Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Geometric morphometrics and paleoproteomics enlighten the paleodiversity of Pongo

  • Jülide Kubat ,

    Roles Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – original draft, Writing – review & editing (JK); (AMB)

    Affiliations CNRS, BABEL, Université Paris Cité, Paris, France, Department of Palaeoanthropology, Senckenberg Research Institute and Natural History Museum Frankfurt, Frankfurt am Main, Germany

  • Ryan Paterson,

    Roles Data curation, Formal analysis, Investigation, Methodology, Validation, Visualization, Writing – review & editing

    Affiliation Globe Institute, University of Copenhagen, Copenhagen, Denmark

  • Ioannis Patramanis,

    Roles Data curation, Formal analysis, Investigation, Methodology, Software, Validation, Visualization, Writing – review & editing

    Affiliation Globe Institute, University of Copenhagen, Copenhagen, Denmark

  • Graeme Barker,

    Roles Resources, Writing – review & editing

    Affiliation McDonald Institute for Archaeological Research, University of Cambridge, Cambridge, United Kingdom

  • Fabrice Demeter,

    Roles Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing

    Affiliations Lundbeck Foundation GeoGenetics Centre, Globe Institute, University of Copenhagen, Copenhagen, Denmark, UMR 7206 Eco Anthropologie, Muséum National d’Histoire Naturelle, CNRS, Paris, France

  • Arnaud Filoux,

    Roles Resources, Writing – review & editing

    Affiliation Palaeontological Research and Education Centre, Mahasarakham University, Mahasarakham, Thailand

  • Ottmar Kullmer,

    Roles Resources, Writing – review & editing

    Affiliation Department of Palaeoanthropology, Senckenberg Research Institute and Natural History Museum Frankfurt, Frankfurt am Main, Germany

  • Meaghan Mackie,

    Roles Data curation, Investigation, Writing – review & editing

    Affiliations Globe Institute, University of Copenhagen, Copenhagen, Denmark, The Novo Nordisk Foundation Center for Protein Research, University of Copenhagen, Copenhagen, Denmark

  • Tomas Marques-Bonet,

    Roles Resources, Writing – review & editing

    Affiliations Institute of Evolutionary Biology (UPF-CSIC), PRBB, Barcelona, Spain, Catalan Institution of Research and Advanced Studies (ICREA), Passeig de Lluís Companys, Barcelona, Spain, CNAG-CRG, Centre for Genomic Regulation (CRG), Barcelona Institute of Science and Technology (BIST), Barcelona, Spain, Institut Català de Paleontologia Miquel Crusafont, Universitat Autònoma de Barcelona, Barcelona, Spain

  • Nguyen Thi Mai Huong,

    Roles Resources

    Affiliation Anthropological and Palaeoenvironmental Department, Institute of Archaeology, Ha Noi, Vietnam

  • Nguyen Anh Tuan,

    Roles Resources

    Affiliation Anthropological and Palaeoenvironmental Department, Institute of Archaeology, Ha Noi, Vietnam

  • Sytha Pheng,

    Roles Resources

    Affiliation Royal University of Fine Arts, Phnom Penh, Cambodia

  • Jessica Rippengal,

    Roles Resources

    Affiliation McDonald Institute for Archaeological Research, University of Cambridge, Cambridge, United Kingdom

  • Friedemann Schrenk,

    Roles Resources, Writing – review & editing

    Affiliation Department of Palaeoanthropology, Senckenberg Research Institute and Natural History Museum Frankfurt, Frankfurt am Main, Germany

  • Viengkeo Souksavatdy,

    Roles Resources

    Affiliation Department of Heritage, Ministry of Information, Culture and Tourism, Vientiane, Lao People’s Democratic Republic

  • Lim Tze Tshen,

    Roles Resources, Writing – review & editing

    Affiliation Department of Geology, Faculty of Science, Universiti Malaya, Kuala Lumpur, Malaysia

  • Athiwat Wattanapituksakul,

    Roles Resources

    Affiliation Palaeontological Research and Education Centre, Mahasarakham University, Mahasarakham, Thailand

  • Wei Wang,

    Roles Resources

    Affiliation Institute of Cultural Heritage, Shandong University, Qingdao, China

  • Clément Zanolli,

    Roles Conceptualization, Data curation, Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – review & editing

    Affiliation CNRS, MCC, PACEA, UMR 5199, Univ. Bordeaux, Pessac, France

  • Enrico Cappellini,

    Roles Funding acquisition, Methodology, Project administration, Resources, Supervision, Writing – review & editing

    Affiliation Globe Institute, University of Copenhagen, Copenhagen, Denmark

  •  [ ... ],
  • Anne-Marie Bacon

    Roles Conceptualization, Funding acquisition, Project administration, Resources, Supervision, Writing – review & editing (JK); (AMB)

    Affiliation CNRS, BABEL, Université Paris Cité, Paris, France

  • [ view all ]
  • [ view less ]


Pleistocene Pongo teeth show substantial variation in size and morphology, fueling taxonomic debates about the paleodiversity of the genus. We investigated prominent features of the enamel-dentine-junction junction (EDJ)–phylogenetically informative internal structures–of 71 fossil Pongo lower molars from various sites by applying geometric morphometrics and conducted paleoproteomic analyses from enamel proteins to attempt to identify extinct orangutan species. Forty-three orangutan lower molars representing Pongo pygmaeus and Pongo abelii were included for comparison. The shape of the EDJ was analyzed by placing five landmarks on the tip of the main dentine horns, and 142 semilandmarks along the marginal ridges connecting the dentine horns. Paleoproteomic analyses were conducted on 15 teeth of Late Pleistocene Pongo using high-resolution tandem mass spectrometry. The geometric morphometric results show variations in EDJ shape regarding aspects of the height and position of the dentine horns and connecting ridges. Despite the issue of molar position and sample size, modern molars are distinguished from fossil counterparts by their elongated tooth outline and narrowly positioned dentine horns. Proteomic results show that neither a distinction of P. pygmaeus and P. abelii, nor a consistent allocation of fossil specimens to extant species is feasible. Based on the EDJ shape, the (late) Middle to Late Pleistocene Pongo samples from Vietnam share the same morphospace, supporting the previous allocation to P. devosi, although substantial overlap with Chinese fossils could also indicate close affinities with P. weidenreichi. The hypothesis that both species represent one chronospecies cannot be ruled out. Two fossil specimens, one from Tam Hay Marklot (Laos, Late Pleistocene), and another from Sangiran (Java, Early to Middle Pleistocene), along with some specimens within the Punung sample (Java), exhibit affinities with Pongo abelii. The Punung fossils might represent a mix of early Late Pleistocene and later specimens (terminal Pleistocene to Holocene) related to modern Pongo. The taxonomy and phylogeny of the complete Punung sample needs to be further investigated.

1. Introduction

A large geographical and temporal distribution of extinct orangutans (Pongo) is evidenced by the rich Early to Late Pleistocene fossil record ranging from southern China and continental Southeast Asia (Vietnam, Laos, Cambodia, Thailand, Peninsular Malaysia) to insular Southeast Asia–Sumatra, Java, and Borneo [15]. In a recent revision of the Pleistocene Pongo species diversity, Harrison et al. [6, 7] proposed to assign all Early to Late Pleistocene specimens from southern China and Vietnam to Pongo weidenreichi, encompassing in one single species all taxa defined previously [3]. Late Pleistocene specimens from the Sibrambang in the Padang highlands (Central Sumatra) were recognized as Pongo palaeosumatrensis [1], whereas the Late Pleistocene orangutans from Niah Cave (Sarawak, Malaysian Borneo) were assigned to the modern species Pongo pygmaeus [810]. Other extinct samples from Indonesia were tentatively attributed to Pongo javensis from the Late Pleistocene site of Punung (Central Java) and Pongo duboisi from the Late Pleistocene Lida Ayer and Djambu sites in the Padang highlands (Central Sumatra) byHarrison [6] based on Drawhorn’s work [11].

Extant species can be largely distinguished based on several features related to hair color, cheek flanges, various cranial features, DNA, and geographical distribution [6, 1217]. To date, only three Pongo species remain in the Ponginae subfamily, namely P. abelii and P. tapanuliensis in Sumatra, and P. pygmaeus in Borneo [5, 1720]. The reasons for this impoverishment of diversity and decline in geographical distribution of Pongo are linked to the loss of suitable habitats caused by multiple factors such as climatic change during the Pleistocene and anthropic impacts such as hunting and deforestation [21]. The slow reproduction rates and prolonged dependence of infants on their mothers makes them additionally vulnerable when facing these impacts [22]. This might also have caused the decrease in Pongo populations in the Holocene.

Soft tissue, hair, and/or DNA have not been recovered so far in fossil orangutans and cranial remains are rare. In the absence of these features, the most diagnostic elements in the fossil records are teeth. However, it was proposed that mixed samples of isolated teeth of extant species such as P. abelii and P. pygmaeus would be impossible to separate from each other solely based on their external morphology [6]. Therefore, if even well-identified extant species are difficult to separate based on dental traits, the validity of fossil Pongo species identification raises questions, given that most fossilized remains are isolated teeth. Furthermore, occlusal wear during the lifetime of an individual and post-mortem taphonomy can alter the shape of the crown, which makes it especially difficult to identify species based on the macromorphology of molars [23, 24].

Extinct Pongo exhibits a high degree of variation in tooth size and morphology [5, 6]. Several attempts to distinguish taxa at the (sub)species level have been proposed based on odontometric measurements and non-metric features recorded on teeth [3, 6, 7]. Criteria such as relative tooth size, degree of crenulation on the occlusal surfaces of molars and premolars, size of dental crown areas, and variability of buccolingual and mesiodistal tooth dimensions, have been applied to identify morphotypes relating to paleospecies [1, 3, 6, 7, 11, 25]. However, any of these criteria cannot be used as secure diagnostic features due to the continuous gradient of these traits among fossil specimens and/or to the large morphodimensional overlap between Pongo samples [5].

Hence, the characterization of these species may significantly benefit from geometric morphometrics of internal structures like the enamel-dentine-junction (EDJ). The EDJ is protected by the enamel and, therefore, generally less affected by taphonomic damage and occlusal wear than the surface of the crown [23, 24, 26, 27]. This structure has also been recognized as a reliable taxonomic proxy, even at the subspecies level [23, 26, 27].

In addition, paleoproteomic analyses of dental enamel can be performed as a complementary approach. Ancient proteins from fossil tooth enamel are proven to clarify species identifications and phylogenetic relationships, notably when ancient DNA is not preserved or accessible [2835]. Proteins are more resistant to diagenesis than DNA and recent analyses prove that they can be successfully retrieved from dental enamel of Early Pleistocene fossils [36, 37].

Here, we applied geometric morphometrics on 71 mandibular molars of Pongo (Table 1 and S1 Table) from various Pleistocene sites and museum collections (Punung and Sangiran from Indonesia; Lang Trang, Tham Om, Hang Hum, Mai Da Dieu, Duoi U’Oi from Vietnam; Tam Hay Marklot and Tam Hang South from Laos; Ganxian from China; Chinese Apothecary collection housed at the Senckenberg Research Institute and Museum, Frankfurt). In addition, 43 modern Pongo mandibular molars were included in the dataset with both species P. pygmaeus and P. abelii represented. Paleoproteomics of dental enamel was performed on 15 Late Pleistocene teeth from Vietnam, Laos, Malaysia, and Thailand. Four of these specimens are mandibular molars and thus included in the geometric morphometric analyses.

Table 1. Fossil specimens included in the geometric morphometric analyses.

