Professional Documents
Culture Documents
doi:10.1038/nature10574
Despite decades of research, the roles of climate and humans in driving the dramatic extinctions of large-bodied mammals during the Late Quaternary period remain contentious. Here we use ancient DNA, species distribution models and the human fossil record to elucidate how climate and humans shaped the demographic history of woolly rhinoceros, woolly mammoth, wild horse, reindeer, bison and musk ox. We show that climate has been a major driver of population change over the past 50,000 years. However, each species responds differently to the effects of climatic shifts, habitat redistribution and human encroachment. Although climate change alone can explain the extinction of some species, such as Eurasian musk ox and woolly rhinoceros, a combination of climatic and anthropogenic effects appears to be responsible for the extinction of others, including Eurasian steppe bison and wild horse. We find no genetic signature or any distinctive range dynamics distinguishing extinct from surviving species, emphasizing the challenges associated with predicting future responses of extant mammals to climate and human-mediated habitat change.
Towards the end of the Late Quaternary, beginning around 50,000 years ago, Eurasia and North America lost approximately 36% and 72% of their large-bodied mammalian genera (megafauna), respectively1. The debate surrounding the potential causes of these extinctions has focused primarily on the relative roles of climate and humans25. In general, the proportion of species that went extinct
was greatest on continents that experienced the most dramatic climatic changes6, implying a major role of climate in species loss. However, the continental pattern of megafaunal extinctions in North America and Australia approximately coincides with the first appearance of humans, suggesting a potential anthropogenic contribution to species extinctions3,5.
1 Centre for GeoGenetics, University of Copenhagen, ster Voldgade 57, DK-1350 Copenhagen K, Denmark. 2Center for Macroecology, Evolution and Climate, Department of Biology, University of Copenhagen, Universitetsparken 15, DK-2100 Copenhagen , Denmark. 3Smithsonian Tropical Research Institute, Tupper Building, 401 Balboa, Ancon, Punama, Republica de Panama. 4Departments of Integrative Biology and Statistics, University of California, Berkeley, 4098 VLSB, Berkeley, California 94720, USA. 5Department of Biology, University of Copenhagen, Ole Maaloes Vej 5, DK-2200, Denmark. 6 7 School of Biological Sciences, University of Sydney, New South Wales 2006, Australia. Center for the Study of the First Americans, Department of Anthropology, Texas A&M University, College Station, Texas 77843, USA. 8Department of Sociology and Anthropology, Missouri State University, 901 South National, Springfield, Missouri 65807, USA. 9Department of Evolutionary Biology, Uppsala University, 75236 Uppsala, Sweden. 10Conservation and Evolutionary Genetics Group, Estacion Biologica de Donana (EBD-CSIC), Avenida Americo Vespucio, 41092 Seville, Spain. 11Department of Ecology and Evolutionary Biology, University of California, Los Angeles, California 90095, USA. 12Laboratory of Genomic Diversity, National Cancer Institute, Building 560, Room 11-33, Frederick, Maryland 21702, USA. 13 Department of Earth and Atmospheric Sciences, University of Alberta, Edmonton, Alberta T6G 2E3, Canada. 14Government of Yukon, Department of Tourism and Culture, Yukon Palaeontology Program, PO Box 2703 L2A, Whitehorse, Yukon Territory Y1A 2C6, Canada. 15Stafford Research Inc., 200 Acadia Avenue, Lafayette, Colorado 80026, USA. 16Department of Earth and Environment, Mount Holyoke College, 50 College Street, South Hadley, Massachusetts 01075, USA. 17School of Earth and Environment, University of Leeds, Woodhouse Lane, Leeds, West Yorkshire LS2 9JT, UK. 18School of Geographical Sciences, University of Bristol, University Road, Bristol BS8 1SS, UK. 19Diamond and Precious Metals Geology Institute, Siberian Branch of Russian Academy of Sciences, 39 Prospect Lenina, 677891 Yakutsk, Russia. 20Royal Alberta Museum, Edmonton, Alberta T5N 0M6, Canada. 21The Manitoba Museum, Winnipeg, Manitoba R3B 0N2, Canada. 22North-East Science Station, Pacific Institute for Geography, Far East Branch of Russian Academy of Sciences, 2 Malinovy Yar Street, 678830 Chersky, Russia. 23Museum of Natural and Cultural History, 1224 University of Oregon, Eugene, Oregon 974031224, USA. 24Institute of Plant and Animal Ecology, Ural Branch of the Russian Academy of Sciences, 8 Marta Street, 202, 620144 Ekaterinburg, Russia. 25Moscow State University, Vorobevy Gory, 119899 Moscow, Russia. 26State Key Laboratory of Biogeology and Environmental Geology, China University of Geosciences, Wuhan, Hubei 430074, China. 27University of Kansas Museum of Natural History, University of Kansas, Lawrence, Kansas 66045, USA. 28Park Museum Management Program, National Park Service, 1201 Oakridge Drive, Suite 150, Fort Collins, Colorado 80525, USA. 29Natural History Museum, Rotterdam, c/o Gudumholm 41, 2133 HG Hoofddorp, Netherlands. 30Bioinformatics Research Centre (BiRC), Aarhus University, C.F. Mllers Alle 8, DK-8000 Aarhus C, Denmark. 31 sidium Stuttgart, Landesamt fur Denkmalpflege, Stromeyersdorfstrasse 3, D-78467 Konstanz, Germany. 32Zoological Institute of Russian Academy of Sciences, Universitetskaya nab. 1, Regierungspra 199034 Saint-Petersburg, Russia. 33Christian-Albrechts-University of Kiel, Institute for Nature and Resource Conservation, Department of Landscape Ecology, Olshausenstrasse 40, 24098 Kiel, Germany. 34 Institute of Ecology and Evolution, Russian Academy of Sciences, 33 Leninsky Prospect, 119071 Moscow, Russia. 35San Bernardino County Museum, Division of Geological Sciences, 2024 Orange Tree Lane, Redlands, California 92374, USA. 36Departments of Biomathematics and Human Genetics, David Geffen School of Medicine, University of California, Los Angeles, Los Angeles, California 90095, USA. 37 Department of Biostatistics, UCLA School of Public Health, University of California, Los Angeles, Los Angeles, California 90095, USA. 38Museum of Cultural History, University of Oslo, St. Olavsgate 29, Postboks 6762 St. Olavsplass, 0130 Oslo, Norway. 39Australian Centre for Ancient DNA, The University of Adelaide, South Australia 5005, Australia. 40Department of Biology (Area 2), The University of York, Wentworth Way, Heslington, York YO10 5DD, UK. 41Department of Biology, The Pennsylvania State University, 326 Mueller Laboratory, University Park, Pennsylvania 16802, USA. 42Department of Anthropology, University of Utah, 271N1400E, Salt Lake City, Utah 84112-0060, USA. 43Museo de Historia Natural de San Rafael, (5600) Parque Mariano Moveno, San Rafael, Mendoza, Argentina. *These authors contributed equally to this work. {Deceased.
1 7 NO V E M B E R 2 0 1 1 | VO L 4 7 9 | N AT U R E | 3 5 9
RESEARCH ARTICLE
Demographic trajectories of different taxa vary widely and depend on the geographic scale and methodological approaches used3,5,7. For example, genetic diversity in bison8,9, musk ox10 and European cave bear11 declines gradually from approximately 50,00030,000 calendar years ago (kyr BP). In contrast, sudden losses of genetic diversity are observed in woolly mammoth12,13 and cave lion14 long before their extinction, followed by genetic stability until the extinction events. It remains unresolved whether the Late Quaternary extinctions were a cross-taxa response to widespread climatic or anthropogenic stressors, or were a species-specific response to one or both factors15,16. Additionally, it is unclear whether distinctive genetic signatures or geographical range-size dynamics characterize extinct or surviving speciesquestions of particular importance to the conservation of extant species. To disentangle the processes underlying population dynamics and extinction, we investigate the demographic histories of six megafauna herbivores of the Late Quaternary: woolly rhinoceros (Coelodonta antiquitatis), woolly mammoth (Mammuthus primigenius), horse (wild Equus ferus and living domestic Equus caballus), reindeer/ caribou (Rangifer tarandus), bison (Bison priscus/Bison bison) and musk ox (Ovibos moschatus). These taxa were characteristic of Late Quaternary Eurasia and/or North America and represent both extinct and extant species. Our analyses are based on 846 radiocarbon-dated mitochondrial DNA (mtDNA) control region sequences, 1,439 directly dated megafaunal remains and 6,291 radiocarbon determinations associated with Upper Palaeolithic human occupations in Eurasia. We reconstruct the demographic histories of the megafauna herbivores from ancient DNA data, model past species distributions and determine the geographical overlap between humans and megafauna over the past 50,000 years. We use these data to investigate how climate change and anthropogenic impacts affected species dynamics at continental and global scales, and contributed to the extinction of some species and the survival of others.
42 kyr BP Woolly rhinoceros Extinct 30 kyr BP 21 kyr BP 6 kyr BP
Woolly mammoth NA
Horse
Reindeer
Bison
Musk ox
Figure 1 | Modelled potential ranges of megafauna species at 42, 30, 21 and 6 kyr BP. Ranges were modelled using the megafauna fossil record and palaeoclimatic data for temperature and precipitation; ice sheet extent was not included as a co-variable. Range measurements were restricted to the regions for which fossils were used to build the models, rather than all potentially suitable Holarctic area. NA, not available.
noteworthy that both estimated modelled ranges and genetic data are derived from a subset of the entire fossil record (Supplementary Information sections 1 and 3). Thus, changes in effective population size and range size might change with the addition of more data, especially from outside the geographical regions covered by the present study. However, we expect that the reported positive correlation will prevail when congruent data are compared. The best-supported models of changes in effective population size in North America and Eurasia during periods of dramatic climatic change over the past 50,000 years are those in which populations increase in size (Fig. 3 and Supplementary Information section 3). This is true for all taxa except bison. However, the timing is not synchronous across populations. Specifically, we find highest support for population increase beginning approximately 34 kyr BP in Eurasian horse, reindeer and musk ox (Fig. 3a). Eurasian woolly mammoth and North American horse increase before the Last Glacial Maximum (LGM) approximately 26 kyr BP. Models of population increase in woolly rhinoceros and North American woolly mammoth fit equally well before and after the LGM, and North American reindeer populations increase later still. Only North American bison shows a population decline (Fig. 3b), the intensity of which probably swamps the signal of global population increase starting at approximately 35 kyr BP identified in the skyride plot (Fig. 2a). These increases in effective population size probably reflect responses to climate change. By 34 kyr BP, the relatively warmer Marine Isotope Stage (MIS) 3 interstadial marked the transition to cold, arid full-glacial conditions of MIS 2, which began approximately
ARTICLE RESEARCH
Horse (5 years) log10[Area (km2)]
7.5
Reindeer (4 years)
humans, as the key driver of these species-specific and, in some cases, continent-specific demographic changes. This conclusion is supported by the significant correlations between modelled range sizes and effective population sizes (Fig. 2).
6.5
7.0
Modes of extinction
Both woolly rhinoceros and woolly mammoth suffered global extinctions during the Late Quaternary. Neither shows evidence of a decline in genetic diversity leading to their extinction at either continental or global scales (Supplementary Figs 3.2 and 3.6). However, the fossil records of the two species differ: woolly rhinoceros remains widely distributed across Eurasia until it disappears from the fossil record approximately 14 kyr BP (Supplementary Fig. 2.2), whereas the woolly mammoth range retreats northwards during its last millennia (Supplementary Figs 2.3 and 5.2c, d). We find increased isolationby-distance preceding extinction (Supplementary Fig. 3.1 and Supplementary Information section 3), suggesting that populations of both species became increasingly fragmented, although the results are not statistically significant for woolly mammoth. The high and sustained levels of genetic diversity in these species might reflect the fixation of multiple distinct haplotypes in increasingly isolated and diminishing subpopulations. For woolly mammoth, this pattern is also supported by fossil evidence24. Our data suggest similar possibilities of increased isolation-bydistance before the extinctions of musk ox in Eurasia (approximately 2.5 kyr BP25,26) and of steppe bison in the north of the North American plains, which potentially survived until only a few hundred years ago8 (Supplementary Fig. 3.1). Such fragmentation is commonly observed in wide-ranging species undergoing population decline, owing to populations aggregating in patches of high-quality habitat27. In contrast, we find low levels of isolation-by-distance in wild horse and in Eurasian and North American reindeer, suggesting these populations remained relatively panmictic over time.
log10[Ne*]
6 5 4
Musk ox (2 years)
log10[Ne*]
6.5 7 6 5 4 50 40 30 20 10 0 50 40 30 20 10 0
7.0
Figure 2 | Temporal changes in global genetic diversity and range size in horse, bison, reindeer and musk ox. The x-axis is in calendar years; the y-axis is the product of effective population size and generation time (Net). Generation times are given in parentheses. Comparable estimates of associated range sizes (square kilometres) are from Fig. 1. The temporal span of the radiocarbon-dated samples used in each approach is shown as vertical lines below each panel; each line represents one dated individual.
Eurasia
0.31 0.24
5 30K
North America
NA 0.21 0.24 0.24
0.29
NA 0.73
Musk ox
5 30K
NA
40
40
0.50
Bison
0.44
0.30
Reindeer
Figure 3 | Best-supported demographic models inferred by approximate Bayesian computation model-selection. a, Eurasia; b, North America. Grey dots on the time axis indicate periods with range size estimates. Yellow dots indicate the periods of demographic increase or decline, which were tested against each other in the approach. White values inside coloured bars reflect support for the best-supported model (for example, Eurasian woolly mammoth, increase at 26 kyr BP). The intensity of increase or decline (for example, 310) and effective population size at the time of the youngest sample (for example, 5,000 individuals) are shown. We indicate in grey cases where multiple models received similar levels of support.
To evaluate the potential role of humans in the local and global megafauna extinctions, we measured the following: (1) the spatial overlap between the modelled range of each megafauna species and the Eurasian Palaeolithic archaeological record at 42, 30 and 21 kyr BP; (2) the presence of megafauna remains in Palaeolithic archaeological assemblages from Europe (4818 kyr BP) and Siberia (4112 kyr BP); and (3) variation in fossil abundance and the temporal and spatial distributions of known Palaeolithic archaeological sites and the Eurasian megafauna fossil record at 1,000-year intervals. For the last category, we added 1,557 indirectly dated megafaunal remains to the 1,439 directly dated specimens to increase sample sizes. Although associated with greater age-estimate uncertainties, the integrity of each of the indirectly dated samples was evaluated before inclusion following the guidelines listed in Supplementary Information section 5. Woolly rhinoceros and Eurasian woolly mammoth experience a fiveto tenfold increase in effective population size between 34 kyr BP and 19 kyr BP (Fig. 3), at least 10,000 years after first human contact as inferred from the overlap between estimated ranges and archaeological sites (Supplementary Figs 1.2 and 1.5). This result directly contradicts models of population collapse from human overkill (blitzkrieg)2 or infectious diseases following the first human contact (hyperdisease)28. We find no evidence that Palaeolithic humans greatly impacted musk ox populations, in agreement with previous conclusions that humans were not responsible for the extinction of musk ox in Eurasia10. Musk ox remains are found in only 1% of European archaeological sites and 6% of Siberian sites, and do not overlap noticeably in range with Palaeolithic humans in either Europe or Siberia (Fig. 4). However, the decline in the potential range of musk ox by 60% between 21 and 6 kyr BP (Fig. 1), the increase in isolation-by-distance at 19 kyr BP (Supplementary Fig. 3.1 and Supplementary Table 3.3) and the positive correlation between climate-driven range size and genetic diversity (Fig. 2b) all point
1 7 NO V E M B E R 2 0 1 1 | VO L 4 7 9 | N AT U R E | 3 6 1
0.26 0.35
0.21
0.48
RESEARCH ARTICLE
towards climate as the main driver of musk ox population dynamics, including the decrease in genetic diversity after the LGM (Fig. 2a). The importance of climate is further supported by the physiology of musk ox, which might be a more sensitive indicator of environmental warming than the other species. Musk ox has extreme temperature sensitivity and is unable to tolerate high summer temperatures; the 10 uC summer isotherm approximates the southern limit of its present-day range29. We find little regional overlap between Palaeolithic humans and woolly rhinoceros in Siberia after the LGM (that is, after 20 kyr BP); the species is found in fewer than 11% of Siberian archaeological sites during this time (Fig. 4). This suggests that woolly rhinoceros was not a common prey species for humans, and that overhunting is an unlikely explanation for their extinction in Siberia. However, we note that geographical overlap existed between humans and woolly rhinoceros in Europe during the two millennia preceding extinction (Fig. 4), and therefore cannot exclude the hypothesis that humans influenced the final collapse of the species in this region. The continued presence of woolly rhinoceros in the fossil record throughout Siberia and parts of Europe up until the species extinction event (Supplementary Fig. 2.2) suggests that the final collapse of the species was synchronous across its range. The data from woolly mammoth are inconclusive about the causes of extinction. We find that the range of Eurasian woolly mammoth
a
Archaeofaunal sites
overlaps continuously with humans throughout the Palaeolithic (Fig. 4), in agreement with previous results based on a more limited data set30. Woolly mammoth remains are found in 40% and 35% of all European and Siberian Palaeolithic sites, respectively, and mammoth subsistence hunting by Clovis peoples in North America has been documented31. However, the prevalence of woolly mammoth in Siberian sites declines after the LGM (43% of sites before 19 kyr BP compared with 30% after; Fig. 4). This decline could indicate a northward range shift of woolly mammoth ahead of humans30 (Fig. 5.2c, d), an increasing scarcity of woolly mammoths in southern Siberia or an increasing human preference for other prey species. In wild horse, the large mid-Holocene range of over 9 million km2 (Fig. 1 and Supplementary Table 1.3) suggests the potential for a large Eurasian population at this time, and is not consistent with climate driving the final disappearance of the species in the wild. Rather, the decline in genetic diversity observed after the LGM in horse and bison, and to a lesser degree in reindeer (Fig. 2), might reflect the impact of expanding human populations in Europe and Asia. The presence of the three species in the archaeological record suggests that their populations are more likely to have been influenced by humans. Bison and horse are the most common megafauna herbivores found in archaeological sites (Fig. 4), with horse present in 58% and 66% of European and Siberian sites, respectively. Furthermore, horse shows extensive geographical
b
Siberia
Archaeofaunal sites
20 1.0
Europe
Spatial overlap
1.0
Spatial overlap
Woolly 60 rhinoceros
30
0.5
10
0.5
Woolly mammoth
60
0 1.0
0 20
0 1.0
30
0.5
10
0.5
0 60
0 1.0
0 20
0 1.0
Horse
30 0.5 10 0.5
0 60
0 1.0
0 20
0 1.0
Reindeer
30 0.5 10 0.5
0 60
0 1.0
0 20
0 1.0
Bison
30
NA
0.5
NA
10
0.5
0 60
0 1.0
0 20
0 1.0
Musk ox
30 0.5 10 0.5
0 44
0 12 44
12
0 44
0 12 44
12
Figure 4 | Spatial and temporal association between megafauna and Upper Palaeolithic humans. a, Europe; b, Siberia. Column graphs represent known cultural occupations containing at least one of the six species, averaged over 2,000-year time bins; data span 4818 kyr BP for Europe and 4112 kyr BP for Siberia. Open bars indicate the number of archaeofaunal sites, filled bars
3 6 2 | N AT U R E | VO L 4 7 9 | 1 7 NO V E M B E R 2 0 1 1
represent the frequency of each species in the binned assemblages. Area graphs show the fraction of megafauna surface area shared with humans at 1,000-year intervals, calculated from the mean 6one standard deviation of latitude and longitude; data represented in Supplementary Fig. 5.2. Graphs use coordinates of data associated with both direct and indirect dates.
ARTICLE RESEARCH
overlap with humans in both Europe and Siberia after the LGM, although large population sizes might have insulated horses to some extent from the effects of selective hunting by humans. In bison, the pre-human decline in genetic diversity starting approximately 35 kyr BP and the strong correlation between range size and genetic diversity (Fig. 2) indicate climate as a main driver of demographic change. This conclusion is supported by the fivefold decline in effective population size (Fig. 3) and increased isolation-bydistance approximately 11 kyr BP in North America (Supplementary Fig. 3.1 and Supplementary Table 3.3). The timing of these demographic changes coincides with the pronounced climatic shifts associated with the Pleistocene/Holocene transition32, although they also coincide with fossil evidence of growing populations of potential competitors such as Alces and Cervus33. The accelerated rate of decline in genetic diversity after approximately 16 kyr BP (Fig. 2) is coincident with the earliest known human expansion in the Americas23, and the significant presence of bison in 77% of the Siberian archaeological assemblages points to their popularity as a prey species (Fig. 4). Reindeer are the most abundant of the six taxa today. As with horse, they show continuous geographical overlap with Palaeolithic humans in Eurasia (Fig. 4). Reindeer are common in both European and Siberian Palaeolithic assemblages, are found in 67% of Siberian sites after the LGM and were an important prey species for humans in both Eurasia and North America34. Unlike bison and horse, the potential range of reindeer declines by 84% between 21 and 6 kyr BP (Fig. 1 and Supplementary Table 1.3). Despite the apparently detrimental influences of both humans and climate change, wild and domestic reindeer currently number in the millions across the Holarctic35. Although individual populations are affected by changing climate36, the species is not currently under threat of extinction. The success of reindeer may be explained by high fecundity37 and ecological flexibility38. In addition, continued low levels of isolation-by-distance suggest high mobility and near-panmixia of populations over millennia (Supplementary Fig. 3.1 and Supplementary Table 3.3).
past 50,000 years; (2) serial-coalescent simulations and the approximate Bayesian computation model-selection approach39 to assess demographic change in Eurasia and in North America, and in the global data set; (3) isolation-by-distance to investigate changes in population structure over time in the two continental subpopulations. Palaeoclimatic estimates of precipitation and temperature were used to model the potential geographical range of each species at 42, 30, 21 and 6 kyr BP, using only contemporaneous radiocarbon-dated megafauna fossils (63 kyr) for each period. Range measurements were restricted to Holarctic regions for which fossils were used to build the models. Using a Bayesian hierarchical modelling framework, these changes in range size were compared with changes in effective population size estimated from the Bayesian skyrides. To assess the spatial and temporal association between humans and megafauna, we (1) analysed variations in fossil abundance and spatial and temporal overlap between the human Upper Palaeolithic and megafauna fossil records in Europe and Siberia, (2) inferred the area of overlap between the human data from (1) and the megafauna ranges at 42, 30 and 21 kyr BP, and (3) assembled a list of the cultural occupations in Europe and Siberia with megafauna presence, to determine which taxa were directly associated with Palaeolithic humans. For details on methods see Supplementary Information.
Full Methods and any associated references are available in the online version of the paper at www.nature.com/nature. Received 22 April; accepted 16 September 2011. Published online 2 November 2011. 1. Barnosky, A. D., Koch, P. L., Feranec, R. S., Wing, S. L. & Shabel, A. B. Assessing the causes of Late Pleistocene extinctions on the continents. Science 306, 7075 (2004). Martin, P. S. in Quaternary Extinctions: A Prehistoric Revolution (eds Martin, P. S. & Klein, R. G.) 364403 (Univ. Arizona Press, 1984). Alroy, J. A. multispecies overkill simulation of the end-Pleistocene megafaunal mass extinction. Science 292, 18931896 (2001). Stuart, A. J., Kosintsev, P. A., Higham, T. F. G. & Lister, A. M. Pleistocene to Holocene extinction dynamics in giant deer and woolly mammoth. Nature 431, 684689 (2004). Koch, P. L. & Barnosky, A. D. Late Quaternary extinctions: state of the debate. Annu. Rev. Ecol. Evol. Syst. 37, 215250 (2006). Nogues-Bravo, D., Ohlemuller, R., Batra, P. & Araujo, M. B. Climate predictors of Late Quaternary extinctions. Evolution 64, 24422449 (2010). Haile, J. et al. Ancient DNA reveals late survival of mammoth and horse in interior Alaska. Proc. Natl Acad. Sci. USA 106, 2236322368 (2009). Shapiro, B. et al. Rise and fall of the Beringian steppe bison. Science 306, 15611565 (2004). Drummond, A. J., Rambaut, A., Shapiro, B. & Pybus, O. G. Bayesian coalescent inference of past population dynamics from molecular sequences. Mol. Biol. Evol. 22, 11851192 (2005). Campos, P. F. et al. Ancient DNA analyses exclude humans as the driving force behind Late Pleistocene musk ox (Ovibos moschatus) population dynamics. Proc. Natl Acad. Sci. USA 107, 56755680 (2010). Stiller, M. et al. Withering away25,000 years of genetic decline preceded cave bear extinction. Mol. Biol. Evol. 27, 975978 (2010). Barnes, I. et al. Genetic structure and extinction of the woolly mammoth, Mammuthus primigenius. Curr. Biol. 17, 14 (2007). Debruyne, R. et al. Out of America: ancient DNA evidence for a new world origin of Late Quaternary woolly mammoths. Curr. Biol. 18, 13201326 (2008). Barnett, R., Yamaguchi, N., Barnes, I. & Cooper, A. The origin, current diversity, and future conservation of the modern lion (Panthera leo). Proc. R. Soc. B 273, 21592168 (2006). Guthrie, R. D. Rapid body size decline in Alaskan Pleistocene horses before extinction. Nature 426, 169171 (2003). Grayson, D. K. Deciphering North American Pleistocene extinctions. J. Anthropol. Res. 63, 185214 (2007). Andrewartha, H. G. & Birch, L. C. The Distribution and Abundance of Animals (Univ. Chicago Press, 1954). Borregaard, M. K. & Rahbek, C. Causality in the relationship between geographic distribution and species abundance. Q. Rev. Biol. 85, 325 (2010). Minin, V. N., Bloomquist, E. W. & Suchard, M. A. Smooth skyride through a rough skyline: Bayesian coalescent-based inference of population dynamics. Mol. Biol. Evol. 25, 14591471 (2008). Zazula, G. D. et al. Ice age steppe vegetation in east Beringia. Nature 423, 603 (2003). Zazula, G. D., Froese, D. G., Elias, S. A., Kuzmina, S. & Mathewes, R. W. Arctic ground squirrels of the mammoth-steppe: paleoecology of Late Pleistocene middens (,24 00029 450 14C yr BP), Yukon Territory, Canada. Quat. Sci. Rev. 26, 9791003 (2007). Pitulko, V. V. et al. The Yana RHS site: humans in the Arctic before the last glacial maximum. Science 303, 5256 (2004). Goebel, T., Waters, M. R. & ORourke, D. H. The Late Pleistocene dispersal of modern humans in the Americas. Science 319, 14971502 (2008).
1 7 NO V E M B E R 2 0 1 1 | VO L 4 7 9 | N AT U R E | 3 6 3
2. 3. 4.
5. 6. 7. 8. 9.
Conclusions
We find that neither the effects of climate nor human occupation alone can explain the megafauna extinctions of the Late Quaternary. Rather, our results demonstrate that changes in megafauna abundance are idiosyncratic, with each species (and even continental populations within species) responding differently to the effects of climate change, habitat redistribution and human encroachment. Although reindeer remain relatively unaffected by any of these factors on a global scale, climate change alone explains the extinction of Eurasian musk ox and woolly rhinoceros, and a combination of climatic and anthropogenic effects appears to be responsible for the demise of wild horse and steppe bison. The causes underlying the extinction of woolly mammoth remain elusive. We have shown that changes in habitat distribution and population size are intrinsically linked over evolutionary time, supporting the view that populations of many species will decline in the future owing to climate change and habitat loss. Intriguingly, however, we find no distinguishing characteristics in the rate or pattern of decline in those species that went extinct compared with those that have survived. Our study demonstrates the importance of incorporating lessons from the past into rational, data-driven strategies for the future to address our most pressing environmental challenges: the ongoing global massextinction of species and the impacts of global climate change and humans on the biodiversity that remains.
10.
20. 21.
METHODS SUMMARY
Our data comprise 846 radiocarbon-dated ancient mitochondrial DNA sequences, 1,439 directly dated and 1,557 indirectly dated megafauna specimens, and 6,291 dated remains associated with Upper Palaeolithic humans in Eurasia. For population genetic analysis, we used the following: (1) the Bayesian skyride approach20 to estimate the global demographic trajectory of each species over the
22. 23.
RESEARCH ARTICLE
24. Stuart, A. J., Sulerzhitsky, L. D., Orlova, L. A., Kuzmin, Y. V. & Lister, A. M. The latest woolly mammoths (Mammuthus primigenius Blumenbach) in Europe and Asia: a review of the current evidence. Quat. Sci. Rev. 21, 15591569 (2002). 25. Vereshchagin, N. K. Prehistoric hunting and the extinction of Pleistocene mammals in the USSR. Proc. Zool. Inst. Russ. Acad. Sci. 69, 200232 (1971). 26. Kuznetsova, T. V., Sulerzhitsky, L. D., Siegert, C. & Schirrmeister, L. (2001) in La Terra degli Elefanti (eds Cavarretta, G. Giola, P., Mussi, M. & Palombo, M. R.) The World of Elephants, Proc. 1st Int. Congr. 289292 (2001). 27. Wilson, R. J., Thomas, C. D., Fox, R., Roy, D. B. & Kunin, W. E. Spatial patterns in species distributions reveal biodiversity change. Nature 432, 393396 (2004). 28. MacPhee, R. D. E. & Marx, P. A. in Natural Change and Human Impact in Madagascar (eds Goodman, S. M. & Patterson, B. D.) 169217 (Smithsonian Institution Press, 1997). 29. Tener, J. S. Muskoxen in Canada: a biological and taxonomic review. Canadian Wildlife Service Monograph Series No. 2 (1965). 30. Ugan, A. & Byers, D. A global perspective on the spatiotemporal pattern of the Late Pleistocene human and woolly mammoth radiocarbon record. Quaternary Int. 191, 6981 (2008). 31. Surovell, T. A. & Waguespack, N. M. How many elephant kills are 14? Clovis mammoth and mastodon kills in context. Quaternary Int. 191, 8297 (2008). 32. Zielinski, G. A. & Mershon, G. R. Paleoenvironmental implications of the insoluble microparticle record in the GISP2 (Greenland) ice core during the rapidly changing climate of the PleistoceneHolocene transition. Geol. Soc. Am. Bull. 109, 547559 (1997). 33. Guthrie, R. D. New carbon dates link climatic change with human colonization and Pleistocene extinctions. Nature 441, 207209 (2006). 34. Farnell, R. et al. Multidisciplinary investigations of alpine ice patches in southwest Yukon, Canada: paleoenvironmental and paleobiological investigations. Arctic 57, 247259 (2004). 35. Williams, T. M. & Heard, D. C. World status of wild Rangifer tarandus population. Rangifer 1 (special issue), 1928 (1986). 36. Joly, K., Klein, D. R., Verbyla, D. L., Rupp, T. S. & Chapin, F. S. Linkages between largescale climate patterns and the dynamics of Arctic caribou populations. Ecography 34, 345352 (2011). 37. Skogland, T. The effects of density-dependent resource limitation on the demography of wild reindeer. J. Anim. Ecol. 54, 359374 (1985). 38. Leader-Williams, N. Reindeer on South Georgia Ch. 1, 318 (Cambridge Univ. Press, 1988). 39. Beaumont, M. in Simulations, Genetics and Human Prehistory (eds Matsumura, S., Forster, P. & Renfrew, C.) 134154 (McDonald Institute for Archaeological Research, 2008). Supplementary Information is linked to the online version of the paper at www.nature.com/nature. Acknowledgements This paper is in memory of our friend and colleague Andrei Sher, who was a contributor to this study. Dr Sher died unexpectedly, but his major contributions to the field of Quaternary science will be remembered and appreciated for many years. We are grateful to A. Lister and T. Stuart for guidance and discussions. We thank T. B. Brandt, B. Hockett and A. Telka for laboratory help and samples, and L. M. R. Thrane for his work on the megafauna locality database. Data taken from the Stage 3 project were partly funded by grant F/757/A from the Leverhulme Trust, and a grant from the McDonald Grants and Awards Fund. B.S. was supported by NSF ARC-0909456. We acknowledge the Danish National Research Foundation, the Lundbeck Foundation, the Danish Council for Independent Research and the US National Science Foundation for financial support. Author Contributions E.W. conceived and headed the overall project. C.R. headed the species distribution modelling and range measurements. E.D.L. and J.T.S. extracted, amplified and sequenced the reindeer DNA sequences. J.B. extracted, amplified and sequenced the woolly rhinoceros DNA sequences; M.H. generated part of the woolly rhinoceros data. J.W., K.-P.K., J.L. and R.K.W. generated the horse DNA sequences; A.C. generated part of the horse data. L.O., E.D.L. and B.S. analysed the genetic data, with input from R.N., K.M., M.A.S. and S.Y.W.H. Palaeoclimate simulations were provided by P.B., A.M.H, J.S.S. and P.J.V. The directly dated spatial latitudinal/longitudinal megafauna locality information was collected by E.D.L., K.A.M., D.N.-B., D.B. and A.U.; K.A.M. and D.N.-B. performed the species distribution modelling and range measurements. M.B. carried out the geneclimate correlation. A.U. and D.B. assembled the human Upper Palaeolithic sites from Eurasia. T.G. and K.E.G. assembled the archaeofaunal assemblages from Siberia. A.U. analysed the spatial overlap of humans and megafauna and the archaeofaunal assemblages. E.D.L., L.O., B.S., K.A.M., D.N.-B., M.K.B., A.U., T.G. and K.E.G. wrote the Supplementary Information. D.F., G.Z., T.W.S., K.A.-S., G.B., J.A.B., D.L.J., P.K., T.K., X.L., L.D.M., H.G.M., D.M., M.M., E.S., M.S., R.S.S., T.S., E.S., A.T., R.W. and A.C. provided the megafauna samples used for ancient DNA analysis. E.D.L. produced the figures. E.D.L, L.O. and E.W. wrote most of the manuscript, with input from B.S., M.H., D.N.-B., K.A.M., M.T.P.G., C.R., R.K.W, A.U. and the remaining authors. Author Information Mitochondrial DNA sequences are deposited in GenBank under accession numbers JN570760JN571033. Reprints and permissions information is available at www.nature.com/reprints. The authors declare no competing financial interests. Readers are welcome to comment on the online version of this article at www.nature.com/nature. Correspondence and requests for materials should be addressed to E.W. (ewillerslev@snm.ku.dk).
