ARTICLE https://doi.org/10.1038/s41467-020-15969-4 OPEN Isotopic evidence for initial coastal colonization and subsequent diversification in the human occupation of Wallacea Patrick Roberts 1,2 ✉, Julien Louys 3, Jana Zech1, Ceri Shipton 4,5, Shimona Kealy 4,5, Sofia Samper Carro4,6, Stuart Hawkins4,5, Clara Boulanger 5,7, Sara Marzo1, Bianca Fiedler1, Nicole Boivin1, Mahirta5,8, Ken Aplin4,9 & Sue OʼConnor 4,5 ✉ 1234567890():,; The resource-poor, isolated islands of Wallacea have been considered a major adaptive obstacle for hominins expanding into Australasia. Archaeological evidence has hinted that coastal adaptations in Homo sapiens enabled rapid island dispersal and settlement; however, there has been no means to directly test this proposition. Here, we apply stable carbon and oxygen isotope analysis to human and faunal tooth enamel from six Late Pleistocene to Holocene archaeological sites across Wallacea. The results demonstrate that the earliest human forager found in the region c. 42,000 years ago made significant use of coastal resources prior to subsequent niche diversification shown for later individuals. We argue that our data provides clear insights into the huge adaptive flexibility of our species, including its ability to specialize in the use of varied environments, particularly in comparison to other hominin species known from Island Southeast Asia. 1 Department of Archaeology, Max Planck Institute for the Science of Human History, 07745 Jena, Germany. 2 School of Social Science, The University of Queensland, St Lucia, QLD 4072, Australia. 3 Australian Research Centre for Human Evolution, Environmental Futures Research Institute, Griffith University, Nathan, QLD 4111, Australia. 4 School of Culture, History and Language, College of Asia and the Pacific, The Australian National University, Canberra, ACT 2600, Australia. 5 ARC Centre of Excellence for Australian Biodiversity and Heritage, Australian National University, Canberra, ACT 2600, Australia. 6 Centre d’Estudis del Patrimoni Arqueologic, Facultat de Lletres, Universitat Autònoma de Barcelona, 08193 Bellaterra, Spain. 7 Muséum National d’Histoire Naturelle, Département Homme et Environment, CNRS UMR 7194, Histoire Naturelle de l’Homme Préhistorique Paris, France. 8 Department of Archaeology, Universitas Gadjah Mada, Yogyakarta 55281, Indonesia. 9Deceased: Ken Aplin. ✉email: roberts@shh.mpg.de; sue.oconnor@anu.edu.au NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 R ecent, high-profile studies of symbolic material culture environments, lacking in reliable terrestrial protein and carbohy- (e.g., ref. 1), technological complexity (e.g., ref. 2), fossil drate resources13,16,17. Significantly, while these islands are home morphology and chronology (e.g., ref. 3), and genetics4 are to some of the earliest firm evidence for H. sapiens east of Africa demonstrating an increasingly complex and dynamic picture of and the Middle East c. 45 ka (refs. 13,18,19), fossil and artifact finds the capacities and interactions of different hominin populations have also suggested the presence of earlier members of the genus in the Late Pleistocene (126–12 ka), particularly in Asia. If we are Homo on the island of Flores from ~1 Ma (refs. 20,21), Luzon from to determine the ‘uniqueness’ of Homo sapiens, the last extant 0.7 Ma (ref. 22), and Sulawesi from ~0.2 Ma (ref. 23). Although hominin on the face of the planet, it is becoming apparent that we zooarchaeological records have provided some insights into the must examine how its ecological adaptations differed from those ecological niches of different hominin populations in Wallacea9,24, of other members of the genus Homo5,6. It has been suggested more direct assessments of overall hominin resource reliance and that Late Pleistocene populations of H. sapiens expanding across palaeoenvironmental change in the region have been lacking. the globe were able to not only flexibly exploit varied, and often In this paper, we examine the adaptations of the earliest extreme, environments—including deserts, tropical rainforests, known fossil members of our species in Wallacea by means of high-altitude settings, and deep-sea maritime habitats—but also isotopic analysis of archaeological human tooth enamel from specialize in the occupation of them, enabling our species as a two islands (Timor and Alor; Fig. 1). Timor has yielded the whole to proliferate even while local communities may sometimes earliest dated material culture and fossil evidence for H. have failed6. By contrast, earlier and contemporaneous Homo sapiens in Wallacea at the sites of Asitau Kuru (formerly Jer- species expanding into Eurasia in the Early and Middle Pleisto- imalai) and Laili18,19. At the former, faunal remains and cul- cene (2.6 Ma–126 ka) made generalized use of forest and grass- tural artifacts suggest Late Pleistocene human reliance on land mosaics7,8, potentially making them vulnerable to more marine shellfish and fish, obtained in part through offshore extreme Late Pleistocene environmental changes (e.g., ref. 3) and fishing13. Laili also provides evidence for early reliance on unable to survive on islands depauperate in large terrestrial marine resources18. This stands in contrast to the generalized fauna9. mixed grassland and woodland adaptations associated with Testing this hypothesis is particularly timely given recent finds other hominins in the region9,24,25. However, human reliance that imply other hominin species may have ventured into chal- on pelagic fishing at Asitau Kuru has been questioned26. lenging adaptive settings4,10. Wallacea provides an ideal ‘island Moreover, there remains the possibility that giant rat taxa, laboratory’ setting in which to do so in the increasingly with proposed preferences for closed forest environments and palaeoanthropologically significant Southeast Asian region. Wal- an adult body weight of up to 6 kg, represented significant food lacea is an isolated series of islands that was never connected to resources; and they have been identified in early coastal and the neighboring Pleistocene landmasses of Sunda or Sahul, inland archaeological contexts in Alor and Timor (e.g., ref. 27). necessitating water crossings to reach9,11–15. These islands have Insights into the environments present on Wallacea, as well as been hypothesized as hosting depauperate island forest human reliance on different ecosystems is difficult to resolve Fig. 1 Maps showing the location of the studied sites within Wallacea. Asitau Kuru, Lene Hara, Matja Kuru 1 and 2 (Timor), Makpan, and Tron Bon Lei (Alor). 