Nuclear Tracks and Radiation Measurements, 14(1–2): 237–241. DOI: https://doi.org/10.1016/1359-0189(88)90071-4
Hoffmann, D and Mangini, A. 2003. “A method for coupled ESR/U-series dating of teeth showing post-depositional U-loss.” Quaternary Science Reviews, 22(10–13): 1367–1372. DOI: https://doi.org/10.1016/S0277-3791(03)00032-5
Hutt, G, Jaek, I and Tchonka, J. 1988. Optical dating: K-feldspars optical response stimulation spectra. Quat. Sci. Rev., 7: 381–385. DOI: https://doi.org/10.1016/0277-3791(88)90033-9
Kaifu, Y. 2017. Archaic hominin populations in Asia before the arrival of modern humans: their phylogeny and implications for the “Southern Denisovans”. Curr. Anthropol., 58(S17): S418–S433. DOI: https://doi.org/10.1086/694318
Jacobs, Z, Duller, GA and Wintle, AG. 2006. Interpretation of single grain De distributions and calculation of De. Radiation Measurements, 41: 264–277. DOI: https://doi.org/10.1016/j.radmeas.2005.07.027
Le Fur, S, Fara, E, Mackaye, HT, Vignaud, P and Brunet, M. 2009. The mammal assemblage of the hominid site TM266 (Late Miocene, Chad Basin): ecological structure and paleoenvironmental implications. Naturwissenschaften, 96(5): 565–574. DOI: https://doi.org/10.1007/s00114-008-0504-7
Lisiecki, LE and Raymo, ME. 2005. A Pliocene-Pleistocene stack of 57 globally distributed benthic d18O records. Paleoceanography, 20: PA1003, DOI: https://doi.org/10.1029/2004PA001071
López-García, JM, Berto, C and Peresani, M. 2019. Environmental and climatic context of the hominin occurrence in northeastern Italy from the late Middle to Late Pleistocene inferred from small-mammal assemblages. Quaternary Science Reviews, 216: 18–33. DOI: https://doi.org/10.1016/j.quascirev.2019.05.025
Louys, J. 2008. Quaternary extinctions in southeast Asia. In Mass extinction, 159–189. Springer, Berlin, Heidelberg. DOI: https://doi.org/10.1007/978-3-540-75916-4_13
Louys, J and Roberts, P. 2020. Environmental drivers of megafauna and hominin extinction in Southeast Asia. Nature, 586: 402–406. DOI: https://doi.org/10.1038/s41586-020-2810-y
Louys, J and Turner, A. 2012. Environment, preferred habitats and potential refugia for Pleistocene Homo in Southeast Asia. Comptes Rendus Palevol, 11(2–3): 203–211. DOI: https://doi.org/10.1016/j.crpv.2011.03.003
Louys, J, Curnoe, D and Tong, H. 2007. Characteristics of Pleistocene megafauna extinctions in Southeast Asia. Palaeogeography, Palaeoclimatology, Palaeoecology, 243(1–2): 152–173. DOI: https://doi.org/10.1016/j.palaeo.2006.07.011
Louys, J, Meloro, C, Elton, S, Ditchfield, P and Bishop, LC. 2015. Analytical framework for reconstructing heterogeneous environmental variables from mammal community structure. Journal of human evolution, 78: 1–11. DOI: https://doi.org/10.1016/j.jhevol.2014.11.001
Louys, J, Kealy, S, O‘Connor, S, Price, GJ, Hawkins, S, Aplin, K, Rizal, Y, Zaim, J, Mahirta, Tanudirjo, DA, Santoso, WD, Hidayah, AR, Trihascaryo, A, Wood, R, Bevitt, J and Clark, T. 2017. Differential preservation of vertebrates in Southeast Asian caves. International Journal of Speleology, 46: 379–408. DOI: https://doi.org/10.5038/1827-806X.46.3.2131
Louys, J, Duval, M, Price, G, Westaway, K, Zaim, Y, Rizal, Y, Aswan, Puspaningrum, M, Trihascaryo, A, Breitenbach, S, Kwiecien, O, Cai, Y, Higgins, P, Albers, P, De Vos, J and Roberts, P. 2021a. Speleological and environmental history of Lida Ajer cave, western Sumatra. Philosophical Transactions of the Royal Society B. DOI: https://doi.org/10.1098/rstb.2020.0494