3 6 4 | N AT U R E | VO L 4 7 9 | 1 7 N O V E M B E R 2 0 1 1
ARTICLE RESEARCH
METHODS
Data. Mitochondrial DNA sequences and accelerator mass spectrometry radiocarbon dates were collected from the past and present geographical ranges of six megafauna herbivores from Eurasia and North America: woolly rhinoceros (Coelodonta antiquitatis), woolly mammoth (Mammuthus primigenius), horse (wild Equus ferus and living domestic Equus caballus), reindeer/caribou (Rangifer tarandus), bison (Bison priscus/Bison bison) and musk ox (Ovibos moschatus) (Supplementary Fig. 2.1 and Supplementary Information sections 2 and 3). Our data comprise 846 radiocarbon-dated ancient mitochondrial DNA sequences (274 of which are new), 1,439 directly dated megafauna specimens (357 of which are new) and 6,291 dated remains associated with Upper Palaeolithic humans in Eurasia. In one analysis of the spatial and temporal association between humans and megafauna detailed below, we included an additional 1,557 indirectly dated megafaunal remains. Species distribution modelling. We assessed changes in potential range size of each species over the past 50,000 years using 829 radiocarbon-dated megafauna fossils calibrated with the IntCal09 calibration curve40 and palaeoclimatic estimates of precipitation and temperature41. Potential ranges were estimated for the four periods for which palaeoclimatic data are available, 42, 30, 21 and 6 kyr BP, using only contemporaneous fossils (63 kyr) for each period (Supplementary Fig. 1.2). We compared temporal changes in potential range size (from species distribution models) and genetic diversity (from Bayesian skyrides19) during the past 50 kyr BP to assess the relation between these independent proxies of population size. If climate were a major driver of changes in population size, we would expect these two measures to be positively correlated. Estimating past ranges using species distribution models can be affected by an incomplete or biased fossil record as well as inaccuracies in the palaeoclimate simulations used in the models; uncertainties associated with these issues are depicted in our estimates of range size and how it correlates to genetic diversity (Supplementary Fig. 4.3). Range measurements were restricted to regions for which fossils were used to build the models, rather than all potentially suitable Holarctic areas. Fossil localities represent a subset, rather than an exhaustive search, of the literature available, and modelled ranges consequently represent a subset of the entire past distribution of the species. Too few fossils were available to estimate the potential ranges of woolly rhinoceros and woolly mammoth at 6 kyr BP, as the former was extinct and the latter was restricted to two island populations. Thus, too few periods with range estimates for these two species precluded statistical comparison with the genetic data, which spanned 50,000 years. For further details see Supplementary Information sections 1 and 4. Ancient genetic analysis. We used three analytical approaches capable of incorporating serially sampled data to reconstruct the past population dynamics of each megafauna herbivore species. (1) The Bayesian skyride approach19 estimates changes in genetic diversity through time as a proxy for effective population size, and was used to estimate the global demographic trajectory of each species. Because these data sets comprise samples from both a broad temporal and geographical extent, it is likely that they violate, at least during some of their evolutionary history, the assumption of panmixia made by the coalescent models currently implemented in BEAST42. However, the skyride makes the least stringent prior assumptions among these coalescent models, and therefore is the most likely to accommodate the temporal changes in structure that might characterize each of these species. (2) Serial-coalescent simulations and the approximate Bayesian computation model-selection approach39 were used to test for demographic change in the continental subpopulations (Eurasia and North America) and in the global data set. Time points were chosen to represent midpoints between the four periods (42, 30, 21 and 6 kyr BP) for which we modelled potential megafauna ranges, and periods of dramatic climatic changes: the beginning (26 kyr BP) and end (19 kyr BP) of the LGM, the onset of the Younger Dryas (12.9 kyr BP) and the beginning of the Holocene (11 kyr BP). (3) Isolation-by-distance was used to test for changes in population structure over time in the continental subpopulations. Note that as with the species distribution models, the demographic events inferred from the ancient DNA data are conditional upon the samples included in the analysis. Hence, although we use the broad geographical terms of Eurasia and North America, the regions are limited to the localities covered by the sequenced samples (Supplementary Fig. 2.1). For further details on the genetics data see Supplementary Information section 2. For further details on the statistical analysis see Supplementary Information section 3. Spatial association between megafauna and Palaeolithic humans. The presence of humans within the range of a species might directly or indirectly influence the capacity of the species to occupy that habitat. As a proxy for human impact, we assessed the spatial and temporal association between humans and megafauna using three approaches. (1) We compiled the human Upper Palaeolithic fossil record (5012 kyr BP), including 6,291 radiocarbon determinations associated with human occupations in Europe and Siberia. We analysed variations in fossil abundance and spatial and temporal overlap at 1,000-year intervals between humans and the megafauna fossil record. To increase sample sizes for this particular analysis, we augmented the 1,439 directly dated megafauna specimens with an additional 1,557 indirectly dated megafaunal remains. Although associated with greater ageestimate uncertainties, the integrity of each indirectly dated sample was evaluated before inclusion following the guidelines listed in Supplementary Information section 5. (2) We inferred the area of overlap between the archaeological record from (1) and the megafauna ranges at 42, 30 and 21 kyr BP estimated using species distribution models. (3) We assembled a list of 380 cultural occupations in Europe (4818 kyr BP) and 98 sites in Siberia (4112 kyr BP) with megafauna presence, to determine which taxa were directly associated with Palaeolithic humans. For further details see Supplementary Information section 5.
40. Reimer, P. J. et al. IntCal09 and Marine09 radiocarbon age calibration curves, 050,000 years cal BP. Radiocarbon 51, 11111150 (2009). 41. Nogues-Bravo, D., Rodrguez, J., Hortal, J., Batra, P. & Araujo, M. B. Climate change, humans, and the extinction of the woolly mammoth. PLoS Biol. 6, e79 (2008). 42. Drummond, A. J. & Rambaut, A. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7, 214 (2007).
SUPPLEMENTARY INFORMATION
This file contains six Supplementary Information sections, which include Supplementary Data,
doi:10.1038/nature10574
Supplementary Figures, Supplementary Tables, and Supplementary Discussion. SuPPLementarY InFormatIon SECTION OVERVIEW Section S1: Section S2: Section S3: Section S4: Section S5: Species distribution modelling (page 2)
Megafauna ancient DNA extraction, amplification and sequencing (page 24) Megafauna ancient DNA sequence analysis (page 31) Gene-climate correlation (page 52) Temporal and spatial overlap of humans and megafauna (page 62)
References for S1S5 (page 79) Section S6: Data tables and sample information (page 87)
w w w. n a t u r e . c o m / n a t u r e | 1
doi:10.1038/nature10574
WWW.NATURE.COM/NATURE | 2
doi:10.1038/nature10574
species through the late Quaternary. In practice, SDMs reconstruct species geographic distributions by relating species presence records (in this case, fossil locality data) to a set of environmental predictors (e.g., temperature and rainfall) to map a species geographic range using a geographic information system (GIS)13. Strong enthusiasm for incorporating SDMs in a variety of biological studies has resulted in intense scrutiny of the methods theoretical assumptions14. Paramount is recognising the difference between a species fundamental niche, the full set of conditions in which a species can survive long-term, and the realised niche, the subset of the fundamental niche that is actually occupied at a given time5 and upon which SDMs are based. SDMs are generated using climatic data15, but a species realised niche is also determined by other factors (such as barriers to dispersal). Projecting an SDM onto past or future climate surfaces, as is common in climate change studies, may ignore those limits while assuming a species will exist in all places with favourable climatic conditions, and that the niche is static through evolutionary timeassumptions which need to be explored for many species14 . Further, combinations of climatic variables with no analogues in other time periods may result in underestimation of a species ecological and geographic range in past or future projections16. Therefore, range size estimates to be compared with the genetic data (results in Fig. 2 in main text) were modelled using only locality and climate data from the same time periods (42, 30, 21 and 6 kyr BP); SDMs from one time period were not projected onto earlier or later periods, and range measurements were restricted to regions for which fossils were used to build the models, rather than all potentially suitable Holarctic area. This approach allowed us to circumvent assumptions regarding climatic niche stasis through time, as well as the effects of dispersal limitations which might have prevented species from reaching areas of otherwise suitable habitat. 1.2 Palaeoclimate data Late Quaternary climatic conditions are simulated using Atmospheric-Ocean coupled General Circulation Models (hereafter AOGCMs). An AOGCM is a set of equations simulating the dynamics of the ocean and the atmosphere under certain environmental conditions (i.e, CO2 concentration, ice sheet extent) to provide estimates of past climatic parameters (e.g., rainfall or temperature). Each AOGCM differs slightly in both the absolute values of estimated climatic conditions and in the geographical distribution of those conditions. To assess the effect of AOGCM choice on species distributions and the subsequent relationship between range size and effective population size, we simulated past climatic conditions using two different AOGCMs: GENESIS2 and HadCM3.
WWW.NATURE.COM/NATURE | 3
doi:10.1038/nature10574
1.2.1 GENESIS2 Four GENESIS2 simulations were used: two for Marine Isotope Stage 3 (MIS 3), one for the Last Glacial Maximum (LGM; ~21 kyr BP) and one for the mid-Holocene (~6 kyr BP). The Marine Isotope Stage 3 (MIS 3) simulations represent the warmer middle part (~42 kyr BP) and colder later part (~30 kyr BP) of MIS 3. Carbon dioxide levels were specified at 200 ppm for the MIS 3 and LGM simulations17 and 280 ppm for the mid-Holocene simulation18. Sea surface temperatures (SSTs) for the MIS 3 and LGM simulations were taken primarily from CLIMAP19, with modifications from GLAMAP-2000 and other sources20. SSTs for the mid-Holocene simulation were prescribed at present-day values21. In all cases, insolation was calculated using orbital parameters22,23. All simulations were spun up to equilibrium; results are 10-year averages. Comparison of GENESIS2 model output to proxy data shows that temperatures in Europe are accurate to within 1C for the mid-Holocene and 2C for the LGM24. For Oxygen Isotope Stage 3, 25 found that GENESIS2 temperatures in southern Europe agree well with proxy data but were 34C too warm in Northern Europe. Comparison of present day GENESIS2 model output with observations suggests that variability in Europe can be extrapolated to all of northern Eurasia and North America26. We are unaware of any GENESIS2 model-data comparisons for palaeoprecipitation. However, for present-day northern Eurasia and North America the model is accurate to within 1 mm day-1 when compared to observations26. Atmospheric carbon dioxide boundary conditions are well-constrained27,28. 1.2.2 HadCM3 A second set of climate model outputs for the same Quaternary periods, the warmer middle part (~40 kyr BP) and colder later part (~32 kyr BP) of MIS 3, LGM and the Mid-Holocene were derived from the Hadley Centre Coupled Climate Model Version 3 (HadCM3). The simulations form part of the ensemble presented in 29. The particulars of HadCM3 are well documented30. HadCM3 was one of the first coupled atmosphere-ocean climate models which required no flux corrections, even for simulations of a thousand years or more31. The climate model consists of a linked atmospheric model, ocean model and sea ice model. In HadCM3 the horizontal resolution of the atmosphere model is 2.5 degrees latitude by 3.75 degrees longitude. This gives a grid spacing at the equator of 278 km in the north-south direction and 417 km east-west and is approximately comparable to a T42 spectral model resolution. The atmospheric model consists of 19 layers. The spatial resolution over the ocean in is 1.25 1.25 and the model has 20 layers. The atmospheric
WWW.NATURE.COM/NATURE | 4
doi:10.1038/nature10574
model has a time step of 30 minutes and includes a radiation scheme that can represent the effects of minor trace gases32. A parameterization of simple background aerosol climatology is also included33. The convection scheme is that of 34. A land-surface scheme includes the representation of the freezing and melting of soil moisture. The representation of evaporation includes the dependence of stomatal resistance on temperature, vapour pressure and CO2 concentration35. The ocean model includes the use of the Gent-McWilliams mixing scheme36. There is no explicit horizontal tracer diffusion in the model. The horizontal resolution allows the use of a smaller coefficient of horizontal momentum viscosity leading to an improved simulation of ocean velocities. The sea ice model is a simple thermodynamic scheme and contains parameterizations of ice drift and leads (Polynyas37). For the MIS 3, LGM and mid-Holocene, orbital parameters are taken from 23. Atmospheric concentrations of CO2 were taken from the Vostok ice core record38 and CH4, and N2O were taken from EPICA39. All ice-core data were on the same EDC3 timescale40. Ice-sheet reconstructions are developed from the ICE5G model41, which includes a detailed evolution of the ice thickness, extent and continental isostatic rebound for the whole period from the LGM to the modern at 500-year intervals. Using standard linear interpolation techniques, this dataset was used to calculate, at the scale of the climate model, the total continental elevation (including the direct thickness of the ice sheets plus the effects of isostatic adjustment), bathymetry (including isostatic changes), ice-area extent, and land sea mask for the LGM and mid-Holocene. To ensure consistency with pre-industrial boundary conditions an anomaly-based method was used to calculate palaeogeographic boundary conditions. In this method, for anomalies of a particular time-slice, palaeogeography minus pre-industrial ICE-5G data are then added to our model preindustrial geographical boundary conditions. The geographical extent and heights of the major ice sheets prior to the LGM were based on 42, which included a calculation of the pre- and post-glacial ice. Singarayer and Valdes29 used the SPECMAP43 record of d18O history to constrain the evolution of the volume of land ice from the last interglacial up to the LGM. Each simulation was integrated for approximately 200 years, a sufficient period of time to bring the surface climatology to equilibrium, with the final 30 years used to calculate the required climatological mean. The version of HadCM3 used does not include interactive vegetation, so all simulations use the same pre-industrial vegetation boundary condition. Similarly, aerosol loading in the model is unchanged and does not account for changes in dust during the cycle.
WWW.NATURE.COM/NATURE | 5
doi:10.1038/nature10574
Using HadCM3, we obtain reasonable estimates of the global temperature glacial/interglacial range as well as trends in polar regions, although the magnitude of change at high latitudes is underestimated, as other similar models have found for the LGM29. HadCM3 produces a pattern of cooling for the LGM which are broadly consistent with the findings from simpler models and palaeoclimatic data44. HadCM3 performance in simulating both the LGM and mid-Holocene have been evaluated and are recognised as being either good or generally comparable with other climate models run for the same time intervals45. 1.2.3 Palaeoclimate variables From both AOGCM datasets, three variables (Supplementary Figure S1.1) were selected to describe the potential distribution of each species: mean temperature of the coldest month (C), mean temperature of the warmest month (C), and annual precipitation (mm). Temporal trends in these climatic variables are similar between the two AOGCMs (Supplementary Figure S1.2). Palaeoclimatic simulations for GENESIS2 are at a 22-degree spatial resolution; palaeoclimatic simulations for HadCM3 are resampled at a 22-degree resolution. This conservative variable set was selected to balance the number of species occurrences versus the number of climatic variables used to calibrate species realised climatic niches. Using many climatic variables to model the potential distribution of a species using few presence records (e.g., fossil localities) is likely to lead to model over-fitting, yielding a misrepresentation of the geographical distribution of the modelled species46. Given the constraint of using only a small number of climatic variables, we aimed to capture the upper and lower thermal limits of each species, as well as a moisture variable. Previous applications of SMDs have used similar limited sets of climatic variables as predictors of megafauna distributions3. All figures presented in the main text were generated using climate variables from GENESIS2.
WWW.NATURE.COM/NATURE | 6
doi:10.1038/nature10574
mm
0 - 100 100 - 150 150 - 200 200 - 250 250 - 300 300 - 500 500 - 750 750 - 1,000 1,000 - 1,500 1,500 - 2,000
0 42 kyr BP- 100 100 - 150 150 - 200 200 - 250 250 - 300 300 - 500 500 - 750 750 - 1,000 1,000 - 1,500
1,500 - 2,000
30 kyr BP
21 kyr BP
6 kyr BP
C
-70 - -50 -50 - -40 -40 - -30 -30 - -20 -20 - -10 -10 - 0 0 - 10 10 - 20 20 - 30 30 - 45
-70 - -50 -50 - -40 -40 - -30 -40 - -20 -20 - -10 -10 - 0 0 - 10 10 - 20 20 - 30 30 - 45
Supplementary Figure S1.1. Palaeoclimatic data (GENESIS2) for annual precipitation (mm), average temperature of the warmest month (C), average temperature of the coldest month (C) and the fossil record were used to estimate species potential ranges.
Supplementary Figure S1.2. Average climatic temporal trends across the Holartic based on GENESIS, left, and HadCM3, right, AOGCMs. Green (lower) lines indicate mean temperature of
WWW.NATURE.COM/NATURE | 7
doi:10.1038/nature10574
the coldest month (C), red (upper) lines indicate mean temperature of the warmest month (C), and blue bars indicate annual precipitation (mm). 1.3 Megafauna locality data For each species, 14C-dated fossil localities from Eurasia and North America were obtained for the following calendar time intervals: 4539, 3327, 2418 and 93 kyr BP. Radiocarbon dates (uncalibrated 14C dates) were calibrated into calendar years using the IntCal09 calibration curve47 using the OxCal 4.1 online calibration resource (https://c14.arch.ox.ac.uk). The 829 fossil localities included data from sequenced specimens (Supplementary TableS6.2-6.4), supplemented with fossil localities from the literature (Supplementary Table S6.1); The majority of the 829 localities used for the Species Distribution Models (SDMs) were compiled from synthetic works, rather than the original papers in which the dates were first presented, so we were unable to evaluate our data by fossil context, dated material or dating method48. However, the vast majority (98%) of the 829 localities used were from directly dated (standard or AMS) animal remains; thus, if we assume that most dates were from bone collagen, dung, hide or hair, we anticipate that most of our directly dated samples would rank an 11 or 12 on the scale proposed by 48 . Given the breadth of the time bins used6,000 calibrated yearsthe difference between dating methods should have little impact on our results, as this is significantly greater than the differences between conventional and AMS dates observed by 48. The list of localities is not and was not intended to be exhaustive, but was meant to cover at least those regions with genetic data to enable comparison of the estimates of potential range size and estimated effective population size. Therefore, modelled species ranges are not intended to fully represent past species distributions in great detail, and may under-represent the species actual range in areas for which we have little data. Fossil localities are indicated in Supplementary Fig. S1.3. Data were only included where the literature contained explicit geographic coordinates or detailed site descriptions which could be located at http://toolserver.org/~geohack/. As each set of geographical coordinates relates to a specific dated fossil, localities are duplicated where more than one dated fossil has been found. For woolly mammoth, all known fossils dating from 93 kyr BP are from Wrangel Island49,50, with the exception of one known specimen from St. Paul Island51. This fossil distribution, consisting of two unique localities, is insufficient data with which to generate species distribution models, and so
WWW.NATURE.COM/NATURE | 8
doi:10.1038/nature10574
no distribution is presented for the woolly mammoth at 6 kyr BP, even though the species was not yet extinct.
Woolly rhinoceros
Woolly mammoth
42 kyr BP
42 kyr BP
30 kyr BP
30 kyr BP
21 kyr BP
21 kyr BP
Supplementary Figure S1.3. Megafauna potential range at 42, 30, 21 and 6 kyr BP estimated from palaeoclimatic data (GENESIS2) and dated fossils for each species, represented by white dots.
WWW.NATURE.COM/NATURE | 9
doi:10.1038/nature10574
Range measurements were restricted to regions for which fossils were used to build the models, rather than all potentially suitable Holarctic area. Contemporaneous Palaeolithic human sites for each period are represented by black dots. No or too few fossils were available for woolly rhinoceros and mammoth to estimate their ranges at 6 kyr BP.
Horse
42 kyr BP
30 kyr BP
21 kyr BP
6 kyr BP
WWW.NATURE.COM/NATURE | 10
doi:10.1038/nature10574
Reindeer
42 kyr BP
30 kyr BP
21 kyr BP
6 kyr BP
WWW.NATURE.COM/NATURE | 11
doi:10.1038/nature10574
Bison
42 kyr BP
30 kyr BP
21 kyr BP
6 kyr BP
WWW.NATURE.COM/NATURE | 12
doi:10.1038/nature10574
Musk ox
42 kyr BP
30 kyr BP
21 kyr BP
6 kyr BP
WWW.NATURE.COM/NATURE | 13
doi:10.1038/nature10574
1.4 Species distribution modelling 1.4.1 Model algorithms Species distribution model (SDM) projections are sensitive to the different statistical techniques used to describe and project species potential ranges. Mahalanobis Distance (MD)52, a strict presence-only method, was used to model the 42, 30, 21 and 6 kyr BP distributions of each species. MD is a simple envelope technique that provides a strict presence-only measure of environmental distance, which is calculated in relation to an optimum climatic point, defined as the centroid for all occurrence points in the total climatic space. The distance between this "optimum" and the observed climatic values for each species presence is inverse to the suitability of the climate at that site. MD produces an ellipsoidal envelope around the climatic optimum space by taking into account the covariance among climatic variables. While a variety of methods are available for modelling species distributions, certain characteristics of the fossil record help to narrow the range of algorithms from which to select. First, the fossil record provides information about the presence of species, but not about their absence, and so presence-absence algorithms must be discarded. Other well-known, more complex algorithms such as GARP53 or Maxent54 could be used because they generate pseudo-absences against which to test the models. However, different procedures for calculating pseudo-absences can yield significantly different modelled species ranges55,56. In addition, bias in the fossil record is not simply a result of sampling effort, as with extant species, but of unevenly distributed geomorphological conditions affecting the fossilisation and persistence of remains through time, making selection of the study area on which to calculate pseudo-absences far more challenging for fossil data than for extant species for which the extent of the distribution is often known to some degree55. After these methods are discarded, the remaining suitable algorithms are restricted to simple presence-only methods based on environmental distances to a climatic optimum, which have been shown to handle bias in the fossil record better than more complex algorithms57. Specifically, MD has been shown to perform better than other presence-only methods in a recent comparative study58. It has been successfully used for palaeobiology studies3,57,59 and is specifically recommended57 for modelling potential species distributions using the fossil record.
WWW.NATURE.COM/NATURE | 14
doi:10.1038/nature10574
1.4.2 Model implementation and performance Species distributions used to estimate range size (see below) were modelled using only locality and climate data from the same time periods (42, 30, 21 and 6 kyr BP). A minimum of five dated fossil localities per species/time period were used to build the models60 (Supplementary Table S1.1). Modelling was implemented using the openModeller cross-platform modelling interface61. Each species/time period was modelled with both methods using all available locality data to build the model using both GENESIS2 and HadCM3. To assess the impact of using a limited subset of the known fossil record on model performance for each species, we performed ten independent model runs in which a different randomly selected 75% of the data were used to build the model and 25% were used to test it each time. This evaluation does not assess the accuracy of model predictionsindependent evaluation data (e.g., genetic data or additional fossils) would be required for this purpose. Rather, it provides a measure of internal consistency among repeated runs. Model performance was assessed using the Area Under the (Receiver Operating Characteristic) Curve (AUC; see 62 for a review on the advantages and disadvantages of using the AUC as a performance measure). Scores >0.75 are typically considered adequate for species distribution modelling63. Random sub-set model runs were performed using GENESIS2. Species distribution models yielded consistently high AUC values under testing, with only two species/time period yielding a coefficient of variation in AUC greater than 5% (bison and musk ox, 42 kyr BP; Supplementary Table S1.2). This suggests that model performance was relatively robust.
WWW.NATURE.COM/NATURE | 15
doi:10.1038/nature10574
Supplementary Table S1.1. Total number of samples used to build the SDM for each species/ time period. Some localities had multiple dated samples, and the number of unique localities is given in parentheses.
42 kyr BP 30 kyr BP 21 kyr BP 6 kyr BP Woolly rhinoceros Woolly mammoth Wild horse Reindeer Bison Musk ox 35 (29) 78 (50) 35 (21) 16 (10) 15 (8) 8 (7) 34 (28) 112 (77) 42 (29) 33 (23) 16 (9) 25 (10) 27 (24) 97 (78) 53 (24) 35 (12) 10 (7) 49 (10) n/a n/a 32 (26) 46 (18) 15 (7) 16 (12)
Supplementary Table S1.2. AUC scores for ten model runs using a unique 75% and 25% of the data for building and testing the model, respectively (GENESIS2). Mean AUC, standard deviation and coefficient of variation (CV) are indicated. Woolly rhinoceros 42 kyr BP 30 kyr BP 21 kyr BP 6 kyr BP 0.97 0.95 0.97 n/a 0.96 0.97 0.97 n/a 0.97 0.95 0.98 n/a 0.94 0.97 0.98 n/a 0.96 0.94 0.97 n/a 0.96 0.96 0.98 n/a 0.95 0.97 0.96 n/a 0.97 0.98 0.98 n/a 0.93 0.97 0.98 n/a 0.96 0.97 0.99 n/a 0.96 0.96 0.98 n/a 0.01 0.01 0.01 n/a 1.40% 1.30% 0.86% n/a Woolly mammoth 42 kyr BP 30 kyr BP 21 kyr BP 6 kyr BP 0.98 0.97 0.95 n/a 0.98 0.98 0.96 n/a 0.97 0.95 0.97 n/a 0.95 0.96 0.96 n/a 0.94 0.96 0.97 n/a 0.9 0.96 0.96 n/a 0.97 0.95 0.98 n/a 0.92 0.97 0.95 n/a 0.94 0.97 0.94 n/a 0.98 0.96 0.95 n/a 0.95 0.96 0.96 n/a 0.03 0.01 0.01 n/a 2.93% 0.99% 1.25% n/a Reindeer 42 kyr BP 30 kyr BP 21 kyr BP 6 kyr BP 0.98 0.98 0.96 0.98 0.96 0.97 0.98 0.98 0.95 0.98 0.94 0.98
run01 run02 run03 run04 run05 run06 run07 run08 run09 run10 Mean StDev CV
Wild horse 42 kyr BP 30 kyr BP 21 kyr BP 6 kyr BP run01 0.98 0.97 0.98 0.97 run02 0.99 0.98 0.94 0.99 run03 0.99 0.98 0.96 0.99
WWW.NATURE.COM/NATURE | 16
doi:10.1038/nature10574
0.98 0.99 0.98 0.99 0.97 0.99 0.99 0.99 0.01 0.72%
0.98 0.96 0.97 0.96 0.97 0.97 0.97 0.97 0.01 0.76%
0.98 0.99 0.96 0.95 0.97 0.99 0.96 0.97 0.02 1.74%
0.99 0.98 0.96 0.98 0.96 0.99 0.99 0.98 0.01 1.27%
0.99 0.95 0.96 0.99 0.99 0.94 0.98 0.97 0.02 1.97%
0.98 0.91 0.97 0.98 0.97 0.98 0.97 0.97 0.02 2.20%
0.98 0.98 0.97 0.97 0.99 0.96 0.97 0.97 0.01 1.46%
run01 run02 run03 run04 run05 run06 run07 run08 run09 run10 Mean StDev CV
Bison 42 kyr BP 30 kyr BP 21 kyr BP 6 kyr BP 0.96 0.98 0.99 0.96 1 0.99 0.94 0.91 0.94 1 0.93 0.96 0.82 0.99 0.87 0.95 0.94 1 0.98 0.91 0.95 1 0.94 0.96 1 0.99 0.99 0.96 1 0.98 0.98 0.94 0.83 1 1 0.96 0.93 0.89 0.98 0.99 0.94 0.98 0.96 0.95 0.06 0.03 0.04 0.02 6.92% 3.39% 4.17% 2.58%
Musk ox 42 kyr BP 30 kyr BP 21 kyr BP 6 kyr BP 0.99 0.98 1 1 0.97 0.98 0.99 1 0.82 0.98 0.99 0.99 0.98 0.99 0.99 0.93 0.94 0.98 0.96 0.99 0.83 0.98 0.95 0.97 0.99 0.98 1 0.99 0.96 1 0.94 0.98 0.96 0.99 0.95 0.99 0.97 0.99 0.99 0.99 0.94 0.99 0.98 0.98 0.06 0.01 0.02 0.02 6.69% 0.72% 2.38% 2.09%
1.4.3 Measuring range size To calculate modelled range size for each species/time period, the continuous suitability values mapped by openModeller were converted into deciles and the upper decile (suitability >0.9), the area of most suitable climate conditions, was used to map modelled range size (following a similar approach to 3). Assuming that only the areas with the highest suitability constituted the potential range is a conservative approach which should prevent overestimation of range sizes. Because all species were not present throughout the Holarctic for all periods, the Holarctic was divided into 3 regions: Europe, Asia and North America. Modelled ranges were cropped to exclude non-land areas and to match those regions for which fossil localities were used to generate the models (e.g., if no fossils were available from North America for a given period, as for the woolly rhinoceros, climatically suitable range from North America was excluded from the range size estimate). This ensured that aDNA and species distribution modelling methods were testing hypotheses relating to
WWW.NATURE.COM/NATURE | 17
doi:10.1038/nature10574
the geographic space actually occupied by each species at a given time as indicated by the fossil record. Cropping was conducted in R64 using the package sp65. Species ranges were then measured (in square kilometres) using IDRISI Taiga (Clark Labs, Worcester, MA, USA). Distributions for the six megafauna herbivores, reconstructed using SDMs for the periods 42, 30, 21 and 6 kyr BP, contracted in size from 30 kyr BP to the present for all species, although the severity of contraction varies substantially among taxa (Supplementary Table S1.3, Supplementary Fig. S1.3). While the absolute area of species ranges modelled using GENESIS2 and HadCM3 differ, as expected, trends of range expansion and contraction through time are consistent between AOGCMs (Supplementary Figure S1.4). Estimates of range size based on HaDCM3 also shows a significant correlation with estimated effective population size (Section S4; Supplementary Figure S4.3).
Supplementary Table S1.3. Area of potential species range (climatic suitability 0.9.), rounded to the nearest 50,000 km2. Potential range modelled with GENESIS2 Woolly rhinoceros Woolly mammoth Wild horse Reindeer Bison Musk ox 42 kyr BP 30 kyr BP 21 kyr BP 6 kyr BP 34,900,000 32,600,000 19,350,000 n/a 26,050,000 42,900,000 39,500,000 n/a 19,400,000 26,250,000 16,000,000 9,200,000 12,250,000 26,250,000 17,800,000 2,750,000 6,800,000 8,700,000 3,700,000 2,950,000 12,300,000 27,750,000 22,550,000 9,250,000
Potential range modelled with HadCM3 Woolly rhinoceros Woolly mammoth Wild horse Reindeer Bison Musk ox 42 kyr BP 30 kyr BP 21 kyr BP 6 kyr BP 19,400,000 18,600,000 12,650,000 n/a 12,150,000 25,900,000 22,600,000 n/a 15,450,000 24,000,000 10,650,000 4,100,000 6,950,000 22,350,000 10,250,000 4,900,000 11,800,000 14,900,000 1,850,000 3,350,000 6,650,000 12,900,000 9,200,000 6,150,000
WWW.NATURE.COM/NATURE | 18
doi:10.1038/nature10574
Supplementary Figure S1.4. Estimated potential range size in km2 for all 6 species at 42, 30, 21 and 6 kyr BP, modelled using the GENESIS2 (solid line) and HadCM3 (dotted line) AOGCMs. While the absolute area of species ranges differ, temporal trends in range size are consistent between AOGCMs.
WWW.NATURE.COM/NATURE | 19
doi:10.1038/nature10574
1.4.4 Sensitivity of range size estimates to fossil record uncertainty Species Distribution Models are sensitive to the initial conditions used to calibrate the models. When modelling the past distributions of extinct and extant species, initial conditions include the historical climatic data and the distribution of the fossil record. Above we address the effect of using different AOGCMs for estimating range size, but the fossil record is an incomplete and often biased representation of the past distribution of species which can also bias our results Therefore, to incorporate this uncertainty in the estimation of potential range size into the correlation between effective population size and geographic range size (Supplementary Information section S4) we performed ten additional independent model runs using GENESIS2, in which a different randomly selected 90% set of the localities was used to build the model each time. Modelled ranges from the 90% random sub-set runs were cropped and measured as described in S1.4.2. We found a positive correlation between changes in the size of available habitat and genetic diversity for the four species for which we have range estimates spanning all four time-points (although the correlation was not statistically significant for reindeer: p = 0.101; Supplementary Information section S4 ). 1.4.5 Sensitivity of range sizes to radiocarbon dating error Radiocarbon dates associated with indirectly-dated fossils are considered less reliable than those for which the specimen of interest itself is dated48 although in some cases (e.g., reindeer, 21 kyr BP, Spanish localities) indirectly-dated fossils can represent important extensions of a species geographic distribution for which directly-dated fossils are unavailable. Furthermore, for reindeer, ten indirectly-dated specimens from North America were included in the analysis, as the published DNA sequences from the samples66 were included in the genetic analysis. To examine whether the incorporation of 16 indirectly-dated fossils (ten reindeer, six bison) is likely to have influenced the detected trends in range size through time, we re-ran the models for these species excluding the indirectly-dated specimens (from Supplementary Tables S6.1 and S6.4) using GENESIS2. While the absolute area of species ranges differ, as in the random sub-set model runs, consistent temporal trends of range expansion and contraction through time are detected with these 16 specimens included and excluded (Supplementary Figure S1.5). Incorporating these new measurements into the correlation analysis did not affect the strength of the correlation between range size and effective population size, which was still significant (Supplementary Table S4.).
WWW.NATURE.COM/NATURE | 20
doi:10.1038/nature10574
Supplementary Figure S1.5. Estimated potential range size in km2 for reindeer and bison at 42, 30, 21 and 6 kyr BP, modelled using all (solid line) and only directly-dated (dotted line) fossil localities (GENESIS2). While the absolute area of species ranges differ, temporal trends in range size are consistent between fossil datasets used. 1.5 Human presence within modelled ranges To calculate the density of Palaeolithic human fossil sites (bones, artefacts and charcoal) within the modelled geographic range for each species/time period, human fossil localities from the calendar time intervals 4539, 3327 and 2418 kyr BP were overlaid on species modelled ranges (GENESIS2) for 42, 30 and 21 kyr BP, respectively. Human radiocarbon data were derived from the INQUA Palaeolithic Radiocarbon Database v. 11 for Europe67 and from 68 for Siberia. Details on these radiocarbon determinations and their selection can be found in the associated citations and in Supplementary Information section S5. The most recent period (6 kyr BP) was excluded because localities were only compiled for >9 kyr BP in some data sets. Palaeolithic human localities (Supplementary Fig. S1.3) were mapped on top of SDM results in IDRISI Taiga to identify grid cells within each species measured range in which humans were present. Grid cells were then converted into area (km2) to calculate the extent of each species range occupied by humans (Supplementary Fig. S1.6).