2 NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 ARTICLE using traditional zooarchaeological methods alone due to preservation biases and the role of nonhuman predators in site taphonomy (e.g., ref. 28). Here, we apply stable carbon (δ13C) and oxygen (δ18O) isotope analyses to human and faunal tooth enamel from six Late Pleis- tocene/Holocene archaeological sequences (Fig. 1) on Timor and Alor, in order to determine the varying reliance of early human colonisers of Wallacea on tropical forest and terrestrial versus marine resources. Stable carbon isotope analysis of faunal (including hominin) tooth enamel in tropical regions has been used to assess the proportion of C3-dominated woodland/forest and C4 grassland biomass in diets29–31. In regions such as Pleistocene Wallacea, where some researchers have suggested that tropical forests dominated terrestrial environments32, with grasslands considered largely absent, the most significant driver of terrestrial stable carbon isotope variation will be the canopy effect, whereby low light and respired CO2 cause forest-dwelling plant biomass and its consumers to have more negative δ13C values than their counterparts in more open habitats30,31. Meanwhile, marine producer biomass has higher δ13C than all C3 terrestrial plants33,34, enabling marine consumers to be dis- tinguished from terrestrial C3 consumers35. Based on research done in East Africa30, Sri Lanka31, and Japan35, including extensive modern studies30, we expect preindustrial humans relying completely on tropical forest, open C3 resources, and marine resources to have tooth enamel δ13C values of c. −14‰, c. −11‰, and c. −4‰, respectively. Stable oxygen isotope (δ18O) measurements from animal tooth enamel provide additional paleoecological information about water and food, and have also been argued to distinguish terrestrial from marine consumers36. Based on existing, published, and available chronological information, the Late Pleistocene–Holocene deposits of Asitau Kuru, Matja Kuru 1 and 2, Lene Hara, Makpan, and Tron Bon Lei provide a unique suite of human and associated faunal samples spanning the earliest fossil appearance of H. sapiens in Wallacea, through the Last Glacial Maximum, and across the Terminal Pleistocene–Holocene transition18,19. They also cover both coastal and hinterland habi- tats (Fig. 1). Ample terrestrial and marine animal remains also allow us to build the first detailed paleoecological and palaeoen- Fig. 2 δ13C measurements for human and faunal tooth enamel from the vironmental records for Pleistocene Wallacea and test assump- islands of Alor and Timor analyzed in this study. Data shown by phases tions in relation to: (1) pure C3 terrestrial environments on Timor developed on the basis of existing and published stratigraphic, and and Alor in the past; (2) the δ13C distinction between available chronological information (Supplementary Note 1). Boxplots (showing terrestrial and marine resources; and (3) environmental shifts median and interquartile range—with outliers indicated) of terrestrial fauna across the Pleistocene–Holocene boundary proposed elsewhere are shown in green (black diamond points) and those of marine fauna in in Southeast Asia (e.g.,5,37). The preservation of a subsection of the blue (black square points). Human samples are shown as white symbols analysed tooth enamel samples was also checked using depending on site (circle, Asitau Kuru; diamond, Matja Kuru 1 and 2; square, Fourier transform infrared spectroscopy (FTIR) as per Roberts Lene Hara; triangle, Makpan; and inverted triangle, Tron Bon Lei). Source et al.31,38. data for Fig. 2 can be found in the accompanying Source Data file in Table 1 Our extensive faunal baseline demonstrates that terrestrial and (Fig. 2). marine environments can be clearly distinguished isotopically in Wallacea on the basis of stable isotope analysis of fossil tooth Results enamel. We show that a tooth of the earliest preserved H. sapiens Sites, samples, and chronology. The detailed stratigraphic and fossil found from the region c. 42–39,000 years ago shows that this chronological information for the six sites studied (Supplementary individual made significant use of coastal resources. From 20,000 Note 1) has enabled division of the human and faunal samples into years ago, human populations show an increasing reliance on occupation phases at each site (Figs. 2–6). To display and compare interior, terrestrial environments on the islands of both Timor our data on a broader scale, we have also divided the human and and Alor at a time of increasing forest expansion in Island faunal data into broader island phases of occupation for Timor and Southeast Asia more generally, though some individuals continue Alor, respectively, based on the stratigraphic and chronometric to intensively use marine resources. We argue that our data further information (Supplementary Note 1, Supplementary Tables 6 and demonstrates the huge adaptive flexibility of our species, 7, Fig. 2). For the sites of Asitau Kuru, Lene Hara, Matja Kuru 1, acutely visible as it rapidly and persistently colonized Wallacean and Matja Kuru 2 on Timor, this system includes four broad environments. Its ability to specialize in the use of more phases: a Late Pleistocene pre-LGM phase (46,000–29,000 years extreme environments seems to stand in contrast to other hominin ago), a Terminal Pleistocene phase (20,000–11,001 years ago), an species known from Island Southeast Asia based on current Early and Middle Holocene phase (11,000–4001 years ago), and a evidence. Late Holocene Neolithic phase (4000–0 years ago). For the sites of NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 Makpan and Tron Bon Lei on Alor, the phasing system includes becoming obviously different between 15,000–7400 cal. years BP three broad phases: a Late Pleistocene pre-LGM phase and 4000–0 cal. years BP (Fig. 2). These results suggest that C4 (40,000–21,000 years ago), a Terminal Pleistocene to Middle resources may have been available to some terrestrial fauna in the Holocene phase (15,000–7400 years ago), and a Late Holocene earliest phase of human occupation on Alor, with their presence Neolithic phase (4000–0 years ago). The specific associated dates declining through time. However, it is also possible that elevated for the human samples within these broader phases are discussed rat δ13C could be a product of early access to marine resources39. in the main text where appropriate. Finally, separation of the combined dataset of marine fauna for Timor and Alor (n = 105), and Kruskal–Wallis analysis and pairwise comparison (Kruskal–Wallis chi-squared = 17.975, df = Stable isotope analysis of archaeological tooth enamel. Faunal 8, p = < 0.05) showed significant δ13C differences between reef δ13C from the Late Pleistocene–Holocene sequences of Asitau taxa such as Balistidae (Triggerfishes) and more wide-ranging Kuru, Matja Kuru 2, Makpan, and Tron