Our aim is to obtain morphological and molecular data to characterize the diversity in fossil Pongo and explore their phylogenetic relationship to modern species. We aim particularly to verify if our results obtained with geometric morphometrics of the EDJ on a set of lower molars and paleoproteomics of dental enamel complement each other and fit with the distinction of extinct species established in previous studies [6]. Furthermore, we aim to characterize enamel protein sequences of extinct species and compare them with modern orangutan protein sequences.

2. Material and methods

2.1 Materials

Our analysis surveyed Pongo samples from continental and insular sites in Southeast Asia. A total of 114 teeth were analyzed in this study, including 71 fossils and 43 teeth from extant species. The details of their location and chronology, along with the original allocations of Pongo specimens are shown in Table 1. The fossil sample (Table 1 and S1 Table) comprises 35 teeth from Vietnam, 25 from Indonesia, 9 from China, one from Malaysia and one from Laos, all lower molars. Within the modern molars 20 are allocated to Pongo abelii and 23 to Pongo pygmaeus, all with known molar positions on the tooth row. For paleoproteomics, fifteen fossil teeth were sampled (Table 2). The sample set, either premolar, molar or canine, permanent or deciduous, includes six teeth from Laos (Tam Hay Marklot n = 5, Tam Hang South n = 1), three from Vietnam (Duoi U’Oi n = 2, Hoà Binh n = 1), four from Malaysia (Niah Cave), one from Indonesia (Sangiran) and one from Thailand (Tham Prakai Phet).

2.2 Fossil sites

2.2.1 China.

The Ganxian Cave located in the Bubing Basin, Guangxi Zhuang Autonomous Region, is dated between 362 ± 78 ka and 168.9 ± 2.4 ka by Uranium–series (U-series) and coupled Electron Spin Resonance (ESR) and /U-series dating [38]. Specimens from this site have been allocated to Pongo weidenreichi [38]. The Chinese Apothecary collection comprises isolated teeth of several specimens attributed to Sinanthropus officinalis (Homo erectus), Hemanthropus peii. Gigantopithecus and fossil Pongo. The teeth were purchased by Gustav Heinrich Ralph von Koenigswald in traditional Chinese drugstores, and therefore their provenance and dating remain unknown [39, 40].

2.2.2 Vietnam.

The Lang Trang cave system is composed of multiple breccia deposits. The fauna has been biochronologically dated to ~80 ka, whereas the ESR dates for the cave breccias gave an age interval of 480–146 ka, which makes this site difficult to characterize chronologically. The molars used in this study have first been assigned to Pongo pygmaeus [4143]. The Pongo specimens from two fossil assemblages, Tham Om (province of Nghe An) and Hang Hum (province of Yên Bai) in northern Vietnam, are supposed to be 300–200 ka and 140–80 ka respectively, based on biochronology [3, 44]. The Duoi U’Oi site located in the Hoà Binh province is dated to 70–60 ka by Optically stimulated Luminescence (OSL) and U-series dating and teeth have been identified as Pongo pygmaeus [45]. No dates are reported for the Pongo specimen from Hoà Binh, since radiocarbon dating was inconclusive [4]. The Mai Da Dieu Rockshelter is located in the limestone mountains of Thanh Hoa Province. Pongo fossils derived from layer 1 with radiocarbon dates of 30–24 ka [46]. The Thung Lang specimens [47] have never been dated.

2.2.3 Laos.

The specimens from Laos come from Tam Hay Marklot Cave, in which a set of mammal teeth have been dated to 38.4–13.5 ka by Uranium-thorium (U-Th) [48] and from Tam Hang South dated to 94–60 ka using OSL and U-series methods on fossiliferous deposits [49, 50]. Both sites are located in the Huà Pan province in northeastern Laos. Molars are identified as Pongo pygmaeus.

2.2.4 Thailand.

Tham Prakai Phet cave is located in northeast Thailand and the occurrence of fossil Pongo is biochronologically estimated to the Late Pleistocene. Specimens are attributed to Pongo sp. [51].

2.2.5 Malaysia.

The Late Pleistocene human occupation at Niah Cave, Sarawak (Malaysian Borneo) is dated back to 50–45 ka by radiocarbon dating of charcoal and U-series analyses on bone. Specimens associated with it are allocated to Pongo pygmaeus [810].

2.2.6 Indonesia.

The specimens from Indonesia come from the Early to Middle Pleistocene Sangiran Dome, Java. The hominid bearing layers at the Sangiran Dome cover an age range of 1.5 Ma to 500 ka by 40Ar/39Ar, fission-track and Uranium-lead (U–Pb) dating [52, 53]. The exact provenance of the Pongo material from Sangiran is not documented, hence it is not possible to attach a secure date [54, 55]. The Punung specimens studied here were collected by Gustav Heinrich Ralph von Koenigswald in the early 1930s in the Punung I and II localities [56]. The Late Pleistocene Punung fauna represents a tropical rainforest fauna for the first time on Java and is dated to 128–117 ka using luminescence and U-series methods on sediments from the nearby Gunung Dawung site, named Punung III [57]. Fossils from Punung are allocated to Pongo pygmaeus [58]. Drawhorn [11] suggested to assign these specimens to Pongo javensis, which was echoed by Harrison et al. [6]. However, Pongo javensis as a new species was never formally published. The site, located in the Gunung Sewu karst region in East Java, has been recently reanalyzed with a new interpretation of fossiliferous deposits [59].

The specimens from Indonesia and the Chinese Apothecary Collection are housed at the Senckenberg Research Institute and Museum Frankfurt am Main, Germany. Pongo molars from several Vietnamese Pleistocene sites are all housed at the Institute of Archaeology in Hanoi, Vietnam. The new faunal assemblages from Laos are curated at the National Museum in Vientiane, Laos and the National Natural History Museum in Paris, France for the old collection of Thung Lang. Fossil specimens from Ganxian Cave are curated at the Natural History Museum and Anthropology Museum of Guangxi, China. Fossils from Niah Cave were housed at the University of Cambridge, UK, when they were studied but have since been returned to Sarawak Museum, Malaysia. No permits were required for the described study, which complied with all relevant regulations.

2.3 Methods

2.3.1 Morphological analyses.

X-ray microcomputed tomography (X-μCT). All specimens were scanned using the X-ray microfocus sources (X-μCT) (Phoenix Nanotom S 180; General Electric Company) with the following parameters: 100–160 kV, 0.11–90 μA, 0.14–0.36° between each projection and a voxel size ranging between 0.02 and 0.04 mm.

Segmentation of microCT data. Segmentation and landmarking were performed in Avizo v.7.0 (FEI Visualization Sciences Group). Dentine surfaces of each specimen were segmented using the magic wand tool with manual corrections locally where contrast was low. In cases where resolution and contrast were too weak, the interpolation tool was used. Interpolations were also utilized for reconstructions of morphological structures such as dentine horn tips in cases of damage due to taphonomy or occlusal wear (see Zanolli et al. [63]). Volumetric models were generated for each specimen.

Molar position assessment. As noted by Hooijer [1], Badoux [56], Groesbeek [64] and Harrison [65], and confirmed by our observations, it was not possible to reliably determine the serial position of the isolated molars of our fossil samples. Nevertheless, we tried to identify M3s among the isolated fossil teeth to minimize any potential bias in the phylogenetic analyses as M3s show typical traits such as a marked distal tapering and low cusps and dentine horns, as well as the absence of an interproximal distal contact facet. However, the latter feature is not necessary always reliable as there can be the presence of a M4 (with a frequency of 7–13% in extant orangutans [56, 66], and in case of unerupted or just erupted M1 or M2, since the distally positioned teeth are not erupted, there is no interproximal distal contact facet. Based on majority consensus among three observers (JK, CZ, AMB) molar positions were assessed (S3 Fig). Analyses with all molar positions combined in one single CVA analyses are shown in S4 Fig to compare how this affects the taxonomic signal.

Geometric morphometric analyses. We mirrored all left molars in the sample, so that all teeth are virtually considered as right antimeres to ensure that (semi)landmarking of each specimen is homologous. We placed landmarks and semilandmarks in the same order as follows: Protoconid—Metaconid—Entoconid—Hypoconulid—Hypoconid—Protoconid. Five landmarks were placed on the surface of the tip of the main dentine horns combined with 142 semilandmarks along the marginal ridges connecting the dentine horns of the enamel-dentine junction (EDJ) using the Bspline tool. Generalized Procrustes analyses, principal component analyses (PCA) and canonical variate analyses (CVA) based on Procrustes shape coordinates and geographical origin as groups were carried out using the software R version 4.1.1 [67]. Groups were defined as China, Vietnam and Indonesia for the fossil specimens, and Sumatran orangutans (Pongo abelii) and Bornean orangutan (Pongo pygmaeus) for the living species. Considering the ambiguity of metameric position for a number of specimens, we conducted geometric morphometric analyses using various (sub)sets of specimens that include: (i) M1 and M2 for fossil Pongo based on our molar position classification and M1 and M2 of modern Pongo, (ii) M3s for fossil Pongo based on our molar position classification (including specimens for which the M3 position is ambiguous and could represent M2s) and M3s of modern Pongo, and (iii) all molar positions available for fossil and modern Pongo. Since CVA computation requires the number of variables to be much smaller than the number of specimens, we computed the CVA based on a subset of the first PC scores (12 PCs for the M1–M2 analysis, and 15 PCs for the M3 analysis) showing the highest degree of correct classification (screening the correct classification results and selecting the minimum number of PC scores enabling to reach the optimum of correct classification) [68]. This choice of the PC scores subset is a compromise between including a sufficient proportion of overall shape variation and limiting the number of variables to avoid unrealistic and unstable levels of discrimination.

Specimens from Sangiran and those used for paleoproteomic analyses were excluded from the CVA and projected a posteriori in the plots because the number of these specimens per locality is limited and computing a CVA with less than 5 specimens would not provide reliable results. Cross-validations were run to check that CVA results do not show spurious group discrimination by looking at the cross-validated classification accuracy (S2 and S3 Tables). Cross-validated CVA plots were inspected as well to compare if there is broad agreement on group distribution in the morphospace (S1 Fig). We tested allometry based on the Pearson correlation coefficient between the centroid size and CV1, CV2 and CV3, respectively. In addition, we calculated the squared correlation coefficient between centroid size and canonical variates.

2.3.2 Paleoproteomics.

Protein extraction. Ancient protein extraction was carried out in clean laboratory facilities at the University of Copenhagen dedicated to the extraction of ancient biomolecules. A sterilized drill was used to remove flakes of enamel from intact molars and premolars. One specimen (SMF 8864) was embedded in resin for enamel histological analyses in a previous study [69] prior to laboratory analysis. This specimen (SMF 8864) was coated with two different resins, Crystalbond 509 (SPI Supplies) soluble in acetone which directly covered the teeth and a second layer coating with epoxy resin (EpoThin 2; Buehler), which is not soluble [69]. The specimen embedded in resin was cut and submerged in acetone for 3 hours to remove the resin. The specimen was washed with distilled water and subsequently with 10% trifluroacetic acid (TFA) for 2 minutes to remove residues and left for drying for 12 hours.

For each specimen, approximately 25 mg of enamel was sampled from the non-occlusal portion of the crown. Care was taken to remove any residual dentine remaining on enamel flakes following drilling. These enamel flakes were subsequently crushed to a rough powder, which was demineralized using 10% TFA at 4°C for 24 hours. Samples were subsequently processed using a digestion-free protocol developed for recovering degraded enamel proteins [36]. C-18 StageTips [70] were used to collect and desalt peptides. Negative extraction blanks were processed alongside each batch of samples, to control for contamination.