WWW.NATURE.COM/NATURE | 21
doi:10.1038/nature10574
50
50 50 40 40 30 30 20 20 10 10
42 kyr BP
50
50
30 kyr BP 40
50
21 kyr BP
50
40
Million km2
40
40
40
30
30
30
30
30
20
20
20
20
20
10
10
10
10
10
00
Supplementary Figure S1.6. Overlap (km2) between megafauna and humans at 42, 30 and 21 kyr BP. Palaeolithic human localities were mapped on top of species ranges (GENESIS2), and grid cells in which humans were present were converted into area to calculate the extent of each species range occupied by humans. Column height indicates estimates of megafauna range size; the black portion represents area of human/megafauna overlap. 1.6 Discussion The key goal for this portion of our study was to relate changes in the megafauna species distributions to estimates of effective population size from the Bayesian skyride models (see Supplementary Information section S4). We used SDMs to estimate the potential range size for each species at four periods for which we have palaeoclimatic data, using a subset of the fossil record. The fossil list was not intended to be an exhaustive survey of all known locations for each species; rather, we targeted the fossil record for data within the same regions for which genetic data were sampled, so that SDMs and Bayesian skyride models could be explicitly compared. Modelled distributions are therefore unlikely to capture the full known distributions for some species (e.g., bison). We have taken a conservative approach to modelling the distributions of each species; by using only contemporaneous data to build each model, rather than projecting a species modelled climatic niche from one period to the next, we are able to avoid two potential pitfalls for SDMs: species-
WWW.NATURE.COM/NATURE | 22
doi:10.1038/nature10574
climate equilibrium and climatic niche stability through time14. Species-climate equilibrium is the assumption that a species is in equilibrium with its climate, or that a species will exist in all places in which climatic conditions are favourable for its long-term survival69. However, many factors other than climate shape a species ecological niche; these include barriers to dispersal, interactions with other species, and historical contingency70, and any of these factors may result in a species distribution being out of equilibrium with its climatic niche. For example, moist conditions on the Bering land bridge have been implicated as a barrier to dispersal to steppe-tundra species such as the woolly rhinoceros, which was adapted to drier conditions71, even though suitable habitat likely existed on both sides of the strait. Likewise, niche stability through time is the assumption that a species maintains the same climatic niche, with no niche evolution (e.g., behavioural or physiological adaptation) taking place between periods of interest. While this assumption is likely to be true for some species, resulting in either extinction or tracking of suitable conditions under periods of climate change, it will not hold true for all species. Although our fossil data are limited for some species for certain periods, we maintain that building a discrete SDM for each time period, rather than projecting the distribution from those periods for which we have more data, is more relevant for comparison with the genetic data because it represents a species realised distribution for a given time, rather than a potentially incomplete measure of the species climatic niche.
WWW.NATURE.COM/NATURE | 23
doi:10.1038/nature10574
WWW.NATURE.COM/NATURE | 24
doi:10.1038/nature10574
Supplementary Figure S2.1. Polar view of the Holarctic, with Eurasia to the right, indicating the DNA sample localities of the six megafauna species. New data sets were generated for woolly rhinoceros, horse and reindeer; further information in Supplementary Tables S6.2, S6.3, S6.4.
WWW.NATURE.COM/NATURE | 25
doi:10.1038/nature10574
Supplementary Fig. 2.2. The woolly rhinoceros fossil record, including the 136 new radiocarbon dates generated for this study, plotted against latitude (top) and longitude (bottom) coordinates . Data are presented in Supplementary Tables S6.1 and S6.2. The species was present throughout Siberia right up until its disappearance from the fossil record. X-axis in calendar years BP, 1-sigma errors of the calibrated dates are included.
WWW.NATURE.COM/NATURE | 26
doi:10.1038/nature10574
Supplementary Fig. 2.3. The woolly mammoth fossil record plotted against latitude (top) and longitude coordinates (bottom). The species was present throughout Siberia right up until its disappearance from the fossil record. X-axis in calendar years BP, 1-sigma errors of the calibrated dates are included.
WWW.NATURE.COM/NATURE | 27
doi:10.1038/nature10574
2.3 Sequence generation Stringent ancient DNA protocols were followed to avoid contamination from modern DNA and to assure reliability of results. All DNA extractions and PCR set-ups were performed in a dedicated ancient DNA facility isolated from multi-copy PCR work. PCR amplification, cloning and sequencing were performed at a separate DNA facility. Ancient DNA sequences were obtained using the extraction procedures reported in 73. DNA was PCR amplified using overlapping fragments ranging in length from 80560 bp, depending on the condition of the specimen, and the species being sequenced. Primers were designed to span the entire HVR-1 of the mitochondrial control region. PCR amplifications were performed in 25 l volumes, using 1xPCR buffer, 2 mM of MgSO4, 2.0 mg/ml Bovine Serum Albumin (BSA), 0.4 M of each primer, 1 M of dNTPs, and 5U of High Fidelity Platinum Taq (Invitrogen, Carlsbad, CA). Cycling conditions were: 94C for 2 min, 5070 cycles of 94C for 30 sec, 4363C for 30 sec, and 68C for 45 sec, followed by 72C for 7 min. We included blank extraction controls and blank PCR controls in each reaction. Primer sequences and PCR-annealing temperatures are listed in Supplementary Table S2.1. PCR products were subsequently purified with either the Invitek PCRapace PCR Purification Kit (Invitek, Berlin, Germany) or the QIAquick PCR purification kit (Qiagen, Valencia, CA), according to manufacturers instructions. At least two independent PCRs were carried out for each fragment, and the products were either direct sequenced or cloned using TOPO TA cloning kit for sequencing (Invitrogen), with a minimum of six clones sequenced for each fragment. The overlapping of the PCR fragments resulted in a high degree of sequence replication. The sequences were obtained through the commercial service offered by Macrogen (Macrogen, Seoul, South Korea) and by inhouse sequencing at Department of Biology, University of Copenhagen, Denmark. DNA sequences were subsequently edited by eye and aligned using Se-Al version 2.0A11 (A. Rambaut, University of Edinburgh). To investigate and account contamination and for errors caused by damage or sequencing, 79% of all consensus sequences were replicated. The data sets resulted in 274 new mtDNA sequences, including 55 woolly rhinoceros, 115 wild horse and 104 reindeer. Sequences have been submitted to GenBank with the accession numbers JN570760-JN571033, corresponding sample numbers are listed in Supplementary Tables S6.2, S6.3 and S6.4.
WWW.NATURE.COM/NATURE | 28
doi:10.1038/nature10574
Supplementary Table S2.1. Primer sequences and annealing temperatures for woolly rhinoceros, horse and reindeer.
Woolly rhinoceros Annealing temp. (C) 55 55 55 55 55 55 55 55 55 55 55 55 55 55 55 Annealing temp. (C) 56 56 56 56 56 56 60 60 56 63 56 56 Annealing temp. (C) 51 48
Primer name 1AF/WR15422F 1aF/WR15500F2 1BF/WR15562F 1AR/WR15600R 1CF2/WR15714F2 1BR3/WR15727R3 1cR/WR15792R 1CR/WR15812R 1bzF(1AR) 1bzR(1CF) 2AF/WR16614F 2aF/WR16635F 2BF/WR16704F2 2AR3/WR16724R3 2bR/WR16813R
Primer sequence (5-3) CCCTAACTTCACCATCAACACCC CACTCCCTTCTTAAACCASAAG TATACCAGGTATGTATATCG CATGCTTATATGCATGGGGC TTGATTRATATTGCATAGTAC GACTYRAATGGGGTATGTACG CGCGGCTTGGTGATTAAGCGC GAGAGGGTTGATGATTTCCC GCCCCATGCATATAAGCATG GTACTATGCAATATYAATCAAC GCCAAACCCCAAAAACAAG GACTAGGTATATAATTACACGC CCCTTCTTTTGATACCAACATGC CTAGAGGGGTAYGAGTCTAYGTG GCTACATTAACAGGTGTATTTG
Horse
Primer pairs L1 H1 L2 H2 L3 H3 L4 H3 EQ168-187F EQCR2bii EQCR3ai EQCR3bi EQCR1F EQCR136R EQCR51F EQCR210R EQCR163F EQCR296R EQCRend184F EQCRend342R EQ4F EQ4R EQ5F EQ5R
Primer sequence (5-3) GCCATCAACTCCCAAAGCT ACATGCTTATTATTCATGG CCCACCTGACATGCAATAT TGTTGACTGGAAATGATTTG TCGTGCATACCCCATCCAA CCTGAAGTAGGAACCAGATG CCATGAATAATAAGCATGT CCTGAAGTAGGAACCAGATG CGTGCATTAAATTGTTTGCC CATGGGAGGTGATATGCGTG CGTGCATACCCCATCCAAGTC GAACCAGATGCCAGGTATAG TCCTCGCTCCGGGCCCAT TGTGAGCATGGGCTGATTAGTC CTGGCATCTGGTTCTTTCTTCAG CTTTGACGGCCATAGCTGAGT ACTGTGGTTTCATGCATTTG TTGCTGATGCGGAGTAATAA ATCTTGCCAAACCCCAAAAACAAG TCTAGGGGGATGCCTGTCTATGG CATCAACACCCAAAGCTGAA CGAYGTACATAGGCCATTCAT CATACCCACCTGACATrCAA GACTTGGATGGGGTATGCAC
Reindeer
WWW.NATURE.COM/NATURE | 29
doi:10.1038/nature10574
Rtp3_F Rtp3_Rshort 98F 231R 67F 228R 38F 219R 172F 350R 186F 373R 272F 476R 442_F 603_R
TCCACAAAATTCAAGAGCCTT TAGCCGTACAGGACCATA AAGTTCTAATTAAACTATTCCCTG ATATAATATGGCTATTGAGTGC TATAGCYCCACTATCAACACCC ATATAAYATGGCTATTGAGTGC CCAATCTCCCTAAGACTCAAGG CTGTATTAAATTHTTRAAGGTTTTTRGA AAAAACCTTYAADAATTTAATACAGT TGGGRYATRTARTTTAATGTACTATTAT CCTTCARGAATTTAATACAGTTCTGC CARGTACTTGCTTATAAGCATGGGG GGTCCTGTACGRYTATAGTAC CCCCTAGATCACGAGCT GYCAACATGCGTATCCCG GCCCTGAAGAAAGAACC
50 43 47 49 46 52 40 50
WWW.NATURE.COM/NATURE | 30
doi:10.1038/nature10574
doi:10.1038/nature10574
bases and shorter alignments. The final mtDNA CR data sets were: woolly rhinoceros (55 seq, 348bp), mammoth (82 seq, 705bp), horse (151 seq, 288bp; 136 seq when modern domestics are excluded), reindeer (162 seq, 415bp), bison (140 seq, 549bp) and musk ox (128 seq, 633bp) (Supplementary Table S3.1), with the data sets of mammoth, bison and musk ox being reduced from those previously reported73,77,78 (Supplementary Table S3.2). Supplementary Table S3.1. Summary statistics of the six megafauna DNA data sets. The Eurasian horse data set is represented both with and without the modern domestic samples. Table includes information on time bins used in the ABC and isolation-by-distance (IBD) analyses and number of sequences included in each time bin. Note that the temporal span of the time bins and sample size of each differs between the two analytical methods, as we used a minimum of three samples per time bin in ABC and a minimum of three sample localities per time bin in IBD. Because we had fewer sample localities than samples, time bins from the ABC analysis were pooled in the IBD analysis in some instances. S represents the number of polymorphic sites, represents nucleotide diversity. The IBD p-values were determined from a randomization test with the next time-bin.
WOOLLY RHINOCEROS Region West of 90E Time bin (kyr BP) 0-19 19-26 >34 0-19 19-26 26-34 >34 pan-Eurasian Seq # 4 5 4 9 10 9 14 S 24 16 12 29 42 29 29 (per locus) 0.037 0.020 0.019 0.030 0.045 0.024 0.017 Summary Statistics Tajima D -0.072 -0.840 0.187 -0.184 0.210 -1.045 -1.548 Haplotypic diversity 1 0.9 1 1 0.867 0.972 0.824 Fst 0.265 0.276 0.500 -
IBD (within region) Correlation coefficient -0.08 0.07 0.19 0.13 p 0.160 0.400 0.323
East of 102E
0-19 13 41 0.038 -0.054 1 19-26 15 42 0.042 0.564 0.924 26-34 9 19 0.024 -1.045 0.972 >34 18 39 0.024 -1.548 0.895 Sequence Length: 348bp Mutation Rate Prior: Normal 0.00103355,0.000325 (substitution per generation per locus) Generation Time: 7 years Summary Statistics Seq # 13 9 12 14 3 S 19 4 11 30 10 (per locus) 0.007 0.003 0.004 0.011 0.009 Tajima D -1.067 1.766 -1.080 -0.742 ND Haplotypic diversity 1 0.806 0.939 0.956 1
WOOLLY MAMMOTH Continent Europe Time bin (kyr BP) 0-19 19-26 26-34 >34 0-19
IBD (within continent) Fst 0.273 0.255 0.531 0.260 Correlation coefficient 0.20 -0.17 0.14 0.42 0.41 p 0.007 0.048 0.321 -
America
WWW.NATURE.COM/NATURE | 32
doi:10.1038/nature10574
19-26 5 12 0.008 0.050 0.9 26-34 8 9 0.004 -0.564 0.893 >34 18 23 0.008 -0.810 0.967 Sequence Length: 705bp Mutation Rate Prior: Normal 0.001053,0.0003666 (substitution per generation per locus) Generation Time: 20 years Summary Statistics Time bin (kyr BP) 0-11 11-19 19-26 26-34 >34 Seq # 31 8 4 8 5 S 37 17 12 15 13 (per locus) 0.02668 0.01984 0.02199 0.01637 0.02083 Tajima D -0.4135 -0.6626 -0.3269 -0.9471 -0.2793 Haplotypic diversity 0.989 1 1 0.964 1
0.64 0.08
0.032 0.003
IBD (within continent) Fst 0.264 0.223 0.335 0.321 0.131 0.245 0.335 0.321 0.131 0.141021 0.2857 -0.4023 0.04187 -0.0647 0.1739 0.0223 0.074 0.024 0.371 0.029 0.091 Correlation coefficient p
0-19 10 21 0.021 -0.848 1 19-26 4 12 0.022 -0.327 1 26-34 8 15 0.016 -0.947 0.964 >34 5 13 0.021 -0.279 1 America 11-19 23 33 0.027 -0.474 0.972 19-26 52 38 0.024 -0.587 0.949 26-34 18 20 0.025 0.997 0.922 >34 8 19 0.029 0.780 0.929 Sequence Length: 288bp Mutation Rate Prior: Normal 0.0003988,0.00017 (substitution per generation per locus) Generation Time: 5 years Summary Statistics
REINDEER
IBD (within continent) Correlation coefficient 0.21 0.04 -0.12 -0.01 0.05 0.1290 -0.3215 0.6973
Nucleotide diversity Time bin Seq Haplotypic (per locus) (kyr BP) # diversity Continent S Tajima D Europe 0-11 31 37 0.017 -0.884 0.994 19-26 14 32 0.020 -0.831 0.989 19-26 19 39 0.021 -0.947 1 26-34 12 27 0.017 -0.994 1 >34 10 23 0.014 -1.255 1 America 0-11 58 54 0.016 -1.415 0.976 Nov-26 8 15 0.012 -0.631 0.964 26-34 6 19 0.018 -0.590 1 >34 4 15 0.020 0.395 1 Sequence Length: 415bp Mutation Rate Prior: Normal 0.00060125,0.00016 (substitution per generation per locus) Generation Time: 4 years Summary Statistics
IBD (within continent) Fst$ 0.528 0.243 0.235 0.139 0.210 Correlation coefficient 0.03 0.10 0.21 0.48 0.14 p 0.110 0.259 0.168 0.033
Time bin Seq (per Haplotypic (kyr BP) # locus) diversity S Tajima D >17 7 29 0.0205 -0.2882 1.000 0-11 55 48 0.0159 -0.5781 0.877 11-19 27 51 0.0259 0.2812 0.986 19-26 16 50 0.0228 -0.7125 0.992 26-34 13 42 0.0223 -0.4288 0.987 >34 22 58 0.0223 -0.9194 0.996 Sequence Length: 549bp Mutation Rate Prior: Normal 0.00102443,0.00025 (substitution per generation per locus) Generation Time: 3 years $ between each American time-bin and the unique Eurasian time-bin
WWW.NATURE.COM/NATURE | 33
doi:10.1038/nature10574
Time bin Seq (per Haplotypic (kyr BP) # locus) diversity S Tajima D 0-11 5 36 0.025 -0.643 1 11-19 17 60 0.028 -0.045 0.985 19-26 31 92 0.022 -1.477 1 26-34 22 57 0.015 -1.597 1 34-50 3 22 0.023 ND 1 America 0-26 50 87 0.010 -2.434 0.913 Sequence Length: 633bp Mutation Rate Prior: Normal 0.00097355,0.0002 (substitution per generation per locus) Generation Time: 2 years $ between each Eurasian time-bin and the unique American time-bin
Supplementary Table S3.2. Differences in sample size between original published (unfiltered) and filtered data sets of woolly mammoth, bison and musk ox. Unfiltered Sample Seq size length Woolly mammoth Bison 160 220 705 685 Filtered Sample size 82 140 Seq length 705 549 Reduction of data set Sample Sample size size % 78 80 34 49 36 21 Seq length 0 136 49 Seq length %* 0 20 7
Musk ox 162 682 128 633 * Relative to the sequence length before filtering 3.2 Genetic analysis Summary statistics
We grouped the sequence data sets of each species into geographic and temporal bins to calculate summary statistics for the serial coalescent simulations. Sequences were separated into Eurasia or North America and assigned to sequential time bins within each continent (Supplementary Table S3.1). Because woolly rhinoceros was found exclusively in Eurasia, we (i) analysed a single Eurasian time-series; and (ii) grouped samples west of 90E and east of 102E in two separate geographic units, as no wolly rhinoceros have ever been recovered in between. Summary statistics were computed using Arlequin 3.579 (Tajimas D and Fst) and dnaSP v580 (number of segregating sites, nucleotide diversity per site and haplotypic diversity). Time bins were selected to test for different population models of demographic expansion or decline at various time points during the past 50,000 years and to test for possible shifts in geographic structure using isolation-by-distance. Due to variations among species in the temporal coverage of samples, the
WWW.NATURE.COM/NATURE | 34
doi:10.1038/nature10574
number of sequences in each continent and within each time bin differed among data sets. Time bins were designed to include a minimum number of three sequences each (average = 15.8, range 358). To avoid over-representation of modern domestic horse sequences in subsequent analyses and to minimise computation time for serial-coalescent simulations and Bayesian skyrides, we generated random subsets of the data available on GenBank. In horse, we generated ten random data sets of 13 sequences from the subset of 140 sequences selected from Genbank. Two of the new sequences generated in this study were recovered from Neolithic Eurasian horses (specimens JW191 and JW25; Supplementary Table S6.3) and were added to the domestic data set, and we subsequently estimated summary statistics independently for each of the ten data sets. The final summary statistic vector for Eurasian Neolithic horses (time bin 110 kyr BP) was determined as the average of the summary statistics recovered from the ten independent data sets of 15 sequences. The procedure used for the incorporation of modern reindeer is discussed in section S3.1 above. Isolation-by-distance (IBD) We tested for temporal changes in the level of isolation-by-distance (IBD) within each species by calculating the correlation between pairwise genetic and geographic distances within and between consecutive time bins (Supplementary Table S3.3). Each time bin included a minimum of three geographically distinct sample localities; hence, the number of sequences in each time bin differed among species due to different sampling regimes (Supplementary Fig. S3.1). Of note, the temporal span of the time bins differed from those used in the ABC analysis because some of the ABC time bins were pooled for the IBD analysis due to small locality number; e.g., we pooled samples from >34 kyr BP (n = 4) and 3426 kyr BP (n = 6) in North American reindeer, as they represented two localities each. Geographic distances were estimated from latitude and longitude using the haversine formula and a spherical-earth approximation, ignoring hills. We corrected the pairwise-distance estimates among sequences for differences in calendar age of the samples following 81 and used the mutation rate average used in the serial-coalescent simulations (Supplementary Table S3.1). Hence, if two specimens were separated in time and space, only geographic distance and not sample age contributed to their genetic distance, assuming a constant clock and constant population size. Correlation coefficients were estimated in the R statistical package64 for each time bin and the significance of change in correlation between two successive time bins was tested through a randomisation approach using 10,000 pseudo-replicates. Briefly, for two successive time bins with
WWW.NATURE.COM/NATURE | 35
doi:10.1038/nature10574
NA and NB sequences, pseudo-replicates were generated randomly by sampling (without replacement) NA and NB sequences from the merged pool of NA and NB sequences, and the difference in correlation coefficients was recorded. The re-sampling procedure provided an empirical estimate of the difference between correlation coefficients, assuming no change in population structure; the observed difference in correlation coefficients was compared against the re-sampling distribution to test for significance at the 5%-level (one-sided test). Pseudo-replicates were generated using a home-made Perl script and distributions were analysed using R. Supplementary Table S3.3. Changes in isolation-by-distance through time for the continental populations in Eurasia and North America and for the gobal data (EurA + NA). The correlation coefficients observed between geographic distances and pairwise genetic distances are reported for each megafauna species and for each time bin where a minimum number of five sequences were available (top line of each population panel). Genetic distances were estimated after correcting for time differences between sequence pairs. The significance of the changes in correlation coefficients between two successive time periods was tested through a randomization procedure and corresponding p-values are indicated. Significant tests (5%-level) are marked with an asterisk. Time bins are in thousands of years before present.
Species Woolly rhinoceros Continent Eurasia >34 0.46 34-26 0.38 26-19 0 19-11 11-0 0.42 0.038*
0.031* Woolly mammoth Eurasia 0.128 0.82 0.14 -0.09 0.37 0.231
0.172 North America 0.203 0.02 0.55 0.017* 0.44 0.65 0.47 0.452 0.35 0.49 0.238
EurA + NA
0.111 0.157 Horse Eurasia 0 0.345 0.23 0.22 0.17 0.357 North America 0.07 0.02 0.197 0.05
EurA + NA
0.13 0.466
WWW.NATURE.COM/NATURE | 36
doi:10.1038/nature10574
Reindeer
Eurasia
0.02
EurA + NA
0.20
0.472 Musk ox Eurasia 0.146 0.02 0* EurA + NA 0.02 0.06 0.31 0.07 0.016* 0.129 0.31 0* 0.076 0.123 0.128 0.40 0.04 0.02 0.77 -0.19 0.20 0.014*
Bison
North America
0.12
0.387 0.17
0.02
EurA + NA
0.30 0*
0.77
WWW.NATURE.COM/NATURE | 37
doi:10.1038/nature10574
Supplementary Figure S3.1. Isolation-by-distance analysis of continental populations in Eurasia (EurA) and North America (NA). Correlation between genetic (calculated as nucleotide diversity corrected for temporal age) and geographic distance among samples within each time bin is indicated by the colour intensity, with values shown in Supplementary Table S3.3. The darker the colour, the stronger the correlation between geographic and genetic distance. Sample size of each time bin is indicated and an asterisk marks significant change in isolation-by-distance between consecutive bins. Crosses mark species extinction times; mainland (older) and island (younger) mammoth extinctions are included.
WWW.NATURE.COM/NATURE | 38
doi:10.1038/nature10574
Approximate Bayesian Computation (ABC) and model selection Serial-coalescent simulations (1,000,000 iterations per model) were performed using Bayesian Serial SimCoal (http://www.stanford.edu/group/hadlylab/ssc/) on a series of nine to 16 population models per species (Supplementary Fig. S3.2, Supplementary Table S3.4). Simulations were run on three data sets: (i) Eurasia; (ii) North America; and (iii) global. For woolly rhinoceros, simulations were run only for Eurasia, as the species was never found in North America. We tested a total of 217 models, resulting in 217 million serial-coalescent simulations. The K2P+ mutation model was used in all simulations, using the average of the posterior distributions for kappa (transition/transversion ratio) and alpha (gamma shape) estimated by BEAST. Similarly, we used the normal posterior distributions recovered from BEAST as a prior for the mutation rates (Supplementary Table S3.1). Using the age of first reproduction as a rough proxy for generation time, we assumed that generation times were: woolly rhinoceros (seven years, based on extant rhinoceros species), mammoth (20 years, based on extant elephants78), wild horse (five years, based on Przewalskis horse82), reindeer (four years83), bison (three years84) and musk ox (two years73). In all models, the effective population size at first generation was randomly sampled at each iteration from a uniform prior ranging from 1,000 to 100,000 individuals. The first model consisted of a panmictic deme with constant effective population size (Supplementary Fig. S3.2). In a second series of models, we simulated an instantaneous demographic expansion (uninformative prior, up to 10-fold) or decline (uninformative prior, down to 0.1-fold) occurring at a fixed and unique time in the past (34, 26, 19, 12.9, 11 kyr BP). These time points were chosen for the following reasons. They represented midpoints between the periods from which we have palaeoclimatic data and potential range size estimates of each species (42, 30, 21, 6 kyr BP). Also, some of them represented periods of putative climatic change, such as the beginning of the Last Glacial Maximum (LGM; 26 kyr BP), the end of the LGM (19 kyr BP), the onset of the Younger Dryas (12.9 kyr BP), and the beginning of the Holocene (11 kyr BP). Two additional models, an instantaneous expansion at 26 kyr BP followed by a decline at 19 kyr BP, and the reverse scenario (population decline at 26 kyr BP followed by an expansion at 19 kyr BP), were considered; these models were introduced to mimic demographic events possibly driven by climatic changes around the time of the LGM. With a final set of models, we aimed a final set of models at testing population subdivision between continents at 19 kyr BP (the end of the LGM) or at 11 kyr BP (the inundation of the Bering land bridge, which put an end to gene flow between continents). For woolly rhinoceros, which did not colonise the Americas, we assumed subpopulations to be west of 90E and east of 102E. Alternatively, we assumed that population subdivision was of an older date (uniform prior, 6075 kyr BP), followed by different episodes of isolation and migration between continents. No
WWW.NATURE.COM/NATURE | 39
doi:10.1038/nature10574
symmetry in migration rates was assumed and migration frequencies were randomly sampled from a uniform distribution of 0-0.01 per generation. Approximate Bayesian Computation (ABC) analyses were performed for each model using nucleotide diversity, Tajimas D, haplotypic diversity, and Fst (except in single-continent analyses) as a vector of summary statistics. We used a tolerance region of 0.1% of all simulations and the R makepd4() function. Note that, to match the Bayesian Serial SimCoal output, the observed haplotypic diversity values were converted and multiplied by a factor (n-1)/n, with n being the number of sequences considered in a given time bin. Finally, the posterior probability of all NM models was estimated using categorical regression and the R calmod() function following Beaumont85. This procedure takes advantage of the weighted regression framework and treats a model indicator as a categorical variable that can take values ranging from 1 to NM. R functions for ABC and categorical regression are available online at http://www.rubic.rdg.ac.uk/~mab/stuff/. Support values of the models tested in each continental population and the global species data sets are shown in Supplementary Table S3.4. To test if the number of samples within each time bin influenced the associated estimate of nucleotide diversity, we used Spearmans ranked correlation coefficient and the R statistical package64. Similarly, we tested the correlation between the temporal distribution of samples and nucleotide diversity. This was done by calculating the temporal distance from each sample within a time bin to the median sample age of that time bin, and correcting for sample size. Plots of nucleotide diversity against sample size and temporal span are shown in Supplementary Figs S3.3 and S3.4. We did not find any correlation.
WWW.NATURE.COM/NATURE | 40
doi:10.1038/nature10574
Supplementary Figure S3.2. Simulated demographic models tested against the observed data using the ABC model-selection approach. A maximum of 16 models were tested per species. Migrations between continents is indicated in light grey. Results in Supplementary Table S3.4.
WWW.NATURE.COM/NATURE | 41
doi:10.1038/nature10574
1 2 3 4 5
Supplementary Table S3.4. Simulated demographic models tested against the observed data using the ABC model-selection approach. Models were run for Eurasia, North America and the global data set. A maximum of 16 different models were simulated for each species; two-population models were only analysed with the global data and encompass Eurasia and North America as separate populations, with or without migration (Supplementary Fig. S3.2). Support values for the different models are shown, and sum to 1 across all models within each data set. Values > 0.2 are in bold. Horses were analysed both with and without domestics.
Eurasia Horse + dom Woolly rhinoceros Woolly mammoth Reindeer Musk ox North America Woolly mammoth Reindeer Global Horse + dom Woolly rhinoceros* Woolly mammoth Reindeer Musk ox 0.00 n/a 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.01
Horse
Horse
Horse
Bison
Decline
0.09 0.26 0.31 0.24 n/a 0.00 0.00 0.01 n/a n/a 0.02
0.07 0.14 0.26 0.21 0.18 0.02 0.01 0.01 0.01 0.02 0.04
0.08 0.34 0.22 0.14 0.11 0.00 0.00 0.02 0.02 0.02 0.04
0.06 0.35 0.25 0.17 0.12 0.00 0.00 0.00 0.00 0.01 0.03
0.06 0.44 0.11 0.08 0.19 0.00 0.00 0.01 0.01 0.01 0.07
0.03 0.73 0.08 0.02 0.04 0.00 0.00 0.00 0.01 0.01 0.07
0.10 0.21 0.24 0.24 n/a 0.03 0.02 0.03 0.02 n/a 0.05
0.04 0.27 0.48 0.11 n/a 0.00 0.00 0.01 n/a n/a 0.09 0.00
0.03 0.12 0.20 0.29 0.30 0.00 0.00 0.00 0.00 0.00 0.02 0.03
0.02 0.00 0.00 0.00 n/a 0.00 0.01 0.05 0.37 0.50 0.05 0.00
0.04 0.26 0.35 0.08 n/a 0.00 0.00 0.00 n/a n/a 0.01 0.02 0.21 0.02 0.00 n/a n/a
0.00 0.01 0.01 0.01 0.01 0.00 0.00 0.00 0.00 0.00 0.01 0.00 n/a n/a n/a n/a 0.00
0.04 0.29 0.36 0.09 0.09 0.00 0.00 0.01 0.02 0.01 0.09 0.00 n/a n/a n/a n/a 0.00
0.05 0.29 0.34 0.10 0.07 0.00 0.00 0.00 0.01 0.00 0.09 0.00 n/a n/a n/a n/a 0.01
0.05 0.44 0.09 0.07 0.17 0.00 0.00 0.01 0.01 0.03 0.06 0.02 n/a n/a n/a n/a 0.00
0.02 0.00 0.00 0.00 0.00 0.00 0.00 0.04 0.58 0.22 0.08
0.07 0.04 0.02 0.01 0.02 0.00 0.06 Two-population model, Eurasia only. No migration
0.00 0.00 n/a n/a n/a n/a n/a WWW.NATURE.COM/NATURE | 42 n/a n/a n/a 0.00 0.26
Bison
doi:10.1038/nature10574
Woolly Rhino
0.05 0.04 0.03 0.02 0.01 0.00 0
Woolly Mammoth
0.47 0.05 0.04 0.03 0.02 0.01 0.00 0 5 Pi (per site) 0.82
Pi (per site)
10
15
20
25
30
10
15
20
25
Nb. samples
Nb. samples
Pi (per site)
10
20
30
40
50
60
10
20
30
40
50
Nb. samples
Nb. samples
Reindeer
0.05 0.04 0.03 0.02 0.01 0.00 0 0.97
Bison
0.05 0.04 0.03 0.02 0.01 0.00 0 Pi (per site) 0.96
Pi (per site)
10
20
30
40
50
60
10
20
30
40
50
Nb. samples
Nb. samples
Musk ox
0.05 0.04 0.03 0.02 0.01 0.00 0 Pi (per site) 0.14
10
20
30
40
50
Nb. samples
Supplementary Fig. S3.3. Plots of sample size against nucleotide diversity () for each time bin; data from Supplementary Table S3.1. Dots represent Eurasia, triangles represent North America. In woolly rhinoceros, which is not present in North America, the two populations are from west of 90E (dots) and east of 102E (triangles); crosses represent the pooled Eurasian data. The p-value of Spearmans correlation test is indicated in the top right-hand corner of each plot, and is based on all data points within each species.
60
WWW.NATURE.COM/NATURE | 43
60
60
30
doi:10.1038/nature10574
Woolly Rhino
0.05 0.04 0.03 0.02 0.01 0.00 0
Woolly Mammoth
0.07 0.05 0.04 0.03 0.02 0.01 0.00 0 1000 Pi (per site) 0.59
Pi (per site)
1000
2000
3000
4000
5000
2000
3000
4000
Pi (per site)
1000
2000
3000
4000
5000
1000
2000
3000
4000
Reindeer
0.05 0.04 0.03 0.02 0.01 0.00 0 0.44
Bison
0.05 0.04 0.03 0.02 0.01 0.00 10000 20000 0 Pi (per site) 0.32
Pi (per site)
30000
Musk ox
0.05 0.04 0.03 0.02 0.01 0.00 0 Pi (per site) 0.14
1000
2000
3000
4000
Supplementary Fig. S3.4. Plots of temporal span of samples within each time bin (average pairwise distance is in calendar years) against nucleotide diversity (); data from Supplementary Table S3.1. The temporal span was calculated by summing the distance between each sample within a time bin and the temporal median, and dividing by the number of samples. Dots represent Eurasia, triangles represent North America. In woolly rhinoceros, which is not present in North America, the two populations are from west of 90E (dots) and east of 102E (triangles); crosses represent the pooled Eurasian data. The p-value of Spearmans correlation test is indicated in the top right-hand corner of each plot, and is based on all data points within each species.