Bon Lei (Figs. 2–6) show taxa, such as Scaridae (Parrotfishes) (Supplementary Table 10), a δ13C division in terms of terrestrial and marine fauna (Fig. 2, indicating the potential utility of isotopic analysis of teeth to Supplementary Figs. 13 and 14, Supplementary Data 1). For the distinguish fish from different marine niches (see also ref. 40). island of Timor, terrestrial fauna recovered from the coastal site Our large, robust faunal baseline enables the long-term of Asitau Kuru (Fig. 3) and inland site of Matja Kuru 2 (Fig. 4) ecological niches of Late Pleistocene/Holocene human foragers have δ13C ranges of −14.9 to −7.9‰ (mean = −11.6 ± 1.8‰) to be directly determined for Asitau Kuru, Matja Kuru 2, and −14.9 to −9.0‰ (mean = −12.0 ± 1.4‰), respectively. By Makpan, and Tron Bon Lei, as well as the additional sites of Lene contrast, the marine fauna from Asitau Kuru (Fig. 3) has a δ13C Hara and Matja Kuru 1 where only human samples were available range of −8.3 to 3.7‰ (mean = −4.1 ± 3.0 ‰). The situation is (Figs. 2–6; n = 26). Sampled human δ13C and δ18O ranges slightly more complex on Alor (Fig. 2) where terrestrial fauna between −14.1 and −5.6‰ and −6.2 to −3.2‰, respectively from Makpan (Fig. 5) has δ13C ranging from −19.2‰ to −3.1‰ (Fig. 2, Supplementary Data 2). The earliest human sample in the (mean = −8.7 ± 3.5‰), and marine fauna from Makpan (Fig. 5) study, and the earliest recovered from Wallacea, from context B63 and Tron Bon Lei (Fig. 6) have δ13C ranging from −6.4 to 4.9‰ at Asitau Kuru, dated to c. 42,440–38,853 cal. years BP, has a δ13C (mean = −3.2 ± 2.9‰) and −8.9 to 4.7‰ (mean = −5.4 ± value of −5.6‰ (Fig. 3). This is indicative of a high reliance on 3.5‰), respectively. marine resources, given the lack of any evidence for C4 resources A Shapiro–Wilk test indicated that the δ13C (p = < 0.05) and in this part of Timor at this time and zooarchaeological evidence δ18O (p = < 0.05) of the entire faunal dataset (n = 223) were non- for abundant marine resources (Fig. 2). The majority of the normally distributed. Mann–Whitney–Wilcoxon tests demon- remaining humans sampled from the Terminal Pleistocene and strated the δ13C (W = 9257, p = < 0.05) and δ18O (W = 5264, Holocene contexts of Asitau Kuru (Fig. 3), Lene Hara (Fig. 3), p = < 0.05) of fauna to be significantly different between Timor Matja Kuru 1 and 2 (Fig. 4), and the site of Makpan (Fig. 5) on and Alor. Consequently, the faunal δ13C and δ18O datasets of the Alor, have δ13C values between −14.1‰ and −9.6‰, indicating islands (Timor n = 111; Alor n = 112) were separated for human reliance on a mixture of terrestrial tropical forest subsequent analyses unless otherwise specified. For both Timor resources and more open C3 environments (Fig. 2). and Alor, Shapiro–Wilk tests found the resulting δ13C and δ18O Indeed, Fig. 6 demonstrates that even the δ13C values of datasets for each island to be non-normally distributed (p = < humans excavated from Tron Bon Lei (Supplementary Note 1), 0.05). Mann–Whitney–Wilcoxon tests found marine and terres- with an assumed economic and cultural reliance on marine trial fauna to be significantly different in terms of δ13C on both resources, have a range of between −12.5‰ and −9.6‰. With Timor (W = 2657, p = < 0.05) and Alor (W = 2423, p = < 0.05). perhaps the exception of the individual from square C with a No difference was found between terrestrial and marine faunal value of −9.6‰, who likely demonstrates some contribution of groups in terms of δ18O on either Timor (W = 1507, p = > 0.05) marine or C4 resources, this shows that terrestrial C3 resources or Alor (W = 1375, p = > 0.05). made up the majority of the diets of these individuals. Never- In terms of terrestrial palaeoenvironmental conditions and theless, it is clear that two Terminal Pleistocene/Holocene changes in Late Pleistocene–Holocene Wallacea, our data directly individuals at Makpan, Alor (−8.1‰; 15,000−11,000 cal. years confirms the C3 forest–woodland preferences for now-extinct BP; Fig. 5), and Matja Kuru 2, Timor (−5.6‰; Fig. 4; giant rat taxa on both Timor and Alor (Figs. 3–5; as per ref. 27). 11,000–4000 cal. years BP) incorporated a significant proportion Separation of the terrestrial fauna dataset from Timor (n = 76) of C4/marine and marine resources into their diets, respectively, and Kruskal–Wallis analysis shows there to be no significant δ13C based on associated faunal data (Fig. 2), suggesting that diets were differences by phase on this island (Kruskal–Wallis chi-squared diverse between individuals and societies during the Terminal = 3.157, df = 3, p = > 0.05). Although significant differences in Pleistocene and Holocene. δ18O were noted between phases (Kruskal–Wallis chi-squared = 13.647, df = 3, p = <0.05), pairwise comparison failed to draw out any specific differences (>0.05) (Supplementary Table 8). For the Fourier transform infrared spectroscopy. Full results of the Alor terrestrial faunal dataset (n = 42), the situation is more infrared indices of samples subjected to FTIR analysis are shown complicated. Here, a Kruskal–Wallis test (Kruskal–Wallis chi- in Supplementary Data 3. All of the fossil and modern enamel squared = 14.386, df = 2, p = <0.05) followed by pairwise com- samples displayed classic enamel FTIR spectra (Supplementary parison demonstrated significant δ13C differences between phases Fig. 15). No additional bands from secondary carbonate (e.g. A (40,000–21,000 years ago) and C (4000–0 years ago), and B calcite at 710 cm−1; ref. 41) were observed in the spectra of the (15,000–7400 years ago) and C (4000–0 years ago; Supplementary fossil samples (Supplementary Fig. 15). Boxplots of API, BPI, Table 9). No significant differences were found for δ18O WAMPI, PCI, and BAI for the groups of ‘Modern’, ‘Fossil (Kruskal–Wallis chi-squared = 2.321, df = 2, p = > 0.05). Human’, ‘Fossil Terrestrial Fauna’, and ‘Fossil Marine Fauna’ are The difference between the islands is driven by the fact that in shown in Supplementary Fig. 16. Broadly, there are minimal the earliest phase of occupation on Alor (40–21,000 cal. years BP) changes between the groups with fossil samples perhaps having there is an overlap between terrestrial and marine fauna in δ13C marginally lower API, slightly higher BPI, and higher BAI than (Figs. 2 and 5). From this point, the δ13C of terrestrial fauna the ‘Modern’ sample (Supplementary Fig. 16). Analysis of var- declines through the phases, with terrestrial and marine fauna iance (ANOVA) and post-hoc Tukey pairwise comparisons 4 NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 ARTICLE Fig. 3 Isotope data from Asitau Kuru (Timor). Stable carbon (δ13C) and oxygen (δ18O) isotope data from terrestrial and marine faunal tooth enamel samples, with human samples shown