LC-MS/MS. StageTips were eluted into a 96-well MS plate using 30 μL of 40% acetonitrile (ACN) 0.1% formic acid (FA). To remove the ACN and concentrate the samples, samples were vacuum-centrifuged until approximately 3 μL of sample remained. Finally, samples were resuspended in 10 μL of 5% ACN 0.1% TFA, except for SMF 8864, which was resuspended in 6 μL instead based on observations of peptide concentrations at this site from previous tested suid specimens (not reported). Liquid chromatography-tandem mass spectrometry (LC-MS/MS) was used to analyse the samples, using established protocols for paleoproteomic samples [36, 71]. Samples separation was completed on a 15 cm column (75 μm inner diameter) in-house laser pulled and packed with 1.9 μm C18 beads (Dr. Maisch, Germany) on an EASY-nLC 1200 (Proxeon, Odense, Denmark) connected to an Exploris 480 Orbitrap mass spectrometer (Thermo Scientific, Bremen, Germany). 0.5 μL was injected for all samples except SMF 8864, for which 3 μL was used, based on the total ion current of small test injections (0.5 μL injections over 15 min). Buffer A was milliQ water and the peptides were separated with increasing buffer B (80% ACN and 0.1% FA) with a 77 min gradient, increasing from 5% to 30% in 50 min, 30% to 45% in 10 min, 45% to 80% in 2 min, held at 80% for 5 min and decreased back to 5% in 5 min and held for 5 min. Flow rate was 250 nL/min. An integrated column oven was used to maintain the temperature at 40°C. A wash blank, using 5% ACN 0.1% TFA, was run in between each sample to hinder cross-contamination.

The Exploris 480 recorded full scan mass spectra (MS1) at a resolution of 120,000 over the m/z range 350–1400 with a target AGC of 300% and a maximum injection time of 25 ms. Spray voltage was set to 2kV, the heated capillary at 275°C, and the S-lens RF level was at 40%. HCD-generated product ions (MS2) were recorded in data-dependent top 10 mode with a maximum ion injection time set to 118 ms and a target ACG value of 200%, recorded at a resolution of 60,000. HCD collision energy was set at 30% and the isolation window was 1.2 m/z with the dynamic exclusion set to 20 s.

Data analysis. For an initial database search, a broad taxonomic database was constructed by translating those proteins commonly encountered in enamel from publicly-available genomic data for catarrhines [33, 37]. These sequences were supplemented with sequences from various other common bone and dentine proteins available on Uniprot for P. abelii and P. pygmæus. A second reference database on common laboratory contaminants was also included in each database.

Proprietary.raw files generated by the mass spectrometer were searched using MaxQuant (v1.6.3.4) [72], for the main protein sequence reconstruction. Raw files were additionally searched using PEAKS, via the spider algorithm [73], and PFind, via its open search strategy to potentially discover single amino acid polymorphisms (SAPs) [74]. Initial database searches with MaxQuant were run against the two aforementioned databases. A minimum Andromeda score of 60 was set for both modified and unmodified peptides, with a minimum length of seven amino acids and a maximum of 25 amino acids. The default tolerance settings were used– 20 ppm for the first search and 4.5 ppm for the final search, with a fragment mass tolerance of 20 ppm. The peptide false-discovery rate (FDR) was set to 1.0%. Delta score was set at zero for both modified and unmodified peptides. No fixed post-translational modifications (PTMs) were set. The following variable modifications were included: glutamine and asparagine deamidation, oxidation of methionine, proline and tryptophan, phosphorylation of serine and threonine, and pyroglutamic acid from N-terminal glutamic and aspartic acid.

Similar search parameters were selected in PFind and PEAKS. In PFind, the peptide FDR remained 1%, but the protein FDR was set to 10%, and the peptide mass window was set at 350–5000. In PEAKS, the average local confidence was set to 90%. Potential SAPs discovered in PFind and PEAKS were then included in manually-constructed alternative protein sequences, to be included in a final validatory run using MaxQuant alongside the initial hominid-only database. Besides the updated database, the final MaxQuant run used the same parameters as the initial database searches, but also included an additional PTM (ornithine conversion from arginine).

Common laboratory contaminants and reverse hits from the FDR calculation were removed from consideration for all further aspects of the analysis. Peptide sequences matching to spectra with poorly-covered ion series were discarded prior to sequence reconstruction. Protein sequences were reconstructed using MAFFT v.7 [75], specifically the–addfragments option, for which individual peptide sequences were aligned to the Pongo abelii sequences for each protein, prior to generating a consensus sequence. Deamidation was calculated using a python script on the evidence.txt output file from the final MaxQuant run, as described in Mackie et al. [72] and Cappellini et al. [36]. Figures were constructed using GGplot2 [76].

Protein reference dataset for phylogenetic analyses. We created a reference dataset for 8 proteins common in dental enamel: alpha-2-HS-glycoprotein (AHSG), albumin (ALB), ameloblastin (AMBN), amelogenin (AMEL), amelotin (AMTN), collagen alpha-1(XVII) (COL17A1), enamelin (ENAM), and matrix metalloproteinase-20 (MMP20), that includes all extant hominid genera. Additionally, the proteins collagen alpha-1 (I) (COL1A1) and collagen alpha-2 (I) (COL1A2), which are common in dentine, were included in the same dataset in case traces of dentine were accidentally co-extracted.

Since protein data for Pongo and other non-human great apes is scarce, we translated the proteins of interest from publicly available genomic data [17, 7780]. This translated dataset includes 27 Pongo individuals from all 3 extant species of the Pongo genus (15 P. pygmaeus, 11 P. abelii, 1 P. tapanuliensis) and captures much of the within-genus protein diversity of these 9 proteins. We combined this reference dataset with the protein sequences from the 15 fossil Pongo samples.

Protein phylogenetic analysis. We created a multiple sequence alignment (MSA) for each one of these 9 proteins using MAFFT [70]. We then identified all positions in the alignment, covered by at least one fossil sample, with either an isoleucine (I) or leucine (L) amino acid. These two amino acids are isobaric and are impossible to differentiate using the mass spectrometric procedure described here [33]. For those positions, if all modern samples had one of the two amino acids, we parsimoniously switched the fossil samples to the same. If both amino acids are present within the modern individuals, we switched all samples of the alignment, both modern and fossil, to a leucine. Finally, we concatenated the 9 MSAs into a single alignment and used it as input for two phylogenetic software: PhyML (Maximum likelihood) [81] and MrBayes (Bayesian) [82]. Using these two software tools, we generated two species-phylogenetic trees.

3. Results

3.1 Geometric morphometrics

3.1.1 Extant Pongo.

CVA analyses with M1 and M2 combined. In the canonical variate analysis (CVA) (Fig 1A), CV1 (axis X) and CV2 (axis Y) represent 42.11% and 32.41% of the total variance, respectively. The CVA plot shows that most of the P. pygmaeus molars are grouping on the positive part of CV1, whereas the P. abelii molars appear clearly distinct on the negative part. Along this axis, P. pygmaeus molars group together with Indonesian fossil molars, whereas the majority of fossils from Vietnam and China cluster on the negative part together with P. abelii. The shape change along CV1 indicates higher and more inward curved dentine horns for Vietnamese and Chinese specimens and P. abelii. P. pygmaeus and Indonesian molars show lower dentine horns, especially less protuberant hypoconid.

Fig 1.

a. Canonical variate analyses (CVA) of the enamel-dentine-junction (EDJ) based on M1s and M2s. Specimens from the Sangiran Dome (Java, Indonesia) and Laos are projected a posteriori to the analyses due to the limited number of available specimens for these sites. All these specimens are highlighted with respective labels (SMF8864, SMF8858 and SMF8879, Sangiran Dome, Java; ML5, Tam Hay Marklot, Laos). EDJ shape changes and outline of crowns along canonical variates are illustrated. The maximum of extreme shape change along CV1 is illustrated in blue, while the minimum is shown in red, for CV2: maximum = green, minimum = orange; for CV3 maximum = purple, minimum = black. Specimens SMF 8864 (Sangiran Dome), and ML5 (Tam Hay Marklot, Laos) are included in the paleoproteomic analyses as well. 1b. CVA of the EDJ based on M3s. One specimen from Malaysia is projected a posteriori to the analyses due to the limited number of available specimens for this site and highlighted with the respective label (S2258; Niah Cave, Malaysian Borneo). EDJ shape changes and outline of crowns along canonical variates are illustrated. The maximum of extreme shape change along CV1 is illustrated in blue, while the minimum is shown in red, for CV2: maximum = green, minimum = orange; for CV3 maximum = purple, minimum = black. Specimens DU1177 (Duoi U’Oi, Vietnam) and S2258 (Niah Cave, Malaysian Borneo) are included in the paleoproteomic analyses as well.

A more evident separation of extant and fossil specimens occurs along CV2 where extant species cluster in the positive part whereas fossil groups are in the negative area. However, some little overlaps exist between fossils from Vietnam and Indonesia and modern molars, especially P. pygmaeus, while Chinese molars range show no overlap with ranges of modern taxa.

Along CV2, the outline of the tooth changes. It shifts from an elongated, almost rectangular, and narrow crown mesiodistally expressed through buccal and lingual dentine horns. Dentine horns are narrower and more closely arranged in modern Pongo molars compared to the more peripheral dentine horns in the fossil molars with a rounder and shorter enamel-dentine junction (EDJ) outline. The shape change along CV1 is mainly driven by the height of the dentine horns, while CV2 is mainly determined by the position of the dentine horns and by the shape of the outline.

The degree of cross-validated correct classification (reported in S2 and S3 Tables, illustrated in S1 Fig) for P. abelii and P. pygmaeus is 77% and 69%, respectively, which shows that the two species of living Pongo included in this study are relatively well distinguished from each other and from fossil groups. The overall classification accuracy is 67% indicating a general separation of groups, even if some overlap exists. There is no significant allometric trend between fossil and modern specimens regarding the grouping of molars along CV1 and CV3 (p = 0.89 and p = 0.44, respectively). Along with CV2 there is little allometry with a p-value of < 0.05 and r2 = 0.22.

3.1.2 Fossil Pongo.

Fossils from the Middle to Late Pleistocene sites in Vietnam share the same morphospace along all 3 CVs and cluster on the negative area (Fig 1A). The cross-validated classification results show that 63% of Vietnamese fossils were correctly classified with substantial overlap to Chinese and Indonesian fossils. The overlap especially concerning China and Vietnam is evident in CV1 vs. CV2 (Fig 1A) and in the cross-validated classification results, which show that 44% of Chinese fossils were correctly classified, whereas 33% were identified as Vietnamese fossils (S2 Table). In CV2 vs. CV3 (Fig 1A; CV3 representing 15.79% of the total variance) fossils from China separate to some extent from Vietnam indicating some shape differences captured along CV3. Moreover, the Chinese sample is showing a high variation along CV3 due to the large morphospace occupied, which might indicate high variability of the Chinese group (Fig 1A). The majority of the specimens group on the far positive part along CV3, which is characterized by slightly sharper dentine horns with a prominent entoconid as opposed to the far negative area along CV3 where most of the Vietnamese fossils group, which show a relatively low entoconid expression. Three specimens from Ganxian (GX010, GXI007, GX3012) partially overlap with some specimens from Vietnam. The rest of the Middle Pleistocene Ganxian and Chinese Apothecary specimens fall together in the positive space of CV3. The Punung specimens overlap with fossil Pongo from Vietnam and China. Several specimens from Punung fall within the range of P. pygmaeus as well. Indonesian fossils show 74% of correct classification.