5000
WWW.NATURE.COM/NATURE | 44
40000
1000
2000
3000
4000
5000
5000
5000
doi:10.1038/nature10574
Bayesian skyride plots To explore the evolutionary history of the six species, we estimated the genealogical relationships of the sequences using the Bayesian phylogenetic inference package BEAST v1.5.486, which allows the simultaneous estimation of demographic and evolutionary parameters. For each data set we performed analyses both with and without the post-mortem damage (PMD) model, which accounts for additional substitutions at the terminal branches that may be due to DNA damage87. For each model, we assume the HKY+ model of nucleotide substitution and the strict molecular clock, with the evolutionary rate calibrated using the age (calibrated radiocarbon date or sampling date) of each sequence in the data set. For all analyses, two MCMC chains were run for 30200 million iterations each, with samples drawn from the posterior every 5,000 iterations. Convergence to stationarity and mixing were evaluated using Tracer88. The first 10% of runs were discarded as burn-in and the remainder of posterior samples from the two runs were combined. As a coalescent prior we assumed the skyride demographic model89, which accommodates uncertainty in the demographic and phylogeographic history of each species. Although in many cases a constant population size model may be the simplest model to use, the skyride model is the most flexible coalescent model currently available in the BEAST package. In addition, the skyride model allows the estimation of theta, which approximates the effective population size, confounded by population structure, throughout the history of the sampled genealogies. Theta is proportional to the effective population size unless there is substantial structure or if sampling is biased90,91,92. For some taxa, other demographic models, such as the constant population size and exponential growth/decline model, were also evaluated. In all cases, comparison between models was performed using Bayes factors93. 3.3 Results and discussion Data sets Although some analyses were more flexible than others in terms of missing data, it was necessary to prune all the data sets in terms both of number of sequences and the number of (nucleotide) sites as described in section 3.1. First, to compare the genetic data and the results of the species distribution models (Supplementary Information section S1), it was necessary to calibrate the radiocarbon dates generated from each fossil to reflect calendar years, and samples > c. 43,000 radiocarbon years before present (BP) were discarded as they fell beyond the IntCal09 calibration curve47. This resulted in a significant decrease in the number of specimens in some data sets. In musk ox, for example, almost all northeast Siberian sequences (which include a distinct genetic clade73), were excluded from our analyses as they could not be calibrated. Second, the Serial
WWW.NATURE.COM/NATURE | 45
doi:10.1038/nature10574
SimCoal software does not accept missing sites, and we therefore filtered the data for these prior to analysis (see section S3.1). To ensure that the results from the three different analytical approaches (IBD, ABC, and BEAST) were directly comparable, we used the filtered data sets in all analyses. For differences in number of sequences and sites between the published and filtered data sets used in our analyses, see Supplementary Table S3.2. Isolation-by-distance We find a significant increase in IBD in Eurasian woolly rhinoceros and musk ox after 19 kyr BP, and in North American bison after 11 kyr BP (Supplementary Fig. S3.1, Supplementary Table S3.3). Although not significant, Eurasian mammoth also show an increase in the correlation between genetic and geographic distance after 19 kyr BP. Eurasian and North American reindeer and Eurasian wild horse show no changes in IBD over time, and IBD decreases in North American horse prior to the LGM. We note that although we interpret high levels of IBD as increased structuring within populations, temporal changes in genetic diversity could also be caused by local extinctions and replacements by genetically divergent populations. Approximate Bayesian Computation and model selection In eight of the nine continental populations, we find maximal support for models of population expansion using the ABC model-selection approach (Supplementary Table S3.4). The intensities of the increases range from 2.5 to 10-fold across populations; distribution plots of intensity estimates for the best-fit model are shown in Supplementary Figure S3.5. Bison is the only species with a well-supported signal of decline; in all other populations, models of decline are supported by posterior probabilities of 0.03, where the support across all given models sums to 1 (Supplementary Table S3.4). The posterior probability values of the expansion models are 410x the values of the models of decline. In woolly rhinoceros and North American mammoth, models of expansion at 35/34, 26 and 19 kyr BP yield similar levels of support, and we are therefore unable to conclude the exact timing of the event. Similarly, we find high levels of support for population expansion at 19 and 11 kyr BP in North American reindeer. In Eurasian reindeer and musk ox, we find high support (0.44 and 0.73, respectively) for an expansion at 34 kyr BP. Interestingly, in the global populations of woolly mammoth and musk ox, we find highest support for two-population models with migration between Eurasia and North America, where the onset of migrations coincides with the timing of expansion in the continental populations (Supplementary Table S3.4). Furthermore, the end of migration at 11 kyr BP in musk ox coincides with the inundation of the
WWW.NATURE.COM/NATURE | 46
doi:10.1038/nature10574
Bering land bridge, which prevented migration between continents. We could not test this timing (11 kyr BP) in mammoth, due to a lack of younger samples from North America. The estimates of effective population size at first generation were similar across populations within species (Fig. 3 in main text). Distribution plots of Ne estimates of the best-fit model are included in Supplementary Figure S3.5. We ran the ancient horse data set both with and without the modern domestic sequences, and results remained consistent (Supplementary Table S3.4).
Intensity
Ne
Supplementary Figure S3.5a. Density plots of the parameter estimates in the ABC modelselection approach in the Eurasian populations. Parameter estimates of intensity and effective population size (Ne) are shown for the model with highest support. (a) Woolly rhinoceros, population increase at 26 kyr BP, (b) Woolly mammoth, population increase at 26 kyr BP, (c) Horse, population increase at 34 kyr BP, (d) Reindeer, population increase at 34 kyr BP, (e) Musk ox, population increase at 34 kyr BP. Support for all demographic models tested are shown in Supplementary Table S3.4.
WWW.NATURE.COM/NATURE | 47
doi:10.1038/nature10574
Intensity
Ne
WWW.NATURE.COM/NATURE | 48
doi:10.1038/nature10574
Intensity
Ne
Supplementary Figure S3.5b. Density plots of the parameter estimates in the ABC modelselection approach in the North American populations. Parameter estimates of intensity and effective population size (Ne) are shown for the model with highest support. (a) Woolly mammoth, population increase at 26 kyr BP, (b) Horse, population increase at 26 kyr BP, (c) Reindeer, population increase at 11 kyr BP, d) Bison, population decline at 11 kyr BP. Support for all demographic models tested are shown in Supplementary Table S3.4.
WWW.NATURE.COM/NATURE | 49
doi:10.1038/nature10574
Bayesian skyride plots All skyrides can be seen in Figure 2 of the main text and in Supplementary Figures S3.6 and S3.7. For bison and musk ox (Fig. 2 in main text), the skyride estimates of changes in diversity through time differ slightly from those reported previously73,78 ,94. This is due to our data-filtering approach, where sequences and missing sites were pruned prior to the analysis. The additional signals of expansion reported in the published studies of bison and musk ox both occurred prior to 43,000 radiocarbon years BP. Because we exclude the older samples, we do not recover the expansion signal. However, as the time period falls outside the scope of our study, which is focused on changes that occur within the most recent 50,000 years, our results are not influenced by this omission. For horses, excluding the domestic sequences resulted in no change to the estimated demographic trajectory (Supplementary Fig. S3.7). For some data sets, visual inspection of the skyride plots suggests that a constant population size demographic model would be a reasonable fit to the data. Bayes Factor tests (data not shown) indicate that constant population size models are a better fit to both the mammoth and woolly rhinoceros. However, because these data sets comprise samples from both a broad temporal and geographic extent, it is likely that they violate, at least during some of their evolutionary history, the assumption of panmixia made by the coalescent models used in BEAST. The skyride plot provides the most flexibility of the coalescent models currently implemented in BEAST, and therefore is the most likely to accommodate the temporal changes in structure that likely characterised each of these species. The results of these analyses (wide confidence intervals and an inability to reject simpler, constant-size model) indicate that several of the data sets simply contain too little evolutionary information to be characterised using this approach.
WWW.NATURE.COM/NATURE | 50
doi:10.1038/nature10574
50
a Woolly rhinoceros 7 b
40
30
20
10
Woolly mammoth 7 6 5 4
Log Ne*
6 5 4
7.5
Log km2
7.0
6.5
50
40
30
kya BP
20
10
6.5
7.0
7.5
50
40
30
kya BP
20
10
Supplementary Figure S3.6. Temporal changes in global effective population size and generation time (Ne * ) and potential range size in (a) woolly rhinoceros and (b) woolly mammoth. Each species panel includes the demographic trajectory of the past 50,000 years inferred from BEAST and the area of potential range size (km2) at 42, 30 and 21 estimated using species distribution models; range sizes could not be calculated at 6 kyr BP due to insufficient fossil localities (Fig. 1 in main text; Supplementary Information section S1). We assume a generation time of seven years in woolly rhinoceros and 20 years in woolly mammoth. Radiocarbon-dated samples used in each approach are shown as vertical lines below each panel; each line represents one dated individual.
50
40
30
a 20
Log Ne*
7 10 6 5 4
b
Log Ne*
7 6 5 4
50
40
30 20 kyr BP
10
Supplementary Figure S3.7. Temporal changes in global effective population size and generation time (Ne * ) in wild horse (a) including modern domestics (b) excluding modern domestics.
years bef re present o
Radiocarbon-dated samples used in the analysis are shown as vertical lines below the skyride in (a). We assume a generation time of five years.
Population siz e
WWW.NATURE.COM/NATURE | 51
doi:10.1038/nature10574
WWW.NATURE.COM/NATURE | 52
doi:10.1038/nature10574
model (all localities), we measured the size of the projected range. Within each of the 5,000 individual linear regressions, the climatic range size at each time point was sampled randomly from the set of potential range sizes.
Supplementary Figure S4.1. Fifty randomly selected skylines for musk ox extracted from BEAST. Though there is an apparent overall trend, there is also considerable variation in the trajectory of each line. The WinBUGS model used a common prior for the regression slopes for all four species, and independent priors for the intercepts. The biological rationale for this model is that although species exhibit positive temporal relationships between range size and abundance, the exact parameters of this relationship have been found to vary between different species95. Thus, points for different species may not simply be pooled. Using a hierarchical model for the regression slope allows a combined analysis of the regression slopes without enforcing a common intercept. The WinBugs model used was:
WWW.NATURE.COM/NATURE | 53
doi:10.1038/nature10574
Model { for (i in 1:N) { pops[i] ~ dnorm(mu[i], tau) mu[i] <- alpha[species[i]] * bioms[i] + beta[species[i]] } for (j in 1:Nspec) { alpha[j] ~ dnorm(mu.alpha, tau.alpha) beta[j] ~ dnorm(0, 1.0E-6) } tau ~ dgamma(0.001, 0.001) sigma <- 1/sqrt(tau) sigma.alpha ~ dunif(0,100) tau.alpha <- 1/(sigma.alpha*sigma.alpha) mu.alpha ~ dnorm(0, 1.0E-6) } In WinBUGS, the Gibbs sampler was run in three individual chains for 50,000 iterations. The first 25,000 iterations were discarded as burn-in, and the remaining 25,000 were thinned by 1:25 to 1,000. Of these three chains of 1,000 values, 200 iterations were picked randomly and saved, resulting in 200 values times 5,000 skylines = 1,000,000 simulated values from the posterior distributions. These values represent a full sample of the posterior distribution for the relationship between the two variables, and incorporate error estimates from both the skyride analysis and the SDM-projected climatic range sizes. 4.2 Results The results of a Bayesian analysis are probability distributions that reflect the degree of belief in the estimated parameters. To make the results comparable to the results of standard frequentist regression tests, we report the proportion of the posterior density of the slopes that is lower than zero. This is comparable to the p-value from a standard one-tailed t-test. The results are summarized in Supplementary Table S4.1 and Supplementary Figure S4.2.
WWW.NATURE.COM/NATURE | 54
doi:10.1038/nature10574
Supplementary Table S4.1. Summary of the posterior parameters of the Bayesian hierarchical model. SE is the standard deviation of the posterior distribution. Here p represents the proportion of values that are below 0, and is comparable to the p-value from a standard one-tailed test. Range data are modelled using the GENESIS2 climatic simulations. Slope 1.335 0.561 1.974 0.889 SE 0.889 0.463 0.983 0.520 p Intercept 0.049 -8.829 0.101 5.452 0.011 -18.013 0.042 -2.957 SE 14.675 7.451 15.124 8.505
Supplementary Figure S4.2. Relationship between Ne* and range size (modelled with GENESIS2), summarized as posterior probability distributions of regression slopes from a Bayesian hierarchical regression. The proportion of probability densities below zero (analogous to a onetailed p-value) can be seen in Supplementary Table S4.1. The distributions of the six megafauna herbivores, reconstructed using SDMs for the periods 42, 30, 21 and 6 kyr BP, decreased in size from 30 kyr BP to the present for all species, although the severity of decline varies substantially among taxa (Supplementary Fig. S1.3). These trends are mirrored by genetic diversity in four of the six species, where there is a strong positive correlation
WWW.NATURE.COM/NATURE | 55
doi:10.1038/nature10574
for horse, bison and musk ox, and a positive relationship for reindeer (Supplementary Fig. S4.3). Of note, due to the large confidence intervals, the slope for reindeer would not be considered significant when the posterior distribution is compared to a one-tailed significance value of 0.05. A positive correlation between range size and genetic diversity is consistent with ecological theory: the relationship between geographic distribution (range size) and species abundance is one of the best-documented patterns in macroecology96. These findings support the validity of climatic range as a proxy for range size, and of genetic diversity as a proxy for effective population size. The observation that effective population size is dependent on climate also strongly supports a role for climate in driving the population dynamics of megafauna species.
4.5
6.5
6.5
7.5
6.5
7.0
7.5
Supplementary Figure S4.3. Correlation between Ne* and range size for the four species for which we had sufficient data for a correlation analysis, modelled using GENESIS2. Genetic log geographic range diversity is shown as the mean and standard deviation of the 1,000 values used in the regression analysis. The mean values of slope and intercept (Supplementary Table S4.1) were used to draw the trendlines. Error bars represent measures of uncertainty: range sizes calculated using ten random 90% subsets of fossil localities per period (horizontal bars) and posterior probabilities of Ne* (vertical bars) (S1 and S4).
WWW.NATURE.COM/NATURE | 56
doi:10.1038/nature10574
To assess whether the analysis was robust with the inclusion of the data available for woolly rhinoceros and woolly mammoth, for which we did not have range size estimates at 6 kyr BP, we repeated the analysis with the three potential range size data points (42, 30, 21 kyr BP) for these two species. As expected, due to wide probability intervals of the skyrides for woolly rhinoceros and woolly mammoth (Supplementary Fig. S3.6), no relationship was detected for these two species, and the results for the remaining four species did not change. 4.3 Assessing effect of AOGCM choice To specifically assess the effect of AOGCM choice on species ranges (horse, reindeer, bison and musk ox) and the subsequent relationship between range size and effective population size, we also measured potential ranges modelled with an alternative Atmospheric-Ocean coupled General Circulation Model, HadCM3. The correlation between these alternative range size estimates and the estimates of genetic diversity is comparable to those for GENESIS2, and are shown in Supplementary Table S4.2 and Supplementary Figures S4.4 and S4.5. The species distribution model was run for the full set of fossil localities only (i.e. no subsampling was performed), and the genetic diversity data was represented by 1,000 different BEAST trees. Supplementary Table S4.2. Summary of the posterior parameters of the Bayesian hierarchical model. SE is the standard deviation of the posterior distribution. Here p represents the the proportion of values that are below 0, and is comparable to the p-value from a standard one-tailed test. Range data are modelled using HadCM3. Slope 0.915 0.714 1.026 1.149 SE 0.540 0.556 0.421 0.785 p Intercept 0.047 -1.637 0.098 3.053 0.009 -3.478 0.048 -6.730 SE 8.754 8.923 6.530 12.529
WWW.NATURE.COM/NATURE | 57
doi:10.1038/nature10574
Supplementary Figure S4.4. Relationship between Ne* and range size (modelled with HadCM3), summarized as posterior probability distributions of regression slopes from a Bayesian hierarchical regression. The proportion of probability densities below zero (analogous to a one-tailed p-value) can be seen in Supplementary Table S4.2.
WWW.NATURE.COM/NATURE | 58
doi:10.1038/nature10574
4.5
4.5
5.0
5.0
Log10
6.0
6.0
6.5
6.5
6.4
6.4
6.6
6.6
6.8
7.0
7.2
7.2
7.4
7.4
Log10 km2
Supplementary Figure S4.5. The relationship between Ne* and range size for the four species for which we had sufficient data for a correlation analysis, modelled using HadCM3. Genetic diversity is shown as the mean and standard deviation of the 1,000 values used in the regression analysis. The mean values of slope and intercept (Supplementary Table S4.2) were used to draw the trendlines. 4.4 Influence of sample distribution and -size on results The positive correlation observed between geographic range and genetic diversity could potentially be caused by the spatial distribution of samples and sample size, so the wider the distribution, the more palaeohabitats covered and consequently the larger the forecasted range. Similarly, the genetic diversity may increase with the number of samples and the geographic range covered. To investigate the effect of sample distribution, we calculated the mean pairwise geographic distance between all samples within each of the four time bins (4539, 3327, 2418 and 93 kyr BP). Geographic distances were calculated between the LAT/LON coordinates using the Haversine formula and ignoring hills. For species present in both North America and Eurasia, pairwise distances between continents were rooted through Beringia (LAT 66.07, LON -168.92) to avoid crossing of the North Pole. To account for differences in sample size between time bins, we
WWW.NATURE.COM/NATURE | 59
doi:10.1038/nature10574
averaged the pairwise distances by the number of samples. However, due to differences in the number of samples from each locality, we also averaged the geographic distances within each time bin by number of unique localities. As a third measure of spatial distribution, we calculated the distance between the two furthermost samples within each timebin. To asses the influence of sample number (disregarding spatial distribution), we estimated the correlation between estimates and sample size. The distribution measures were compared to the mean values of genetic diversity and potential range size estimated for each time bin. To ensure that comparisons with our results were unbiased, we used the same WinBUGS hierarchical regression model to perform the correlations, and summarized our results by the proportion of the posterior distribution of slope values that are below 0 (see above). The results are presented in Supplementary Table S4.3. All posterior probability intervals include 0, indicating that the relationship between population and range size is unaffected by any relationship with the geographic distribution and number of samples. Supplementary Table S4.3. Summary of Bayesian hierarchical linear regression models between measures of sampling (rows) and genetic diversity or range size (columns). Numbers indicate the proporation of the posterior density for the regression slope that lies below 0, and is thus comparable to the "p"-values given for the effective population size - range size correlation and to one-sided p values in frequentist statistics. Potential range size Mean pairwise distance All samples Unique localities Maximum distance Number of samples 0.175 0.096 0.209 0.246 Genetic diversity 0.154 0.149 0.197 0.225
To investigate whether the inclusion of 16 non-directly dates reindeer and bison samples (Supplementary Information section S1) influenced the positive relationship between range size and genetic diversity, we repeated the correlation analysis using GENESIS2 range sizes estimated with those data excluded. The significance of the correlation was not affected (Supplementary Table S4.4).
WWW.NATURE.COM/NATURE | 60
doi:10.1038/nature10574
Supplementary Table S4.4. Summary of the posterior parameters of the Bayesian hierarchical model, with range sizes estimated using only directly-dated fossils for all species. Range data are modelled using GENESIS2. Here p is the proportion of probability densities below zero and is analogous to a one-tailed p-value. Slope Horse Reindeer Bison Musk Ox 1.410 0.512 1.934 0.909 SE 0.830 0.405 0.702 0.473 p 0.039 0.092 0.003 0.029 Intercept SE -10.191 6.158 -17.434 -3.241 13.760 6.577 10.840 7.728
WWW.NATURE.COM/NATURE | 61
doi:10.1038/nature10574
archaeofaunas. In Europe, these are primarily from sources compiled as part of the Stage 3 Project102. In Siberia, they come from sources cited in Supplementary Table S6.5. The indirectlydated faunal material was included for multiple reasons. Firstly, the integrity of these data were good (see handling of samples below) and were comparable to the European human data. Secondly, had we not included the indirectly-dated information, we would ignore large amounts of data indicating the presence of the six megafauna species at particular points in time and in conjunction with humans. If we ignored these because they are not directly dated, our sample of direct material would be biased. Finally, including the indirectly-dated specimens considerably increased our samples sizes; without these data included, musk ox in Europe and bison in Siberia were uninformative, due to the rarity of their occurrences in the fossil record. These indirect dates apply only to determinations of geographic overlap and temporal trends in radiocarbon frequencies as outlined in this section, and are presented within the main text. We also provide a replicate analysis using directly-dated only material for comparison, with similar results (Supplementary Fig. S5.2). The potential range size estimates used in the analysis of overlap between humans and megafauna ranges at 42, 30 and 21 kyr BP used directly-dated materials only
WWW.NATURE.COM/NATURE | 62
doi:10.1038/nature10574
(except 16 indirect dates, discussed in Supplementary Information section S1), and provide another, quasi-independent assessment of these same patterns. Indirectly-dated megafauna specimens from Europe In Europe, data were selected from the Stage 3 database provided they met a number of criteria. 1) Ages had to be associated with the presence of one of the five key taxa being studied in Europe (woolly rhinoceros, woolly mammoth, horse, reindeer and musk ox). 2) All ages had to be clearly identified as being radiocarbon determinations (standard or AMS) or, in a few cases, had to come from organic materials that could reasonably be inferred to have been radiocarbon-dated; luminescence and uranium-thorium dates were excluded. 3) Each radiocarbon age had to have associated latitude and longitude information, lab codes and reported errors less than 10% of mean ages. All ages greater than 45,000 radiocarbon years before present were also excluded. 4) Any radiocarbon determinations that were directly associated with a given taxon were identified and assigned correctly. For example, a deposit containing horse, reindeer and woolly rhinoceros might be placed at 27,000 BP based on the dating of a horse tibia. The horse would be identified as directly dated and tallied with the directly dated specimens, while the other two taxa would be assigned an age of 27,000 BP but be considered indirectly dated. These indirect radiocarbon ages were then combined with the directly-dated material and compared to trends in human radiocarbon frequencies. Note because of the temporal focus of the Stage 3 Project, the data on early European faunas (prior to 2018 kyr BP) is actually somewhat richer than that for later periods. Archaeofaunal data from Europe and Siberia A further subset of purely archaeological faunas was also identified and their temporal trends examined (Figure 4 in main text). In the case of the Stage 3 data from Europe, radiocarbon ages were tallied only when associated with archaeological stone tool industries. To make the data comparable to the Siberian dataset, information was summarized by occupation (n=380) rather than individual radiocarbon age, with occupations assigned primarily on the basis of excavation layers identified in the Stage 3 database. When detailed information on individual sites and excavations was lacking, multiple radiocarbon ages within occupations were averaged, but not weighted or pooled. The Siberian data were compiled from 98 radiocarbon-dated cultural occupations of 68 Upper Palaeolithic archaeological sites, each of which listed at least one of the six megafuna species. The
WWW.NATURE.COM/NATURE | 63
doi:10.1038/nature10574
data and references are listed in Supplementary Table S6.5. An important note is that the radiocarbon dates presented are again not necessarily direct dates on a megafauna specimen, but are dates from materials (e.g., charcoal or even some other animal bone) associated with the listed species. The sites range in age from c. 4112 thousand calendar years ago (kyr BP) and geographically come from the Ob, Yenisei, and Lena River basins as well as the eastern Transbaikal and far northeast Russia. All of these cultural occupations are interpreted to represent Upper Palaeolithic occupations. Middle Palaeolithic occupations were excluded. There are a few other notable Upper Palaeolithic assemblages with rich faunal records that were not included in the analysis; these are sites or occupations with no associated radiocarbon ages (e.g., the later Upper Palaeolithic occupations at the Krasnyi Iar sites, layer 8 at Diuktai Cave), only infinite dates (e.g., Makarovo-4), or problematic radiocarbon chronologies (e.g., Mogochino, Studenoe-1). All taxonomic identifications were done by primary investigators of the archaeological sites. In most cases, published archaeological reports describing these sites provide only kitchen lists of faunal taxa present, while detailed statistics like number of individual specimens present (NISP) or minimum number of individuals (MNI) are not reported (although there are some notable exceptions, for example103,104,105,106; information included in Supplementary Table S6.5). Identifications reported in the primary literature are often at the genus level, which is not a problem for Coelodonta, Mammuthus, Bison, and Rangifer, but it means that some of the remains identified as Equus could include E. hemionus, the Asiatic wild ass. Similarly for musk ox, multiple species may occur, including Ovibos moschatus and O. pallantis (106,107). Dates were gathered from the primary literature, including dates on animal bone, charcoal and other organic materials associated with Palaeolithic artifacts or features. Dates from non-cultural layers (e.g., from above or below a cultural layer) or from problematic materials (e.g., soil organics) were omitted, as were obviously aberrant dates (i.e., those that were clearly discordant from other dates in the same occupation, or were not in accord with other occupations of the same site). Multiple dates from the same occupations were averaged, using the method described by 108. This was done to keep occupations with multiple dates (for example, Malta, which has 13 radiocarbon dates for the same cultural layer) from weighing more heavily in the analysis than occupations with single dates. Latitude and longitude data were obtained using published descriptions of site locations and Google Earth.
WWW.NATURE.COM/NATURE | 64
doi:10.1038/nature10574
Human data Siberian human occupations come from 109. The authors provide list of 516 georeferenced radiocarbon dates from 129 archaeological sites, along with summaries by individually-dated component. The radiocarbon ages for each occupational component were used here (n=233). European human radiocarbon ages come from the INQUA Palaeolithic Radiocarbon Database, v. 1167. These data represent over 7,000 radiocarbon determinations from more than 1,500 sites. Due to the size and diverse nature of the INQUA data set, radiocarbon ages were not aggregated by archaeological occupation as done in Siberia. The data were cleaned for obvious errors, however, including: a) using only ages between 45,000 and 7,500 14C years BP, b) excluding all ages without latitude and longitude data, c) excluding all ages without associated 14C errors, d) excluding all ages without lab codes, e) excluding dates of palaeontological deposits and f) excluding any determination with an error greater than 10% of the mean. A handful of discordant dates were also removed, mostly very young dates from purportedly old deposits. Although most sites have only a handful of ages, a few sites have many. These few sites are long, continuously occupied stratigraphic sequences, and the multiple 14C determinations generally span the EpiPalaeolithic to Middle Palaeolithic (c. 4512,000 14C years BP) rather than being clustered in individual occupations. They are not biased geographically, and include sites in England, Spain, France, Germany and Russia. This yielded a total of 5,875 ages from over 1,461 sites in Europe. These include those produced by both anatomically modern Homo sapiens sapiens and archaic Homo sapiens neanderthalensis, the latter typically being associated with Middle Palaeolithic, Mousterian stone-tool industries. Although the Neandertals disappeared from most of Europe by 30,000 years ago, current interpretations suggest they were top-tier predators and should therefore have impacted megafauna populations when and where they were present110,111,112. While we were unable to aggregate the European human data as done in Siberia, we were able to assess the use of individual determinations by comparing patterns in the data provided by68. In that case, frequencies of individual 14C determinations and dated components per 500-year interval are very tightly correlated (r=.907, t=17.2, df=64, p<.001; Supplementary Fig. S5.1) and the same is expected to hold true for Europe.
WWW.NATURE.COM/NATURE | 65
doi:10.1038/nature10574
Supplementary Figure S5.1. Comparison of 14C trends by aggregation method. Data points represent the binned, calibrated radiocarbon probabilities per 500-year interval for each data set (dated occupation; individual 14C determinations). Periods with high numbers of individual radiocarbon dates also have high numbers of independent, dated occupations. 5.3 Data analysis Summarizing radiocarbon frequencies and geographic data To compare frequencies of different taxonomic groups, radiocarbon ages for all dated components, individual radiocarbon ages (European humans), or dated megafauna were summed in 500calendar-year intervals from 50,0000 BP (blocks 0-499 BP; 500-999 BP; 1,000-1,499 BP; etc.). Rather than using dated midpoints, the various radiocarbon ages were first calibrated using Calib 6.0 and the IntCal09 calibration curve113. A script was then used to collapse the year-by-year probabilities output by Calib 6.0 (options, write distribution files = yes) into 500-year blocks using R64. This allowed us to use moderately-sized time intervals without worrying about larger errors associated with many determinations, the correct assignment of ages that fall close to interval divisions, or the non-normal distribution of the underlying, calibrated dates. Mean latitudes and longitudes and their standard deviations were also calculated using the probability of a determination falling into a 500-year interval to weight its contribution. Thus dates that were highly likely to fall in a particular interval contributed heavily to estimates of the average latitude/longitude, while a determination that had a small possibility of falling in an interval
WWW.NATURE.COM/NATURE | 66
doi:10.1038/nature10574
contributed little. Graphs of mean latitude and longitude for dated faunas and archaeological sites from Europe and Siberia are shown in Supplementary Figure S5.2. The area shared by humans and faunas at specific time intervals was approximated by identifying the latitudinal and longitudinal extent of the region held in common between them per interval (using mean lat/long 1sd), calculating its area, and expressing it as a percentage of the area occupied by the a given taxon (again at mean lat/long 1sd ). These data and variation in location is discussed in the primary text and shown in Figure 4. Timespan of comparison Radiocarbon data were initially binned and mean latitudes and longitudes calculated across the entire IntCal09 sequence (50,0000 calendar years BP), but trends in calibrated radiocarbon frequencies were compared across a smaller time range. First, no data were considered for periods beyond 45 kyr BP. Underlying 14C data were too few for most taxa and regions, and the reliability of such dates more questionable. The younger end of the sequence was held at 12 kyr BP, the period by which most of the megafauna other than reindeer had substantially declined and the frequency of Palaeolithic occupations also drops off. Supplementary Figures S5.3 and S5.4 show the temporal distribution of binned, calibrated radiocarbon frequencies for the two regions using directly dated faunas only and with indirectly dated faunas included.
WWW.NATURE.COM/NATURE | 67
doi:10.1038/nature10574
a
Latitude
Woolly rhinoceros
90
Woolly mammoth
Horse
Reindeer
Bison
Musk ox
60
n/a
30 70
Longitude
25 -20 50 40 30 20 10 kyr BP 90
n/a
b
Latitude
60
n/a
30 70
Longitude
25 -20 50 40 30 20 10 kyr BP
n/a
c
Latitude Longitude
d
Latitude Longitude
Supplementary Figure S5.2. Plots showing latitudinal and longitudinal overlap between megafauna (coloured shading) and human (grey shading) calibrated radiocarbon ages, A. Europe direct dates only, B. Europe indirect dates added, C. Siberia direct dates only, D. Siberia indirect dates added. Error bars are 1s.d.
WWW.NATURE.COM/NATURE | 68
doi:10.1038/nature10574
Supplementary Figure S5.3a. European radiocarbon frequencies, humans and directly dated faunas only.
Supplementary Figure S5.3b. European radiocarbon frequencies, indirectly dated faunas included.
WWW.NATURE.COM/NATURE | 69
doi:10.1038/nature10574
Supplementary Figure S5.4a. Siberian radiocarbon frequencies, humans and directly dated faunas only. The first panel shows the paeleoclimate data for the corresponding period.
WWW.NATURE.COM/NATURE | 70
doi:10.1038/nature10574
Supplementary Figure S5.4b. Siberian radiocarbon frequencies, indirectly dated faunas included. The first panel shows the paeleoclimate data for the corresponding period.
WWW.NATURE.COM/NATURE | 71
doi:10.1038/nature10574
5.4 Comparing radiocarbon frequencies of humans and megafauna We compare radiocarbon frequencies among taxa after accounting for the possible effects of taphonomy and climate, following methods outlined in 97,98 and 114. Our chosen climate proxy was Shackletons 24MD952042 North Atlantic benthic 18O core, corrected to the SFCP2005 timescale115,116 (variable, 18O; first panel in Fig. S5.4a,b). This is the same fine-grained proxy used in the previously cited studies, is appropriate for comparing trends in geographically widespread locations and mirrors but is less noisy than the Greenland ice-cores. To incorporate issues of taphonomy and preservation in the human data, we used the model proposed by 117. Their formula calculates an expected number of observations at time t for a constant, random phenomenon suffering solely taphonomic losses. We converted this to an estimate of the proportion of original observations surviving at each 500-year interval midpoint (12,250 calendar years BP; 12,750 calendar years BP; etc.) as Nt/N0. These values scale between 0 and 1, decline in a curvilinear fashion, and were used as predictors in our regression models (variable, Surv). Fits were slightly better than simple log transforms, with somewhat more variation across intermediate values but a reduced tendency to overestimate extremes. While this approach differs from that advocated by the original authors, we argue that it is more appropriate in cases such as ours, where similar taphonomic histories cannot be assumed for all sequences118. As an ancillary measure, we also calculated the absolute difference (variable, AbsDif) between radiocarbon and calendar years for each interval using the IntCal09 curve. For example, the period from 2928.5 kyr BP corresponds to 24,14623,743 14C years BP (ignoring error). The corresponding radiocarbon interval is thus 403 years long, or 97 years shorter than the corresponding calendar span. This interval would have a measure of -97. Exactly equal intervals would have a value of 0 assigned, etc. These measures were meant to provide a check on possible over- or underrepresentation of parts of the calendar due to differences in atmospheric 14C production. A file containing the benthic 18O sequence, calculated survivorship and 14C/calendar differences is provided in the supplemental archive (climtaph.csv). Comparison among taxa in Siberia and Europe were based on partial correlations after controlling for the three variables just mentioned. This was done using the residuals from a multiple regression of calibrated radiocarbon frequency (variable, 14C; the binned sums) against each main effect and allowing for a survivorship by climate interaction (14C~ Surv + 18O + AbsDif +Surv:18O). We then looked at how human frequencies compared with those of other faunas. In Europe, comparisons
WWW.NATURE.COM/NATURE | 72
doi:10.1038/nature10574
using indirectly dated faunas were done only after excluding 676 radiocarbon determination from the INQUA human database that were also found in the Stage 3 archaeofaunas. While not perfect, the overall behaviour of the models is good to very good in most cases, although the actual amount of variation explained by the taphonomic and climate variables varies substantially and the residuals in some taxa and regions still deviate from normal. Examples of such deviations includes higher than expected frequencies of human occupations at the PleistoceneHolocene boundary, high frequencies of several indirectly-dated European faunas around 32 kyr BP (Fig. S5.3b), and the irregular behavior of several directly dated taxa such as European musk ox and, to a lesser extent, European horse and Siberian bison. In the latter cases this clearly results from the substantial gaps in the 14C record (Supplementary Fig. S5.3a) and one should be cautious when drawing inferences from these data. Partial correlation coefficients for humans versus European and Siberian faunas are presented in Supplementary Table S5.1. After controlling for the possible influence of climate and taphonomic losses, there appears to be little relationship between humans and any of the faunas other than mammoth and perhaps Siberian musk ox (Figure S5.5a,c). There is also little relationship among changes in the frequencies of dated faunas. Again, part of this is due to the small sample sizes of many directly dated taxa. If the indirectly dated materials are included, changes in the frequency of dated faunas and humans in Europe become more positively correlated (Supplementary Table S5.1b; Supplementary Fig. S5.5b,d) and higher yet for OIS 3 material only (4522 kyr BP). Correlations among faunas other than musk ox also become very high (.85 <= r <= .95). In Siberia, residual radiocarbon frequencies of humans and faunas other than musk ox also become moderately but positively correlated, echoing previously reported patterns identified for humans and woolly mammoths97,98. The relationship with humans reflects the incorporation of archaeological faunas, particularly the OIS 3 faunas in Europe, while the higher correlations among faunas reflect the fact that many of these animals regularly co-occur within particular archaeological deposits. The frequencies of archaeological faunas and non-faunal archaeological deposits remain highly correlated throughout OIS 3, including both broad increases from 45 kyr BP to roughly 30 kyr BP and general declines between 30 and 22 kyr BP (r=.82), and it would be interesting to see if these continued past the Last Glacial Maximum. While suggestive of some common driver for humans and faunas, this strong, positive relationship cannot be read as absolute proof that both humans and faunas were more common at certain times, since people might be expected to preferentially incorporate larger animals even if the latter were in decline.