as white circles, displayed by site phases (see Supplementary Note 1). support this, finding no significant differences in A-site carbo- (F(3,55) = 6.232, p < 0.05), with ‘Fossil Marine Fauna’ and ‘Fossil nation (F(3,55) = 1.194, p > 0.05), B-site carbonation (F(3,55) = Terrestrial Fauna’ being significantly different from ‘Modern’ 1.387, p > 0.05), PCI (F(3,55) = 1.845, p > 0.05), or WAMPI (F samples (Supplementary Table 11). (3,55) = 1.245, p > 0.05) between the different sample groups. By Poor preservation of skeletal material has been suggested contrast, BAI does show a difference between the groups elsewhere in tropical rainforest environments, though such NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 Fig. 4 Isotope data from Matja Kuru 1 and 2 (Timor). Stable carbon (δ13C) and oxygen (δ18O) isotope data from terrestrial faunal tooth enamel samples from Matja Kuru 2 (Timor), with human samples shown as white diamonds, displayed by site phases (see Supplementary Note 1). Regional phasing has been used for the 11,000–4000 cal. years BP grouping so individuals from Matja Kuru 1 and 2 can be combined. The grouping of 16,000–11,000 cal. years BP is only represented by two human individuals from Matja Kuru 1 as there is no occupation at Matja Kuru 2 at this time (see Supplementary Note 1). 6 NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 ARTICLE Fig. 5 Isotope data from Makpan (Alor). Stable carbon (δ13C) and oxygen (δ18O) isotope data from terrestrial and marine faunal tooth enamel samples, with human samples shown as white triangles, displayed by site phases (see Supplementary Note 1). comments have mainly focused on organic bone material42. The of a problem in the context of more porous materials, such as fossil enamel FTIR spectra produced here are virtually indis- bone43,44. Subtle differences noted in faunal enamel apatite tinguishable from modern spectra (Supplementary Figs. 15 and during fossilization noted here, including increased BAI and 16). The precipitation of carbonate minerals is likely to be more decreased A-carbonate on phosphate index, have also been NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 Fig. 6 Isotope data from Tron Bon Lei (Alor). Stable carbon (δ13C) and oxygen (δ18O) isotope data from terrestrial and marine faunal tooth enamel samples, with human samples shown as white inverted triangles, displayed by site phases (see Supplementary Note 1). demonstrated in other studies, including within the tropics of demonstrate broader use of interior environments, including South Asia, and have been argued to be a product of the reduction closed tropical forest habitats. This may be considered surprising, in organic material within the apatite matrix through time31,38,45. particularly given the ongoing presence of fish and shellfish9, the Such change is not considered to have major impacts on overall symbolic burial of an individual at Tron Bon Lei with fishhooks enamel apatite structure or stable carbon and oxygen isotope (Supplementary Note 1), as well as archaeological evidence for measurements from enamel45, as further suggested by the increased transfer of material culture between islands from this preservation of expected ecological differences in the enamel time46,47. However, tooth enamel δ13C reflects the whole diet of here and elsewhere29,31. an individual, and our data highlights the necessity of paying more attention to the contribution of plant (and terrestrial ani- mal) resources to human diets on tropical islands, particularly as Discussion those populations become more established—something also Our stable isotope data allow us to directly assess the ecological recently urged even for the study of the later Lapita expansion in reliance on different categories of resources that accompanied our the Pacific48,49. species’ arrival and subsequent settling of Wallacea. The earliest This study presents the first detailed palaeoenvironmental humans to arrive in this part of the world seemingly specialized in information for Late Pleistocene–Holocene Wallacea, directly the use of coastal resources. While we cannot currently distin- associated with records of human behavior. Our data indicates guish between pelagic and other forms of offshore resource use13, that, on Timor, tropical forest environments remained prevalent and our sample size from this early period is limited, we can be throughout the past 45,000 years, only decreasing during the Late confident that the δ13C value for this individual indicates a reli- Holocene with the arrival of human-induced deforestation during ance on marine resources. Following 20,000 cal. years BP, a clear the Iron Age27,50. There is no evidence for the presence of C4 diversification in human resource use across Wallacea emerges. grassland environments in the vicinity of any of the sites studied. While some coastal reliance is indicated by one individual at By contrast, on Alor, C4 resources may have been available to some Matja Kuru 2, and perhaps also Makpan, most individuals small mammals, and presumably also humans, during the earliest 8 NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 ARTICLE period of occupation (40,000–21,000 years ago). From this point and changes following the arrival of ‘Neolithic’ material culture during the Holo- onward, these resources disappear as tropical forest environments cene (Supplementary Note 1, Supplementary Tables 6 and 7). Although bone collagen is frequently the primary tissue used in the isotopic expanded across the Terminal Pleistocene–Holocene boundary. determination of ancient diets, it is often degraded and nearly impossible to extract While it remains possible that elevated δ13C for some fauna in the from archaeological remains in the humid tropics, particularly those dating back to earlier phase represents consumption of marine resources, this the Pleistocene57,58. This is the case for the sites studied here. By contrast, tooth palaeoenvironmental pattern of increasing tropical forests during enamel consists primarily of an inorganic fraction extremely resilient to the Terminal Pleistocene has also been documented elsewhere in postmortem diagenesis59, meaning that it is the fossil tissue of choice for tropical dietary reconstruction58,60. Tooth identification and analyses were conducted at the both Island and Mainland Southeast Asia5. In these cases, the Australian National University (ANU). Fish, reptile, and non-murid mammal expansion of tropical forest during the Terminal Pleistocene and identifications were facilitated through comparisons with specimens from the ANU Early Holocene has also been associated with increasingly specia- Archaeology and Natural History Osteology Laboratory reference collection. Murid lized human hunting