The molars from Sangiran (Indonesia), Niah Cave (Malaysia) and Tam Hay Marklot (Laos) were projected a posteriori to the CVA plot due to the limited number of specimens. One molar from Laos labeled “ML5” in Fig 1A falls close to the P. abelii group, a result supported by the typicality probability (Table 3). Specimen SMF 8858 from the Sangiran Dome groups with fossils from China, and Vietnam, whereas SMF 8864 falls in the morphospace shared by fossils from Punung, Indonesia and China. Both Indonesian specimens are well discriminated from the extant species. Typicality probabilities show a higher affinity of SMF 8858 to the Chinese group, and of SMF 8864 to the Indonesian group (Table 3). Specimen SMF 8879 does not show particular affinity to any of the fossil groups, but exhibits a close proximity to P. abelii, which is also suggested by the typicality probabilities.

Table 3. Typicality probabilities of the investigated specimens.

CVA analyses of M3. The CVA analyses for M3s follow the trend seen in M1 and M2s regarding the grouping of extant and fossil specimens. However, Chinese fossils are missing because of the lack of identifiable M3s in the sample (Fig 1B; CV1 representing 50.31% of the total variance and CV2 representing 35.67%). P. abelii and Vietnamese fossils cluster in the negative area along CV1 with Indonesian fossils and P. pygmaeus grouping on the positive part. These results are similar to the previous plots where M1 and M2 are analyzed. The shape change along CV1 exhibits more inward curved dentine horns with slightly more prominent distal dentine horns for specimens from Vietnam and P. abelii than for specimens from Indonesia and P. pygmaeus. The shape change along CV2 shows a more rectangular and elongated EDJ with more closely placed dentine horns for modern molars, whereas fossil specimens exhibit a shorter and rounder EDJ outline. There is no significant allometric trend between fossil and modern specimens regarding the grouping of molars along CV1 and CV2 (p = 0.5 and p = 0.17, respectively).

The degree of cross-validated correct classification (reported in S4 and S5 Tables, illustrated in in S1 Fig) for Pongo abelii and Pongo pygmaeus is 43% and 43%, respectively. Two species of living Pongo included in this study are somehow distinguished from each other and from fossil groups but the cross-validated classification results for M1–M2 showed higher percentages. The overall classification accuracy is 50% indicating substantial overlap, however fossils from Punung were well separated from all other groups with 83%. The specimen from Niah Cave (Malaysia) labeled “S2258” projected a posteriori (Fig 1B) clusters close to Vietnamese specimens along CV1 and Indonesian specimens along CV3. The typicality probability for this specimen shows the highest affinity to Vietnamese fossils (S2 Table).

3.2 Paleoproteomics

The high likelihood of authenticity of proteins retrieved from fossil samples is suggested by the high rates of deamidation, which are similar across all specimens in the analysis (Fig 2A and S6 Table). Moderate rates of arginine to ornithine conversion are also observed in the ancient samples (Fig 2B), though a noticeably higher rate is observed in the Middle Pleistocene Sangiran specimen SMF 8864, reflecting its greater antiquity. In spite of the increased level of degradation observed in its paleoproteomic profile, SMF 8864 preserves a suite of enamel proteins comparable to those of the other Late Pleistocene specimens, and an amino acid coverage approaching the median of the Late Pleistocene specimens (Fig 3). In addition to the typical suite of enamel proteins (AMBN, AMELX, AMTN, ENAM, MMP20) and serum albumin (ALB), a high abundance of collagen peptides (COL1A1 and COL1A2) is observed in a pair of samples (S1181and S2254). The presence of COL1A1 and COL1A2 is likely due to traces accidentally co-extracted from dentine in these samples (Fig 3).

Fig 2.

(a) Deamidation rates for each fossil specimen. Note range of values (80–100) (see S6 Table for data). (b) Arginine to ornithine ratio for each fossil specimen. Proportion of arginines that have been modified in the form of conversion to ornithine, by sample (see S6 Table for data).

Fig 3. Amino acid coverage per protein.

Bar plot displaying number of amino acids covered by protein for each specimen.

3.2.1 Sequence variation.

After inspecting the alignments of the Pongo clade, we were able to identify few single amino acid polymorphisms (SAPs) within that clade (Fig 4). Although no SAP is exclusive to any species of Pongo, some of them appear with different frequency within each species. We were able to identify a single SAP in AMBN (Position 174 in the Pongo abelii canonical Ensembl transcript of AMBN—ENSPPYT00000017210.2) present only in the fossil samples and all the other non-Pongo hominids (Fig 4). This position was covered in only 2 out of the 15 ancient samples we analyzed, with one originating from the Early to Middle Pleistocene Sangiran site (Java) and the other from the Late Pleistocene site of Duoi U’Oi (Vietnam). This SAP should be interpreted cautiously due to its restricted occurrence and its limited value on broader evolutionary inferences.

Fig 4. Amino acid sequence variation within Pongo.

The first column represents the position on the respective protein, where 2 alternative variants can appear in the genus (a single amino acid polymorphic site). These variants appear in different frequencies across extant Pongo species and fossil Pongo specimens (variant 1 = orange, variant 2 = blue, if no position is covered = black).

3.2.2 Protein phylogenetic results.

Both the Bayesian tree (Fig 5) from MrBayes and the maximum likelihood tree (S2 Fig) from PhyML capture the established phylogeny of the four extant hominid genera, chimpanzees (Pan), gorillas (Gorilla), orangutans (Pongo) and humans (Homo) with gibbons (Nomascus) as an outgroup. In the Bayesian tree, all the fossil Pongo samples are placed in the same clade as all modern Pongo samples with high support, whereas for the Maximum Likelihood tree this placement has moderate support. In both trees there is no statistical support for any substructure within the Pongo genus, including no separation of the modern species (P. pygmaeus, P. abelii, P. tapanuliensis) or the modern samples from the fossil ones.

Fig 5. Phylogenetic tree.

Bayesian phylogenetic tree created by using MrBayes based on dental protein sequences of fossil Pongo samples and several hominid species (Gorilla gorilla, Gorilla beringei graueri, Pan paniscus, Pan troglodytes schweinfurthii, Pan troglodytes ellioti, Pongo pygmaeus, Pongo abelii, Pongo tapanuliensis and Homo sapiens). Nomascus leucogenys was added as an outgroup. Both modern and fossil Pongo samples are grouped into a single clade with high confidence and without any substructure.

4. Discussion

4.1 Does the analysis of prominent features of the EDJ confirm species concepts in the genus Pongo?

4.1.1 Molar position assessment.

Previous studies of external size and morphology of molars demonstrated that metameric variation in Pongo is too high to accurately determine the serial position of molars among isolated teeth [1, 56, 64, 65]. Our inter-observer test to identify molar position confirmed these previous results, which led us to run geometric morphometric analyses using some subsets of molar positions (M1–M2 on the one hand, and M3 on the other), as well as with all molar positions combined. Our results show that, whether the molars are tentatively discriminated by position or combined together, there is a broad agreement among grouping of extant and fossil groups and similar proximities between each other in the morphospace (S4 and S5 Figs). The overlap between groups is however considerably higher when all molar positions are combined, which leads to a less distinct taxonomic signal, but the statistical power is higher since there are more specimens, and the risk of tooth misattribution is avoided.

However, running geometric morphometric analyses separately for each molar position can also lead to biased results if molar positions are not correctly identified for isolated fossil specimens. Our molar type assessment revealed a tendency to assign uncertain cases to the M2 position. As a result, M2 assignments were overrepresented in our dataset. Furthermore, in several cases there was no consensus among all three observers and M3 were assigned much less frequently than the other positions (see S3 Fig for the results of the molar type assignments). Although the absence of interproximal distal contact facet is generally a diagnostic feature of M3s, it can be absent on M1 if M2 is not yet emerged, likewise on M2 if M3 is not emerged. There is also the possibility of the presence of a M4 in Pongo, which is relatively high, around 7–13% in extant species [56, 66] and similarly high in extinct Pongo [1]. Accordingly, a M3 would have a distal wear facet when an M4 is present. Our results show that, even when mixing molar positions, the different groups represented by extant and fossil specimens show broad agreement across all CVA plots suggesting morphometric distinction of extant Pongo species from fossil groups, despite the uncertainty of molar positions.

4.1.2 Extant Pongo.

The large spatio-temporal distribution and marked dental morphological diversity of fossil Pongo were linked by numerous scholars to the variability of extant species. Hooijer [1] described a high diversity in extant Pongo dentition and argued that fossil teeth from Early to Middle Pleistocene southern China and Late Pleistocene Sumatra fall within the morphological range of P. pygmaeus, hence allocating these fossils to subspecies, P. p. weidenreichi and P. p. palaeosumatrensis, respectively. Kahlke [2] questioned the validity of both subspecies due to the lack of “clearly defined morphological and metrical differences” and “inadequate characterization” especially of P. p. palaeosumatrensis. Ho et al. [83] suggested merging both fossil subspecies to P. p. weidenreichi. Based on morphology, even P. abelii was classified as a subspecies of P. pygmaeus [13, 15]. Later, genetic evidence revealed the molecular distinction of Bornean and Sumatran orangutans, resulting in the elevation of P. abelii to a full species [84]. Despite the genetic separation of extant Pongo species, the outer crown morphology of P. abelii and P. pygmaeus shows very little differences, hence isolated teeth of both species mixed together were reported to be impossible to separate [6, 85]. Identifying species of Pongo and determining their phylogenetic relationship solely based on the external morphology of isolated dental material remain unsatisfactory due to the large overlap and similarities of dental traits between extinct and extant species [6, 7].

According to our results, the morphometric analysis of the EDJ of P. abelii and P. pygmaeus shows distinction both between species and, to some extent, to fossil molars from different geographic origins as well (Indochinese and Sundaic areas). However, the fossil sample set comprises more specimens than our modern groups. Many of the wild orangutan specimens in museum collections were obtained a century ago, thus location and Pongo species identification (particularly the identification of P. abelii from P. tapanuliensis within Sumatran orangutans) are not always certain. A bigger sample size is needed to adequately capture the differences between P. abelii, P. pygmaeus and the recently described Pongo tapanuliensis [17]. Capturing the morphospace of extant species would greatly enhance the characterization of fossil molars as well. Despite the constraint due to the sample size, there is a general trend in the distribution of fossil groups relative to modern groups, which will be further discussed.

4.1.3 Fossil Pongo from insular Southeast Asia.

Based on morphometric traits, our results show that some fossils from Punung, Indonesia, overlap partially with P. pygmaeus but not with the P. abelii sample (Fig 1A), although there is overlap with both modern species when considering CV3 (Fig 1B). Some molars from Punung might derive from deposits younger than those of the Late Pleistocene (128–117 ka; [57]) intruding the stratigraphic fossiliferous layers, as recently demonstrated by Kaifu et al. [59]. These teeth seem to be morphometrically very similar to modern species as previously suggested by Badoux [56].

In our analysis, the rest of the Punung material groups in part with fossils from Vietnam and China (Fig 1A). Even though the historical Punung collection was previously considered to be morphologically and metrically identical to Pongo pygmaeus [56], a further examination of dental traits established morphometric differences to modern species [6, 11]. Indeed, the relatively smaller size of the Punung dentition compared to that of Late Pleistocene taxa from Sumatra, Borneo, and mainland Southeast Asia led some authors to propose an allocation to a distinct species P. javensis [6, 11]. Although this new taxon P. javensis was suggested for the Pongo remains from Punung by Drawhorn [11], it was never formally published following the rules of ICZN [86]. Our analysis shows very little allometric influences based on the shape differences of the EDJ morphology along CV2, but only for M1 and M2. The EDJ morphology of Punung fossils shows some distinction to fossils from continental Southeast Asia.