Louys, J, Zaim, Y, Rizal, Y, Puspaningrum, M, Trihascaryo, A, Price, GJ, Petherick, A, Scholtz, E and DeSantis, LR. 2021b. Sumatran orangutan diets in the Late Pleistocene as inferred from dental microwear texture analysis. Quaternary International, 603: 74–81. DOI: https://doi.org/10.1016/j.quaint.2020.08.040
Martínez-Navarro, B. 2010. Early Pleistocene faunas of Eurasia and hominin dispersals. In Out of Africa I, 207–224. Springer, Dordrecht. DOI: https://doi.org/10.1007/978-90-481-9036-2_13
Mason, VC, Helgen, KM and Murphy, WJ. 2019. Comparative phylogeography of forest-dependent mammals reveals Paleo-forest corridors throughout Sundaland. Journal of Heredity, 110(2): 158–172. DOI: https://doi.org/10.1093/jhered/esy046
Meckler, AN, et al. 2012. Interglacial hydroclimate in the tropical West Pacific through the Late Pleistocene. Science, 336: 1301–1304. DOI: https://doi.org/10.1126/science.1218340
Morwood, MJ, Soejono, RP, Roberts, RG, Sutikna, T, Turney, CS, Westaway, KE, … and Fifield, LK. 2004. Archaeology and age of a new hominin from Flores in eastern Indonesia. Nature, 431(7012): 1087–1091. DOI: https://doi.org/10.1038/nature02956
Murray, AS, Buylaert, JP, Thomsen, KJ and Jain, M. 2009. The effect of preheating on the IRSL signal from feldspar. Radiat. Meas. 44: 554–559. DOI: https://doi.org/10.1016/j.radmeas.2009.02.004
O’Connor, S., et al. 2017. Cave stratigraphies and cave breccias: Implications for sediment accumulation and removal models and interpreting the record of human occupation. Journal of Archaeological Science, 77: 143–159. DOI: https://doi.org/10.1016/j.jas.2016.05.002
Petraglia, M. 2017. Hominins on the move: An assessment of anthropogenic shaping of environments in the Palaeolithic. Human dispersal and species movement: From prehistory to the present, 90–118. DOI: https://doi.org/10.1017/9781316686942.005
Potts, R and Teague, R. 2010. Behavioral and environmental background to ‘Out-of-Africa I’and the arrival of Homo erectus in East Asia. In Out of Africa I, 67–85. Dordrecht: Springer. DOI: https://doi.org/10.1007/978-90-481-9036-2_5
Prescott, JR and Hutton, JT. 1994. Cosmic-ray contributions to dose rates for luminescence and ESR dating: large depths and long-term time variations. Radiat. Meas. 23: 497–500. DOI: https://doi.org/10.1016/1350-4487(94)90086-8
Price, GJ and Webb, GE. 2006. Late Pleistocene sedimentology, taphonomy and megafauna extinction on the Darling Downs, southeastern Queensland. Australian Journal of Earth Sciences, 53(6): 947–970. DOI: https://doi.org/10.1080/08120090600880842
Rhodes, EJ. 2015. Dating sediments using potassium feldspar single-grain IRSL: initial methodological considerations. Quaternary International, 362: 14–22. DOI: https://doi.org/10.1016/j.quaint.2014.12.012
Roberts, HM. 2012. Testing Post-IR IRSL protocols for minimising fading in feldspars, using Alaskan loess with independent chronological control. Radiation Measurements, 47: 716–724. DOI: https://doi.org/10.1016/j.radmeas.2012.03.022
Sarr, AC, et al. 2019. Subsiding Sundaland. Geology, 47: 119–122. DOI: https://doi.org/10.1130/G45629.1
Shackelford, L, Demeter, F, Westaway, KE, Duringer, P, Bacon, A-M, Ponche, JL, Sayavongkhamdy, T, Zhao, J-X, Barnes, L, Boyon, M, Sichanthongtip, P, Sénégas, Karpoff, A-M, Patole-Edoumba, E, Coppens, Y and Braga, J. 2018. Additional evidence for early modern human morphological diversity in Southeast Asia at Tam Pa Ling, Laos. Quat. Int. 466: 93–106. DOI: https://doi.org/10.1016/j.quaint.2016.12.002
Shao, Q, Bahain, J-J, Dolo, JM, Falguères, C. 2014. Monte Carlo approach to calculate US-ESR age and age uncertainty for tooth enamel. Quaternary Geochronology, 22: 99–106. DOI: https://doi.org/10.1016/j.quageo.2014.03.003