of arboreal and semi-arboreal mammals and identifications were facilitated through comparisons with archaeological and fossil specimens collected from previous ANU expeditions, material held in KA’s private the use of tropical forest plants5. collections, the Australian National Wildlife Collection of the Commonwealth Our isotopic evidence supports human colonization models of Scientific and Industrial Research Organisation National Facilities and Collections, Wallacea and Australia that suggest a rapid, initial coastal colo- and descriptions and illustrations in Aplin and Helgen61 and Glover62. nization, followed by later inland settlement12,14, at least with Tooth enamel records an isotopic dietary signature for the period of enamel formation that will vary depending on the species and tooth sampled. For humans, regards to the sites studied here. This mode of colonization the longest isotopic signature is provided by third molars that can develop any time is distinct from isotopic and material evidence from the Wet Zone between 7 and 13 years of age, and the mid to late teenage years63,64. While rainforests of Late Pleistocene Sri Lanka31,51 and archaeological collagen δ13C is also biased toward protein components of the diet65, tooth enamel evidence from the Niah Caves in Borneo52 that indicate dedi- δ13C reflects that of the whole diet during formation. Due to the rarity of fossil human remains in Pleistocene archaeological sites, we sampled all of the available cated, specialized tropical forest foraging by early human popu- human teeth present (Supplementary Data 2). The result is one of the largest lations in these regions from 45,000 years ago. This further collections of human tooth enamel data that spans the Late Pleistocene/Holocene highlights the potential role of sophisticated seafaring in the (see refs. 31,60 for comparison in Sri Lanka) and the first for Island Southeast Asia. human colonization of eastern Wallacea and Australasia12,13,53. A While some of these teeth come from skeletons that have been aged and sexed, the later, increased focus on terrestrial resources or nearshore coastal majority are loose teeth. Subtle variations in δ13C and δ18O between teeth could occur as a result, particularly for teeth formed during ‘weaning’66. However, overall resources during the Terminal Pleistocene and Holocene has also this minor ‘noise’ will not hinder the testing of the major δ13C distinctions between been argued on the basis of zooarchaeological evidence from closed canopy C3 resources, C3 open settings, C4 resources, and marine resources. Timor, and elsewhere in Island Southeast Asia18,54. This sub- sistence pattern occurs alongside an increase in occupation Stable isotope analysis of archaeological tooth enamel. All teeth or teeth intensity across Wallacea from the Terminal Pleistocene, as well fragments were cleaned using air-abrasion to remove any adhering external as an increase in formalized long-distance trading networks, and material. Enamel powder for bulk analysis was obtained using gentle abrasion with is likely representative of exchange between settled groups in the a diamond-tipped drill along the full length of the buccal surface, in order to ensure a representative measurement for the entire period of enamel formation. All region46,47. enamel powder was pretreated to remove organic or secondary carbonate con- The adaptive flexibility visible in the colonization of almost all taminates. This method followed established protocols that have been applied of the Earth’s continents by our species in the Late Pleistocene elsewhere to Pleistocene tooth enamel in the tropics, where it has proven to be stands in stark contrast to the adaptations of other hominin effective31,60,67,68, enabling future comparison between datasets. Samples were washed in 1.5% sodium hypochlorite for 60 min, followed by three rinses in pur- species3,6,9,24. While there is some evidence that other hominins ified H2O and centrifuging, before 0.1 M acetic acid was added for 10 min, followed may have made water crossings20–23 or ventured into high- by another three rinses in purified H2O. Samples were then lyophilized for 24 h. altitude environments10, where present, existing zooarchaeologi- Following reaction with 100% phosphoric acid, gases evolved from the samples cal/paleontological and palaeoenvironmental evidence suggests a were measured by stable carbon and oxygen isotope analysis using a Thermo Gas Bench 2 connected to a Thermo Delta V Advantage Mass Spectrometer at MPI- general, albeit diverse, focus on mixed grassland and woodland SHH. δ13C and δ18O values were compared against International Standards environments, with dispersals and contractions often reliant on (IAEA-603 (δ13C = 2.5; δ18O = −2.4); IAEA-CO-8 (δ13C = −5.8; δ18O = −22.7); climatically driven environmental change24,55. Future isotopic USGS44 (δ13C = −42.2)) and in-house standard (MERCK (δ13C = −41.3; δ18O = analysis and more detailed zooarchaeological work are required to −14.4)) using Isodat 3.0 software from Thermo Electron Corporation. Replicate analysis of MERCK carbonate standards suggests that machine measurement error test this distinction, both in Wallacea and beyond. However, there is c. ±0.1‰ for δ13C and ±0.2‰ for δ18O. While each sample was measured only is clear evidence that different populations of H. sapiens were able once to preserve material for future analyses, as is standard for this approach, to specialize in a variety of extreme environments even as our overall measurement precision and reproducibility for tooth enamel samples on species as a whole generalized in the use of multiple settings6. this machine setup was studied through the measurement of repeat extracts from This flexibility, perhaps supported by unique capacities of inno- an in-house bovid tooth enamel standard (n = 20; δ13C = −12.4 ± 0.2‰; δ18O = −8.0 ± 0.3‰). vation, technological sophistication, and social communication (e.g., ref. 56), enabled adaptation to a variety of conditions, not Fourier transform infrared spectroscopy. While reliable δ13C- and δ18O-based just through space but also through time, that would eventually dietary and environmental indicators have been demonstrated across millions of leave us the last hominin standing. years29, protocols to check the structural preservation of fossil tooth enamel samples remain important (see ref. 69). This is particularly the case in tropical forest environments with ion-rich soils and high hydrological activity. One means Methods to check enamel preservation is the application of FTIR, which absorbs radiation at Sites, samples, and chronology. We sampled available human and animal teeth discrete vibrational frequencies related to the presence and crystallographic