Due to these various overlaps with modern and fossil groups, it is difficult to determine the evolutionary relationships of the Punung sample. Either some of the specimens could belong to the same taxa as the fossils from Vietnam and China (assigned to P. devosi and P. weidenreichi, respectively), or they could represent an endemic taxon (cf. P. javensis) as proposed by Drawhorn [11]. The former assumption would lead to a scenario where a group close to P. weideinreichi/P. devosi evolved in Java leading to a dental reduction compared to the continental populations but remaining morphologically and genetically compatible with them. In the second scenario, they would have evolved into a separate geographical variant. Based on the morphospecies concept, as commonly used in paleontological studies, the distinct range of EDJ morphology of a great part of this Punung sample (vs China and Vietnam) would indicate the existence of this endemic taxon (P. javensis).

Size reduction could be either the result of an evolutionary trend in insular environments due to limited resources, or the result of the founder effect of a population of small-sized individuals. The Punung fossil-bearing layers were deposited between 128 ka and 117 ka which corresponds to the early period of the Last Interglacial, when Java might still have had some remaining connections to neighboring land masses, which favored the expansion range of Pongo [57, 58]. This timing is consistent with paleoclimatic records for the western region of Java, which indicate shifts towards very warm and humid conditions c. 126 ka in the Bandung basin, Java [58, 87, 88]. However, other reconstructions of global sea level fluctuations also suggest sea level rises around 120 ka separating Java from continental Asia, Borneo and Sumatra [89], and it was proposed that these insular conditions might have caused a selective pressure in favor of dental size reduction [5, 85]. A more extensive comparative modern and fossil dataset is required to verify, if fossils from Punung, showing an affinity to continental Southeast Asian fossils, are a distinct paleospecies or fall within the modern species range of variation.

Further samples from Indonesia such as the Early to Middle Pleistocene Pongo remains from the Sangiran Dome are the oldest known fossils from the Sunda region (dated to 1.5–0.5 Ma). It is likely that the orangutan populations like that recorded at Sangiran were widespread across Sundaland in the Middle Pleistocene. These populations are considerably older than the Punung fossils. The typicality probabilities groups for SMF 8858 and SMF 8864 show affinities to all fossil groups with the latter demonstrating a high affinity to Punung fossils. SMF 8879 has an affinity to P. abelii and needs to be investigated more closely. The typicality probabilities of the Sangiran samples are not uniform, which suggests a high variability.

However, the small sample size complicates the assignment and characterization of Pongo from the Sangiran Dome and its relationship to other fossils from continental Southeast Asia and Sunda. It is unclear if Pongo from the subsequent Punung site could be either a continuation of Pongo from Sangiran, both located on Java, or the result of a later expansion from Late Pleistocene continental Southeast Asia. Our data do not provide a conclusive statement regarding the relationship of Pongo from Punung and Sangiran. A re-examination focusing on the Punung fossil collections is required, in order to separate the possibly terminal Pleistocene to Holocene material from the early Late Pleistocene specimens, which might convolute the morphometric and phylogenetic characterization of the Punung material.

The specimen S2258 from the Late Pleistocene Niah Cave, Borneo shows a high affinity to fossils from Vietnam based on typicality probabilities (Table 3). This tooth was previously reported as identical in size and morphology to modern P. pygmaeus [6]. More specimens are needed to test the relationship between Niah Cave samples, Vietnamese samples and Pongo pygmaeus.

4.1.4 Fossil Pongo from mainland Asia.

Fossils from China, Baikong and Mohui cave dated to 2.2–2.0 Ma are the oldest known fossils identified as Pongo [6, 85, 90, 91]. Hooijer [1] assigned Early to Middle Pleistocene orangutans discovered in southern China to P. pygmaeus weidenreichi based on the large size of teeth. Harrison et al. [6] elevated Pongo weidenreichi to species level based on morphometric traits such as lower crowned canines and relatively large second and third molars compared to modern species. Our results show no overlap between the range of P. pygmaeus and that of fossil molars from China but little overlap with P. abelii. Futhermore, most of the fossils from China cluster on the negative morphospace along CV1 with a higher affinity to fossils from Vietnam.

Fossil teeth from Chinese sites bought in drug stores at Manila and Hong Kong by von Koenigswald [39, 40] were allocated to several taxa such as Hemanthropus peii, Sinanthropus officinalis (Homo erectus) and fossil Pongo. The Pongo teeth identified are thought to have come from Pleistocene caves in southern China based on preservation and associated faunas. These teeth are described as larger than those of modern Pongo [1, 39]. However, specimens CA805, CA806 and CA807 were originally assigned to Sinanthropus officinalis [40]. Smith et al. [26] reevaluated the taxonomy of isolated teeth from the Chinese apothecary collection and showed a correspondence of these specimens to Pongo based on dental periodicity, EDJ shape and enamel thickness. Only CA806 was assigned to “likely Homo” by Smith et al. [26] due to the low periodicity of 8 days, resembling periodicities reported in Homo while Pongo shows higher periodicities. But it was noted by Smith et al. [26] that the EDJ shape resembles Pongo rather than Homo. Our results confirm that specimens from the Chinese apothecary collection share the same morphospace as specimens from Ganxian (identified as P. weidenreichi; [38]). Although, Chinese samples are showing a high variation along CV3 with several specimens from Ganxian grouping closer to Vietnamese fossils, the others group closer with molars from the Chinese Apothecary Collection. This is reflected in the cross-validated classification results for Chinese specimens as well, which demonstrates a high affinity to fossils from Vietnam. Nevertheless, if CV1/CV2 is considered, Ganxian and Chinese apothecary collection fossils group close together. Hence, all fossils from China could belong to a similar morphotype, even if Pongo from the Chinese Apothecary collection is difficult to characterize due to the lack of the exact provenance and age.

Schwartz et al. [3] examined fossils from various Middle to Late Pleistocene sites in Vietnam and established several (sub)species such as P. hooijeri (Tham Khuyen), P. pygmaeus ciochoni (Lang Trang), P. pygmaeus devosi (Hang Hum), P. pygmaeus kahlkei (Tham Khuyen) and P. pygmaeus fromageti (Tham Om). They also described a new ape genus Langsonia in northern Vietnam. Cameron [92] argued that the large mainland taxa (from China and Vietnam) cluster together and are separated from smaller island taxa of Borneo and Sumatra. Following Schwartz et al. [3], Cameron [92] confirmed the distinctiveness of P. p. ciochoni (Lang Trang) from other Pongo taxa. Our results, however, show morphological similarities of all fossils from Vietnam as they cluster together in one morphospace including P. p. ciochoni.

It was suggested that the diversity observed by Schwartz [25] reflects the high morphological variability within Pleistocene Pongo rather than multiple (sub)species or a new genus [4, 65]. Later, Harrison et al. [6] reexamined the (sub)species created by Schwartz et al. [3] and concluded that there are not sufficient morphological differences to justify these (sub)species or a new genus. This interpretation is supported by our result, although the identification of subspecies in the fossil record might be difficult, considering that a distinction at the species level is already confused by various conflicting species distinctions [14, 6, 7, 11, 56, 83].

Harrison et al. [6] allocated Middle Pleistocene specimens from China and Vietnam to P. weidenreichi and Late Pleistocene specimens from the same area to P. devosi, thus re-allocating all (sub)species defined by Schwartz et al. [3] to junior synonyms. P. devosi is described as relatively smaller than P. weidenreichi and might have replaced the latter during the Late Pleistocene. It was recently debated whether P. devosi should be considered as a distinct species or a junior synonym to P. weidenreichi based on a bigger sample size of fossil orangutan teeth from China [7]. In their study, Harrisonet al. [7] proposed that P. weidenreichi and P. devosi may be regarded as one single species since P. weidenreichi exhibited a temporal trend of size decrease in the Late Pleistocene. Furthermore, it was argued that throughout the Early to Late Pleistocene, Pongo’s distinctive morphological features remained unaltered, justifying the demotion of P. devosi to a synonym of P. weidenreichi [7, 93]. Our results indicate substantial overlap between fossils from China and Vietnam along CV1 vs. CV2 (Fig 1A), also indicated by the cross-validated classification results (S2 Table), but with some morphometric distinction along CV3. More Chinese specimens are needed to determine whether P. devosi and P. weidenreichi represent a single chronospecies or separate species.

The greater size of Early to Middle Pleistocene Pongo fossils from China compared to their modern counterparts and the temporal trend of diminution of tooth size from the Pleistocene to the Holocene were observed in several studies [83, 92, 94, 95]. Other authors show that reduction in tooth size throughout the Quaternary might not strictly follow the chronology [5, 51, 85, 92]. This is in line with our results indicating that shape changes in the EDJ are not mainly tied to dental size.

Fossil Pongo specimens from Tham Prakai Phet (Thailand) dated to the Late Pleistocene show a bigger size compared to P. devosi, suggesting an affinity to the species P. weidenreichi [51]. We observe the same trend for other fossils from the mainland. Specimens from the Late Pleistocene Batu Cave (66–33 ka) in Peninsular Malaysia are larger than specimens from the Middle Pleistocene site of Badak Cave C (500 ka) proving that Pongo did not experience a steady decline in tooth size over time. As previously hypothesized [51], two forms of Pongo might have existed during the Late Pleistocene, a big morph in the center of the continent (represented by the Thai and Malay specimens) and a smaller one represented by Vietnamese specimens assigned to P. devosi.

Pongo remains from the Late Pleistocene site in Laos (Tam Hay Marklot Cave) were reported as Pongo sp. [45, 48, 50]. Our results show that the specimen “ML5” shows a high affinity to the P. abelii sample and to fossils from the same area in Vietnam (Table 3). Due to the limited sample size, a definite assignment is not feasible. Additional samples are required to clarify the relationships between, on one hand, Late Pleistocene Pongo from Tam Hay Marklot and P. abelii, and on the other hand, the relationships between fossils from the mainland.

In conclusion, there is morphometric variation within fossil Pongo from continental SEA, pointing to the distinction of fossil morphotypes of Pongo compatible with the previously published species P. weidenreichi and P. devosi. Whether the Punung specimens represent a mix of early Late Pleistocene attributable to P. weideinreichi/P. devosi or P. javensis and later specimens (terminal Pleistocene to Holocene) related to the living species of Pongo remains unclear. The Punung assemblage is the only one in our fossil sample, which is represented in a large number of specimens from one specific locality. This might be the reason for the higher classification results as opposed to fossils from Vietnam and China.

4.2 Proteomic challenges in Pongo species distinction

The preserved protein regions were not informative to confidently discriminate between the different species of Pongo. There are no species-specific taxonomic markers that may be derived from our experimental protocols. Inspection of multiple sequence alignments, including a broad selection of in silico sequences from hominids such as Pan and Gorilla, reveals that these enamel protein sequences display a smaller intrageneric variability in Pongo than in other hominids. For instance, chimpanzees (Pan) display several taxonomic markers within frequently preserved regions, enabling the differentiation of extant species. This accounts for the clear separation between Pan troglodytes and Pan paniscus in our tree (Fig 5). It is not clear what is driving this lack of variation in enamel protein sequences among extant Pongo species. Recent DNA analyses revealed several admixture events resulting in gene flow between populations throughout the Pleistocene, which ceased completely 20–10 ka due to habitat loss in areas between the species’ ranges and isolation on islands [17]. The lack of variation in protein regions might be a consequence of high admixture within the genus Pongo. Biogeographical events played a key role in facilitating exchange between Pongo populations from the mainland Southeast Asia and the Sunda region (see below 4.3.) [96, 97].