WWW.NATURE.COM/NATURE | 73
doi:10.1038/nature10574
Regardless, these changes are not the clear-cut indicator one would expect if people were negatively impacting these megafauna; more human occupations do not appear to lead to fewer occurences of any taxon. These results are not necessarily surprising given the often limited overlap in their ranges (Fig. 4 in main text, Supplementary Fig. S5.2). Humans and musk ox tend to be found in very different regions of Europe and Siberia, and in Siberia woolly rhino and woolly mammoth appear to shift their ranges northward even as human distribution remains fairly constant through 12,000 kyr BP. Opportunities for humans to impact these taxa may have been limited as a result. These results do not deny humans a role in the eventual extinction of any taxon and will merit reconsideration as additional remains are uncovered. However, they do cast doubt on whether such extinctions occurred solely as a consequence of human impact. Supplementary Table S5.1a. Partial correlations between the frequency of dated humans and faunas per 500-year time period after controlling for climate and taphonomy. Residuals based on the model described in the text using directly dated faunas only. Woolly Rhino Europe Siberia .222 .334 Woolly Mammoth .415 .374 .232 .040 .010 .027 .084 .034 -.249 Horse Reindeer Bison Musk ox
Supplementary Table S5.1b. Partial correlations between the frequency of dated humans and faunas per 500-year time period after controlling for climate and taphonomy. Residuals based on the model described in the text using directly and indirectly dated faunas. Woolly Rhino Europe Siberia .526 .421 Woolly Mammoth .540 .448 .523 .528 .471 .525 .525 .285 -.148 Horse Reindeer Bison Musk ox
WWW.NATURE.COM/NATURE | 74
doi:10.1038/nature10574
Supplementary Figure S5.5a. Scatterplot of residual 14C frequencies for Europe. "ho" horse; "hu" human; "mo" musk ox; "rd" reindeer; "rh" woolly rhino; "wm" woolly mammoth. Directly dated material only.
WWW.NATURE.COM/NATURE | 75
doi:10.1038/nature10574
Supplementary Figure S5.5b. Scatterplot of residual 14C frequencies for Europe. "bi" bison; "ho" horse; "hu" human; "mo" musk ox; "rd" reindeer; "rh" woolly rhino; "wm" woolly mammoth. Indirect dates included.
WWW.NATURE.COM/NATURE | 76
doi:10.1038/nature10574
Supplementary Figure S5.5c. Scatterplot of residual 14C frequencies for Siberia. "bi" bison; "ho" horse; "hu" human; "mo" musk ox; "rd" reindeer; "rh" woolly rhino; "wm" woolly mammoth. Direct dates only.
WWW.NATURE.COM/NATURE | 77
doi:10.1038/nature10574
Supplementary Figure S5.5d. Scatterplot of residual 14C frequencies for Siberia. "bi" bison; "ho" horse; "hu" human; "mo" musk ox; "rd" reindeer; "rh" woolly rhino; "wm" woolly mammoth. Indirect dates included.
WWW.NATURE.COM/NATURE | 78
doi:10.1038/nature10574
REFERENCES
Lomolino, M. V. & Heaney, L. R., eds. Frontiers of Biogeography: New Directions in the Geography of Nature. Sinauer Associates, Sunderland, Massachusetts (2004)
1 2 2
Diniz-Filho, J. A. F. et al. Partitioning and mapping uncertainties in ensembles of forecasts of Diniz-Filho, J. A. F. et al. Partitioning and mapping uncertainties in ensembles of forecasts of species turnover under climate change. Ecography 32, 897-906 (2009) Nogus-Bravo, D., Rodrguez, J., Hortal, J., Batra, P. & Arajo, M. B. Climate change, humans, and the extinction of the woolly mammoth. PLOS Biol. 6, e79 (2008) Grinnell, J. The niche-relationships of the California Thrasher. Auk 34, 427-433 (1917)
4 5
Hutchinson, G. E. Concluding remarks. Cold Spring Harbor Symp. Quant. Biol. 22, 145-159 (1957) Sobern, J. Grinellian and Eltonian niches and geographic distributions of species. Ecol. Lett. 10, 1115-1123 (2007) Whittaker, R. H. Vegetation of the Great Smoky Mountains. Ecol. Monogr. 26, 1-80 (1956)
7 8
Whittaker, R. H. Direct Gradient Analysis. In: Whittaker, R. H., ed. Handbook of Vegetation Science 5: Coordination and Classification of Communities. The Hague: Junk Publishers, 9-50 (1973)
9
Austin, M. P. Models for the analysis of species response to environmental gradients. Vegetatio 69, 35-45 (1987)
Austin, M. P. Spatial prediction of species distribution: an interface between ecological theory and statistical modelling. Ecol. Model. 157, 101-118 (2002) Colwell, R. K. & Rangel, T. F. Hutchinsons duality: the once and future niche. P. Nat. Acad. Sci.106 (Supplement 2), 19651-19658 (2009) Purvis, A., Gittleman, J. L., Cowlishaw, G. & Mace, G. M. Predicting extinction risk in declining species. Proc. R. Soc. Lond. B 267, 1947-1952 (2000) Elith, J., et al. Novel methods improve prediction of species distributions from occurrence data. Ecography 29, 129151 (2006) Nogus-Bravo, D. Predicting the past distribution of species climatic niches. Global Ecol. Biogeogr. 18, 521-531 (2009) Pearson, R. G. & Dawson, T. P. Predicting the impacts of climate change on the distribution of species: are bioclimate envelope models useful? Global Ecol. Biogeogr. 12, 361-371 (2003) Williams, J. W., Jackson, S. T. & Kutzbach, J. E. Projected distributions of novel and disappearing climates by 2100 AD. P. Nat. Acad. Sci. 104, 5738-5742, (2007)
16 15 14 13 12 11
10
WWW.NATURE.COM/NATURE | 79
doi:10.1038/nature10574
17 Sowers, T. & Bender, M. Climate records covering the last deglaciation. Science 269, 210214 (1995) Bonfils, C. J., Lewden, D. & Taylor, K. E. Summary documentation of the PMIP models. Available: http://pmip.lsce.ipsl.fr/docs/. Accessed 27 February 2008 (1998) CLIMAP Project Members. Seasonal reconstructions of the Earths surface at the Last Glacial Maximum. In: Map Series, Technical Report MC-36. Boulder (Colorado): Geological Society of America (1981) Barron, E. J., & Pollard, D. High-resolution climate simulations of Oxygen Isotope Stage 3 in Europe. Quaternary Res. 58, 296309 (2002) Pollard, D., Bergengren, J. C., Stillwell-Soller, L. M., Felzer, B. & Thompson, S. L. Climate simulations for 10000 and 6000 years BP using the GENESIS global climate model. Palaeoclimates- Data Model. 2, 183218 (1998)
22 21 20 19 18
Berger, A. L. Long-term variations of daily insolation and Quaternary climatic changes. J Atmos. Sci. 35, 23622367 (1978)
Berger, A., & Loutre, M. F. Insolation values for the climate of the last 10 million years. Quaternary Sci. Rev. 10, 297317 (1991) Hoar, M. R. Palutikof, J. P. & Thorne, M. C. Model intercomparison for the present day, the midHolocene, and the Last Glacial Maximum over western Europe. J. Geophys. Res. 109, D08104.1D08104.25 (2004) Alfano, M. J., Barron, E. J., Pollard, D., Huntley, B. & Allen, J. R. M. Comparison of climate model results with European vegetation and permafrost during oxygen isotope stage three. Quaternary Res. 59, 97-107 (2003)
26
Thompson, 25 24
23
S. L. & Pollard, D. Greenland and Antarctic mass balances for present and doubled atomospheric CO2 from the GENESIS version-2 global model. J. Climate 10, 871900 (1997)
Barnola, J.-M., Raynaud, D., Korotkevich, Y. S. & Lorius, C. Vostok ice core provides 160,000year record of atmospheric CO2. Nature 329, 408-414 (1987) Neftel, A., Oeschger, H., Staffelbach, T. & Stauffer, B. CO2 record in the Byrd ice core 50,000 5,000 years B.P. Nature 331, 609611 (1988) Singarayer, J. S. & Valdes, P. J. High-latitude climate sensitivity to ice-sheet forcing over the last 120 kyr. Quaternary Sci. Rev. 29, 43-55 (2010) Gordon, C. et al. The simulation of SST, sea ice extents and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments. Clim. Dynam. 16, 147168 (2000) Gregory, J. M. & Mitchell, J. F. B. The climate response to CO2 of the Hadley Centre Coupled AOGCM with and without flux adjustment. Geophys. Res. Lett. 24, 19431946 (1997) Edwards, J. M. & Slingo, A. Studies with a flexible new radiation code.1. Choosing a configuration for a large-scale model. Q. J. Royal Met. Soc. 122, 689719 (1996)
WWW.NATURE.COM/NATURE | 80
27
28
29
30
31
32
doi:10.1038/nature10574
Cusack, S., Slingo, A., Edwards, J. M. & Wild, M. The radiative impact of a simple aerosol climatology on the Hadley Centre GCM. Q. J. Royal Met. Soc. 124, 25172526 (1998) Gregory, D., Kershaw, R. & Inness, P. M. Parameterisation of momentum transport by convection: II. Tests in single column and general circulation models. Q. J. Royal Met. Soc. 123, 11531183 (1997) Cox, P., Betts, R., Bunton, C., Essery, R., Rowntree, P. R., & Smith, J. The impact of new landsurface physics on the GCM simulation and climate sensitivity. Clim. Dynam. 15, 183-203 (1999) Gent, P. R. & McWilliams, J. C. Isopycnal mixing in ocean circulation models. J. Phys. Oceanogr. 20, 150155 (1990) Cattle, H. & Crossley, J. Modelling arctic climate change. Phil. Trans. R. Soc. Lond. A 352, 201213 (1995) Petit, J. R. et al. Climate and atmospheric history of the past 420 000 years from the Vostok Ice Core, Antarctica. Nature 399, 429436 (1999) Spahni, R. et al. Variations of atmospheric methane and nitrous oxide during the last 650 000 years from Antarctic ice cores. Science 310, 13171321 (2005)
40 39 38 37 36 35 34 33
Parrenin, F. et al. The EDC3 agescale for the EPICA dome C ice core. Clim. Past 3, 485497 (2007)
Peltier, W. R. Global glacial isostasy and the surface of the iceage Earth: the ICE-5G (VM2) model and GRACE. Annu. Rev. Earth Planet Sci. 32, 111149 (2004) Peltier, W. R. & Fairbanks, R. G. Global glacial ice volume and Last Glacial Maximum duration from an extended Barbados sea level record. Quat. Sci. Rev. 25, 33223337 (2006) Martinsen, D. G., et al. Age dating and orbital theory of the ice ages: development of a high resolution 0300,000-year chronostratigraphy. Quat. Res. 27, 130 (1987) Hewitt, C. D., et al. The effect of a large freshwater perturbation on the glacial North Atlantic Ocean using a coupled general circulation model. J. Clim. 19, 44364447 (2006) Braconnot, P., et al. Results of PMIP2 coupled simulations of the Mid-Holocene and Last Glacial MaximumPart 1: experiments and large-scale features. Clim. Past 3, 261277, doi:10.5194/cp-3261-2007 (2007) Beaumont, L. J., Hughes, L. & Poulsen, M. Predicting species distributions: use of climatic parameters in BIOCLIM and its impact on predictions of species current and future distributions. Ecol. Mod. 186, 251-270 (2005) Reimer, P. J., et al. IntCal09 and Marine09 radiocarbon age calibration curves, 050,000 years cal BP. Radiocarbon 51, 11111150 (2009)
47 46 45 44 43 42
41
WWW.NATURE.COM/NATURE | 81
doi:10.1038/nature10574
48
Barnosky, A. D. & Lindsey, E. L. Timing of Quaternary megafaunal extinction in South America in relation to human arrival and climate change. Quatern. Int. 217, 10-29 (2010) Kuzmin, Y. V. & Orlova, L. A. Radiocarbon chronology and environment of wooly mammoth (Mammuthus primigenius Blum.) in northern Asia: results and perspectives. Earth Sci. Rev. 68, 133-169 (2004) Vartanyan, S. L., Arslanov, K. A., Karhu, J. A., Possnert, G. & Sulerzhitsky, L. D. Collection of radiocarbon dates on the mammoths (Mammuthus primigenius) and other genera of Wrangel Island, northeast Siberia, Russia. Quaternary Res. 70, 51-59 (2008) Guthrie, R. D. Radiocarbon evidence of mid-Holocene mammoths stranded on an Alaskan Bering Sea island. Nature 429, 746-749 (2004) Farber, O. & Kadmon, R. Assessment of alternative approaches for bioclimatic modelling with special emphasis on the Mahalanobis distance. Ecol. Mod.160, 115-130 (2003) Anderson, R. P., Lew, D. & Peterson, A. T. Evaluating predictive models of species' distributions: criteria for selecting optimal models. Ecol. Mod. 162, 211-232 (2003) Phillips, S. J., Anderson, R. P. & Schapire, R. E. Maximum entropy modelling of species geographic distributions. Ecol. Mod. 190, 231-259 (2006)
55 54 53 52 51 50 49
Lobo, J. M., Jimnez-Valverde, A. & Hortal, J. The uncertain nature of absences and their importance in species distribution modelling. Ecography 33, 103-114 (2010) Lobo, J. M. & Tognelli, M. F. Exploring the effects of quantity and location of pseudo-absences and sampling biases on the performance of distribution models with limited point occurrence data. J. Nat. Conserv. 19, 1-7 (2011)
56
Varela, S. J. M., Lobo, J., Rodrguez, J., & Batra, P. Were the Late Pleistocene climatic changes responsible for the disappearance of the European spotted hyena populations? Hindcasting a species geographic distribution across time. Quaternary Sci. Rev. 29, 2027-2035 (2010) Tsoar, A., Allouche, O., Steinitz, O., Rotem, D. & Kadmon, R. A comparative evaluation of presence-only methods for modelling species distribution. Diversity Distrib. 13, 397-405 (2007) Varela, S., Rodrguez, J. & Lobo, J. M. Is current climatic equilibrium a guarantee for the transferability of distribution model predictions? A case study of the spotted hyena. J. Biogeogr. 36, 1645-1655 (2009) Martnez-Meyer, E., Peterson, A. T. & Hargrove, W. W. Ecological niches as stable distributional constraints on mammal species, with implications for Pleistocene extinctions and climate change projections for biodiversity. Global Ecol. Biogeogr. 13, 305-314 (2004)
61 60 59 58
57
Souza Muoz, M. et al. openModeller: a generic approach to species potential distribution modelling. GeoInformatica 1-25, DOI 10.1007/s10707-009-0090-7(2009)
Lobo, J. M., Jimnez-Valverde, A. & Real, R. AUC: a misleading measure of the performance of predictive distribution models. Global Ecol. Biogeogr. 17, 145-151 (2008)
62
WWW.NATURE.COM/NATURE | 82
doi:10.1038/nature10574
63 Pearce, J. & Ferrier, S. An evaluation of alternative algorithms for fitting species distribution models using logistic regression. Ecol. Mod. 128, 127-147 (2000) R Development Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.Rproject.org (2010) Bivand, R. S., Pebesma, E. J. & Gomez-Rubio, V. Applied spatial data analysis with R. Springer, NY. http://www.asdar-book.org/ (2008)
66
Kuhn, 65 64
T. S., McFarlane, K. A., Groves, P., Mooers, A. O. & Shapiro, B. Modern and ancient DNA reveal recent partial replacement of caribou in the southwest Yukon. Mol. Ecol. 19, 1312 1323 (2010)
INQUA. International Union for Quaternary Research (INQUA) Radiocarbon Palaeolithic Database Europe, v.11. (http://www.kuleuven.be/geography/frg/projects/14cpalaeolithic/download) (2010)
68
67
Hamilton, M. J. & Briggs, B. Archaeological support for the three-stage expansion of modern humans across northeastern Eurasia and into the Americas. PLOS One 5(8), 1-9 (2010)
Arajo, M. B. & Pearson, R. G. Equilibrium of species distributions with climate. Ecography 28, 693-695 (2005) Svenning, J.-C. & Skov, F. Could tree diversity pattern in Europe be generated by postglacial dispersal limitation? Ecol. Let. 10, 453-460 (2007) Elias, S. A. & Crocker, B. The Bering Land Bridge: a moisture barrier to the dispersal of steppetundra biota? Quaternary Sci. Rev. 27, 2473-2483 (2008) Weinstock, J. et al. Evolution, systematics, and phylogeography of Pleistocene horses in the New World: a molecular perspective. PLoS Biol. 3: e241 (2005) Campos, P.F. et al. Ancient DNA analyses exclude humans as the driving force behind late Pleistocene musk ox (Ovibos moschatus) population dynamics. P. Nat. Acad. Sci.107, 56755680 (2010) Gravlund, P., Meldgaard, M., Paabo, S. & Arctander, P. Polyphyletic origin of the small-bodied, high-arctic subspecies of tundra reindeer (Rangifer tarandus). Mol. Phylogenet. Evol. 10, 151159 (1998) Edgar, R. C. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 32, 1792-1797 (2004) Gouy, M., Guindon, S. & Gascuel, O. SeaView version 4: a multiplatform graphical user interface for sequence alignment and phylogenetic tree building. Mol. Biol. Evol. 27, 221224 (2010)
77
Shapiro, 76 75 74 73 72 71 70
69
B. et al. Rise and fall of the Beringian steppe bison. Science 306, 1561-1565 (2004)
WWW.NATURE.COM/NATURE | 83
doi:10.1038/nature10574
78
Debruyne, R. et al. Out of America: ancient DNA evidence for a new world origin of late Quaternary woolly mammoths. Curr. Biol. 18, 13201326 (2008)
Excoffier, L. & Lischer, H. E. Arlequin suite ver 3.5: a new series of programs to perform population genetics analyses under Linux and Windows. Mol. Ecol. Res. 10, 564567 (2010) Librado, P. & Rozas, J. DnaSP v5: A software for comprehensive analysis of DNA polymorphism data. Bioinformatics 25, 14511452 (2009) Depaulis, F., Orlando, L. & Hnni, C. Using classical population genetics tools with heterochroneous data: time matters! PLoS ONE 4, e5541 (2009) Boyd, L. & Haupt, K.A. Przewalski's horse: the history and biology of an endangered species. Albany: State University of New York Press (1984) Boulet, M., Couturier, S., Ct, S. D., Otto, R. & Bernatchez, L. Integrative use of spatial, genetic, and demographic analyses for investigating genetic connectivity between migratory, montane, and sedentary caribou herds. Mol. Ecol. 16, 422342 (2007) Halbert, N. D., Ward, T. J., Schnabel, R. D., Taylor, J.F. & Derr, J.N. Conservation genomics; disequilibrium mapping of domestic cattle chromosomal segments in North American bison populations. Mol. Ecol. 14, 23432362 (2005) Beaumont, M. Joint determination of topology, divergence time and immigration in population trees. pp. 134-154. In Simulations, Genetics and Human Prehistory, S. Matsumura, P. Forster and C. Renfrew (eds.) Cambridge: McDonald Institute for Archaeological Research (2008) Drummond, A. J. & Rambaut, A. BEAST: Bayesian evolutionary analysis by sampling trees. BMC Evol. Biol. 7, 214 (2007) Drummond, A. J., Ho, S. Y. W., Phillips, M. J. & Rambaut, A. Relaxed phylogenetics and dating with confidence. PLoS Biology 4, e88 (2006)
88 87 86 85 84 83 82 81 80 79
Rambaut, A. & Drummond, A. J. Tracer v1.4, Available from: http://beast.bio.ed.ac.uk/Tracer (2007). Accessed April 2010.
Minin, V. N., Bloomquist, E. W. & Suchard, M. A. Smooth skyride through a rough skyline: Bayesian coalescent-based inference of population dynamics. Mol. Biol. Evol. 25, 1459-1471 (2008) Pannell, J. R. Coalescence in a metapopulation with recurrent local extinction and recolonization. Evolution 57, 949-961 (2003) Navascus, M., Depaulis, F. & Emerson, B. C. Combining contemporary and ancient DNA in population genetic and phylogeographical studies. Mol. Ecol. Res. 10, 760-772 (2010) Ho, S. Y. W. & Shapiro, B. Skyline-plot methods for estimating demographic history from nucleotide sequences. Mol. Ecol. Res. 11, 423-432 (2011) Suchard, M. A., Weiss, R. E. & Sinsheimer, J. S. Bayesian selection of continuous-time Markov chain evolutionary models. Mol. Biol. Evol. 18, 1001-1013 (2001)
WWW.NATURE.COM/NATURE | 84
89
90
91
92
93
doi:10.1038/nature10574
94
Drummond, A. J., Rambaut, A., Shapiro, B. & Pybus, O. G. Bayesian coalescent inference of past population dynamics from molecular sequences. Mol. Biol. Evol. 22, 11851192 (2005)
Webb, T. J., Noble, D. & Freckleton, R. P. Abundanceoccupancy dynamics in a human dominated environment: linking interspecific and intraspecific trends in British farmland and woodland birds. J. Anim. Ecol. 76, 123 (2007) Borregaard, M. K. & Rahbek, C. Causality of the relationship between geographic distribution and species abundance. Q. Rev. Biol. 85, 3-25 (2010)
97
Ugan, 96 95
A. & Byers, D. Geographic and temporal trends in proboscidean and human radiocarbon histories during the late Pleistocene. Quaternary Sci. Rev. 26, 30583080 (2007) A. & Byers, D. A global perspective on the spatiotemporal pattern of the Late Pleistocene human and woolly mammoth radiocarbon record. Quatern. Int. 191, 6981 (2008)
98 Ugan,
Dye, T. & Komori, E. A pre-censal population history of Hawai'i. New Zeal. J. Archaeol. 14, 113128 (1992)
100
Ukkonen,
99
P., Arppe, L., Houmark-Nielsen, M., Kjr, K. H. & Karhu, J. A. MIS 3 mammoth remains from Swedenimplications for faunal history, palaeoclimate and glaciation chronology. Quaternary Sci. Rev. 26, 30813098 (2007)
101 Nadachowski,
A., Lipeckia, G., Wojtala, P. & Mikdo, B. (2011). Radiocarbon chronology of woolly mammoth (Mammuthus primigenius) from Poland. Quatern. Int. doi:10.1016/j.quaint.2011.03.011 Andel, T.H. & Davies, W.D. (editors). Neanderthals and Modern Humans in the European Landscape of the Last Glaciation - Archaeological Results of the Stage 3 Project. The McDonald Institute for Archaeological Research, Cambridge (2003)
102 van
Ermolova, N. M. Teriofauna Doliny Angary v Pozdnem Antropogene (Nauka, Novosibirsk, 1978) [in Russian] Germonpre, M. & Lbova, L. Mammalian remains from the Upper Palaeolithic Site Kamenka, Buryatia (Siberia). J. Archaeol. Sci. 23, 35-57 (1996) Ovodov, N. D. in Prirodnaia Sreda i Drevnii Chelovek v Pozdnem Antropogene (Nauka, UlanUde, 1987), pp. 12240 [in Russian] Ineshin, E. M., & Tetenkin, A. V. Chelovek i Prirodnaia Sreda Severa Baikalskoi Sibiri v Pozdnem Pleistotsene. Mestonakhozhdenie Bolshoi Iakor I (Nauka, Novosibirsk, 2010) [in Russian] Mochanov, IU. A. Drevneishie Etapy Zaseleniia Chelovekom Severo-Vostochnoi Azii (Nauka, Novosibirsk, 1977) [in Russian] Long, A., & Rippeteau, B. Testing contemporaneity and averaging radiocarbon dates. Am. Antiquity 39, 205-215 (1974)
108 107 106 105 104
103
WWW.NATURE.COM/NATURE | 85
doi:10.1038/nature10574
109 Hamilton, M. J., Buchanon, B. & ORourke, D. Archaeological Support for the Three-Stage Expansion of Modern Humans across Northeastern Eurasia and into the Americas. PLoS ONE 5, e12472 (2010) Bocherens, H., et al. New isotopic evidence for dietary habits of Neandertals from Belgium. J. Hum. Evol. 40, 497505 (2001) Adler, D., Bar-Oz, G., Belfer-Cohen, A. & Bar-Yosef, O. Ahead of the game: Middle and Upper Paleolithic hunting behaviors in the southern Caucasus. Curr. Anthr. 47, 89118 (2006)
112 111 110
Richards, M., et al. Isotopic dietary analysis of a Neanderthal and associated faunas from the site of Jonzac (Charente-Maritime), France. J. Hum. Evol. 55, 179185 (2008)
113
Stuiver, M. & Reimer, P. J. Extended 14C database and revised CALIB radiocarbon calibration program. Radiocarbon 35, 215230 (1993)
Surovell, T. A. & Brantingham, P. J. A note on the use of temporal frequency distributions in studies of prehistoric demography. J. Archaeol. Sci. 24, 18681877 (2007) Shackleton, N. J., Hall, M. A., &Vincent, E. Phase relationships between millennial-scale events 64,00024,000 years ago. Palaeoceanography 15, 565569 (2000) Shackleton, N. J., Fairbanks, R. G., Chiu, T. C. & Parrenin, F. Absolute calibration of the Greenland time scale: implications for Antarctic time scales and for d14C. Quaternary Sci. Rev. 23, 15131522 (2004) Surovell, T., Finley, J., Smith, G., Brantingham, P. & Kelly, R. Correcting temporal frequency distributions for taphonomic bias. J. Archaeol. Sci. 36, 17151724 (2009) Ballenger, J. & Mabry, J. Temporal frequency distributions of alluvium in the American Southwest: taphonomic, paleohydraulic, and demographic implications. J. Arch. Sci. 38, 13141325 (2011)
118 117 116 115
114
WWW.NATURE.COM/NATURE | 86
doi:10.1038/nature10574
Species BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS
AMS ID OxA-11169 OxA-11271 OxA-11618 Beta-1627 Beta 65662 OxA-11579 OxA-11610 OxA-11624 OxA-11165 OxA-11585 OxA-11589 OxA-11581 OxA-11614 OxA-11583 OxA-11622 OxA-11223 CAMS 53777 OxA-10542 n/a n/a CAMS 53772 OxA-11247 OxA-11139 n/a OxA-12068 OxA-11629 n/a CAMS 53901 OxA-11194 CAMS 53764 OxA-11959
LAT 44.50 44.50 53.28 51.08 66.65 49.63 53.34 51.08 70.52 51.08 49.63 49.63 49.63 51.08 51.08 66.26 70.81 64.06 65.46 64.84 70.81 60.81 64.06 64.84 44.50 69.90 64.84 70.81 71.16 70.81 68.20
LON -108.20 -108.20 -110.00 -114.08 -143.72 -110.21 -113.31 -114.08 -128.35 -114.08 -110.21 -110.21 -110.21 -114.08 -114.08 -161.35 -154.41 -141.89 -147.38 -147.96 -154.41 -149.43 -141.89 -147.96 -108.20 133.90 -147.96 -154.41 153.45 -154.41 157.67
Country U.S.A. U.S.A. Canada Canada U.S.A. Canada Canada Canada Canada Canada Canada Canada Canada Canada Canada U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. Russia U.S.A. U.S.A. Russia U.S.A. Russia
Locality Natural Trap Cave, WY Natural Trap Cave, WY Lloydminster, AB Hitching Post Ranch, Calgary, AB Black R. Yukon Flats, AK Stampede Site, Cypress Hills, AB Edmonton, AB Horse Hills Pit, Edmonton, AB Baillie Island, NWT Tuscany Site, Calgary, AB Stampede Site, Cypress Hills, AB Stampede Site, Cypress Hills, AB Stampede Site, Cypress Hills, AB Tuscany Site, Calgary, AB Tuscany Site, Calgary, AB Elephant Point, AK Ikpikpuk R., North Slope, AK Lost Chicken Cr., Chicken, AK Upper Cleary Cr., Fairbanks, AK Ester Cr., Fairbanks, AK Ikpikpuk R., North Slope, AK Seward Peninsula, Alder Cr., AK Lost Chicken Cr., Chicken, AK Ester Cr., Fairbanks, AK Natural Trap Cave, WY Yana-Indigirka lowland, Siberia Ester Cr., Fairbanks, AK Ikpikpuk R., North Slope, AK Kolyma lowland, Bol. Khomus-Yuryakh R., Siberia Ikpikpuk R., North Slope, AK Kolyma lowland, Siberia
Ref 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
WWW.NATURE.COM/NATURE | 87
doi:10.1038/nature10574
BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS BIS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS
CAMS 53899 OxA-11227 Beta 110938 OxA-11131 AECV:140 2c OxA-11193 CAMS 53758 CAMS 53768 n/a CAMS 53892 OxA-11613 n/a OxA-11991 CAMS 53894 CAMS 53782 CAMS 53900 OxA-11196 CAMS 53893 OxA-11224 CAMS 53769 CAMS 53779 OxA-10683 CAMS 53781 CAMS 53761 OxA-11275 GIN-3243 GIN-10256 Erl-11271 Erl-11268 UtC-11461 Erl-11266 ETH-9346 Erl-11267 Erl-11272 Erl-11265 KIA-35736 UtC-11460 UtC-11462 KIA-35913 KIA-4216 UtC-11453 KIA-30008 KIA 10334 ETH-11029 KIA 10336 K-2651-a
25,980 26,210 26,300 26,460 26,560 27,060 27,400 27,590 27,440 28,120 34,050 33,880 34,470 35,580 35,710 36,320 37,550 37,460 37,810 38,700 38,800 39,200 39,800 39,850 40,800 3,250 4,610 4,686 4,742 4,770 4,783 4,810 4,818 4,840 4,870 4,975 5,010 5,070 5,185 5,270 5,306 5,314 5,319 5,455 5,495 5,550
230 170 300 160 670 190 260 280 790 290 450 1,900 200 550 730 780 400 890 380 1,000 1,100 550 1,200 1,200 600 60 40 46 57 51 46 66 48 47 46 31 51 61 31 41 46 36 35 61 36 96
30,745 30,918 30,937 31,075 31,107 31,331 31,556 31,760 32,035 32,357 39,044 39,065 39,424 40,733 40,808 41,375 42,244 42,248 42,406 43,150 43,245 43,433 43,942 43,976 44,607 3,477 5,387 5,409 5,482 5,509 5,516 5,526 5,528 5,585 5,612 5,696 5,751 5,812 5,940 6,058 6,087 6,090 6,094 6,249 6,295 6,352
201 161 217 112 635 120 286 360 849 436 668 2,118 387 617 747 697 303 680 293 761 839 456 964 968 468 68 95 75 82 75 66 83 62 61 53 52 77 71 34 73 72 64 64 75 44 103
70.81 64.06 65.12 64.06 45.80 63.90 70.81 70.81 64.84 70.81 53.50 65.57 64.05 70.81 70.81 70.81 60.81 70.81 75.37 70.81 70.81 65.04 70.81 70.81 66.65 71.60 71.80 48.15 48.40 48.41 48.13 49.12 48.40 48.15 48.13 48.77 47.80 48.41 54.03 54.26 49.95 54.10 52.76 49.12 52.73 56.15
-154.41 -141.89 -153.34 -141.89 126.67 -139.00 -154.41 -154.41 -147.96 -154.41 -113.10 -148.38 -139.53 -154.41 -154.41 -154.41 -149.43 -154.41 135.59 -154.41 -154.41 -147.11 -154.41 -154.41 -143.72 87.00 129.30 10.95 11.45 9.91 10.96 8.58 11.45 10.95 10.96 2.44 9.10 9.91 10.50 10.76 9.63 10.81 18.15 8.58 18.68 10.11
U.S.A. U.S.A. U.S.A. U.S.A. China Canada U.S.A. U.S.A. U.S.A. U.S.A. Canada U.S.A. Canada U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. Russia U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. Russia Russia Germany Germany Germany Germany Germany Germany Germany Germany France Germany Germany Germany Germany Germany Germany Poland Germany Poland Denmark
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 3 3 4 3 5 3 3 3 6 4 4 7 8 4 6 9 10 9 11
WWW.NATURE.COM/NATURE | 88
doi:10.1038/nature10574
HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS HRS
KIA-35738 KIA-35737 KIA 9561 KIA 10344 KIA 10331 KIA 10335 OxA-1134 OxA-1131 KIA-35735 OxA-8996 OxA-9017 GIN-10233 GIN-10668 GIN-11133 Beta148659 GrA-17351 GIN-3140 GIN-8252 OxA-6927 OxA-1790 GIN-9879 GIN-11132 GIN-6426 Beta148660 GIN-1817a OxA-6928 GIN-3142 GIN-3841b GIN-8219 GIN-5732 GIN-10257 GIN-10232 GIN-11083 GIN-10268 GIN-3141v GIN-9043 GIN-10254 GIN-10667 GIN-10691 GIN-9873 GIN-10269 GIN-10252 GIN-3119 GIN-8221 Beta148622 GIN-10673 GIN-10661 GIN-6430
5,600 5,698 5,794 5,903 5,999 6,111 6,250 7,010 7,385 7,970 15,440 16,380 16,800 17,000 17,950 18,090 18,300 19,100 23,580 23,670 23,850 23,900 24,000 24,690 25,200 25,940 26,400 27,900 28,180 28,300 28,400 34,100 34,200 34,600 34,600 34,700 34,800 35,100 35,000 35,800 35,900 35,800 36,300 36,300 36,770 37,200 38,000 38,100
26 74 71 34 35 36 131 91 36 81 80 120 170 150 60 80 200 120 320 400 700 400 400 110 200 420 300 300 270 400 300 400 500 100 1,200 1,900 700 1,200 100 500 600 100 900 640 610 800 100 800
6,363 6,493 6,593 6,721 6,838 6,988 7,149 7,839 8,223 8,828 18,670 19,527 19,955 20,182 21,416 21,572 21,855 22,787 28,385 28,515 28,737 28,769 28,857 29,516 29,993 30,682 31,004 32,111 32,424 32,601 32,715 39,102 39,255 39,592 39,676 39,846 39,901 40,159 40,173 40,973 41,038 41,050 41,337 41,382 41,735 42,051 42,511 42,670
34 88 84 40 49 76 154 87 63 121 84 177 230 250 108 186 252 260 380 455 773 432 427 181 245 343 210 421 424 596 497 586 690 343 1,302 2,067 781 1,230 332 523 602 202 822 556 462 605 178 612
55.40 57.93 54.25 52.15 54.38 52.81 50.48 37.18 59.50 49.73 42.37 71.80 73.50 73.54 74.60 74.60 73.50 76.00 51.55 51.55 71.80 73.50 70.50 74.50 75.00 51.55 73.50 74.53 72.00 73.00 71.80 71.80 72.30 71.81 73.50 72.82 71.80 73.50 73.50 71.80 71.80 71.81 73.50 72.00 73.04 73.50 73.50 70.50
9.80 27.17 11.03 11.22 16.32 17.70 7.48 -3.24 26.57 2.15 1.84 129.30 142.20 100.49 102.60 108.70 104.00 138.00 -4.24 -4.24 129.30 100.40 120.00 100.50 100.00 -4.24 104.00 100.53 116.00 106.00 129.30 129.30 126.10 129.35 104.00 141.31 129.30 142.20 142.20 129.30 129.30 129.35 104.00 116.00 106.58 142.20 142.20 122.00
Denmark Estonia Germany Germany Poland Poland Germany Spain Estonia Belgium Spain Russia Russia Russia Russia Russia Russia Russia U.K. U.K. Russia Russia Russia Russia Russia U.K. Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia
Tybrind Vig Kpa Rosenhof Eilsleben Dbki Boejewice Niederbieber Cueva de la Carigela Kunda Place Saint-Lambert Montllas open air site, Prats Lena Delta, Bykovsky P N.S.I., Bol. Lyakhovsky Is Bol'shaya Balakhnaya Taimyr Lake Arlakh Lake Bolshaya Balakhnya R Taimyr Kotelniy Is N.S.I. Paviland Cave [Goat's Hole] Paviland Cave [Goat's Hole] Bykovsky Peninsula, N-E Siberia Bol'shaya Balakhnaya Anabar-Olenyok Taimyr L, Cape Sabler Engelgardt L Taimyr Paviland Cave [Goat's Hole] Bolshaya Balakhnya R Taimyr Taimyr L, Cape Sabler Anabar-Olenyok Khatanga R, Kozhevnikov B Bykovsky Peninsula, N-E Siberia Bykovsky Peninsula, N-E Siberia Lena R, Olenekskaya Channel, Byor-Khaya Lena Delta, Bykovsky P Bolshaya Balakhnya R Taimyr Laptev Sea coast (east) Bykovsky Peninsula, N-E Siberia N.S.I., Bol. Lyakhovsky Is N.S.I., Bol. Lyakhovsky Is Bykovsky Peninsula, N-E Siberia Bykovsky Peninsula, N-E Siberia Lena Delta, Bykovsky P Bolshaya Balakhnya R Taimyr Anabar-Olenyok Khatanga Talalakh L N.S.I., Bol. Lyakhovsky Is N.S.I., Bol. Lyakhovsky Is Anabar-Olenyok
6 6 9 9 9 9 12 13 6 14 15 2 2 2 2 2 2 2 16 16 2 2 2 2 2 16 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2
WWW.NATURE.COM/NATURE | 89
doi:10.1038/nature10574
HRS HRS HRS HRS HRS MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM
GIN-4965 GIN-3519 GIN-3823 GIN-11135 GIN-10693 GIN-3518 SOAN111A Ly-434 AA-27374 GIN-5370 GIN-8255 GIN-6023 OxA-719 OxA-716 Lu-358 SOAN4462 GIN6024bis Ki-1130 GIN-8258 SOAN5065 LU-1671 GrA-4891 LU-127 HV-1961 Hela-321 IM-887 SOAN3835 SOAN4804 AA 14866 IERZH-165 GIN-3130 SOAN4461 GIN-6093 GIN-2002 QC-886 SOAN4843 QC-621 SOAN4844 SOAN5044 SOAN5084 SOAN3504 GIN-8983 Gif-6418 Ki-1301 GIN-10908 GIN-7576 SOAN4418 SOAN3503 Gif-6419
38,700 39,600 40,200 40,400 41,000 14,800 14,800 14,850 14,940 15,100 15,000 15,130 15,100 15,100 15,110 15,150 15,200 15,300 15,400 15,420 15,420 15,560 15,660 15,810 15,910 16,000 16,000 16,130 16,168 16,300 16,330 16,320 16,300 16,300 16,565 16,700 16,960 17,050 17,100 17,200 17,220 17,290 17,320 17,400 17,450 17,500 17,610 17,600 17,720
1,000 500 1,200 100 1,600 50 150 350 170 70 70 50 200 250 530 280 80 140 100 215 100 200 180 410 155 300 385 310 209 300 100 450 600 700 270 240 420 160 390 230 245 100 290 150 100 300 200 500 290
43,150 43,709 44,221 44,334 44,994 18,000 18,036 18,063 18,213 18,248 18,250 18,256 18,281 18,283 18,291 18,318 18,438 18,563 18,647 18,654 18,660 18,770 18,846 19,042 19,108 19,161 19,185 19,270 19,280 19,465 19,486 19,528 19,564 19,607 19,763 19,870 20,217 20,248 20,388 20,493 20,530 20,553 20,684 20,752 20,784 20,851 20,957 20,989 21,091
761 417 998 170 1,524 227 266 403 228 164 162 168 226 270 572 302 189 211 115 263 106 227 211 414 178 295 389 334 252 357 164 507 682 805 308 277 537 282 515 367 376 249 396 268 240 389 304 631 416
70.50 73.60 73.30 73.60 73.50 72.12 54.50 44.90 43.00 65.00 72.36 68.45 51.75 52.83 53.33 59.24 68.45 44.15 71.00 61.05 75.00 48.52 52.67 50.07 60.37 62.42 58.50 55.29 64.40 54.10 75.30 59.24 55.92 53.33 51.70 59.24 51.70 59.38 52.85 55.92 52.01 57.23 42.17 44.15 57.22 70.11 57.30 52.01 42.17
134.50 118.00 97.00 101.13 142.20 104.00 80.20 1.02 104.00 171.00 139.73 150.00 33.08 30.97 34.33 62.34 150.00 131.78 -179.00 68.57 138.00 15.68 33.28 8.53 25.43 133.00 81.05 59.29 -147.00 61.40 105.00 62.34 92.33 34.33 36.00 62.34 36.00 62.33 86.68 92.33 86.32 112.25 2.74 131.78 111.83 75.46 112.00 86.32 2.74
Russia Russia Russia Russia Russia Russia Russia France Russia Russia Russia Russia Russia Russia Rus/Ukr Russia Russia Russia Russia Russia Russia Austria Russia Germany Finland Russia Russia Russia Alaska Russia Russia Russia Russia Rus/Ukr Russia Russia Russia Russia Russia Russia Russia Russia Spain Russia Russia Russia Russia Russia Spain
2 2 2 2 2 17 17 18 18 17 17 18 18 18 18 17 18 17 17 17 17 18 18 18 18 17 18 17 19 17 17 17 17 18 18 17 18 17 17 17 17 17 18 17 17 17 17 17 18
WWW.NATURE.COM/NATURE | 90
doi:10.1038/nature10574
MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM
SOAN4463 GIN-5042 GIN-11463 LE-1432A SOAN2211 KI-1055 SOAN3610 Beta148646 OxA-3694 Birm-1460 GIN-3727 TA-121 SOAN3838 GIN-8229 SOAN4945 SOAN4845 GIN-5046 GIN-6099 LU-361 LE-3834 LE-2950 SOAN4815 OxA-697 LE-2946B SOAN3609 OxA-718 GIN-2862 KI-1056 R-2533 Giff-1110 GIN-2859 UCIAMS11211 LU-654A OxA-10122 GIN-2861 SOAN4464 OxA-698 LE-2949 UtC-8137 GIN-3016 LU-688 GIN-7705 OxA-7112 LU-1970 LU-2807 GIN-8259 Hela-281 GIN-8257
17,810 17,780 17,800 17,930 18,050 18,020 18,040 18,190 18,160 18,000 18,300 18,320 18,250 18,500 18,580 18,600 18,680 18,700 18,690 18,930 19,010 18,990 19,000 19,200 19,190 19,200 18,600 19,280 19,400 19,300 19,500 19,530 19,640 19,700 19,700 19,710 19,800 19,860 19,910 19,960 19,970 19,900 19,980 19,990 20,000 22,400 22,420 22,400
320 80 100 100 95 600 175 60 260 1,400 200 280 1,100 120 240 230 120 100 770 320 120 340 300 200 310 350 2,000 600 230 700 200 80 330 500 200 205 350 200 130 80 110 800 220 110 110 300 315 200
21,224 21,266 21,277 21,400 21,524 21,530 21,569 21,722 21,738 21,818 21,855 21,873 21,952 22,087 22,136 22,163 22,292 22,312 22,392 22,641 22,655 22,723 22,733 22,901 22,918 22,936 22,946 23,064 23,106 23,108 23,271 23,355 23,460 23,541 23,561 23,572 23,664 23,737 23,780 23,834 23,852 23,855 23,878 23,880 23,894 27,066 27,090 27,096
465 202 214 163 190 773 281 179 337 1,847 252 333 1,421 224 348 343 194 161 981 438 265 447 400 298 391 435 2,902 737 348 870 337 233 460 636 311 315 474 281 214 184 205 1,045 292 206 206 474 482 403
59.27 70.00 54.50 54.25 56.00 49.63 55.64 73.60 49.42 53.26 55.27 65.02 61.05 75.26 55.85 59.30 71.40 70.00 52.85 55.15 51.29 57.07 50.55 56.58 55.64 45.82 55.20 49.63 44.38 52.44 53.55 39.89 79.90 43.35 53.55 59.38 52.00 51.29 73.53 55.05 79.47 53.00 51.55 75.00 71.00 71.00 62.85 71.00
62.33 125.00 80.20 69.73 65.92 31.40 88.00 100.48 20.17 -3.37 39.45 57.38 68.57 144.00 88.05 62.38 119.00 119.00 33.23 91.10 39.00 63.57 29.23 27.50 88.00 28.58 92.05 31.40 8.98 -2.83 92.00 -98.03 94.58 -5.83 92.00 62.33 33.27 39.00 105.82 90.00 96.75 103.50 -4.24 138.00 -179.00 -179.00 28.62 -179.00
Russia Russia Russia Russia Russia Russia Russia Russia Poland U.K. E_Europ e E_Europ e Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia E_Europ e Russia E_Europ e Russia Russia Italy U.K. Russia Kansas Russia Spain Russia Russia Russia Russia Russia Russia Russia Russia U.K. Russia Russia Russia Finland Russia
Rychkovo Lower Lena River Volchya Griva (2) Gagarino Shikaevka 2 Mezhirich Shestakovo Bolshaya Balakhnya River Oblazowa Cave Cae Gwyn Cave Zaraisk Byzovaya Lugovskoye Faddeyevsky Island Kochegur Berezovy Mys Bur River Amydai River Pogon Tarachikha Kostienki I Komsomolsky Randomyshl Leski Shestakovo Kirillovka Shlenka Mezhirich Arene Candidae, Schicht P7 Condover, Shrosphire Middle Yenisei River Lovewell Reservoir Oktyabrskoi Revolutsii Island Cueto de la Mina Middle Yenisei River Evalga Novgorod-Severskii Kostienki I Bolshaya Balachnya River Chulym River Oktyabrskoi Revolutsii Island Mal'ta (Belaya River) Paviland Cave [Goat's Hole] Kotelny Island Wrangel Island Wrangel Island Nilsia, Syvari Wrangel Island
17 17 17 18 17 18 17 17 20 18 18 18 17 17 17 17 17 17 18 17 21 17 18 18 17 18 17 18 18 18 17 22 17 18 17 17 18 21 17 17 17 17 16 17 17 17 18 17
WWW.NATURE.COM/NATURE | 91
doi:10.1038/nature10574
MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM
LU-104 SOAN4416 SOAN1467 SOAN4177 OxA-7108 LE-2969 LE-2800 GrA-15880 GIN-3089 OxA-4114 SOAN4802 GIN-8888 AA 14868 LE-3276 SOAN1386 Poz-124 GrN-6636 GIN-3232 AA 14864 GrA-5005 LE-3289 LE-3287 IERiZh-176 GrN-13235 Hela-282 LU-104 GIN-2763a GIN-5886 LU-358 LE-3283 Poz-1248 KIGN-397f Poz-1251 LE-2951 SOAN3634 GIN-7992 LE-2946A GIN-1296B AA 14881 GIN-8244 Poz-225 Poz-268 GIN-7166 IERiZh-63 OxA-7111 Beta148639 Beta148651 GrA-10935 SOAN-119 IM-835
22,410 22,480 22,450 22,500 22,620 22,700 22,760 22,750 22,750 22,780 22,860 22,900 23,015 23,010 22,990 23,020 23,040 23,100 23,222 23,180 23,260 23,260 23,300 23,330 23,340 23,430 23,500 23,600 23,660 23,640 23,750 23,670 23,770 23,770 23,760 23,800 23,770 23,800 23,808 23,940 23,980 24,000 24,000 24,000 24,140 24,170 24,250 24,360 24,400 24,400
200 420 200 280 340 250 250 160 150 250 410 240 449 300 170 180 170 200 453 120 680 420 500 110 350 180 300 200 270 320 140 410 160 200 245 150 1,540 400 487 150 280 300 1,100 1,500 400 110 110 150 650 650
27,115 27,136 27,178 27,200 27,293 27,383 27,451 27,463 27,468 27,478 27,549 27,667 27,754 27,798 27,829 27,861 27,888 27,951 28,025 28,036 28,047 28,077 28,114 28,168 28,170 28,230 28,302 28,352 28,451 28,477 28,507 28,517 28,545 28,566 28,573 28,584 28,659 28,665 28,684 28,780 28,840 28,857 28,882 28,904 28,969 28,987 29,066 29,183 29,250 29,250
400 569 385 438 475 383 385 324 320 387 541 393 588 439 334 335 316 323 587 205 838 543 633 176 435 202 351 257 343 1,050 246 465 271 305 335 265 1,824 441 529 254 331 342 1,188 1,745 429 226 234 253 668 668
55.64 57.30 55.64 55.64 51.55 51.29 51.29 55.64 74.03 51.39 57.68 53.00 64.94 51.29 55.64 50.07 50.07 67.35 64.94 48.32 51.42 51.29 61.00 55.64 60.39 52.83 73.06 59.00 53.33 51.29 50.07 54.87 50.07 51.29 55.12 43.00 56.58 74.50 64.87 75.26 50.07 50.07 70.50 66.00 51.55 75.50 72.25 55.90 52.63 72.47
88.00 112.00 88.00 88.00 -4.24 39.00 39.00 88.00 100.00 39.04 62.20 103.50 -147.65 39.00 88.00 19.95 19.95 116.00 -147.65 15.40 39.00 39.00 77.00 88.00 25.18 30.97 102.16 101.30 34.12 39.00 19.95 70.50 19.95 39.00 84.24 33.00 27.50 102.00 -146.84 144.00 19.95 19.95 134.23 67.00 -4.24 100.50 109.75 87.95 85.67 128.42
Russia Russia Russia Russia U.K. Russia Russia Russia Russia Russia Russia Russia Alaska Russia Russia Poland Poland Russia Alaska Austria Russia Russia Russia Russia Finland Russia Russia Russia Russia Russia Poland Russia Poland Russia Russia Russia E_Europ e Russia Alaska Russia Poland Poland Russia Russia U.K. Russia Russia Russia Russia Russia
Shestakovo Mama Tributary, Vitim Basin, Tesa R. Kiya River Tesa River Paviland Cave [Goat's Hole] Kostienki I Kostienki I Shestakovo Baskura Peninsula Kostienki XIV [Markina Gora] Tavda River Mal'ta (Belaya River) Goldstream Kostienki I Shestakovo Krakow-Spadzista Street Krakow-Spadzista Street Tyung River Goldstream Willendorf II Kostenki Kostenki Agansky Uval Shestakovo Helsinki, Toolo Berdyzh Bederbo-Tarida River Middle Angara River Khotylevo II Kostienki I Krakow-Spadzista Street Uspenka Krakow-Spadzista Street Kostienki I Kudelin Kluch Kostyonki Site, near Voronezh Leski Sabler Cape Gilmore Creek Faddeyevsky Island Krakow-Spadzista Street Krakow-Spadzista Street Omolon River, Kular Settlement 430th KM Paviland Cave [Goat's Hole] Trautfetter River Munchirdakh Lake Shestakovo Biya River Sobo-Sise Island
17 17 17 17 23 21 21 17 17 20 17 18 19 21 17 18 18 17 19 16 18 16 18 17 18 18 17 17 18 21 18 17 18 21 17 18 18 17 19 17 18 18 17 18 23 17 18 18 17 17
WWW.NATURE.COM/NATURE | 92
doi:10.1038/nature10574
MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM
Hela-295 K-3806 GIN-7999 LE-2624 SOAN4422 SOAN2712 SOAN4401 GIN-2160 IGAN-73 LU-749B GrA-13506 LE-612 K-3699 GIN-8227 GIN-8246 GIN-6143 GIN-3502 AA 14870 GIN-2210 K-3809 GIN-8532 GrA13238 Ly-1863 GIN-11465 Ox-A6190 Beta148634 GIN-4710B SR-6086 Beta148665 GIN-11127 Mo-215 LU-125 WB7-41 OxA-1205 GIN-1216 GIN-8224 OxA-3607 GIN-3836 PV-0175 GIN-2021b OxA-9039 AA-38235 GIN-3929 GIN-3505 GIN-5880 KI-1051 GIN-4710 K-4192 OxA-5229 Beta148662 GIN-8545 K-3808
24,450 24,400 24,500 24,600 24,600 24,650 24,650 24,900 24,960 24,960 25,040 25,100 25,110 25,180 25,200 25,300 25,300 25,362 25,400 25,480 25,540 25,660 25,800 25,800 25,700 25,800 25,800 26,000 26,100 26,200 26,000 26,470 26,560 26,700 26,700 27,100 27,150 27,300 26,695 27,200 27,460 27,470 27,500 27,500 27,700 27,500 28,000 27,810 27,950 28,270 28,300 28,120
385 900 450 150 730 305 340 500 400 210 200 500 440 150 180 400 600 584 300 560 170 200 700 600 260 130 200 120 170 150 1,600 420 550 550 700 300 350 200 1,300 500 310 310 300 300 500 800 200 610 550 210 350 760
29,256 29,273 29,324 29,442 29,459 29,504 29,504 29,781 29,850 29,860 29,909 29,953 29,962 29,981 29,992 30,108 30,109 30,166 30,214 30,275 30,396 30,509 30,515 30,532 30,542 30,615 30,623 30,777 30,839 30,919 30,951 31,025 31,086 31,193 31,238 31,375 31,427 31,452 31,533 31,574 31,653 31,663 31,685 31,685 32,049 32,096 32,183 32,232 32,316 32,541 32,585 32,600
458 896 513 232 722 410 436 524 418 245 235 492 435 220 235 395 548 527 344 496 245 235 593 497 279 165 202 155 169 152 1,890 294 462 489 701 231 315 179 1,443 510 361 364 360 360 578 861 360 709 674 377 535 871
63.80 61.67 51.29 47.65 57.23 50.50 56.30 74.03 53.33 79.52 52.45 72.50 56.72 75.26 75.26 56.85 70.45 64.40 68.92 56.02 75.26 55.90 47.14 70.72 53.00 72.83 75.26 46.99 74.03 73.60 71.05 52.67 43.11 53.26 73.60 75.26 52.10 73.35 49.35 71.02 56.13 68.33 73.28 70.45 50.00 51.75 75.26 56.28 48.40 76.00 71.00 59.20
23.48 9.68 39.00 31.10 112.25 72.75 90.40 100.00 34.12 96.92 128.11 87.00 10.12 144.00 144.00 53.23 131.00 -147.00 71.00 12.35 144.00 87.95 5.57 135.42 103.50 106.75 144.00 -104.19 100.00 100.48 127.30 33.28 128.91 -1.20 100.48 144.00 -7.50 97.00 117.58 79.20 40.48 161.50 97.88 131.00 38.00 33.08 144.00 8.80 9.77 113.00 66.50 17.73
Finland Norway Russia Russia Russia Kazakhst an Russia Russia Russia Russia Russia Russia Denmark Russia Russia E_Europ e Russia Alaska Russia Denmark Russia Russia France Russia Russia Russia Russia Montana Russia Russia Russia Russia China U.K. Russia Russia Ireland Russia China Russia Russia Russia Russia Russia Russia Russia Russia Denmark Germany Russia Russia Denmark
Lohtaja Kvam Kostienki VIII [Tel'manskaya site] Anetovka Kaverga River Batpak Achinsk Baskura Peninsula Khotylevo II Oktyabrskoi Revolutsii Island Uralovka Pyasina River Hadsund Faddeyevsky Island Faddeyevsky Island Lower Kama River Laptev Sea Coast Cleary Creek Yuribey River Ostrupgaard Faddeevsky Island Shestakovo Gr de la Mere Clochette Yana River Mal'ta (Belaya River) Sopochnaya Faddeyevsky Island Beaver Creek Baskura Peninsula Bolshaya Balakhnya River Chekurovka Yudinovo Mingyuegou Pin Hole Cave Bolshaya Balakhnya River Faddeyevsky Island Shandon Cave Logata River Zhalainuoer Yambuto Lake Sungir' [Vladimir] Maly Anui River Kubalakh River Laptev Sea Coast Sungir', Vladimir Region Mazin Faddeyevsky Island Stengardens Grusgrav 2 Das Geissenklosterle Taymyr Peninsula Yamal Peninsula Ronninge 1
18 18 16 18 17 17 17 17 18 17 17 17 18 17 17 18 17 19 17 18 17 18 16 18 18 17 17 24 17 17 17 18 17 25 17 17 26 17 17 17 27 17 17 17 18 18 17 18 18 18 17 18
WWW.NATURE.COM/NATURE | 93
doi:10.1038/nature10574
MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM
Beta148643 GIN-5696 GIN-8220 KIA-13081 SOAN2222 SOAN3440 OxA-5228 OxA-4235 OxA-6920 GIN-8254 WB-78-42 Birm-466 GIN-8247 GIN-8711 GIN-4434 OxA-1564 GIN-3821 GIN-3503 OxA-1610 GIN-8243 GIN-8223 OxA-10521 LU-504 Lu-879 GIN-8262 GIN-8238 GIN-3425 GIN-3822 SOAN1005 GIN-5751 GIN-8243a GIN-3122 Beta148630 GIN-6141 GIN-5750 Beta148666 GIN-6142 GIN-3231 GIN-3817 GIN-942 GIN-3118 GIN-2763B GIN-3136 GIN-6148 GIN-8250 SOAN1625 GIN-3476 GIN-3831 GIN-1491 Beta148664 GIN3120/P
28,310 28,400 28,400 28,400 27,615 28,525 28,500 34,100 34,100 34,400 34,310 34,500 34,500 34,600 34,700 34,850 35,000 35,000 35,200 35,210 35,800 35,800 35,830 36,000 35,900 36,000 36,000 36,200 36,450 36,600 36,700 36,800 36,950 37,000 37,000 37,080 37,300 37,600 38,300 38,000 38,400 38,500 38,500 38,400 38,500 38,460 38,800 38,900 38,800 39,050 39,100
170 300 340 200 2,015 200 550 840 1,200 400 1,850 500 500 470 400 1,500 500 300 1,600 500 700 690 630 1,550 500 500 500 500 420 500 500 500 450 500 500 460 1,000 400 600 1,500 700 500 600 1,000 900 1,100 400 600 1,300 580 1,000
32,606 32,715 32,719 32,732 32,913 32,917 32,919 39,124 39,141 39,454 39,465 39,594 39,594 39,686 39,768 39,921 40,092 40,114 40,263 40,335 40,908 40,911 40,960 41,025 41,067 41,158 41,158 41,325 41,518 41,617 41,685 41,753 41,851 41,888 41,888 41,939 42,148 42,276 42,761 42,779 42,858 42,883 42,906 42,935 42,981 43,004 43,087 43,212 43,298 43,325 43,432
344 497 550 370 2,611 374 764 1,063 1,357 533 2,026 623 623 588 536 1,572 617 480 1,657 615 712 703 644 1,534 500 477 477 437 335 384 377 373 336 369 369 342 794 303 473 1,288 551 414 486 764 691 842 361 494 1,031 478 758
74.32 62.45 73.55 48.38 53.00 50.11 48.40 52.22 50.42 75.00 43.11 51.87 75.26 70.08 68.45 51.84 73.35 70.45 51.17 75.26 75.63 44.83 68.58 57.91 73.60 75.26 55.00 73.35 71.20 72.10 75.26 75.30 74.42 56.85 73.37 72.50 56.85 21.38 73.35 72.16 73.50 73.06 73.06 56.85 69.60 50.90 73.08 73.35 75.63 74.42 73.30
100.33 150.30 118.50 9.75 104.40 118.00 9.77 -8.58 8.14 138.00 128.91 -1.68 144.00 135.33 150.45 -2.66 97.00 131.00 0.89 144.00 135.83 11.62 147.08 12.04 117.00 144.00 159.00 97.00 150.30 111.00 144.00 105.00 107.58 53.23 110.25 109.00 53.23 55.30 97.00 103.00 105.00 102.16 102.16 53.23 164.80 108.48 98.75 97.00 101.80 107.58 105.00
Russia Russia Russia Germany Russia Russia Germany Ireland Germany Russia China U.K. Russia Russia Russia U.K. Russia Russia U.K. Russia Russia Italy Russia Sweden Russia Russia Russia Russia Russia Russia Russia Russia Russia E_Europ e Russia Russia E_Europ e Belarus Russia Russia Russia Russia Russia E_Europ e Russia Russia Russia Russia Russia Russia Russia
Taymyr Lake Srednekan River Terpyi-Tumus Peninsula Sirgenstein Irkutsk Urtuiskoe Das Geissenklosterle Castlepook Cave Wildscheuer cave Kotelny Island Mingyuegou Little Rissington Faddeyevsky Island Mus-Khaya Duvanny Yar King Arthur's Cave Logata River Laptev Sea Coast Conningbrook Pit Faddeyevsky Island Belkovsky Island Settepolesini di Bodeno Terekhtyakh River Dosebacka mammoth Anabaro-Olenek interfluve Faddeevsky Island Kamchatka River, Nikolka Logata River Shandrin River Anabarka River Faddeyevsky Island Bolshaya Balachnya River Arylakh Lake Lower Kama River Semiriskay River Popigay River Lower Kama River Viliya River, Neman Logata River Khatanga River Bolshaya Balachnya River Bederbo-Tarida River Bederbo-Tarida River Lower Kama River Keinguveem River Kandabaevo Nemu-Dika-Tarida River Logata River Trautfetter River Arylakh Lake Bolshaya Balachnya River
17 17 17 18 17 17 18 26 28 17 18 18 17 17 17 16 17 18 29 17 17 18 17 18 17 17 17 17 17 17 17 18 17 18 17 17 18 18 17 17 17 17 17 18 17 17 17 17 17 17 17
WWW.NATURE.COM/NATURE | 94
doi:10.1038/nature10574
MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MAM MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX
GIN-3071 GIN11127a GrA-13487 GIN-3517 Beta148638 LU-718A LU-718B GIN-3135 GIN-5726A GIN-3804 GIN-11134 GIN-5025 LU-595 GIN1818/P Beta148645 MAG366A GIN-3407 Beta148648 MAG-576 MAG-366B GIN-2744B GIN-25529 OxA-17063 GIN-2945 AAR11733 I-10985 AAR11744 K-3365 K-3531 K-3362 K-3364 K-3366 OxA-17064 AAR12025 AAR12042 AAR11749 I-10919 AAR12082 AAR12085 AAR12081 AAR12084 BI 93435 AAR11711 Beta148653 AAR12061 AAR11728 OxA-17065
39,300 39,300 39,340 39,400 39,560 39,570 39,590 39,800 40,100 40,200 40,200 40,300 40,350 40,500 40,560 40,600 40,600 40,790 41,000 41,000 41,200 2,900 2,918 2,920 3,097 3,280 3,372 3,590 3,670 3,800 3,830 3,870 4,082 4,687 4,753 5,364 6,725 15,020 15,100 15,300 15,380 15,610 15,680 15,710 15,720 15,750 15,770
500 600 1,170 1,000 910 870 770 600 500 600 600 400 880 800 700 700 600 970 900 1,100 1,000 60 28 50 39 90 43 60 80 85 85 85 30 46 49 49 130 90 100 90 100 80 90 50 100 100 55
43,500 43,512 43,630 43,634 43,725 43,725 43,726 43,850 44,069 44,146 44,146 44,239 44,274 44,383 44,427 44,458 44,460 44,614 44,761 44,796 44,921 3,046 3,063 3,073 3,321 3,518 3,613 3,896 4,006 4,195 4,237 4,289 4,577 5,410 5,501 6,148 7,594 18,250 18,259 18,580 18,634 18,768 18,819 18,837 18,863 18,901 18,904
423 483 909 758 690 661 591 481 428 493 493 361 687 628 555 553 483 785 717 958 848 94 58 83 47 103 62 90 115 128 125 123 96 75 79 82 114 166 172 166 125 94 142 120 171 179 157
74.05 73.60 53.00 70.45 75.50 68.45 68.45 73.06 73.54 73.35 73.53 72.10 71.20 75.10 74.53 68.45 55.00 74.42 68.45 68.45 73.06 76.75 74.00 77.63 82.49 64.02 74.00 82.22 83.12 81.98 82.15 81.57 74.00 82.49 75.77 74.00 75.50 74.00 74.00 74.00 74.00 72.00 62.00 74.00 62.00 62.00 74.00
93.10 100.55 128.68 131.00 100.50 158.30 158.30 102.16 114.00 97.00 100.48 111.00 150.30 110.30 100.50 158.30 159.00 107.58 158.30 158.30 102.16 110.50 101.00 104.24 -35.86 -140.87 101.00 -33.37 -33.80 -26.12 -30.15 -61.55 101.00 -35.86 -99.78 101.00 -99.00 101.00 101.00 101.00 101.00 116.00 58.67 101.00 58.67 58.67 101.00
Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Greenlan d Canada Russia Greenlan d Greenlan d Greenlan d Greenlan d Greenlan d Russia Greenlan d Canada Russia Canada Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia
17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 17 30 31 30 31 32 31 32 32 32 32 32 31 31 31 31 32 31 31 31 31 33 31 30 31 31 31
WWW.NATURE.COM/NATURE | 95
doi:10.1038/nature10574
MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX
Beta148653 OxA-17068 AAR12077 GIN-3239 OxA-17072 AAR11747 AAR12052 OxA-17070 AAR11713 AAR12056 AAR11748 GIN-1815 GIN-3140v OxA-17074 AAR12054 AAR11717 BM-725 Beta148628 BI 48628 Beta148628 AAR12086 AAR11766 AAR12080 AAR11753 AAR12060 OxA-17078 Beta148627 BI 48627 AAR12073 Beta148627 AAR11759 Beta148654 Beta148652 Beta148654 OxA-17148 AAR12068 AAR12065 AAR11767 OxA-17075 Beta156194 OxA-17084 GrA-17605 BI 48652 AAR12071 AAR12064 OxA-17081 Beta-13869 AAR11768 AAR12069 OxA-17061 AAR12059
15,800 15,875 16,010 16,080 16,295 16,310 16,810 17,265 17,520 17,660 17,690 17,800 17,800 17,900 17,930 18,100 18,213 18,310 18,310 18,370 18,600 18,630 18,750 18,830 18,960 19,140 19,230 19,230 19,310 19,310 19,570 19,640 19,640 19,710 19,780 19,790 19,840 19,860 19,925 22,370 22,470 22,530 22,550 22,630 23,220 23,430 23,720 23,860 24,000 24,150 24,150
50 60 100 100 60 110 150 65 110 120 120 300 160 65 120 110 310 70 70 70 140 130 120 170 130 70 80 80 140 80 130 70 70 70 75 160 140 130 80 80 90 220 100 180 180 100 80 190 210 110 210
18,949 19,096 19,156 19,199 19,451 19,468 19,967 20,475 20,859 21,056 21,103 21,207 21,236 21,380 21,401 21,624 21,781 21,854 21,854 21,917 22,197 22,237 22,357 22,460 22,594 22,816 22,922 22,922 23,008 23,016 23,404 23,500 23,500 23,578 23,644 23,659 23,706 23,725 23,790 27,029 27,250 27,251 27,297 27,333 28,079 28,230 28,457 28,693 28,855 28,969 28,970
159 149 153 158 124 184 218 229 249 266 265 445 288 119 202 210 388 187 187 196 233 216 207 294 269 242 237 237 267 235 267 179 179 160 155 238 214 206 178 335 291 377 278 325 242 166 177 294 281 224 268
74.54 74.00 74.00 71.60 74.00 62.00 74.00 69.58 62.00 59.42 74.00 75.98 74.54 74.00 59.42 74.00 52.23 74.00 74.54 74.54 74.00 62.00 74.00 74.00 62.00 62.00 74.00 74.54 74.00 74.54 74.00 73.54 74.54 73.54 74.00 74.00 74.00 74.00 74.54 74.54 62.00 72.40 74.00 74.00 74.00 59.42 52.65 62.00 74.00 73.30 62.00
101.64 101.00 101.00 87.00 101.00 58.67 101.00 -139.08 58.67 57.77 101.00 99.78 101.64 101.00 57.77 101.00 -0.82 101.00 101.64 101.64 101.00 58.67 101.00 101.00 58.67 58.67 101.00 101.64 101.00 101.64 101.00 100.49 101.64 100.49 101.00 101.00 101.00 101.00 101.64 101.64 58.67 106.00 101.00 101.00 101.00 57.77 -113.65 58.67 101.00 141.30 58.67
Russia Russia Russia Russia Russia Russia Russia Canada Russia Russia Russia Russia Russia Russia Russia Russia U.K. Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Canada Russia Russia Russia Russia
Taimyr Lake, Cape Sabler Taimyr Taimyr Agapa River, Taimyr Taimyr Medvezhya Cave, Urals Taimyr Herschel Island Medvezhya Cave, Urals Tayn Cave, Urals Taimyr Lower Taimyr R Taimyr Lake, Cape Sabler Taimyr Tayn Cave, Urals Taimyr Clifford Hill, Northamptonshire Taimyr Taimyr L., Taimyr Pen. Taimyr L., S of Sabler C Taimyr Medvezhya Cave, Urals Taimyr Taimyr Medvezhya Cave, Urals Medvezhya Cave, Urals Taimyr Taimyr L., Taimyr Pen. Taimyr Taimyr Lake, Cape Sabler Taimyr Bol'shaya Balakhnaya Taimyr L., coast opposite Kupffer Is Bol'shaya Balakhnaya Taimyr Taimyr Taimyr Taimyr Taimyr Lake, Cape Sabler Taimyr L., coast opposite Kupffer Is Medvezhya Cave, Urals Popygai R Taimyr Taimyr Taimyr Tayn Cave, Urals Gee Pits at Ponoka, AB Medvezhya Cave, Urals Taimyr Bolshoy Lyakhovsky Island Medvezhya Cave, Urals
30 31 31 30 31 31 31 31 31 31 31 30 30 31 31 31 34 33 33 30 31 31 31 31 31 31 33 33 31 30 31 30 33 30 31 31 31 31 31 30 31 30 33 31 31 31 32 31 31 31 31
WWW.NATURE.COM/NATURE | 96
doi:10.1038/nature10574
MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX MOX RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD
AAR12053 AAR11745 OxA-17080 Beta148657 OxA-17082 AAR11770 AAR11721 OxA-17150 AAR11715 AAR11716 Beta148656 AAR11762 AAR-4188 AAR12076 OxA-17077 UtC-10156 AAR12023 AAR11740 AAR11734 Beta173287 BI 93436 GX21987G GX20445G Ua-3104 Ua-3103 OxA-2790 OxA-4012 Ua-3105 GX21285G OxA-3884 Ua-3106 St-13090 GX21986G GX21982G GX20444G GX21984G GX21985G GX21988G n/a OxA-9060 VERA3345 n/a n/a OxA-7500 OxA-7501 OxA-4125 OxA-6808
24,160 24,270 24,310 24,660 24,870 24,940 25,310 25,300 25,490 26,700 27,440 27,500 28,490 34,150 35,830 36,700 38,350 39,000 39,900 40,220 40,270 2,925 3,030 3,140 3,295 3,350 3,435 3,605 3,640 3,665 3,777 3,870 3,925 4,240 4,280 4,425 5,500 5,660 14,930 15,240 15,460 17,320 17,720 17,820 18,220 18,510 18,670
210 160 110 110 110 230 240 110 230 350 150 300 350 600 240 700 900 700 1,000 670 450 80 65 95 95 70 70 105 135 75 70 70 85 85 65 85 110 100 70 100 60 290 290 200 180 200 160
28,977 29,073 29,142 29,491 29,731 29,847 30,091 30,092 30,333 31,177 31,530 31,685 32,855 39,196 41,057 41,684 42,873 43,308 43,966 44,163 44,210 3,085 3,233 3,357 3,534 3,588 3,698 3,920 3,974 3,998 4,158 4,292 4,356 4,756 4,851 5,052 6,295 6,458 18,264 18,516 18,680 20,684 21,091 21,248 21,792 22,053 22,271
268 252 238 173 217 260 292 230 308 228 174 360 586 824 276 551 690 554 770 538 396 116 93 120 110 90 94 148 189 108 114 102 132 130 108 135 129 108 183 196 74 396 416 342 243 281 264
69.58 74.00 62.00 74.14 59.42 74.00 74.00 74.00 74.00 59.42 74.54 62.00 56.02 74.00 74.00 74.92 64.05 69.35 62.00 69.90 73.30 80.25 80.34 80.80 80.80 55.73 55.73 80.80 80.34 55.73 80.80 80.80 80.25 80.25 81.08 80.25 80.25 80.25 43.43 48.45 46.70 42.12 42.12 50.08 50.08 48.25 50.08
-139.08 101.00 58.67 98.44 57.77 101.00 101.00 101.00 101.00 57.77 101.64 58.67 12.18 101.00 101.00 106.58 -139.44 154.97 58.67 -131.17 141.30 53.12 52.46 48.06 48.06 13.50 13.50 48.06 52.46 13.50 48.06 48.06 53.12 53.12 65.05 53.12 53.12 53.12 -5.05 27.47 14.73 2.77 2.77 8.33 8.32 27.17 8.32
Canada Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Denmark Russia Russia Russia Canada Russia Russia Canada Russia Russia Russia Russia Russia Sweden Sweden Russia Russia Sweden Russia Russia Russia Russia Russia Russia Russia Russia Spain Ukraine Austria Spain Spain Germany Germany Moldova Germany
Herschel Island Taimyr Medvezhya Cave, Urals Upper Taimyr R Tayn Cave, Urals Taimyr Taimyr Taimyr Taimyr Tayn Cave, Urals Taimyr Lake, Cape Sabler Medvezhya Cave, Urals Bannebjerg, Helsinge, N. Sjlland Taimyr Taimyr Bikada R Klondike Omoloy R., Yana Lowland Medvezhya Cave, Urals McKinley Bay, Northwest Territories Bol'shoi Lyakhovski I., New Siberian Is. Hooker Island, Frans Josef Land Scot Keltie, Frans Josef Land Alexander Island, Frans Josef Land Alexander Island, Frans Josef Land Harlosa Harlosa Alexander Island, Frans Josef Land Scot Keltie, Frans Josef Land Harlosa Alexander Island, Frans Josef Land Alexander Island, Frans Josef Land Hooker Island, Frans Josef Land Hooker Island, Frans Josef Land Graham Bell Island, Frans Josef Land Hooker Island, Frans Josef Land Hooker Island, Frans Josef Land Hooker Island, Frans Josef Land Tito Bustillo, Asturias Molodovo Griffener Tropfsteinhhle L'Arbreda, Gerona L'Arbreda, Gerona Wiesbaden-Igstadt Wiesbaden-Igstadt Ciuntu Wiesbaden-Igstadt
31 31 31 30 31 31 31 31 31 31 30 31 35 31 31 30 31 31 31 32 33 36 36 36 36 37 37 36 36 37 36 36 36 36 36 36 36 36 38 45 45 38 38 28 28 45 45
WWW.NATURE.COM/NATURE | 97
doi:10.1038/nature10574
RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD RD WR WR WR WR WR WR WR WR WR WR WR WR WR
OxA-6809 GIN-9875 GIN-9888 OxA-4118 OxA-6985 OxA-1490 OxA-8309 OxA-203 OxA-5805 SRR-2104 OxA-5721 SRR-2103 GIN-9869 OxA-6226 OxA-6594 OxA-6227 OxA-5226 OxA-4122 OxA-4855 OxA-4436 OxA-5693 OxA-X2199-15 OxA-3705 OxA-5227 OxA-3984 OxA-7391 OxA-6433 OxA-4236 Erl-6746 OxA-3417 OxA-3406 OxA-11980 OxA-13598 OxA-2032 OxA-4230 OxA-7870 OxA-13888 OxA-11797 GIN-6024 GIN-6020 GIN-9594 A.Lister/A Stuart in prep IM-239 A.Lister/A Stuart in prep GIN-6005 GIN-3209 GIN-6018 A.Lister/A Stuart in prep OxA-3449 OxA-3450 OxA-14715
18,670 18,850 18,900 19,220 19,980 23,340 23,420 23,980 24,560 24,590 24,680 25,360 26,060 26,200 26,320 26,500 26,540 26,600 27,000 27,780 27,820 27,940 28,000 28,050 28,240 28,340 34,950 35,200 35,499 37,200 37,450 37,760 37,900 38,000 38,650 38,800 40,000 40,650 15,130 15,850 19,500
160 360 600 180 220 350 220 320 340 790 360 810 190 600 360 550 460 370 550 400 500 390 370 550 390 420 950 950 436 1,300 1,050 340 1,000 1,000 1,400 1,400 700 500 90 80 120
22,271 22,534 22,630 22,917 23,878 28,170 28,226 28,839 29,388 29,452 29,542 30,138 30,805 30,818 30,934 31,041 31,069 31,117 31,431 32,051 32,162 32,207 32,258 32,409 32,524 32,660 40,033 40,269 40,692 42,106 42,268 42,371 42,580 42,651 43,223 43,329 43,998 44,500 18,266 19,072 23,295 28,327
264 494 753 286 292 435 239 360 438 771 446 727 180 488 250 452 341 249 542 491 598 504 494 693 568 629 1,002 967 527 1,128 833 271 768 767 1,143 1,145 552 408 175 160 271
50.08 71.79 71.79 48.08 51.55 51.67 50.22 46.17 51.23 58.11 48.92 58.11 71.79 51.67 50.22 51.67 48.40 48.08 48.40 50.47 50.47 48.87 52.17 48.40 58.11 50.42 46.17 52.22 51.76 53.26 53.26 53.26 50.47 54.07 52.22 46.17 50.47 53.26 71.16 69.87 62.00 50.45
8.33 129.40 129.40 27.25 -4.25 -4.73 4.83 4.73 -2.67 -4.94 11.83 -4.94 129.40 -4.73 4.90 -4.73 9.77 27.25 9.77 -3.50 -3.50 15.47 -8.42 9.77 -4.94 8.13 4.73 -8.58 10.84 -1.20 -1.19 -1.19 -3.50 -1.87 -8.58 4.73 -3.50 -1.19 153.45 147.58 132.50 5.00 133.75 148.16 157.67 162.17 162.17 31.10 -3.50 -3.50 -3.50
Germany Russia Russia Moldova U.K. U.K. Belgium France U.K. U.K. Germany U.K. Russia U.K. Belgium U.K. Germany Moldova Germany U.K. U.K. Austria Ireland Germany U.K. Germany France Ireland Germany U.K. U.K. U.K. U.K. U.K. Ireland France U.K. U.K. Russia Russia Russia Belgium Russia Russia Russia Russia Russia Russia U.K. U.K. U.K.