from the Late Pleistocene–Holocene deposits of Asitau Kuru, Matja Kuru 1 and 2, environment of key functional groups (see refs. 45,70,71). The polyatomic ions of and Lene Hara on the island of Timor and Makpan and Tron Bon Lei on the island interest are phosphates (PO43−), carbonates (CO32−), and hydroxyl groups (OH−). of Alor. Existing, including published, stratigraphic, and chronological information The observed absorbance bands of enamel can be ascribed to the internal vibrations was compiled, in order to enable the placement of samples in their proper context. of these molecular groups41,72 (Supplementary Table 12). All of the available dates used, stratigraphic information, and accompanying We use the empirical indices from Sponheimer and Lee-Thorp70, and Roche archaeological finds can be found in Supplementary Note 1, Supplementary et al.45 to characterize the crystal-chemical properties of enamel bioapatite Figs. 1–12, and Supplementary Tables 1–7. The available samples were selected on (Supplementary Table 12). The possible presence of calcite was assessed in all the basis of occupation phases noted and published for each archaeological site samples by checking for a peak at 710 cm−1 (refs. 41,70). 14 ‘Fossil Human’, 15 ‘Fossil (Supplementary Note 1, Supplementary Tables 1–7). Based on these established Terrestrial Fauna’, and 15 ‘Fossil Marine Fauna’ samples were subjected to FTIR sequences, samples were also grouped into an overall ‘phasing’ for each of the analysis following pretreatment, in order to determine the remaining potential for islands of Timor and Alor to provide larger sample sizes for the broader evaluation diagenetic structural and compositional modification of enamel after pretreatment. of adaptive context for the arrival of humans on each island, hypothesized Samples were randomly selected to cover a variety of the temporal phases and all of palaeoenvironmental changes across the Terminal Pleistocene/Holocene boundary, the sites studied (Supplementary Data 3). The fossil spectra were compared to those NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 available for 15 modern primate and cervid samples, and historical (late nineteenth 8. Gamble, C. Timewalkers: The Prehistory of Global Colonization (Alan Sutton and early twentieth century) human enamel samples (‘Modern’), from populations Press, Cheltenham, 1993). living in tropical forest environments in Sri Lanka already published by Roberts 9. O’Connor, S. et al. Hominin dispersal and settlement east of Huxley’s Line: the et al.73 (Supplementary Data 3). role of sea-level changes, island size, and subsistence behaviour. Curr. For all samples, powdered enamel was analysed between 400 and 4,000 cm−1 by Anthropol. 58, S567–S582 (2017). FTIR with Attenuated Total Reflectance (FTIR-ATR—Bruker Vertex 70 v) using 10. Chen, F. et al. A late Middle Pleistocene Denisovan mandible from the Tibetan the OPUS 8.5 software from Bruker. Each sample was measured three times. The Plateau. Nature 569, 409–412 (2019). background was subtracted and a baseline correction was carried out using the 11. Bird, M., Taylor, D. & Hunt, C. Palaeoenvironments of insular Southeast Asia OPUS 8.5 software from Bruker. The baselines of the spectra were normalized and during the last glacial period: a savanna corridor in Sundaland? Quat. Sci. Rev. all three spectra of each sample were averaged before calculation of the various 24, 2228–2242 (2005). infrared indices. To ensure better reproducibility of the measurements, only spectra 12. Bird, M. I. et al. Early human settlement of Sahul was not an accident. Nat. Sci. with a minimum absorbance of 0.06 for the highest phosphate band at ~1035 cm−1 Rep. 9, 8220 (2019). were taken into account. The reproducibilities of the indices BPI, API, BAI, and 13. O’Connor, S., Ono, R. & Clarkson, C. Pelagic fishing at 42,000 years before the PCI are ±0.01, ±0.007, ±0.1, and ±0.1, respectively. present and the maritime skills of modern humans. Science 334, 1117–1121 (2011). Statistical analysis. All δ13C and δ18O datasets were tested for normality using 14. Kealy, S., Louys, J. & O’Connor, S. Reconstructing palaeogeography and inter- the Shapiro–Wilk test and histogram observations. Following observations of a lack island visibility in the Wallacean Archipelago during the likely period of Sahul of normality, the significance of δ13C and δ18O variation between the two islands Colonization, 65–45 000 Years Ago. Archaeol. Prospect. 24, 259–272 (2017). (Timor and Alor), and between terrestrial and marine fauna on each island were 15. Norman, K. et al. An early colonization pathway into northwest Australia 70- tested using Mann–Whitney–Wilcoxon tests. The significance of δ13C and δ18O 60,000 years ago. Quat. Sci. Rev. 180, 229–239 (2017). variation between taxa and between the different periods was tested using 16. O’Connell, J. F. & Allen, J. The restaurant at the end of the universe: modeling Kruskal–Wallis tests. Where significant, these tests were followed by a pairwise the colonization of Sahul. Aust. Archaeol. 74, 5–31 (2012). Wilcoxon comparison to determine which groups were significantly different from 17. Samper Carro, S. C. et al. Human maritime subsistence strategies in the Lesser each other. In all cases, the dataset being tested and its size are explicitly stated. All Sunda Islands during the terminal Pleistocene-early Holocene: New evidence statistical analyses were conducted using the free program R software74. from Alor, Indonesia. Quat. Int. 416, 64–79 (2016). All FTIR index values were tested for normality using the Shapiro–Wilk test 18. Hawkins, S. et al. Oldest human occupation of Wallacea at Laili Cave, Timor- and histogram observations. Analysis of variance (ANOVA) followed by post hoc Leste, shows broad-spectrum foraging responses to late Pleistocene Tukey pairwise comparisons were performed for each of the main FTIR indices of environments. Quat. Sci. Rev. 171, 58–72 (2017). enamel apatite (PCI, BAI, BPI, API, and WAMPI—per45 and defined in 19. Shipton, C. et al. A new 44,000-year sequence from Asitau Kuru (Jerimalai), Supplementary Table 12) across the sample groups (i.e., ‘Fossil Marine Fauna’, Timor-Leste, indicates long-term continuity in human behaviour. Archaeol. ‘Fossil Terrestrial Fauna’, ‘Fossil Humans’, and ‘Modern’), in order to determine Anthropol. Sci. 11, 5717–5741 (2019). statistical differences in enamel crystallinity and structure between fossil and 20. Morwood, M. J. et al. Further evidence for small-bodied hominins from the modern samples. All statistical analyses were again conducted using the free Late Pleistocene of Flores, Indonesia. Nature 437, 1012–1017 (2005). program R software74. 