Shao, Q, Bahain, J-J, Falguères, C, Dolo, J-M and Garcia, T. 2012. A new U-uptake model for combined ESR/U-series dating of tooth enamel. Quaternary Geochronology, 10: 406–411. DOI: https://doi.org/10.1016/j.quageo.2012.02.009
Smith, HE, Morley, M and Louys, J. 2020. Taphonomic analyses of cave breccia in Southeast Asia: A review and future directions. Open Quaternary, 6: 13, 1–22. DOI: https://doi.org/10.5334/oq.75
Smith, HE, Price, GJ, Duval, M, Westaway, K, Zaim, J, Rizal, Y, Aswan, Puspaningrum, MR, Trihascaryo, A, Stewart, M and Louys, J. 2021a. Taxonomy, taphonomy and chronology of the Pleistocene faunal assemblage at Ngalau Gupin cave, Sumatra. Quaternary International, 603: 40–63. DOI: https://doi.org/10.1016/j.quaint.2021.05.005
Smith, HE, Bevitt, J, Zaim, J, Rizal, Y, Aswan, Puspaningrum, MR, Trihascaryo, A, Price, GJ, Webb, G and Louys, J. 2021b. High-resolution high-throughput thermal neutron tomographic imaging of fossiliferous cave breccias from Sumatra. Scientific Reports. DOI: https://doi.org/10.1038/s41598-021-99290-0
Thiel, C, Buylaert, JP, Murray, A, Terhorst, B, Hofer, I, Tsukamoto, S and Frechen, M. 2011. Luminescence dating of the Stratzing loess profile (Austria)–Testing the potential of an elevated temperature post-IR IRSL protocol. Quaternary International, 234(1–2): 23–31. DOI: https://doi.org/10.1016/j.quaint.2010.05.018
Thomsen, KJ, Murray, AS, Jain, M and Botter-Jensen, L. 2008. Laboratory fading rates of various luminescence signals from feldspar-rich sediment extracts. Radiat. Meas. 43: 1474–1486. DOI: https://doi.org/10.1016/j.radmeas.2008.06.002
Volmer, R, Hölzchen, E, Wurster, A, Ferreras, MR and Hertler, C. 2017. Did Panthera pardus (Linnaeus, 1758) become extinct in Sumatra because of competition for prey? Modeling interspecific competition within the Late Pleistocene carnivore guild of the Padang Highlands, Sumatra. Palaeogeography, Palaeoclimatology, Palaeoecology, 487: 175–186. DOI: https://doi.org/10.1016/j.palaeo.2017.08.032
Welker, F, Ramos-Madrigal, J, Kuhlwilm, M, Liao, W, Gutenbrunner, P, de Manuel, M, Samodova, D, Mackie, M, Allentoft, ME, Bacon, AM and Collins, MJ. 2019. Enamel proteome shows that Gigantopithecus was an early diverging pongine. Nature, 576(7786): 262–265. DOI: https://doi.org/10.1038/s41586-019-1728-8
Westaway, KE, Louys, J, Awe, RD, Morwood, MJ, Price, GJ, Zhao, JX, Aubert, M, Joannes-Boyau, R, Smith, TM, Skinner, MM, Compton, T, Bailey, RM, van den Bergh, GD, de Vos, J, Pike, AWG, Stringer, C, Saptomo, EW, Rizal, Y, Zaim, J, Santoso, WD, Trihascaryo, A, Kinsley, L and Sulistyanto, B. (2017). An early modern human presence in Sumatra 73,000–63,000 years ago. Nature, 548(7667): 322–325. DOI: https://doi.org/10.1038/nature23452
Wirkner, M and Hertler, C. 2019. Feeding ecology of Late Pleistocene Muntiacus muntjak in the Padang Highlands (Sumatra). Comptes Rendus Palevol, 18(5): 541–554. DOI: https://doi.org/10.1016/j.crpv.2019.03.004
Zanolli, C, Kullmer, O, Kelley, J, Bacon, AM, Demeter, F, Dumoncel, J, Fiorenza, L, Grine, FE, Hublin, JJ, Nguyen, AT and Nguyen, TMH. 2019. Evidence for increased hominid diversity in the Early to Middle Pleistocene of Indonesia. Nature ecology & evolution, 3(5): 755–764. DOI: https://doi.org/10.1038/s41559-019-0860-z
Zhao, J–X, Yu, K–F and Feng, Y–X. 2009. High-precision 238U-234U-230Th disequilibrium dating of the recent past: A review. Quat. Geochronol, 4: 423–433. DOI: https://doi.org/10.1016/j.quageo.2009.01.012