All extant Pongo species share a single SAP at position 174 in AMBN, which is not present in two fossil Pongo specimens (one from the Sangiran site and one from the Duoi U’Oi site). Fossil Pongo displays the same SAP version found in all available great ape taxa, setting them apart from modern Pongo. However, interpreting whether this SAP demonstrates that extant Pongo might be evolutionarily derived compared to fossil Pongo from the Early to Middle Pleistocene should be approached with caution. This is due to its limited occurrence in our fossil dataset and its constrained significance for broader evolutionary conclusions, as evidenced by the grouping of all Pongo fossils and modern specimens in the same clade (Fig 5).

Our protein translations from the genomic data of modern Pongo individuals show that even the full coverage of the enamel proteins cannot clearly separate between the different extant species. This is intriguing, considering that distinguishing between extant Pan and Gorilla species at the enamel protein level is possible. The enamel proteins were also unable to identify groupings within the fossil Pongo specimens. Separation between extant and fossil Pongo might be possible with a higher coverage however, given the identification of at least one separating SAP, or through the recovery of a richer proteome, for example from bone or dentine. Similar to Pongo, within the genus Homo, enamel proteins reveal less information for determining closely related hominin species [33]. This discrepancy in phylogenetic resolution within dental enamel proteins among African great apes, Pongo, and Homo need further investigation.

The EDJ analysis can identify different groups within both the extant and the fossil Pongo sample. This implies that EDJ morphometrics are a more viable method for distinguishing groups of closely related primates, where the enamel proteome offers limited resolution. However, in cases where dental fossil specimens exhibit incomplete morphology due to damage from fossilization processes, the analysis of dental enamel from fossil Pongo specimens could provide opportunities to identify the genus at least. This is particularly important in fossil sites like Sangiran, where a high diversity of hominid species was confirmed including fossil Pongo, Homo erectus, Meganthropus and possibly Gigantopithecus [27]. The convergence in molar outer crown morphology adds complexity to taxonomic classifications, especially when dealing with fragmented specimens. The protein sequences can be utilized in conjunction with the EDJ analysis to confidently place those samples within a wider phylogenetic framework.

4.3 Paleoenvironmental implications on the speciation of Pongo

Sundaland emerged as a continuous landmass connecting the Indonesian archipelago with continental Southeast Asia during the Gelasian Stage (2.59–1.81 Ma). Land bridge formations between Sundaland and mainland Asia during the Early Pleistocene facilitated colonization of these regions by mammalian taxa [88, 98100]. Paleogeographic changes and paleoclimatic oscillations related to spreading and declining of rainforest habitats likely impacted the evolution and speciation of Pongo in Southeast Asia [97, 101]. During the Pleistocene, cool and dry periods which led to a decline and fragmentation in rainforest habitats and the emergence of open savannahs, alternated with warm and wet periods with sea level rises temporarily disconnecting islands and isolating taxa [9698, 102106]. Repeated fluctuations of sea levels contributed to either isolation phases of islands from the Sunda Shelf during which isolated populations of orangutans evolved differently from the continental and other island groups, or intermittent exchanges between continental and insular populations during low sea level events [97]. Sundaland was marked by its low relief and several large rivers. Fragmentation of rainforests due to increase in drier conditions and large rivers dissecting Sundaland hindered the dispersal of orangutan populations and potentially decreased population sizes [97, 101]. It was hypothesized that the formation of a large savannah corridor through Sundaland, starting from the eastern Sumatran highlands and continuing to western Borneo [98], might have been an additional barrier for orangutan dispersal. The emergence of a savannah corridor probably caused the formation of forest refugia in the highlands of Java and Sumatra and northeastern and southern Borneo [98]. Due to these conditions populations of orangutans probably remained ecologically/geographically separated, increasing the probability of genetic drift and local variations. These conditions might have contributed to the high genetic and phenotypic variation of Pongo during the Pleistocene. The divergence of P. tapanuliensis and P. abelii both from Sumatra are reported as 3.38 Ma, which is longer than the divergence of P. pygmaeus from Borneo and P. abelii [17]. Nevertheless, gene flow occurred occasionally after the separation of these species during the Pleistocene affecting the genetic and phenotypic/morphological variation of Pongo. Indeed genetic evidence revealed repeated gene flow events between P. abelii, P. pygmaeus and P. tapanuliensis throughout the Pleistocene [17]. This demonstrates the complex evolutionary history of Pongo and the need for DNA analyses on fossil material to disentangle the evolutionary relationships between fossils from Indochina and insular Southeast Asia and to determine their relationships to modern species.

5. Conclusions

EDJ morphology shows tentative support for previously established fossil Pongo species, which were solely based on linear measurements and morphometric traits of teeth. Our separate analyses for M1/M2s and M3s, along with the combined analysis, highlight that despite the issue of molar position and sample size, (1) fossil Pongo from Vietnam, Indonesia and China show differences to modern species and (2) fossils from Indonesia are different in their EDJ shape to fossils from Vietnam and China. It is clear that our morphometric results are constrained by the small sample size by site and by period (ML5, Tam Hay Marklot, Laos; SMF 8879, SMF 8864, SMF 8858, Sangiran Dome, Java; S2258, Niah Cave, Malaysian Borneo), which is not sufficient to estimate the variability of past populations and their relationships.

The Punung specimens might represent a mix of early Late Pleistocene and later specimens (terminal Pleistocene to Holocene) closely related to extant Pongo. These Late Pleistocene specimens show morphometric differences to fossils from continental Southeast Asia, whether these differences might be attributable to a separate species (such as P. javensis) has to be confirmed with further analyses.

Fossils from Vietnam and China show some morphometric separation, but substantial overlap exist between groups. Larger sample sizes are required to adequately test whether P. weidenreichi (China) and P. devosi (Vietnam) species represent two distinct species or belong to one chronospecies. Paleoproteomic data from dental enamel proteins do not permit a consistent attribution of fossil specimens to either P. pygmaeus or P. abelii, but avoid debates about the appurtenance of isolated fossil specimens to Pongo (versus Homo) with a direct (secure, robust) assignment to the genus [3, 43]. In that respect, EDJ morphometry shows some separation in closely related fossil groups within a genus, where the enamel proteome offers limited resolution. Nevertheless, protein sequences can be utilized with EDJ morphology to confidently place these fossil samples in a broader phylogenetic framework, including other hominid taxa.

Orangutans can be a model for the understanding of human evolution due to known hybridization between modern Pongo species, the high morphological variability in dental remains and the large fossil record, which is unique among other great apes such as Gorilla and Pan. Analyses of the evolutionary history of Pongo and their paleoenvironments are crucial to improve conservational efforts for critically endangered species of Pongo.

Supporting information

S1 Table. List of specimens used in the present study.


S2 Table. Cross-validated classification results in frequencies based on M1–M2.


S3 Table. Cross-validated classification results in percentages based on M1–M2.


S4 Table. Cross-validated classification results in frequencies based on M3.


S5 Table. Cross-validated classification results in percentages based on M3.


S6 Table. Summary statistics of paleoproteomic analyses.


S7 Table. Semilandmark coordinates for each specimen used in this study.


S1 Fig.

a Cross-validated CVA of the EDJbased on M1–M2. Jackknife cross-validation was performed using the same groups as in the CVA (Fig 1A) to validate group discrimination in Fig 1A. b Cross-validated CVA of the EDJ based on M3s. Jackknife cross-validation was performed using the same groups as in the CVA (Fig 1B) to validate group discrimination in Fig 1B.


S2 Fig. Maximum likelihood phylogenetic tree.

Maximum Likelihood phylogenetic tree created by using PhyML based on protein sequences of several great ape species (Gorilla gorilla, Gorilla beringei graueri, Pan paniscus, Pan troglodytes schweinfurthii, Pan troglodytes ellioti, Pongo pygmaeus, Pongo abelii, Pongo tapanuliensis and Homo sapiens) and fossil Pongo samples included. Nomascus leucogenys was added as an outgroup. In this tree both modern and fossil Pongo are placed within the same clade but with moderate support. Some substructure can be observed on the tree but without any concrete support.


S3 Fig. Molar type assignments.

Heatmap of molar type assignments by three observers (JK, CZ, AMB) on a subsample of isolated Pongo fossil molars. Dark blue indicates consent by all 3 observers, medium blue indicates consent among 2 observers, light blue shows one observer assigned the molar to the specific position and light grey means no assignment.


S4 Fig. CVA of the EDJ based on all molar positions combined.


S5 Fig. Cross-validated CVA of the EDJ based on all molar positions combined.


S6 Fig.

a PCA analyses of the EDJbased on M1–M2. b PCA analyses of the EDJ based on M3. c PCA analyses of the EDJ based on all molar positions.


S1 File. Palaeoproteomic phylogenetic analysis.



We express our gratitude to the Werner Reimers Foundation, Bad Homburg (Germany) for making the Gustav Heinrich Ralph von Koenigswald collection available for scientific research as a permanent loan to the Senckenberg Research Institute and Natural History Museum Frankfurt. We thank Alessandro Urciuoli and Jonas Elsborg for their valuable comments on data analyses and interpretation. For access to comparative material, we sincerely thank P. Bayle, A. Mazurier, R. Macchiarelli, and Jean-Jacques Hublin from the Department of Human Evolution of the Max Planck Institute for Evolutionary Anthropology for the access to scans of P. abelii and P. pygmaeus according to the agreement signed with the Directors of the Museum fûr Naturkunde Berlin–Leibniz Institute for Evolution and Biodversity Science (J. Vogel and S. Junker), and the curators of the Phyletic Museum, Jena and the MorphoSource database.