Wiesbaden-Igstadt Lena Delta, Bykovsky P Lena Delta, Bykovsky P Brinzeni I Paviland, Gower Little Hoyle Trou da Somme (Hastiere) Vergisson II Hyena Den, Wookey Hole Reindeer cave Klausenhohlen Reindeer cave Lena Delta, Bykovsky P Hoyle's Mouth Trou Magrite Hoyle's Mouth Geissen Klosterle Cave Brinzeni I Geissenklosterle Kent's Cavern Kent's Cavern, Devon Alberndorf Foley Cave Geissenklsterle Reindeer cave Wildscheuer Cave Vergisson II Castlepook Cave Baumannshhle Cresswell Crags Pin Hole cave Pin Hole cave Kent's Cavern Stump Cross Cave Castlepook Cave Vergisson II Kent's Cavern Pin Hole cave Bolshoi Khomus-Yuriakh River Indigirka River Churapcha Third Cave, Goyet Ikhine 2 Magadan Kolyma River Maly Anui River Maly Anui River Lyuban', Novgorod Region Kent's Cavern, Devon Kent's Cavern, Devon Kent's Cavern, Devon
45 2 2 20 16 39 40 45 41 45 42 45 2 45 45 45 16 20 45 41 41 45 42 16 43 28 45 42 45 43 43 45 45 45 42 45 45 45 44 44 44 45 44 45 44 44 44 45 16 16 46
26,030
200
30,783 30,903
185
63.17 62.77
WWW.NATURE.COM/NATURE | 98
doi:10.1038/nature10574
WR WR WR WR WR WR WR WR WR WR
OxA-14701 OxA-13921 OxA-14201 GIN-6009 OxA-13965 OxA-14196 GIN-6011 GIN-6012 OxA-10804 OxA-15484
35,650 36,040 36,370 37,100 37,200 37,540 39,900 40,000 40,200 40,550
330 330 320 1,100 550 370 500 500 700 400
40,877 41,223 41,469 42,004 42,027 42,235 43,916 43,991 44,148 44,430
378 319 272 920 400 286 422 425 557 343
50.46 50.46 50.46 69.03 50.46 53.29 67.27 67.20 51.29 53.26
-3.50 -3.50 -3.50 156.00 -3.50 -1.19 155.87 132.90 -2.85 -1.20
U.K. U.K. U.K. Russia U.K. U.K. Russia Russia U.K. U.K.
Kent's Cavern, Devon Kent's Cavern, Devon Kent's Cavern, Devon Bolshaya Chukochya River Kent's Cavern, Devon Ash Tree Cave, Derbyshire Dzhelon-Siene Yana River headwaters Picken's Hole, Somerset Robin Hood Cave, Creswell
46 46 46 44 46 46 44 44 46 46
Shapiro, B., et al. Rise and fall of the Beringian steppe bison. Science 306, 1561-1565 (2004)
Sher, A.V., Kuzmina, S.A., Kuznetsova, T.V. & Sulerzhitsky, L.D. New insights into the Weichselian environment and climate of the East Siberian Arctic, derived from fossil insects, plants, and mammals. Quaternary Sci. Rev. 24, 533-569 (2005) Metzger, M., Obermaier, H., Schlager, S., Weber, C. & Steppan, K. Jungsteinzeitliche Wildpferde in Sddeutschland Palogenetik, Morphometrie und Nahrungskologie. Beitrge zur Archozoologie und Prhistorischen Anthropologie 7, 31-40 (2009) Steppan, K. Die Tierknochenfunde aus der Schicht 9 von Sipplingen-Osthafen. In: Kieselbach, P. & Kolb, M. (eds.): Siedlungen der Pfyner Kultur im Osten der Pfahlbaubucht von Sipplingen, Bodenseekreis. Band 2: Naturwissenschaftliche Untersuchungen. Hemmenhofener Skripte 4/2, 8796 (2004) Steppan, K. Mittelholozne Wildpferde am nrdlichen Oberrhein? Beitrge zur Archozoologie und Prhistorischen Anthropologie. Forschungen und Berichte zur Vor- und Frhgeschichte in Baden-Wrttemberg 53, 251-255 (1994) Sommer, R. S., Benecke, N., Lougas, L., Nelle, O. & Schmlcke, U. Holocene colonization pattern of the wild horse (Equus ferus) in Europe: a matter of landscape openness? J. Quat. Sci., accepted. (2011)
Lage, W. Schleifknochen versus Schabbahnknochen: Untersuchungen zur Verwendung steinzeitlicher Langknochen von Grosugern mit konkaven Arbeitsbahnen. Schriften des Naturwissenschaftlichen Vereins fr Schleswig-Holstein 71, 26-40 (2009)
8
Hartz S. & Lbke, H. New Evidence for a Chronostratigraphic Division of the Erteblle Culture and the Earliest Funnel Beaker Culture on the Southern Mecklenburg Bay. In: Kind, C.-J. (Ed.), After the Ice Age. Settlements, subsistence and social development in the Mesolithic of Central Europe. Konrad Theiss Verlag, Stuttgart, 59-74 (2006) Benecke, N. Zu den Anfngen der Pferdehaltung in Eurasien. Aktuelle archozoologische Beitrge aus drei Regionen. Ethnographisch-Archologische Zeitschrift 43, 186-226 (2002)
Steppan, K. Taphonomie - Zoologie - Chronologie - Technologie - konomie. Die Sugetierreste aus den jungsteinzeitlichen Grabenwerken in Bruchsal/Landkreis Karlsruhe. Materialhefte zur Archologie in Baden-Wrttemberg 66. Stuttgart (2003)
10
WWW.NATURE.COM/NATURE | 99
doi:10.1038/nature10574
11 12
Davidsen, K. The Final TRB Culture in Denmark. Akademisk Forlag, Copenhagen (1978)
Street, M. Analysis of Late Palaeolithic and Mesolithic faunal Assemblages in the northern Rhineland, Germany. Ph.D. thesis. University of Birmingham (1993) Hedges, R.E.M., Housley, R. A., Law, I.A., Perry, C. & Gowlett, J.A.J. Radiocarbon Dates Ffrom the Oxford AMS system: Archaeometry Datalist 6. Archaeometry 29, 289-306 (1987) Stevens, R.E. & Hedges, R.E.M. Carbon and nitrogen stable isotope analysis of northwest European HRS bone and tooth collagen, 40,000 BP - present: Palaeoclimatic interpretations. Quaternary Sci. Rev. 23, 977-991 (2004) Ramsey, C.B., Pettitt, P.B., Hedges, R.E.M., Hodgins, G.W.L. & Owen, D.C. Radiocarbon dates from the Oxford AMS system: Archaeometry Datelist 30. Archaeometry 42, 459-479 (2000) Stage Three Project Database (http://www.esc.cam.ac.uk/oistage3/Details/Homepage.html) Accessed 5 Jan 2011 Kuzmin, Y.V. & Orlova, L.A. Radiocarbon chronology and environment of woolly mammoth (Mammuthus primigenius Blum.) in northern Asia: results and perspectives. Earth-Sci. Rev. 68, 133169 (2004) Ugan, A. & Byers, D. Geographic and temporal trends in proboscidean and human radiocarbon histories during the late Pleistocene. Quaternary Sci. Rev. 26, 30583080 (2007) Barnes, I., Shapiro, B., Lister, A., Kuznetsova, T., Sher, A., Guthrie, D. & Thomas, M.G. Genetic structure and extinction of the woolly mammoth, Mammuthus primigenius. Curr. Biol. 17, 10721075 (2007) Hedges, R.E.M., Housley, R.A., Pettitt, P.B., Ramsey, C.B. & Van Klinken, G.J. Radiocarbon dates from the Oxford AMS System: Archaeometry datelist 21. Archaeometry 38, 181-207 (1996) Velichko, A.A. & Zelikson, E.M. Landscape, climate and mammoth food resources in the East European Plain during the Late Paleolithic epoch. Quatern. Int. 126, 137-151 (2005) Holen, S.R. The age and taphonomy of mammoth (Mammuthus) at Lovewell Reservoir, Jewell County, Kansas, USA. Quatern. Int. 169-170, 51-63 (2007) Jacobi, R.M. & Higham, T.F.G. The Red Lady ages gracefully: new ultrafiltration AMS determinations from Paviland. J. Hum. Evol. 55, 898907 (2008)
24 23 22 21 20 19 18 17 16 15 14 13
Hill, C.L. Stratigraphic and geochronologic contexts of mammoth (Mammuthus) and other Pleistocene fauna, Upper Missouri Basin (northern Great Plains and Rocky Mountains), U.S.A. Quatern. Int. 142143, 87106 (2006)
Hedges, R.E.M., Housley, R.A., Law, I.A. & Perry, C. Radiocarbon dates from the Oxford AMS System: Archaeometry datelist 7. Archaeometry 30, 155-164 (1988)
26
25
Woodman, P., McCarthy, M. & Monaghan, N. The Irish Quaternary fauna project. Quaternary Sci. Rev. 16, 129-159 (1997)
WWW.NATURE.COM/NATURE | 100
doi:10.1038/nature10574
Kuzmin, Y. V., Burr, G. S., Jull, A. J. T., and Sulerzhitsky, L. D., AMS 14C age of the Upper Palaeolithic skeletons from Sungir site, Central Russian Plain. Nucl. Instrum. Meth. B 223, 731734 (2004) Hedges, R.E.M., Pettitt, P.B., Ramsey, C.B. & Van Klinken, G.J. Radiocarbon dates from the Oxford AMS System: Archaeometry datelist 26. Archaeometry 40, 437-455 (1998) Hedges, R.E.M., Housley, R.A., Law, I.A. & Bronk, C.R. Radiocarbon dates from the Oxford AMS System: Archaeometry Datelist 9. Archaeometry 31, 207-234 (1989) MacPhee, R.D.E., et. al. Radiocarbon chronologies and extinction dynamics of the Late Quaternary mammalian megafauna of the Taimyr Peninsula, Russian Federation. J. Archaeol. Sci. 29, 1017-1042 (2002) Campos, P. F. et al. Ancient DNA analyses exclude humans as the driving force behind late Pleistocene musk ox (Ovibos moschatus) population dynamics. P. Nat. Acad. Sci. 107, 56755680 (2010) Harington, C.R. Annotated Bibliography of Quaternary Vertebrates of Northern North America. University of Toronto Press, Toronto, Canada. 539 pp. MacPhee, R.D.E., Tikhonov, A.N., Mol, D. & Greenwood, A.D. Late Quaternary loss of genetic diversity in muskox (Ovibos). BMC Evolutionary Biology 5, 49 (2005) Stuart, A. J. Mammalian extinctions in the Late Pleistocene of Northern Eurasia and North America. Biol. Rev. (Camb.) 66, 453-562 (1991) Aaris-Srensen, K. Diversity and dynamics of the mammalian fauna in Denmark throughout the last glacial-interglacial cycle, 115-0 kyr BP. Fossils and Strata 57, 1-59 (2009) Forman, S.L., Lubinski, D. & Weihe, R.R. The Holocene occurrence of reindeer on Franz Josef Land, Russia. Holocene 10, 763-768 (2000) Hedges, R.E.M., Housley, R.A., P.B., Ramsey, C.B. & Van Klinken, G.J. Radiocarbon dates from the Oxford AMS System: Archaeometry Datelist 20. Archaeometry 37, 417-430 (1995) lvarez-Lao, D.J. & Garca, N. Geographical distribution of Pleistocene cold-adapted large mammal faunas in the Iberian Peninsula, Quatern. Int. 233, 159-170 (2011) Hedges, R.E.M., Housley, R.A., Ramsey, C.B. & and Van Klinken, G.J. Radiocarbon dates from the Oxford AMS system: Archaeometry Datelist 17. Archaeometry 35, 305-326 (1993) Ramsey, C.B., Higham, T.H.F., Owen, D.C., Pike, A.W.G. & Hedges, R.E.M. Radiocarbon dates from the Oxford AMS system: Archaeometry Datelist 31. Archaeometry 44, 1-149 (2002) Hedges, R.E.M., Pettitt, P.B., Ramsey, C.B. & Van Klinken, G.J.. Radiocarbon dates from the Oxford AMS system: Archaeometry Datelist 22. Archaeometry 38, 391-415 (1996) Hedges, R.E.M., Pettitt, P.B., Ramsey, C.B. & Van Klinken, G.J. Radiocarbon dates from the Oxford AMS system: Archaeometry datelist 23. Archaeometry 39, 247-262 (1997)
42 41 40 39 38 37 36 35 34 33 32 31 30 29 28
27
WWW.NATURE.COM/NATURE | 101
doi:10.1038/nature10574
Hedges, R.E.M., Housley, R.A., Ramsey, C.B. & van Klinken, G.J. Radiocarbon dates from the Oxford AMS system: Archaeometry Datelist 18. Archaeometry 36, 337-374 (1994) Orlova, L.A., Kuzmin, Y.V. & Dementiev, V.N. A review of the evidence for extinction chronologies for five species of Upper Pleistocene megafauna in Siberia. Radiocarbon 46, 301-314 (2004)
45 46 44
43
Jacobi, R.M., Rose, J., MacLeod, A. & Higham, T.F.G. Revised radiocarbon ages on woolly rhinoceros (Coelodonta antiquitatis) from western central Scotland: significance for timing the extinction of woolly rhinoceros in Britain and the onset of the LGM in central Scotland. Quaternary Sci. Rev. 28, 25512556 (2009)
WWW.NATURE.COM/NATURE | 102
doi:10.1038/nature10574
Supplementary Table S6.2. Woolly rhinoceros (Coelodonta antiquitatis) sample information, listed by calibrated radiocarbon age. Data include radiocarbon age, locality information, and institution currently housing the sample. BC denotes radiocarbon dates beyond the calibration curve. Information on new radiocarbon dates and the GenBank accession numbers of new sequences (JN570760JN570863) are included. Institution abbreviations used in Supplementary Tables S6.2, S6.3 and S6.4 are listed below. AMNH: American Museum of Natural History, New York, USA CGG: Center for GeoGenetics, Natural History Museum, University of Copenhagen, Denmark CMC: Canadian Museum of Civilization, Gatineau, Quebec, Canada CMN: Canadian Museum of Nature, Gatineau, Quebec, Canada EPQ: Department Of Early Prehistory and Quaternary Ecology, Tuebingen GIN RAS: Geological Institute, Moscow, Russian Academy of Sciences, Russia GYW: Government of Yukon, Dept. Turism and Culture, Whitehorse IPAE RAS: Zoological museum of Institute of Plant and Animal Ecology, Ekaterinburg, Russian Academy of Sciences, Russia KIC: Khatanga Ice Cave, Taimyr Peninsula, Russia KU: Kansas University MPI EVA: Max Plank Institute, Leipzig, Germany PIN RAS: Paleontological Institute, Moscow, Russian Academy of Sciences, Russia ZIN RAS: Zoological Institute, St. Petersburg, Russian Academy of Sciences, Russia ZMK: Zoological Museum, University of Copenhagen, Denmark
AMS ID AAR-11027 OxA-20097 OxA-20096 AAR-11028 OxA-15913 AAR-11029 AAR-11042 New date x x x x x x x Sample ID 321 169-38 164-59 313/1084 3658-3 600/398 436-1 Museum GIN RAS PIN RAS PIN RAS GIN RAS PIN RAS GIN RAS GIN RAS Lab ID WR198 WR320 WR319 WR199 WR075 WR200 WR222 New seq JN570894 JN570876 JN570895 14C date 12,190 12,280 12,355 12,460 12,550 12,650 12,675 14C SE 60 45 50 90 50 65 65 IntCal09 date 14,040 14,164 14,384 14,572 14,763 14,961 15,006 IntCal09 SE 141 210 229 259 224 209 205 LAT 61.00 68.70 68.23 70.20 67.64 69.20 59.91 LON 130.00 158.70 161.92 126.00 146.77 123.00 56.35 Country Russia Russia Russia Russia Russia Russia Russia Region East Siberia, Lena R. (middle) Basin NE Siberia, Kolyma Lowland NE Siberia, Kolyma Lowland East Siberia, Lena R. (lower) Basin NE Siberia, Indigirka River East Siberia, Lena R. (lower) Basin European Rissia, east Locality Lena-Amga Ust'-Omolon Molotkovskiy Kamen' Govorovo, Lena Orto-Tirekhtyakh.R. Molodo Gremyachevo
WWW.NATURE.COM/NATURE | 103
doi:10.1038/nature10574
AAR-11048 OxA-15857 OxA-16310 AAR-11053 Lister/Stuart in prep OxA-15854 OxA-16258 AAR-11056 Lister/Stuart in prep OxA-15850 Lister/Stuart in prep GIN-6023 Lister/Stuart in prep OxA-15859 Lister/Stuart in prep AAR-11020 AAR-11022 AAR-11030 OxA-15858 OxA-15806 OxA-20101 AAR-11059 AAR-11033 OxA-16308 AAR-11024 AAR-11039 OxA-16307 OxA-16309
x x x x x x x x x x x x x x x x x x x x x x
65.12 57.45 50.35 69.40 54.00 60.40 66.90 61.20 70.07 55.18 53.00 58.67 56.30 57.00 46.47
172.80 61.45 106.44 155.00 110.10 60.05 148.90 58.38 153.49 58.63 41.50 57.57 42.14 74.50 126.03
50
WWW.NATURE.COM/NATURE | 104
doi:10.1038/nature10574
x x x x -
OxA-16312 OxA-15863 OxA-20109 OxA-20107 OxA-15862 OxA-16302 OxA-15917 AAR-11062 OxA-15856 OxA-20108 AAR-11058 OxA-15860 AAR-11060 AAR-11061 OxA-15911 OxA-15874 AAR-11051 OxA-15916 OxA-16303
x x x x x x x x x x x x x x x x x x x
21851 (6) 21851 (10) 21851 (4) 21851 (1) 21851(5) 10687/1687 21851 (8) GIN-27 4157 21851 (9) GIN-23 4188 GIN-25 GIN-26 F-45 3914-5 Galgan I -1 3491-898 10693
ZIN RAS ZIN RAS ZIN RAS ZIN RAS ZIN RAS ZIN RAS ZIN RAS GIN RAS ZIN RAS ZIN RAS GIN RAS ZIN RAS GIN RAS GIN RAS SIAM PIN RAS PIN RAS PIN RAS ZIN RAS
WR181 WR182 WR178 WR180 WR175 WR110 WR183 WR311 WR143 WR179 WR307 WR161 WR309 WR310 WR061 WR068 WR298 WR088 WR115
JN570889 JN570890 JN570888 JN570886 JN570891 JN570917 JN570867 JN570871 JN570913 JN570880 -
20,480 21,010 21,090 21,160 21,300 21,400 21,560 21,660 21,660 21,690 23,000 23,270 23,270 23,500 24,670 24,860 24,830 24,880 25,040
90 80 90 90 80 100 90 160 90 90 170 100 180 190 110 100 210 110 120
24,436 25,055 25,171 25,268 25,433 25,600 25,823 25,953 25,971 26,011 27,841 28,119 28,123 28,273 29,499 29,707 29,738 29,746 29,914
190 188 190 189 209 229 211 301 201 200 331 181 225 217 175 213 272 217 208
52.86 52.86 52.86 52.86 52.86 62.00 52.86 51.18 58.60 52.86 51.26 57.68 68.32 50.10 70.74 68.19 59.70 68.65 54.60
103.55 103.55 103.55 103.55 103.55 129.70 103.55 108.31 43.69 103.55 107.26 66.20 161.72 119.18 136.21 146.60 161.10 159.15 44.80
Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia
WWW.NATURE.COM/NATURE | 105
doi:10.1038/nature10574
x x x x x x
OxA-15807 AAR-11052
x x
SP1357 Galgan I -2
WR042 WR299
26,990 27,950
180 300
31,300 32,168
110 427
51.41 59.70
39.05 161.10
Russia Russia
OxA-20095
161-135
PIN RAS
WR318
28,160
190
32,393
359
70.56
149.71
Russia
NE Siberia, Indigirka Lowland EUR Russia, NW n/a NE Siberia, Indigirka lower NE Siberia NE Siberia, Indigirka lower NE Siberia, Kolyma Lowland East Siberia, Aldan R. Basin East Siberia, Aldan R. Basin Urals (north) NE Siberia, Indigirka Lowland NE Siberia, Indigirka Lowland NE Siberia, Chukotka NE Siberia, Kolyma Lowland The Urals
OxA-20103 OxA-15853 OxA-15846 AAR-11049 OxA-15912 OxA-20098 AAR-11035 AAR-11036 OxA-15808 OxA-15730 OxA-16323 OxA-15855 OxA-20100 Lister/Stuart in prep
x x x x x x x x x x x x x -
10736 10717 3915-137 202-0203 3657-142 169-86 359/102 359/90 SP1360 3658-13 F-36 35594 171-6 12560/UR.37
ZIN RAS ZIN RAS PIN RAS PIN RAS PIN RAS PIN RAS GIN RAS GIN RAS ZIN RAS PIN RAS SIAM ZIN RAS PIN RAS n/a
WR289 WR132 WR067 WR296 WR074 WR321 WR215 WR216 WR045 WR062 WR080 WR137 WR323 WR252
28,450 29,110 29,260 29,550 30,240 30,350 30,500 30,950 31,070 31,500 32,380 32,690 33,150 Lister/Stuart in prep
180 150 140 350 170 170 250 250 190 200 220 200 220
32,815 33,789 33,955 34,155 34,823 34,874 34,978 35,505 35,607 35,857 36,848 37,176 37,891 38,445
348 346 325 414 137 153 348 419 385 368 301 388 439 847
59.10
31.10
Russia n/a Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia
70.60 72.10 70.60 68.32 62.90 62.90 62.00 68.60 67.94 71.10 69.29 60.40
147.60 142.50 147.60 161.72 134.10 134.10 58.67 147.06 142.42 180.00 154.72 60.05
WWW.NATURE.COM/NATURE | 106
doi:10.1038/nature10574
Lister/Stuart in prep OxA-16687 OxA-20102 OxA-16685 OxA-16686 AAR-11054 AAR-11034 AAR-11043 Oxa-15851 OxA-16314 OxA-15810 OxA-16301 OxA-18755 Lister/Stuart in prep GIN-6012 OxA-20090 OxA-20094 OxA-20106 OxA-15798 OxA-20104 OxA-16306 OxA-15809 OxA-16256 OxA-15804 &15805 OxA-15802 OxA-15848
x x x x x x x x x x x x x x x x x x x x x x x
55.75 71.00 69.00 56.45 69.00 48.56 67.20 68.69 68.52 56.00 51.50 46.47 58.01 50.40 69.28 51.30 51.30 68.65
40.75 136.00 135.00 53.80 162.00 10.19 132.90 161.65 147.10 83.84 6.70 126.03 40.87 106.40 146.85 83.00 83.00 159.15
WWW.NATURE.COM/NATURE | 107
doi:10.1038/nature10574
OxA-20089 OxA-15731 OxA-15910 OxA-15914 OxA-16324 OxA-15915 OxA-16257 OxA-16294 OxA-16295 OxA-16296 OxA-16297 OxA-16299 OxA-16304 AAR-11023 AAR-11026 AAR-11038 AAR-11040 AAR-11045 AAR-11050 AAR-11063 OxA-20092 OxA-20093 OxA-20099 OxA-15799 OxA-15800
x x x x x x x x x x x x x x x x x x x x x x x x x
WWW.NATURE.COM/NATURE | 108
doi:10.1038/nature10574
OxA-15801 OxA-15803 OxA-15947 OxA-16298 AAR-11021 AAR-11025 AAR-11031 AAR-11032 AAR-11037 AAR-11044 AAR-11046 AAR-11055 AAR-11057 AAR-11064
x x x x x x x x x x x x x x
WWW.NATURE.COM/NATURE | 109
doi:10.1038/nature10574
Supplementary Table S6.3. Horse (Equus ferus) sample information, listed by calibrated radiocarbon age. Data include radiocarbon age, locality information, and institution currently housing the sample. Information on new radiocarbon dates and GenBank accession numbers of new sequences (JN570919JN571033) are included. Some DNA sequences were generated from specimens with published radiocarbon dates; these references follow below the table.
AMS ID GIN-10687 New date Collection no BL-O 279-R Museum IEM RAS Lab ID JW25 New seq JN570962 14C date 2,220 14C SE 50 IntCal09 date 2,230 IntCal 09 SE 64 LAT 73.30 LON 141.30 Country Russia Region Novosibirsk Islands, N-E Siberia Lena River Delta, N-E Siberia Alberta, Canada Locality Zimovye River mouth Mamontovy Khayata Grand Prairie Ref 5
OxA-13847 OxA-14270
x -
PIN RAS Royal Alberta Museum AMNH AMNH AMNH AMNH NHMS NHMS AMNH AMNH AMNH AMNH AMNH EPQ IPAE PIN RAS AMNH AMNH AMNH
JW191 JW174
JN570956 DQ007594
5,778 11,200
34 90
6,581 13,092
48 121
71.80 53.50
129.30 -113.50
Russia Canada
1 6
CAMS-119982 AA-37609 AA26819 CAMS-145101 AA26810 OxA-13669 OxA-13670 CAMS-145091 CAMS-145092 AA-37614 AA26829 CAMS-145095 AA26811 OxA-13671 OxA-14363 OxA-13758 AA26809 AA26805 AA26817
x x x x x x x x -
F:AM 142429 A-144-9422 F:AM 143628 A-6159 NHMS n/a NHMS n/a F:AM 143621 F:AM 143622 A-4339 F:AM 143625 Bx334-2870 IPT-6671 EK 994/708 PIN 3658-121 A-276 A-144-6987 A-237-10198
JAL294 JW184 JAL276 JW348 JW177 JW175 JAL249 JAL252 JW356 JAL268 JW579 JW266 JW305 JW194 JW374 JW355 JW585
JN570941 AF326670 JN570955 JN570930 JN570983 DQ007611 DQ007558 DQ007609 DQ007556 JN570922 JN570923 AF326675 JN570991 JN570925 JN571004 JN570963 JN570972 JN570957 JN570993 JN570990 JN571007
12,310 12,380 12,510 12,510 12,560 12,545 12,550 12,560 12,670 12,840 12,860 13,055 13,270 13,845 13,900 13,935 13,940 14,000 14,120
14,338 14,459 14,644 14,681 14,714 14,751 14,752 14,792 15,001 15,381 15,423 15,764 16,182 16,928 16,964 16,990 16,994 17,089 17,219
272 282 292 233 316 226 234 215 196 396 393 331 398 92 102 113 114 203 232
64.40 64.40 64.40 64.40 53.30 47.40 48.20 64.40 64.40 64.40 65.40 64.40 64.40 48.50 59.00 68.30 59.00 64.40 64.50
-147.30 -147.30 -147.30 -147.30 -113.10 8.50 9.40 -147.30 -147.30 -147.30 -147.10 -147.30 -148.00 10.20 58.80 157.70 58.80 -147.30 -147.40
U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. Germany Germany U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. Germany Russia Russia Russia U.S.A. U.S.A.