21. Van den Bergh, G. D. et al. Homo floresiensis-like fossils from the early Middle Pleistocene of Flores. Nature 534, 245–248 (2016). Reporting summary. Further information on research design is available in 22. Ingicco, T. et al. Earliest known hominin activity in the Philippines by 709 the Nature Research Reporting Summary linked to this article. thousand years ago. Nature 557, 233–237 (2018). 23. Van den Bergh, G. D. et al. Earliest hominin occupation of Sulawesi, Data availability Indonesia. Nature 529, 208–211 (2016). All of the data reported in the paper are presented in the main text or in 24. Roberts, P. & Amano, N. Plastic pioneers: Hominin biogeography east of the the Supplementary Notes, Tables, Figures, and Data files. The source data underlying Movius Line during the Late Pleistocene. Archaeol. Res. Asia 17, 181–192 Fig. 2 and Supplementary Fig. 16 are provided as a Source Data file. The source data for (2019). Fig. 2, as well as Supplementary Data Files 1 and 2, also underlie Figs. 3–6. All other data 25. Brumm, A. et al. Age and context of the oldest known hominin fossils from supporting the findings and interpretations of this study are available in existing Flores. Nature 534, 249–253 (2016). publications and the Supplementary Information provided alongside this manuscript. 26. Anderson, A. The antiquity of sustained offshore fishing. Antiquity 87, The faunal and human remains sampled from Timor-Leste are curated in the 879–895 (2013). Archaeology collection at the College of Asia and the Pacific, Australian National 27. Louys, J. et al. New genus and species of giant rat from Alor Island, Indonesia. University, Australia under the site codes J (Asitau Kuru), MK (Matja Kuru), and LH J. Asia-Pac. Biodiv 11, 503–510 (2018). (Lene Hara). These materials are to be returned to Timor-Leste upon construction of a 28. Samper Carro, S. C., Louys, J. & O’Connor, S. Methodological considerations national museum storehouse. The faunal and human remains from Alor are housed in for icthyoarchaeology from the Tron Bon Lei sequence, Alor, Indonesia. the Department of Archaeology, Universitas Gadjah Mada, Indonesia, under the site Archaeol. Res. Asia 12, 11–22 (2017). codes TBL (Tron Bon Lei) and MP (Makpan). All site codes are followed by suffixes of a 29. Lee-Thorp, J. A., Sealy, J. C. & van der Merwe, N. J. Stable carbon isotope ratio single letter denoting the excavation square and a number denoting the spit. differences between bone collagen and bone apatite, and their relationship to diet. J. Archaeol. Sci. 16, 585–599 (1989). 30. Levin, N.E. et al. in The Geology of Early Humans in the Horn of Africa (eds J. Received: 26 January 2020; Accepted: 6 April 2020; Quade & J.G. Wynn.) 215–234 (Geological Society of America Special Paper 446, Boulder, Colorado, 2008). 31. Roberts, P. et al. Fruits of the forest: Human stable isotope ecology and rainforest adaptations in Late Pleistocene and Holocene (~ 36 to 3 ka) Sri Lanka. J. Hum. Evol. 106, 102–118 (2017). 32. Monk, K.A., de Fretes, Y. & Reksodiharjo-Lilley, G. The Ecology of Nusa References Tenggara and Maluku (Periplus Editions, Jakarta, Indonesia, 1997). 1. Aubert, M. et al. Earliest hunting scene in prehistoric art. Nature 576, 442–445 33. Smith, B. N. & Epstein, S. Two categories of 13C/12C ratios for higher plants. (2019). Plant Physiol. 47, 380–384 (1971). 2. Akhilesh, K. et al. Early Middle Palaeolithic culture in India around 385-172 34. Fry, B. Stable Isotope Ecology (Springer, New York, 2006). ka reframes Out of Africa Models. Nature 554, 97–101 (2018). 35. Kusaka, S. et al. Carbon isotope ratios of human tooth enamel record the 3. Rizal, Y. et al. Last appearance of Homo erectus at Ngandong, Java, 117,000- evidence of terrestrial resource consumption during the Jomon period, Japan. 108,000 years ago. Nature 577, 381–385 (2020). Am. J. Phys. Anthropol. 158, 300–311 (2015). 4. Reich, D. et al. Denisova admixture and the first modern human disper- 36. Clementz, M. T. & Koch, P. L. Differentiating aquatic mammal habitat and sals into Southeast Asia and Oceania. Am. J. Hum. Genet. 89, 516–528 foraging ecology with stable isotopes in tooth enamel. Oecologia 129, 461–472 (2011). (2001). 5. Rabett R.J. Human Adaptation in the Asian Palaeolithic (Cambridge 37. Cannon, C. H., Morley, R. J. & Bush, A. B. G. The current refugial rainforests University Press, Cambridge, 2012). of Sundaland are unrepresentative of their biogeographic past and highly 6. Roberts, P. & Stewart, B. A. Defining the ‘generalist-specialist’ niche for vulnerable to disturbance. PNAS 106, 11188–11193 (2009). Pleistocene Homo sapiens. Nat. Hum. Behav. 2, 542–550 (2018). 38. Roberts, P. et al. Fossil herbivore stable isotopes reveal middle Pleistocene 7. Zhu, R. X. et al. Early evidence of the genus Homo in East Asia. J. Hum. Evol. hominin palaeoenvironment in ‘Green Arabia’. Nat. Ecol. Evol. 2, 1871–1878 55, 1075–1085 (2008). (2018). 10 NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-020-15969-4 ARTICLE 39. Polis, G. A. et al. In Food Webs at the Landscape Level (eds Polis, G. A., Power, 67. Sponheimer, M. et al. Hominins, sedges, and termites: New carbon isotope M. E. & Huxel, G. R.) 200–216 (University of Chicago Press, Chicago, Illinois, data from the Sterkfontein valley and Kruger National Park. J. Hum. Evol. 48, USA, 2004). 301–312 (2005). 40. Schoeninger, M. J. & DeNiro, M. J. Nitrogen and carbon isotopic composition 68. Lee-Thorp, J. A. et al. Isotopic evidence for an early shift to C4 resources by of bone collagen from marine and terrestrial animals. Geochim. Cosmochim. Pliocene hominins in Chad. PNAS 109, 20369–20372 (2012). Acta 48, 625–639 (1984). 69. Schoeninger, M. J. et al. Isotopic alteration of mammalian tooth enamel. Int. J. 41. Farmer, V. C. The Infrared Spectra of Minerals (The Mineralogical Society, Osteoarchaeol. 13, 11–19 (2003). London, 1974). 70. Sponheimer, M. & Lee-Thorp, J. A. Isotopic evidence for the diet of an early 42. Tappen, M. Bone weathering in the tropical rain forest. J. Archaeol. Sci. 21, hominid, Australopithecus africanus. Science 283, 368–370 (1999). 667–673 (1994). 71. Nagy, Z. K. et al. Comparative performance of concentration and temperature 43. Lee-Thorp, J. A. & van der Merwe, N. J. Aspects of the chemistry of modern controlled batch crystallizations. J. Process Control 18, 399–407 (2008). and fossil biological apatites. J. Archaeol. Sci. 18, 343–354 (1991). 72. LeGeros, R. Z. Calcium phosphates in Oral Biology and Medicine. Monogr. 44. Michel, V., Ildefonse, P. & Morin, G. Assessment of archaeological bone and Oral. Sci. 15, 1–201 (1991). dentine preservation from Lazaret Cave (Middle Pleistocene) in France. 73. Roberts, P. et al. Historical tropical forest reliance amongst the Wanniyalaeto Palaeogeogr. Palaeoclimatol. Palaeoecol. 