  1. 1. Hooijer DA. Prehistoric teeth of man and of the orangutan from central Sumatra, with notes on the fossil orangutan from Java and Southern China. Zool Meded. 1948;29: 175–301.
  2. 2. Kahlke HD. A review of the Pleistocene history of the orangutan (Pongo Lacépède 1799). Asian Perspectives. 1972;15: 5–14.
  3. 3. Schwartz JH, Long VT, Cuong NL, Kha LT, Tattersall I. A review of the Pleistocene hominoid fauna of the socialist republic of Vietnam (excluding Hylobatidae). Anthropol Pap Am Mus Nat Hist. 1995;76: 1–24.
  4. 4. Bacon A-M, Vu TL. The first discovery of a complete skeleton of a fossil orangutan in a cave of the Hoa Binh Province, Vietnam. J Hum Evol. 2001; 227–241. pmid:11535001
  5. 5. Tshen LT. Biogeographic distribution and metric dental variation of fossil and living orangutans (Pongo spp.). Primates. 2016;57: 39–50.
  6. 6. Harrison T, Jin C, Zhang Y, Wang Y, Zhu M. Fossil Pongo from the Early Pleistocene Gigantopithecus fauna of Chongzuo, Guangxi, Southern China. Quat Int. 2014;354: 59–67.
  7. 7. Harrison T, Zhang Y, Yang L, Yuan Z. Evolutionary trend in dental size in fossil orangutans from the Pleistocene of Chongzuo, Guangxi, Southern China. J Hum Evol. 2021;161: 103090. pmid:34781087
  8. 8. Hooijer DA. The orangutan in Niah cave prehistory. Sarawak Mus J (New Series). 1960;9: 408–421.
  9. 9. Barker G, Barton H, Bird M, Daly P, Datan I, Dykes A, et al. The ‘human revolution’ in lowland tropical Southeast Asia: the antiquity and behavior of anatomically modern humans at Niah Cave (Sarawak, Borneo). J Hum Evol. 2007;52: 243–261. pmid:17161859
  10. 10. Piper P, Rabett R. Vertebrate fauna from the Niah caves. In: Barker G, Farr L, editors. Rainforest foraging and farming in island Southeast Asia: The archaeology of the Niah Caves, Sarawak. Cambridge: McDonald Institute for Archaeological Research; 2016. pp. 401–438.
  11. 11. Drawhorn GM. The systematics and paleodemography of fossil orangutans (Genus Pongo) [dissertation]. Davis (CA): University of California; 1995.
  12. 12. MacKinnon J. Orangutans in Sumatra. Oryx. 1973;12: 234–242.
  13. 13. Groves CP. Systematics of the great apes. In: Swindler DR, Erwin J, editors. Systematics, Evolution, and Anatomy. New York: Alan R. Liss; 1986. pp.187–217.
  14. 14. Groves CP. Primate taxonomy. In: Erwin DH, Funk VA, editors. Washington and London: Smithsonian Institute Press; 2001.
  15. 15. Courtenay J. Inter- or intra-island variation? An assessment of the differences between Bornean and Sumatran orangutans. In: Schwartz JH, editor. Orangutan Biology. Oxford: Oxford University Press; 1988. pp. 19–29.
  16. 16. Weitzel V, Groves CP, Yang CM. A catalogue of primates in the Singapore Zoological reference collection. Raffles Bull Zool. 1988;36: 1–166.
  17. 17. Nater A, Mattle-Greminger MP, Nurcahyo A, Nowak MG, de Manuel M, Desai T, et al. Morphometric, Behavioral, and Genomic Evidence for a New Orangutan Species. Curr Biol. 2017;27: 3576–3577.
  18. 18. Wich SA, Meijaard E, Marshall AJ, Husson S, Ancrenaz M, Lacy RC, et al. Distribution and conservation status of the orangutan (Pongo spp.) on Borneo and Sumatra: how many remain? Oryx. 2008;42: 329–339.
  19. 19. Goossens B, Chikhi L, Jalil MF, James S, Ancrenaz M, Lackman-Ancrenaz I, et al. Taxonomy, geographic variation and population genetics of Bornean and Sumatran orangutans. In: Wich SA, Atmoko SSU, Setia TM, van Schaik CP, editors. Orangutans: geographic variation in behavioural ecology and conservation. New York: Oxford University Press; 2009. pp. 215–224.
  20. 20. Husson SJ, Wich SA, Marshall AJ, Dennis RD, Ancrenaz M, Brassey R, et al. Orangutan distribution, density, abundance and impacts of disturbance. In: Wich SA, Atmoko SSU, Setia TM, van Schaik CP, editors. Orangutans: geographic variation in behavioural ecology and conservation. New York: Oxford University Press; 2009. pp. 77–96.
  21. 21. Spehar SN, Sheil D, Harrison T, Louys J, Ancrenaz M, Marshall AJ, et al. Orangutans venture out of the rainforest and into the Anthropocene. Sci Adv. 2018;4: e1701422. pmid:29963619
  22. 22. van Noordwijk MA, Utami Atmoko SS, Knott CD, Kuze N, Morrogh-Bernard HC, Oram F, et al. The slow ape: High infant survival and long interbirth intervals in wild orangutans. J Hum Evol. 2018;125: 38–49. pmid:30502896
  23. 23. Skinner MM, Gunz P, Wood BA. Discrimination of extant Pan species and subspecies using the enamel–dentine junction morphology of lower molars. Am J Phys Anthropol. 2009;140: 234–243.
  24. 24. Davies TW, Delezene LK, Gunz P, Hublin J-J, Skinner MM. Endostructural morphology in hominoid mandibular third premolars: Discrete traits at the enamel-dentine junction. J Hum Evol. 2019;136: 102670. pmid:31585375
  25. 25. Schwartz JH, Long VT, Cuong NL, Kha LT, Tattersall I. A diverse hominoid fauna from the late Middle Pleistocene breccia cave of Tham Khuyen, Socialist Republic of Vietnam. Anthropol Papers Am Mus Nat Hist. 1994;73: 1–11.
  26. 26. Smith TM, Houssaye A, Kullmer O, Le Cabec A, Olejniczak AJ, Schrenk F, et al. Disentangling isolated dental remains of Asian Pleistocene hominins and pongines. PLoS One. 2018;13: e0204737. pmid:30383758
  27. 27. Zanolli C, Kullmer O, Kelley J, Bacon A-M, Demeter F, Dumoncel J, et al. Evidence for increased hominid diversity in the Early to Middle Pleistocene of Indonesia. Nat Ecol Evol. 2019;3: 755–764. pmid:30962558
  28. 28. Baldauf SL, Roger AJ, Wenk-Siefert I, Doolittle WF. A kingdom-level phylogeny of eukaryotes based on combined protein data. Science. 2000;290: 972–977. pmid:11062127
  29. 29. Welker F, Collins MJ, Thomas JA, Wadsley M, Brace S, Cappellini E, et al. Ancient proteins resolve the evolutionary history of Darwin’s South American ungulates. Nature. 2015;522: 81–84. pmid:25799987
  30. 30. Horn IR, Kenens Y, Palmblad NM, Van der Plas-Duivesteijn SJ, Langeveld BW, Meijer HJM, et al. Palaeoproteomics of bird bones for taxonomic classification. Zool J Linn Soc. 2019;186: 650–665.
  31. 31. Cappellini E, Prohaska A, Racimo F, Welker F, Pedersen MW, Allentoft ME, et al. Ancient Biomolecules and Evolutionary Inference. Annu Rev Biochem. 2018;87: 1029–1060. pmid:29709200
  32. 32. Welker F. Palaeoproteomics for human evolution studies. Quat Sci Rev. 2018;190: 137–147.
  33. 33. Welker F, Ramos-Madrigal J, Gutenbrunner P, Mackie M, Tiwary S, Rakownikow Jersie-Christensen R, et al. The dental proteome of Homo antecessor. Nature. 2020;580: 235–238.
  34. 34. Hendy J. Ancient protein analysis in archaeology. Sci Adv. 2021;7: eabb9314. pmid:33523896
  35. 35. Froment C, Zanolli C, Hourset M, Mouton-Barbosa E, Moreira A, Burlet-Schiltz O, et al. Protein sequence comparison of human and non-human primate tooth proteomes. J Proteomics. 2021;231: 104045. pmid:33189847
  36. 36. Cappellini E, Welker F, Pandolfi L, Ramos-Madrigal J, Samodova D, Rüther PL, et al. Early Pleistocene enamel proteome from Dmanisi resolves Stephanorhinus phylogeny. Nature. 2019;574: 103–107.
  37. 37. Welker F, Ramos-Madrigal J, Kuhlwilm M, Liao W, Gutenbrunner P, de Manuel M, et al. Enamel proteome shows that Gigantopithecus was an early diverging pongine. Nature. 2019;576: 262–265.
  38. 38. Liang H, Liao W, Shao Q, Chen Q, Tian C, Yao Y, et al. New discovery of a late Middle Pleistocene mammalian fauna in Ganxian Cave, Southern China. Hist Biol. 2022; 1–18.
  39. 39. von Koenigswald GHR. Eine fossile Säugetierfauna mit Simia aus Südchina. Proc Roy Acad Amsterdam. 1935;38: 872–879.
  40. 40. von Koenigswald GHR. Gigantopithecus blacki von Koenigswald, a giant fossil hominoid from the Pleistocene of Southern China. Anthropol Pap Am Mus Nat Hist. 1952;43: 310–325.
  41. 41. Long VT, de Vos J, Ciochon RL. The fossil mammal fauna of the Lang Trang caves, Vietnam, compared with Southeast Asian fossil and recent mammal faunas: the geographical implications. Bull Indo-Pac Prehistory Assoc. 1996;14: 101–109.
  42. 42. Ciochon RL, Olsen JW. Palaeoanthropological and archaeological discoveries from Lang Trang caves: a new Middle Pleistocene hominid site from Northern Vietnam. Bull Indo-Pac Prehistory Assoc. 1991;10: 59–73.
  43. 43. Ciochon RL. Divorcing hominins from the Stegodon-Ailuropoda fauna: new views on the antiquity of hominins in Asia. In: Fleagle JG, Shea JJ, Grine FE, Baden AL, Leakey RE, editors. Out of Africa I: The first hominin colonization of Eurasia. Dordrecht: Springer; 2010. pp. 111–126.
  44. 44. Cuong NL. A reconsideration of the chronology of hominid fossils in Vietnam. In: Akazawa T, Aoki K, Kimura T, editors. The evolution of dispersion of hominids in Asia. Tokyo: Hokusen-Sha; 1992. pp. 321–335.
  45. 45. Bacon A-M, Demeter F, Duringer P, Helm C, Bano M, Long VT, et al. The Late Pleistocene Duoi U’Oi cave in Northern Vietnam: palaeontology, sedimentology, taphonomy and palaeoenvironments. Quat Sci Rev. 2008;27: 1627–1654.
  46. 46. Doi NG. Results of recent research into the lithic industries from Late Pleistocene/Early Holocene sites in Northern Vietnam. Bull Indo-Pac Prehistory Assoc. 2005;25: 95–97
  47. 47. Fromaget J. L’Indochine Francaise, sa structure geologique, ses roches, ses mines et leur relation possible avec la tectonique. Bulletin du Service Geologique de L’Indochine. 1941.
  48. 48. Bourgon N, Jaouen K, Bacon A-M, Jochum KP, Dufour E, Duringer P, et al. Zinc isotopes in Late Pleistocene fossil teeth from a Southeast Asian cave setting preserve paleodietary information. Proc Natl Acad Sci USA. 2020;117: 4675–4681. pmid:32071235
  49. 49. Bacon A-M, Duringer P, Antoine P-O, Demeter F, Shackelford L, Sayavongkhamdy T, et al. The Middle Pleistocene mammalian fauna from Tam Hang karstic deposit, Northern Laos: New data and evolutionary hypothesis. Quat Int. 2011;245: 315–332.
  50. 50. Bacon A-M, Westaway K, Antoine P-O, Duringer P, Blin A, Demeter F, et al. Late Pleistocene mammalian assemblages of Southeast Asia: New dating, mortality profiles and evolution of the predator–prey relationships in an environmental context. Palaeogeogr Palaeoclimatol Palaeoecol. 2015;422: 101–127.
  51. 51. Filoux A, Wattanapituksakul A. The Late Pleistocene orangutan from Tham Prakai Phet: new discoveries. Annales de Paléontologie. 2019;105: 287–293.