Fairbanks, Alaska Fairbanks, Alaska Fairbanks, Alaska Fairbanks, Alaska Fairbanks, Alaska Germany, Europe Germany, Europe Fairbanks, Alaska Fairbanks, Alaska Fairbanks, Alaska Fairbanks, Alaska Fairbanks, Alaska Fairbanks, Alaska Germany, Europe Surya, Urals Kolyma lowlands, N-E Siberia Urals Fairbanks, Alaska Fairbanks, Alaska
Goldstream Ester Creek Ester Creek Upper Cleary Fox Petersfels Hohlefels Lower Goldstream Goldstream Ester Creek Fairbanks Creek Goldstream Cripple Creek Vogelherd IV Sur'ya 5 Alyoshkina Zaimka, Loc. 3298 Sur'ya 5 Ester Creek Goldstream
1 3 4 1 4 6 6 1 1 3 4 1 4 1 1 1 4 4 4
WWW.NATURE.COM/NATURE | 110
doi:10.1038/nature10574
OxA-14364 AA26840 OxA-16361 OxA-17691 AA26852 Beta-109267 OxA-14372 CAMS-119977 AA26812 AA26808 CAMS-145109 CAMS-120068 AA26820 CAMS-145104 AA26837 AA26845 CAMS-119968 AA26807 CAMS-145098 AA26839 AA26821 CAMS-120058 OxA-14299 AA26824 AA26801 AA26836 AA26823 AA26816 AA26827 CAMS-120059 CAMS-145097 CAMS-145096 CAMS-145103 CAMS-145108 AA26849
x x x x x x x x x x x x x x x x x -
WWW.NATURE.COM/NATURE | 111
doi:10.1038/nature10574
AA-37611 CAMS-145116 CAMS-145093 AA26841 AA26863 AA26834 CAMS-119983 CAMS-119980 AAR-11188 GIN-10688
x x x x x -
CAMS-145105 AA26813 AA-37608 CAMS-145102 CAMS-145106 AA26862 OxA-14384 CAMS-119978 AA26869 AA26866 AA26857 AA-37613 CAMS-145113 CAMS-145107 OxA-14367 CAMS-145110 AA26850 OxA-17690 OxA-14125 CAMS-119981 OxA-14156
x x x x x x x x x x x x x
F:AM 143632 A-1013 F:AM 143629 F:AM 143634 A-463-3144 EK 994/710 F:AM 142425 A-144-6509 A-559-4332 A-802 F:AM 6171 F:AM 143648 F:AM 143635 EK 994/145 F:AM 143639 A-4-4-2145 2MgV2-16-910 KU-42625 F:AM 142428 KU-51467
AMNH AMNH AMNH AMNH AMNH IPAE AMNH AMNH AMNH AMNH AMNH AMNH AMNH IPAE AMNH AMNH CMC KU AMNH KU
JAL283 JW349 JAL279 JAL287 JW597 JW306 JAL284 JW512 JW351 JW598 JAL310 JAL290 JW338 JAL307 JW601 JW616 JW157 JAL293 JW160
JN570934 JN570984 AF326669 JN570931 JN570936 JN571012 JN570973 JN570935 JN570996 JN570986 JN571013 AF326674 JN570946 JN570938 JN570979 JN570944 JN571015 JN571024 JN570952 JN570940 JN570953
20,150 20,170 20,200 20,210 20,430 20,410 20,430 20,440 20,420 20,545 20,580 20,670 20,720 20,730 20,730 20,810 20,840 21,100 21,130 21,280 21,500
110 329 310 110 90 320 110 100 325 345 160 350 110 90 110 120 350 80 90 100 290
24,080 24,099 24,129 24,130 24,374 24,374 24,375 24,387 24,388 24,546 24,584 24,695 24,712 24,720 24,721 24,792 24,895 25,182 25,229 25,412 25,724
177 412 388 170 185 405 204 196 412 443 230 458 172 161 172 188 471 180 188 217 435
64.40 64.50 65.40 64.40 64.40 64.50 59.00 64.40 64.40 65.40 64.50 64.50 64.40 64.40 59.00 64.40 64.40 67.10 44.50 64.40 44.50
-147.30 -147.40 -147.10 -147.30 -147.30 -147.30 58.80 -147.30 -147.30 -147.10 -147.30 -147.40 -147.30 -147.30 58.80 -147.30 -148.00 -140.50 -108.20 -147.30 -108.20
U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. Russia U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. U.S.A. Russia U.S.A. U.S.A. Canada U.S.A. U.S.A. U.S.A.
1 4 3 1 1 4 1 1 4 4 4 3 1 1 1 1 4 1 1 1 1
WWW.NATURE.COM/NATURE | 112
doi:10.1038/nature10574
CAMS-145114 AA26844 CAMS-119989 AAR-11182 AA26814 AA26848 OxA-14385 OxA-17686 OxA-14307 AA-37612 AAR-11185 AAR-11183 OxA-16428 AAR-11187 AA26859 OxA-17685 AA-37610 OxA-14120 GIN-10253 OxA-14362 OxA-17679 AA-37607 OxA-17680 OxA-14304 OxA-14552 OxA-17681 GIN-10672
x x x x x x x x x x x x x x x x x x -
AAR-11186
YG-29-120
GYW
PH5
JN571031
29,450
350
34,070
416
64.00
-139.20
Canada
WWW.NATURE.COM/NATURE | 113
doi:10.1038/nature10574
x x x x x x -
x x x -
1 1 1 6
This study
Holmes, C.E. Archaeological testing and evaluation of the gerstle River Quarry, East-Central Alaska, 1996. Division of Parks and Outdoor Recreation, Alaska Department of Natural Resources, Office of History and Aracheology Report Number 85, 118 (1998) Vila et al. Widespread origins of domestic horse lineages. Science 291, 474477 (2001) Guthrie, R. D., Rapid body size decline in Alaskan Pleistocene horses before extinction. Nature 426, 169171 (2003)
3 4 5
Sher, A. V., Kuzmina, S. A., Kuznetsova, T. V. & Sulerzhitsky, L. D. New insights into the Weichselian environment and climate of the East Siberian Arctic, derived from fossil insects, plants, and mammals. Quat. Sci. Rev. 24, 533-569 (2005)
6
Weinstock et al. Evolution, Systematics, and Phylogeography of Pleistocene Horses in the New World: A Molecular Perspective. PLoS Biol 3, e241.
WWW.NATURE.COM/NATURE | 114
doi:10.1038/nature10574
Supplementary Table S6.4. Reindeer (Rangifer tarandus) sample information, listed by calibrated radiocarbon age. Data include radiocarbon age, locality information and institution currently housing the sample. BC denotes radiocarbon dates beyond the calibration curve. Information on new radiocarbon dates and GenBank accession numbers of new sequences (JN570760JN570863) are included. GenBank accession numbers of previously published sequences used in the genetic analysis are included. References follow below the table.
AMS ID AA85585 AA83759 AA83779 AA83780 AA83781 AA85594 AA84492 AA83786 AA84475 AA83788 AA84493 AA83789 AA84476 AA83793 AA83794 AA83796 AA83798 AA83799 AA83800 AA83810 AA84483 AA85596 New date x x x x x x x x x x x x x x x x x x x x x x Collection no 3658-142 Ellef Ringnes Isl. 620 618 619 616 614 639 633 628 647 516 552 547 700 749 747 1019 1021 CMN17521 CMN 17521 A-295-5223 Museum PIN RAS Private KIC KIC KIC KIC KIC KIC KIC KIC KIC KIC KIC KIC KIC KIC KIC KIC KIC CMN CMN AMNH Lab ID 289 308 1100 1101 1103 1105 1116 1118 1125 1127 1129 1131 1152 1155 1156 1160 1163 1168 1171 1671 4396 4808 New seq JN570796 JN570797 JN570807 JN570816 AF096422 AF096428 AF096434 AF096443 AY178677 AY178686 AY178688 AY178702 AY178704 AY178714 EU653423 14C date 133 0 0 0 0 0 0 0 0 0 0 0 0 89 0 130 121 0 0 115 220 112 0 0 0 0 0 0 0 0 0 0 0 14C SE 43 0 0 0 0 0 0 0 0 0 0 0 0 49 0 49 49 0 0 44 120 44 0 0 0 0 0 0 0 0 0 0 0 IntCal09 date 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 IntCal09 SE 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 LAT 69.96 78.78 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 74.00 64.08 64.08 64.83 52.81 50.36 52.81 52.09 69.81 75.09 78.82 75.09 75.09 75.09 60.13 LON 147.56 -103.55 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 101.00 -139.43 -139.43 -147.64 -73.44 -85.36 -73.44 -117.08 -142.66 -100.02 18.11 -100.02 -100.02 -100.02 7.48 Country Russia Canada Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Canada Canada Canada Canada Canada Canada Canada Alaska Canada Norway Canada Canada Canada Norway Region Indigirka Canadian Arctic Archipelago Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Taimyr Loy Lake Dawson Area Loy Lake Fairbanks Ck Quebec Southeast Canada Quebec Southwest Canada Fairbanks Canadian Archipelago Svalbard Canadian Archipelago Canadian Archipelago Canadian Archipelago Hardangervidda, Langfjella Ref 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 3 3 3 4 4 4 4 4 4 6
WWW.NATURE.COM/NATURE | 115
doi:10.1038/nature10574
AA85597 AA83785 AA84474 AA85588 AA83783 AA84490 AA83802 B-212897 B-212882 AA83782 AA84472 AA84498 AA83808 AA83797 B-162895 AA83826 AA83835 AA83757 Ua-325 B-217505 AA87167 AA87166 Ua-326 AA83768 AA84462 AA85592 B-162887 B-162889 B-162886
x x x x x x x x x x x x x x x x x x x x -
WWW.NATURE.COM/NATURE | 116
doi:10.1038/nature10574
B-162888 B-162893 B-212884 AA83758 B-162894 B-162885 B-162897 B-212894 AA84495 B-152439 B-162892 B-162882 B-227525 B-227526 B-162884 B-212895 AA83790 Ua-328 AA84494 B-162890 B-212896 -
x x x x x -
WWW.NATURE.COM/NATURE | 117
doi:10.1038/nature10574
x x x x x
AA84465 OxA-21396 AA83769 AA83778 AA83767 AA84497 AA83765 AA83775 AA87053 AA83806 AA83815 AA83774 AA84487 AA84491 AA83761 AA83773 AA83777 OxA-21397
x x x x x x x x x x x x x x x x x x
85E- 31608 (17) n/a 57E- 34768 104E- 34768 45E- 18993 (1) CMN 44455 43E- 18993 (5) 86E- 31608 (12) 17157 (2) MF(1537)10 A-194-5925 81E- 31608 (322) n/a 612 27E- 34768 79E- 31608 (3) 96E- 34768 5215
896 Zin-08 868 915 856 n/a 854 897 Zin-05 1658 1683 892 10411 n/a 838 890 907 1695
JN570786 JN570795 JN570780 JN570790 JN570778 JN570777 JN570787 JN570794 JN570763 JN570843 JN570785 JN570760 JN570773 JN570784 JN570789 JN570848
12,790 13,100 13,150 13,220 13,390 13,930 14,140 14,215 14,310 14,430 14,610 14,761 15,020 15,940 16,010 16,190 16,340 16,565
130 50 130 130 140 150 150 85 150 170 160 96 170 180 200 190 200 70
15,241 15,870 15,971 16,103 16,448 17,021 17,228 17,297 17,408 17,549 17,774 17,962 18,250 19,120 19,152 19,304 19,504 19,721
378 332 388 386 358 183 213 179 227 257 281 243 209 188 202 244 278 150
59.25 60.00 62.00 62.00 60.53 64.08 60.53 59.25 55.98 50.09 64.94 59.25 50.09 74.00 62.00 59.25 62.00 64.83
57.46 113.80 58.67 58.67 57.67 -139.43 57.67 57.46 92.74 19.91 -147.67 57.46 19.91 101.00 58.67 57.46 58.67 -147.65
Russia Russia Russia Russia Russia Canada Russia Russia Russia Poland USA Russia Poland Russia Russia Russia Russia USA
Urals, Central East Siberia, south Urals, Northern Urals, Northern Ural, Chusovaya river, cave Dyrovataya Dawson Area Dominion Creek Ural, Chusovaya river, cave Dyrovataya Urals, Central Baikal Mammoth cave Engineer Creek, Alaska Urals, Central Mammoth Cave Taimyr Urals, Northern Urals, Central Ural, Medvezhya cave, 1960, Kuzmina Gold-Hill, Alaska
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
WWW.NATURE.COM/NATURE | 118
doi:10.1038/nature10574
AA85583 AA85580 AA83801 AA83813 AA84481 AA85578 AA85579 AA84473 AA83818 AA83776 AA84469 AA85590 AA84480 AA84477
x x x x x x x x x x x x x x
AA84501 AA83812 AA85582 AA83762 AA84468 AA83763 AA85595 AA83764 AA84461 AA85593 AA84463 AA85587 AA83824 OxA-21399 AA83811 AA84471
x x x x x x x x x x x x x x x x
550 CMN47679 200-337 38E- 21838 (770) 611 39E- 21838 (155) 651 40E- 21838 (100) 484 875/24 494 465 CMN 37930 915/93, 71 CMN49615 629
n/a 1673 244 849 1108 850 1148 851 484 875-24 494 465 4409 915-93-71 1672 1119
JN570841 JN570825 JN570774 JN570802 JN570775 JN570815 JN570776 JN570835 JN570781 JN570837 JN570833 JN570852 JN570791 JN570840 JN570809
19,430 19,720 20,500 20,810 20,840 20,940 21,202 21,220 21,820 22,410 22,690 23,120 23,320 23,670 23,720 23,850
280 330 320 330 350 340 349 340 280 403 300 440 450 120 510 480
23,154 23,566 24,486 24,851 24,895 25,024 25,377 25,397 26,194 27,056 27,365 27,897 28,143 28,403 28,590 28,726
394 458 409 444 471 465 483 472 467 562 431 577 572 199 569 516
74.00 64.03 71.79 51.90 74.00 51.90 74.00 51.90 70.54 59.27 70.54 70.54 64.08 58.00 64.08 74.00
101.00 -140.73 129.40 103.60 101.00 103.60 101.00 103.60 158.91 62.21 158.91 158.91 -139.43 69.00 -139.43 101.00
Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Russia Canada Russia Canada Russia
Taimyr 60 mile Loc5 Lena Delta Siberia, Baikal area Taimyr Siberia, Baikal area Taimyr Siberia, Baikal area Sibirien, Yakutien West Siberia, southwest Sibirien, Yakutien Sibirien, Yakutien Dawson Area Loc. 45 West Siberia, center Hester Creek, Yukon Taimyr
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
WWW.NATURE.COM/NATURE | 119
doi:10.1038/nature10574
AA84486 AA83827 AA83784 AA84470 AA83821 AA83807 AA84466 AA85584 AA83772 AA83787 AA83828 AA83791 AA84467 AA85581 AA87052 AA84460 AA83819 AA87051 AA83795 AA84488 AA84464 AA85591 AA83809 OxA-21398 AA84479 OxA-21395 AA83760 AA83814 AA84484 AA83820 AA84499 AA83755
x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x
WWW.NATURE.COM/NATURE | 120
doi:10.1038/nature10574
AA83756 AA85577 AA85586 AA84489 AA83816 AA83817 AA84482 AA84503 AA84504 AA83822 AA83823
x x x x x x x x x x x
AA83825 AA84485
x x
CMN CMN
4410 4413
>39,700 >40,400
Infinite Infinite
64.07 64.03
-139.43 -140.73
Canada Alaska
1 1
This study Meldgaard, M. The Greenland caribou zoogeography, taxonomy, and population dynamics. Meddelelser om Grnland Bioscience 29, 188 (1986) Dueck, G.S. Genetic relationships and phylogeography of woodland and Barrenground caribou. M.Sc. thesis, University of Alberta, Canada (1998)
Flagstad, . & Red, K. H. Refugial origins of reindeer (Rangifer tarandus L.) inferred from mitochondrial DNA sequences. Evolution 57, 658670 (2003) Kuhn, T. S., McFarlane, K. A., Groves, P., Mooers, A. O. & Shapiro, B. Modern and ancient DNA reveal recent partial replacement of caribou in the southwest Yukon. Mol. Ecol. 19, 13121323 (2010) Red, K.H. et al. Genetic analyses reveal independent domestication origins of Eurasian reindeer. Proc. R. Soc. B 275, 18491855 (2008)
WWW.NATURE.COM/NATURE | 121
doi:10.1038/nature10574
Supplementary Table S6.5. Megafaunal taxa present in Upper Palaeolithic archaeological sites of Siberia, by calibrated radiocarbon age. Table includes information on the minimum number of megafauna individuals reported at each site (NISP). The data are presented in Figure 4 of the main text. References are listed below table.
Site Ushki, layer 6 Ust'-Timpton, layer 6 Kaminnaia Cave, layer 11a Kaminnaia Cave, layer 11b Oshurkovo, layer 2 Makarovo-2, layer 3 Makarovo-2, layer 4 Bol'shoi Iakor'1, layer 4B Bol'shoi Iakor'1, layer 5 Berelekh Ust'-Kiakhta17, layer 5 Maininskaia, east layer 2-1 Maininskaia, west, layer A1 Ust'-Mil'-2, layer A Bol'shoi Iakor'1, layer 6 Tashtyk-1, layer 1
C Age 10354 10421 10683 10860 11108 11707 11950 11970 12050 12078 12143 12173 12110 12200 12380 12413
14
14
Cal BP 12208 12298 12593 12741 12993 13581 13809 13833 13912 13919 13988 14035 14064 14169 14483 14494
Cal BP () 97 119 170 436 106 274 73 247 174 62 108 202 380 332 359 256
Woolly rhinoceros
Woolly mammoth
Horse x
Reindeer
Bison x
Musk ox
NISP n/a 2
LAT 56.17 58.7 51.2 51.2 51.96 54 54 57.82 57.82 70.43 50.35 52.98 52.98 59.64 57.82 54.61
LON 159.95 127.12 84.6 84.6 107.49 105.84 105.84 113.98 113.98 143.94 106.45 91.5 91.5 133.1 113.98 90.99
Ref. 11 24 9 9 32 32 32 17 17 24 31 33 33 24 17 2
x x x x x x* x x x x x? x x x x x x x x x x x x x x? x x x? x x? x?
79 426 n/a n/a n/a 31 ~361 1003 n/a 57 158 n/a ~153 78
x x
WWW.NATURE.COM/NATURE | 122
doi:10.1038/nature10574
Bol'shoi Iakor'1, layer 7 Diuktai Cave, layer 7a Verkholenskaia Gora, layer 3 Bol'shoi Iakor'1, layer 8 Kokorevo-3 Bol'shoi Iakor'1, layer 9 Kokorevo-2 Listvenka, layer 8 Ui-2, layer 3 Bol'shaia Slizneva, layer 7 Novoselovo-6 Kokorevo-1, layer 2 Golubaia-1, layer 3 Diuktai Cave, layer 7v Listvenka, layer 10 Ui-2, layer 2 Divnyi-1 Maininskaia, east, layer 4 Maininskaia, east, layer 3 Bol'shaia Slizneva, layer 8 Diuktai Cave, layer 7b
12380 12480 12570 12630 12690 12700 12710 12750 12910 12930 12913 12993 13065 13110 13200 13260 13220 13217 13302 13540 13317
250 99 180 230 140 140 71 140 54 60 135 238 89 90 110 270 150 124 86 500 58
14502 14604 14729 14871 14992 15016 15061 15138 15420 15462 15523 15722 15800 15895 16077 16079 16090 16102 16334 16388 16412
438 265 389 528 382 387 218 402 271 291 381 508 356 358 371 489 401 382 332 775 293 x x x x x x x x* x x
x x x x x x x x x x x x x x x x x x x x x x x x x x x
57.82 59.3 52.36 57.82 54.93 57.82 54.93 55.95 52.97 55.96 55.07 54.92 52.98 59.3 55.95 52.97 55.08 52.98 52.98 55.96 59.3
113.98 132.6 104.28 113.98 90.94 113.98 90.94 92.4 91.49 92.6 91.11 90.93 91.51 132.6 92.4 91.49 91.31 91.5 91.5 92.6 132.6
17 24 13 17 1 17 1 3 33 35 2 2 5 24 3 33 22 33 33 35 24
x x
x x
x x
226 598 n/a 8 n/a 9481 >67 7** 163 n/a 29 n/a 65 73 n/a 70
x x x
x* x* x x* x
WWW.NATURE.COM/NATURE | 123
doi:10.1038/nature10574
Tashtyk-2, layer 2 Afontova Gora2 Listvenka, layer 12 Listvenka, layer 6 Kokorevo-1, layer 3 Volch'ia Griva Novoselovo13, layer 1 Biriusa-1 Listvenka, layer 9 Kokorevo-4a, layers 5-3 Chernoozer'e 2 Kurtak-3, excav. 1 Tashtyk-4, layer 2 Oznachennoe-1 Listvenka, layer 7 Novoselovo-7 Kokorevo-4b, layer 2 Maininskaia, east, layer 1 Ust'-Menza-2, layer 17 Verkhne Troitskaia, layer 6 Sokhatino-4 Maininskaia, east, layer 5
13550 13373 13437 13677 13633 13679 14097 14238 14307 14320 14500 14652 14700 14713 14750 15190 15460 15500 15900
320 46 126 280 45 74 190 52 78 330 50 66 150 67 250 93 320 150 313
16495 16537 16550 16740 16787 16818 17199 17319 17396 17469 17654 17825 17896 17898 17966 18330 18689 18717 19099
540 232 310 468 91 100 236 163 180 392 167 165 280 197 347 190 365 154 298 x x x x
x x* x*
x x x x x x x x x x x x x
n/a ~100 n/a n/a >97 1392 n/a n/a n/a n/a ~300 n/a n/a n/a n/a 928 97 39 n/a
54.61 55.99 55.95 55.95 54.92 55.19 55.07 55.85 55.95 54.94 55.74 55.15 54.61 53.08 55.95 55.07 54.94 52.98 50.23
90.99 92.79 92.4 92.4 90.93 75.64 91.11 92.22 92.4 90.96 73.99 91.56 90.99 91.42 92.4 91.11 90.96 91.5 108.63
2 6 3 3 2 28 22 21 3 4 28 22 2 5 3 2,22 4 33 19
x* x x* x* x x x
x x x x
x x
x x x x x x x x
x?
99 226 124
x x
x x x
72 n/a 107
24 26 33
WWW.NATURE.COM/NATURE | 124
doi:10.1038/nature10574
Listvenka, layer 19 Listvenka, layer 15 Ezhantsy, layer 3 Ui-1, layer 2 Shikaevka-2 Tomsk Krasnyi Iar, layer 6 Shlenka Tarachikha Kunalei, comp. 3 Mal'ta Buret' Novoselovo13, layer 3 Alekseevsk Igeteiskii Log Ust'-Kova, middle comp. Kashtanka, layer 1 Arta-2, layer 3 Anui-2, layer 11 Kuilug Khem1, layer 4 Sabanikha Kurtak-4, layer 1 Priiskovoe Ikhine-2, layer 2b Tolbaga Yana RHS
17013 17080 17150 17995 18050 18300 19100 19193 19543 21100 21157 21190 22000 22415 22426 22477 22635 23200 23431 23600 23979 24150 25825 26177 27346 27895
192 485 345 75 95 1000 100 89 56 300 37 100 700 480 171 185 174 2000 1547 400 217 137 290 143 146 54
20201 20391 20446 21454 21524 21973 22779 22879 23378 25241 25247 25306 26510 27047 27157 27217 27337 28181 28277 28435 28837 28970 30636 30904 31465 31943
318 605 474 138 190 1284 253 243 209 423 155 201 925 640 372 363 321 2757 1852 464 289 235 267 152 142 249 x x
x x x x x x x x x x x
x* x* x x*
x x x x x x
n/a n/a 415 173 314 n/a n/a ~1500 n/a n/a 66** >61 n/a n/a 134 x x x x >10,000 269 n/a <9 n/a n/a n/a n/a 126 494 2380
55.95 55.95 60.48 52.97 55.9 56.46 53.6 55.2 55.05 50.64 52.83 52.99 55.07 57.84 53.57 50.3 55.14 51.19 51.39 51.99 54.61 55.15 50.15 63.11 51.21 70.72
92.4 92.4 135.15 91.49 65.8 84.93 103.4 91.93 91.04 107.64 103.55 103.51 91.11 108.34 103.48 100.11 91.52 112.3 84.68 92.96 90.96 91.56 108.32 133.62 109.32 135.42
3 3 24 33 28 28 8 22 22 18 13 13 22 36 23 34 12 10 10 30 22 22 20 24 27 29
x x x x x*
x x x
x x x x x
x x x
x x x x x x x
x x
x x x x
x* x x x x x x x x* x x x
x x x x x x x
x x x
WWW.NATURE.COM/NATURE | 125
doi:10.1038/nature10574
Ikhine-2, layer 27800 500 32143 2g Malaia Syia 28118 196 32339 Ust'-Mil' 2, 28250 354 32524 layer B Nepa-1 28410 295 32729 Kamenka, 28563 143 32989 Complex B Kamenka, 29308 166 34006 Complex A Ikhine-2, layer 29667 439 34234 2v Voennyi 29700 500 34254 Gospital Masterov 30743 845 35489 Kliuch, comp 1 Ust'-Mil' 2, 31800 386 36294 layer V Kurtak-4, layer 32137 248 36653 2 Varvarina Gora 32379 411 37003 Podzvonkaia 35249 500 40381 *May include Equus hemionus. Radiocarbon age is average of two or more. References for faunal data.
594 363 527 492 288 330 492 563 909 523 311 575 614
x x x x x
x x*
x x
x x
63.11 54.41 63.11 59.99 51.74 51.74 63.11 52.28 51.4 63.11 55.15 51.58 50.21
133.62 89.44 133.62 108.37 108.3 108.3 133.62 104.25 110.65 133.62 91.56 108.12 107.32
24 25 24 15 14 14 24 23 16 24 22 27 7
x x
x x x x x x x x x x x x* x* x x x
x x x
1978 32 n/a 18
x x x
Abramova, Z. A. Paleolit Eniseia: Afontovskaia Kultura [in Russian] (Nauka, Leningrad, 1979a) [in Russian] Abramova, Z. A. Paleolit Eniseia: Kokorevskaia Kultura [in Russian] (Nauka, Leningrad, 1979b) [in Russian]
Akimova, E. V. et al. Paleolit Eniseia: Listvenka [in Russian] (Institut Arkheologii i Etnografii Sibirskogo Otdeleniia Rossiiskoi Akademii Nauk, Krasnoiarsk, 2005) [in Russian]
WWW.NATURE.COM/NATURE | 126
doi:10.1038/nature10574
4 5 6 7
Astakov, S. N. PozdnePaleoliticheskaia stoianka Kokorevo 4. Sovetskaia Arkheologiia 2, 288-294 (1966) [in Russian] Astakhov, S. N. Paleolit Tuvy (Sibirskoe Otdelenie, Akademiia Nauk SSSR, Krasnoiarsk, 1986) [in Russian] Astakhov, S. N. Paleolit Eniseia: Paleoliticheskie Stoianki Afontovoi Gore v G. Krasnoiarske (RAN, St. Petersburg, 1999) [in Russian].
Caiwe, N. et al. Le Paleolithique superior ancient de Siberie. Les fouilles de Frans Steenhoudt et de Vasily Tashak en Bouriatie. Bulletin Musees R DArt DHist 64, 129-149 (1993) [in Russian] Derevanko, A. P. The Paleolithic of Siberia: New Discoveries and Interpretations (University of Illinois Press, Urbana and Chicago, 1998)
8 9
Derevianko, A. P. et al. Arkheologiia, Geologiia i Paleogeografiia Pleistotsena i Golotsena Gornogo Altaia (Izdatelstvo Instituta Arkheologii i Etnografii SO RAN, Novosibirsk, 1998) [in Russian]
10
Derevianko, A. P. et al. Prirodnaia Sreda i Chelovek v Paleolite Gornogo Altaia (Izdatelstvo Instituta Arkheologii i Etnografii SO RAN, Novosibirsk, 2003) [in Russian]
11
Dikov, N. N. Arkheologicheskie Pamiatniki Kamchatki, Chukotki i Verkhnei Kolymy (Aziia na Styke s Amerikoi v Drevnosti) (Nauka, Moscow, 1977) [in Russian]
Drozdov, N. I., Chekha, V. P., & Artemev, E. V. in Arkheologiia, Geologiia i Paleogeografiia Paleoliticheskikh Pamiatnikov Iuga Srednei Sibiri (Severo-Minusinskaia Vpadina, Kuznetskii Alatai i Vostochnyi Saian (ed Derevianko, A. P., Drozdov, N. I., & Chekha, V. P.), (RAN, Krasnoiarsk, 1992) [in Russian]
13 14
12
Ermalova, N. M. Teriofauna Doliny Angary v Pozdnem Antropogene (Nauka, Novosibirsk, 1978) [in Russian]
Germonpre, M. & Lbova, L. Mammalian remains from the Upper Paleolithic site of Kamenka, Buryatia (Siberia). J. Archaeolog. Science 23, 35-57 (1996)
15
Goebel, T. in Entering America: Northeast Asia and Beringia before the Last Glacial Maximum (ed Madsen, D. B.) (University of Utah Press, Salt Lake City, pp. 311-356 (2004)
Goebel , T., Waters, M. R., & Meshcherin, M. N. Masterov Kliuch and the early Upper Paleolithic of the Transbaikal, Siberia. Asian Perspectives 39, 47-70 (2000)
16
WWW.NATURE.COM/NATURE | 127
doi:10.1038/nature10574
Ineshin, E. M. & Tetenkin, A. V. Chelovek i Prirodnaia Sreda Severa Baikalskoi Sibiri v Pozdnem Pleistotsene. Mestonakhozhdenie Bolshoi Iakor I (Nauka, Novobisirsk, 2010) [in Russian]
18 19
17
Konstantinov, M. V. in Arkheologicheskii Poisk (Severnaia Azii) (Nauka, Novosibirsk, pp. 16-24, 1980) [in Russian]
Konstantinov, M. V. Kamennyi Vek Vostochnogo Regiona Baikalskoi Azii (Izdatelstvo ION BNTS SO RAN/Izdatelstvo CHGPI Im. N. G. Chernyshevskogo, Ulan-Ude/Chita, 1994) [in Russian]
20 21 22
Konstantinov, M. V. & Konstantinov, A. V. in Problemy Khronologii i Pervobytnaia Iuzhnoi Sibiri (Nauka, Barnaul, pp. 13-15, 1991) [in Russian] Kuzmina , I. & Sinitsyna, G. in Pervoe Mezhdunarodnoe Mamontovoe Soveshchanie (St. Petersburg, 1995) [in Russian]
Lisitsyn, N. F. Pozdnii Paleolit Chulymo-Eniseiskogo Mezhdurechia (Trudy Tom II, Institut Istorii Materialnoi Kultury, RAN, St. Petersburg, 2000) [in Russian]
23
Medvedev, G. I., Savelev, N. A., Svinin, V. V. Stratigrafiia, Paleogeografiia i Arkheologiia Iuga Srednei Sibiri (Irkutskii Gosudarstvennyi Universitet, Irkutsk, 1990) [in Russian] Mochanov, I. U. A. Drevneishie Etapy Zaseleniia Chelovekom Severo-Vostochnoi Azii (Nauka, Novosibirsk, 1977) [in Russian]
24 25
Muratov, V. M., Ovodov, N. D., Panychev, V. A., & Safarova, S. A. in Arkheologiia Severnoi Azii (Nauka, Novosibirsk, pp. 33-48, 1982) [in Russian]
26 27 28 29 30
Okladnikov, A. P. & Kirillov, I. I. Iugo-Vostochnoe Zabaikale v Epokhy Kamnia i Rannei Bronzy (Nauka, Novosibirsk, 1980) [in Russian] Ovodov, N. D. in Prirodnaia Sreda i Drevnii Chelovek v Pozdnem Antropogene (Nauka, Ulan-Ude, pp. 122-140, 1987) [in Russian] Petrin, V. T. Paleoliticheskie Pamitaniki Zapadno-Sibirskoi Ravniny (Nauka, Novosibirsk, 1986) [in Russian] Pitulko, V. V. et al. Yana RHS site: humans in the Arctic before the last glaciation. Science 303, 52-56 (2004)
Semenov, V. A., Vasilev, S. A., Zaitseva, G. I., Kilunovskaya, M. E., & Kasparov, A. K. Kuilug-Khem 1: a new Paleolithic cave site in Tuva (south Siberia, Russia). Current Research in the Pleistocene 22, 9-11 (2005)
WWW.NATURE.COM/NATURE | 128
doi:10.1038/nature10574
Tashak, V. I. in 100 Let Gunnskoi Arkheologii. Nomadizm Proshloe, Nastoiashchee v Globalnom Kontekste i Istoricheskoi Perspective. Gunnskii Fenomen. (Tezisy Dokladov) II Chast. Rossiiskaia Akademiia Nauk, Sibirskoe Otdelenie, Ulan-Ude, pp. 61-64, 1996) [in Russian]
32 33 34 35
31
Tseitlin, S. M. Geologiia Paleolita Severnoi Azii (Nauka, Moscow, 1979) [in Russian] Vasilev, S. A. Pozdnii Paleolit Verkhnego Eniseia (RAN, IIMK, St. Petersburg, 1996) [in Russian] Vasil'evskii, R. S., Burilov, V. V., & Drozdov, N. I. Arkheologicheskie Pamiatniki Severnogo Priangaria (Nauka, Novosibirsk, 1988) [in Russian]
Vdovin , A. S., Iamskikh, A. F., Iamskikh, G. I. U., & Ovodov, N. D. in Arkheologiia, Geologii i Paleogeografiia Paleoliticheskikh Pamiatnikov Iuga Srednei Sibiri (ed Derevianko, A. P., Drozdov, N. I., & Chekha, V. P.) (Nauka, Krasnoiarsk, pp. 21-34, 1992) [in Russian] Zadonin, O. V., Khomik, S. N., & Krasnoshchekov, V. V. in Problemy Arkheologii i Etnografii Sibiri i Dalnego Vostoka, 1. Nauka, Krasnoiarsk, pp. 45-48, 1991) [in Russian]
36
WWW.NATURE.COM/NATURE | 129