126, 109–119 (1996). (Vedda) of Sri Lanka: an isotopic perspective. Hum. Ecol. 46, 435–444 (2018). 45. Roche, D. et al. Preservation assessment of Miocene-Pliocene tooth enamel 74. R Core Team. R: A language and environment for statistical computing (R from Tugen Hills (Kenyan Rift Valley) - through FTIR, chemical and stable- Foundation for Statistical Computing, Vienna, Austria, 2013). isotope analyses. J. Archaeol. Sci. 37, 1690–1699 (2010). 46. Reepmeyer, C. et al. Kisar, a small island participant in an extensive maritime obsidian network in the Wallacean Archipelago. Archaeol. Res. Asia 19, Acknowledgements 100139 (2019). For permission to conduct fieldwork, we thank the Secretaria do Estado da Arte e 47. Shipton, C. et al. Shell Adzes, Exotic Obsidian, and Inter-Island voyaging in Cultura, Timor-Leste, and the Indonesian Ministry of Research, Technology, and Higher the early and Middle Holocene of Wallacea. JICA https://doi.org/10.1080/ Education (RISTEK) Foreign Research Permit Division (S.O’.C. 1172/FRP/E5/Dit.KI/V/ 15564894.2019.1581306 (2019). 2016). This project was funded by the Max Planck Society, a European Research Council 48. Maxwell, J. J. et al. Timing and importance of arboriculture and agroforestry Starter Grant awarded to P.R. (no. 850709), an Australian Research Council Laureate in a temperate East Polynesia Society, the Moriori, Rekohu (Chatham Island). Fellowship awarded to S.O’.C. (FL120100156), and the Australian Research Council Quat. Sci. Rev. 149, 306–325 (2016). Centre of Excellence for Australian Biodiversity and Heritage (CE170100015). We would 49. Tromp, M. et al. Exploitation and utilization of tropical rainforests indicated like to thank the landowners and villagers of Alor and Timor-Leste, staff and students in dental calculus of ancient Oceanic Lapita culture colonists. Nat. Hum. from the Universitas Gadja Mada, Pusat Penelitian Arkeologi Nasional, and Balai Behav. https://doi.org/10.1038/s41562-019-0808-y (2020). Arkeologi Bali for their assistance in the field. We also thank CartoGIS ANU for their 50. O’Connor, S. & Aplin, K. A matter of balance: an overview of Pleistocene assistance. occupation history and the impact of the Last Glacial Phase in East Timor and the Aru Islands, eastern Indonesia. Archaeol. Ocean 42, 82–90 (2007). Author contributions 51. Wedage, O. et al. Specialized rainforest hunting by Homo sapiens 45,000 years P.R., J.L., C.S., S.K., and S.O’C. designed the research; P.R., J.L., J.Z., C.S., S.K., S.S.C., S. ago. Nat. Commun. 10, 739 (2019). H., C.B., S.M., B.F., K.A., and S.O’C. collected the data; P.R., J.L., J.Z., C.S., S.K., S.S.C., S. 52. Barker, G., Farr, L. Archaeological Investigations in the Niah Caves, Sarawak H., C.B., and S.O’C. analyzed the data; P.R., J.L., J.Z., C.S., S.K., S.S.C., S.H., C.B., S.M., B. (The Archaeology of the Niah Caves, Sarawak 2) (McDonald Institute for F., N.B., M.M., K.A., and S.O’C. wrote the paper. Archaeological Research, Cambridge, 2016). 53. Balme, J. Of boats and string: the maritime colonisation of Australia. Quat. Int. 285, 68–75 (2013). Competing interests 54. Boulanger, C. et al. Coastal subsistence strategies and mangrove swamp The authors declare no competing interests. evolution at Bubog I Rockshelter (Ilin Island, Mindoro, Philippines) from the Late Pleistocene to the mid-Holocene. JICA 14, 584–604 (2019). 55. Louys, J. & Turner, A. Environment, preferred habitats and potential refugia Additional information Supplementary information is available for this paper at https://doi.org/10.1038/s41467- for Pleistocene Homo in Southeast Asia. C. R. Palevol 11, 203–211 (2012). 020-15969-4. 56. Boyd, R., Richerson, P. J. & Henrich, J. The cultural niche: why social learning is essential for human adaptation. PNAS 108, 10918–10925 (2011). Correspondence and requests for materials should be addressed to P.R. or S.O. 57. Krigbaum, J. Neolithic subsistence patterns in northern Borneo reconstructed with stable carbon isotopes of enamel. J. Anthropol. Archaeol. 22, 292–304 Peer review information Nature Communications thanks John Krigbaum and the other, (2003). anonymous, reviewers for their contribution to the peer review of this work. Peer 58. Krigbaum, J. Reconstructing human subsistence in the West Mouth (Niah reviewer reports are available. Cave) Sarawak) burial series using stable isotopes of carbon. Asian Perspect. 44, 73–89 (2005). Reprints and permission information is available at http://www.nature.com/reprints 59. Wang, Y. & Cerling, T. A model of fossil tooth and bone diagenesis: implications for paleodiet reconstruction from stable isotopes. Palaeogeogr. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in Palaeoclimatol. Palaeoecol. 107, 281–289 (1994). published maps and institutional affiliations. 60. Roberts, P. et al. Direct evidence for human reliance on rainforest resources in late Pleistocene Sri Lanka. Science 347, 1246–1249 (2015). 61. Aplin, K. P. & Helgen, K. M. Quaternary murid rodents of Timor Part I: new material of Coryphomys buehleri Schaub, 1937, and description of a second Open Access This article is licensed under a Creative Commons species of the genus. Bull. Am. Mus. Nat. 341, 1–80 (2010). Attribution 4.0 International License, which permits use, sharing, 62. Glover, I. Archaeology in Eastern Timor, 1966–67. Terra Australis 11 adaptation, distribution and reproduction in any medium or format, as long as you give (Department of Prehistory, Research School of Pacific Studies (Australian appropriate credit to the original author(s) and the source, provide a link to the Creative National University, Canberra, 1986). Commons license, and indicate if changes were made. The images or other third party 63. Hillson, S. Dental Anthropology (Cambridge University Press, Cambridge, material in this article are included in the article’s Creative Commons license, unless 1996). indicated otherwise in a credit line to the material. If material is not included in the 64. Scheid, R.C. Woelfel’s Dental Anatomy 7th edn (Lipincoot, Williams & article’s Creative Commons license and your intended use is not permitted by statutory Wilkins, Philadelphia, 2007). regulation or exceeds the permitted use, you will need to obtain permission directly from 65. Ambrose, S.H. & Norr, L. In Prehistoric Human Bone: Archaeology at the the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Molecular Level (eds Lambert, B. & Grupe, G.) 1–37 (Springer, Berlin, 1993). licenses/by/4.0/. 66. Wright, L. E. & Schwarcz, H. P. Stable carbon and oxygen isotopes in human tooth enamel: Identifying breastfeeding and weaning in prehistory. Am. J. Phys. Anthropol. 106, 1–18 (1998). © The Author(s) 2020 NATURE COMMUNICATIONS | (2020)11:2068 | https://doi.org/10.1038/s41467-020-15969-4 | www.nature.com/naturecommunications 11