  52. 52. Larick R, Ciochon RL, Zaim Y, Sudijono , Suminto , Rizal Y, et al. Early Pleistocene 40Ar/39Ar ages for Bapang formation hominins, Central Jawa, Indonesia. Proc Natl Acad Sci USA. 2001;98: 4866–4871. pmid:11309488
  53. 53. Matsu’ura S, Kondo M, Danhara T, Sakata S, Iwano H, Hirata T, et al. Age control of the first appearance datum for Javanese Homo erectus in the Sangiran area. Science. 2020;367: 210–214.
  54. 54. Grine FE, Franzen JL. Fossil hominid teeth from the Sangiran dome (Java, Indonesia). Cour Forsch Inst Senckenb. 1994;171: 75–103.
  55. 55. Sondaar PY. Faunal evolution and the mammalian biostratigraphy of Java. Cour Forsch Inst Senckenb. 1984;69: 219–235.
  56. 56. Badoux DM. Fossil mammals from two deposits at Punung(Java) [dissertation]. Utrecht: Utrecht University; 1959.
  57. 57. Westaway KE, Morwood MJ, Roberts RG, Rokus AD, Zhao J-X, Storm P, et al. Age and biostratigraphic significance of the Punung rainforest fauna, East Java, Indonesia, and implications for Pongo and Homo. J Hum Evol. 2007;53: 709–717.
  58. 58. Storm P, de Vos J. Rediscovery of the Late Pleistocene Punung hominin sites and the discovery of a new site Gunung Dawung in East Java. Senckenb Lethaea. 2006;86: 271–281.
  59. 59. Kaifu Y, Kurniawan I, Yurnaldi D, Setiawan R, Setiyabudi E, Insani H, et al. Modern human teeth unearthed from below the ∼128,000-year-old level at Punung, Java: A case highlighting the problem of recent intrusion in cave sediments. J Hum Evol. 2022;163: 103122.
  60. 60. Kha LT. On fossil hominid teeth from Tham Om (Nghé Tinh). New Archaeol Rep. 1977; 24–27.
  61. 61. von Koenigswald GHR. Neue Pithecanthropus Funde 1936–1938; Ein Beitrag zur Kenntnis der Prae-Hominiden. Wet Meded Dienst Mijnb Ned Indie. 1940;28: 1–232.
  62. 62. de Vos J. The Pongo faunas from Java and Sumatra and their significance for biostratigraphical and paleoecological interpretations. Proc Royal Soc B. 1983;86: 417–425.
  63. 63. Zanolli C, Martinón-Torres M, Bernardini F, Boschian G, Coppa A, Dreossi D, et al. The Middle Pleistocene (MIS 12) human dental remains from Fontana Ranuccio (Latium) and Visogliano (Friuli-Venezia Giulia), Italy. A comparative high resolution endostructural assessment. PLoS One. 2018;13: e0189773. pmid:30281595
  64. 64. Groesbeek BJ. The Serial Position of the Trinil Upper Molars. Anthropol Sci. 1996;104: 107–129.
  65. 65. Harrison T. Archaeological and ecological implications of the primate fauna from prehistoric sites in Borneo. Bull Indo-Pac Prehistory Assoc. 2000;20: 133–146.
  66. 66. Swindler DR. Fourth molars in the Anthropoidea. Dental Anthropology Journal. 2002;16: 26–8.
  67. 67. Core Team R. R: A language and environment for statistical computing [software]. Vienna, Austria: R foundation for statistical computing. 2021.
  68. 68. Hastie T, Tibshirani R, Friedman JH. The elements of statistical learning: data mining, inference, and prediction. New York: Springer; 2009. pp. 1–758.
  69. 69. Kubat J, Nava A, Bondioli L, Dean MC, Zanolli C, Bourgon N, et al. Dietary strategies of Pleistocene Pongo sp. and Homo erectus on Java (Indonesia). Nat Ecol Evol. 2023;7: 279–89.
  70. 70. Rappsilber J, Mann M, Ishihama Y. Protocol for micro-purification, enrichment, pre-fractionation and storage of peptides for proteomics using StageTips. Nat Protoc. 2007;2: 1896–1906. pmid:17703201
  71. 71. Mackie M, Rüther P, Samodova D, Di Gianvincenzo F, Granzotto C, Lyon D, et al. Palaeoproteomic profiling of conservation layers on a 14th century Italian wall painting. Angew Chem Int Ed Engl. 2018;57: 7369–7374. pmid:29603563
  72. 72. Cox J, Mann M. MaxQuant enables high peptide identification rates, individualized p.p.b.-range mass accuracies and proteome-wide protein quantification. Nat Biotechnol. 2008;26: 1367–1372. pmid:19029910
  73. 73. Han Y, Ma B, Zhang K. SPIDER: software for protein identification from sequence tags with de novo sequencing error. J Bioinform Comput Biol. 2005;3: 697–716. pmid:16108090
  74. 74. Chi H, Liu C, Yang H, Zeng W-F, Wu L, Zhou W-J, et al. Open-pFind enables precise, comprehensive and rapid peptide identification in shotgun proteomics. BioRxiv. 2018; 285395.
  75. 75. Katoh K, Standley DM. MAFFT multiple sequence alignment software version 7: improvements in performance and usability. Mol Biol Evol. 2013;30: 772–780. pmid:23329690
  76. 76. Wickham H, Chang W, Wickham MH. Package ‘ggplot2’. Create elegant data visualisations using the grammar of graphics Version. 2016;2: 1–189.
  77. 77. Byrska-Bishop M, Evani US, Zhao X, Basile AO, Abel HJ, Regier AA, et al. High-coverage whole-genome sequencing of the expanded 1000 Genomes Project cohort including 602 trios. Cell. 2022;185: 3426–3440. pmid:36055201
  78. 78. Prado-Martinez J, Sudmant PH, Kidd JM, Li H, Kelley JL, Lorente-Galdos B, et al. Great ape genetic diversity and population history. Nature. 2013;499: 471–475. pmid:23823723
  79. 79. Prüfer K, de Filippo C, Grote S, Mafessoni F, Korlević P, Hajdinjak M, et al. A high-coverage Neandertal genome from Vindija cave in Croatia. Science. 2017;358: 655–658. pmid:28982794
  80. 80. Mafessoni F, Grote S, de Filippo C, Slon V, Kolobova KA, Viola B, et al. A high-coverage Neandertal genome from Chagyrskaya cave. Proc Natl Acad Sci USA. 2020;117: 15132–15136. pmid:32546518
  81. 81. Guindon S, Dufayard J-F, Lefort V, Anisimova M, Hordijk W, Gascuel O. New algorithms and methods to estimate maximum-likelihood phylogenies: assessing the performance of PhyML 3.0. Systematic biology. 2010;59: 307–321. pmid:20525638
  82. 82. Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Höhna S, et al. MrBayes 3.2: efficient Bayesian phylogenetic inference and model choice across a large model space. Syst Biol. 2012;61: 539–542. pmid:22357727
  83. 83. Ho CK, Zhou GX, Swindler DR. Dental evolution of the orangutan in China. J Hum Evol. 1995;10: 249–264.
  84. 84. Xu X, Arnason U. The mitochondrial DNA molecule of Sumatran orangutan and a molecular proposal for two (Bornean and Sumatran) species of orangutan. J Mol Evol. 1996;43: 431–437. pmid:8875856
  85. 85. Ibrahim YK, Tshen LT, Westaway KE, Cranbrook EO, Humphrey L, Muhammad RF, et al. First discovery of Pleistocene orangutan (Pongo sp.) fossils in Peninsular Malaysia: biogeographic and paleoenvironmental implications. J Hum Evol. 2013;65: 770–797.
  86. 86. Kraus O, Ride WDL. International Code of Zoological Nomenclature: discussion draft of the proposed fourth edition. Syst Entomol. 1995;20: 375–377.
  87. 87. van der Kaars WA, Dam MAC. A 135,000-year record of vegetational and climatic change from the Bandung area, West-Java, Indonesia. Palaeogeogr Palaeoclimatol Palaeoecol. 1995;117: 55–72.
  88. 88. van den Bergh GD, de Vos J, Sondaar PY. The Late Quaternary palaeogeography of mammal evolution in the Indonesian Archipelago. Palaeogeogr Palaeoclimatol Palaeoecol. 2001;171: 385–408.
  89. 89. Louys J. Quaternary extinctions in Southeast Asia. In: Elewa AMT, editor. Mass Extinction. Berlin and Heidelberg: Springer; 2008. pp. 159–189.
  90. 90. Takai M, Zhang Y, Kono RT, Jin C. Changes in the composition of the Pleistocene primate fauna in Southern China. Quat Int. 2014;354: 75–85.
  91. 91. Wang W, Liao W, Li D, Tian F. Early Pleistocene large-mammal fauna associated with Gigantopithecus at Mohui Cave, Bubing Basin, South China. Quat Int. 2014;354: 122–130.
  92. 92. Cameron DW. Morphometric evolutionary trends in the dental complex of Pongo. Primates. 2001;42: 253–266.
  93. 93. Liao W, Harrison T, Yao Y, Liang H, Tian C, Feng Y, et al. Evidence for the latest fossil Pongo in Southern China. J Hum Evol. 2022;170: 103233.
  94. 94. Pei WC. Fossil mammals from the Kwangsi caves. Bull Geol Soc China. 1935;14: 413–425.
  95. 95. Huong YGA. The study of some fossil orangutan from Guangdong and Guangxi. Acta Anthropologica Sinica. 1987;6: 272.
  96. 96. Tougard C. Biogeography and migration routes of large mammal faunas in South–East Asia during the Late Middle Pleistocene: focus on the fossil and extant faunas from Thailand. Palaeogeogr Palaeoclimatol Palaeoecol. 2001;168: 337–358.
  97. 97. Harrison T, Krigbaum J, Manser J. Primate Biogeography and Ecology on the Sunda Shelf Islands: A Paleontological and Zooarchaeological Perspective. In: Lehman SM, Fleagle JG, editors. Primate Biogeography: Progress and Prospects. Boston (MA): Springer; 2006. pp. 331–372.
  98. 98. Bird MI, Taylor D, Hunt C. Palaeoenvironments of insular Southeast Asia during the Last Glacial period: a savanna corridor in Sundaland? Quat Sci Rev. 2005;24: 2228–2242.
  99. 99. Voris HK. Maps of Pleistocene sea levels in Southeast Asia: shorelines, river systems and time durations. J Biogeogr. 2000;27: 1153–1167.
  100. 100. Reid MJC, Switzer WM, Alonso SK. Evolutionary history of orangutan plasmodia revealed by phylogenetic analysis of complete mtDNA genomes and new biogeographical divergence dating calibration models. Am J Primatol. 2022;84: e23298. pmid:34227139
  101. 101. Arora N, Nater A, van Schaik CP, Willems EP, van Noordwijk MA, Goossens B, et al. Effects of Pleistocene glaciations and rivers on the population structure of Bornean orangutans (Pongo pygmaeus). Proc Natl Acad Sci USA. 2010;197: 21376–21381.
  102. 102. Heaney LR. A synopsis of climatic and vegetational change in Southeast Asia. Clim Change. 1991;19: 53–61.
  103. 103. Louys J, Meijaard E. Palaeoecology of Southeast Asian megafauna-bearing sites from the Pleistocene and a review of environmental changes in the region. J Biogeogr. 2010;37: 1432–1449.
  104. 104. Meijaard E. Mammals of Southeast Asian islands and their Late Pleistocene environments. J Biogeogr. 2003;30: 1245–1257.
  105. 105. Wurster CM, Bird MI, Bull ID, Creed F, Bryant C, Dungait JAJ, et al. Forest contraction in north equatorial Southeast Asia during the Last Glacial Period. Proc Natl Acad Sci USA. 2010;107: 15508–15511. pmid:20660748
  106. 106. Salles T, Mallard C, Husson L, Zahirovic S, Sarr A-C, Sepulchre P. Quaternary landscape dynamics boosted species dispersal across Southeast Asia. Commun Earth Environ. 2021;2: 1–12.