Ancient Jomon genome sequence analysis sheds light on migration patterns of early East Asian populations

Anatomically modern humans reached East Asia more than 40,000 years ago. However, key questions still remain unanswered with regard to the route(s) and the number of wave(s) in the dispersal into East Eurasia. Ancient genomes at the edge of the region may elucidate a more detailed picture of the peopling of East Eurasia. Here, we analyze the whole-genome sequence of a 2,500-year-old individual (IK002) from the main-island of Japan that is characterized with a typical Jomon culture. The phylogenetic analyses support multiple waves of migration, with IK002 forming a basal lineage to the East and Northeast Asian genomes examined, likely representing some of the earliest-wave migrants who went north from Southeast Asia to East Asia. Furthermore, IK002 shows strong genetic affinity with the indigenous Taiwan aborigines, which may support a coastal route of the Jomon-ancestry migration. This study highlights the power of ancient genomics to provide new insights into the complex history of human migration into East Eurasia.

representative individual of the Jomon culture. Previous studies have been sequenced in depth other individuals (even older) of the Jomon culture, but from the northernmost island of Japan (Hokaido) and not from the main island of Japan. The main questions of the study are whether the Jomon were the direct descendants of the Upper Paleolithic first migrants into the Japanese archipelago and to determine if the Jomon culure is associated with the dispersal route from either the northern or the southern parts of Himalaya Mts. After sequencing the specimen genome at a considerable depth (~1.85X) for aDNA standards regarding the expected preservation at humid and warm climates, they performed several population genetics methods and found that, indeed, this individual forms a lineage basal to the rest of the ancient or present-day East Eurasians examined in this study and that the route of migration that this lineage followed was toward East Asia from Southeast Asia. Validity: No major flaws were found to prohibit the publication of the manuscript. However, revisions are needed, as shown in the next sections in more detail. Originality and significance: The originality and significance are obvious as in most ancient DNA studies that examine the genome of missing links and representative individuals of culture around the world. However, this individual has be sequenced in the same depth at 1.85X in a previous study (McColl et al. 2018 Science DOI: 10.1126/science.aat3628) and I guess the wetlab part is the same per se for the both of them. Thus, the wetlab part (DNA extraction, libraries etc) is not original, despite that the authors provide many details in contrast to the 2018 paper that just cites a published protocol. This means that in this paper the authors use previously published genome data despite being the same team who published the 2018 paper. On the other hand, the 2018 paper does not mention anything about data availability. Thus, I have to rise my skepticism if this paper should be considered original in the term of generating new genome data. Maybe it should be made clear to the reader that the genomic data presented here are coming from a previous study and that in this paper the authors provide detailed M&M regarding this specimen. The bioinformatics analyses, though, are indeed original, as the authors use a different dataset and also address different scientific questions. Data & methodology: In this part I will focused on the reproducibility of the analyses taking into account the supplementary info, too. The wetlab part is properly described. However, the sequence data are not publicly available (nothing is mentioned in this manuscript about data availability) so noone can replicate the analyses. I suggest to inform if they going to be uploaded in a public database or if can someone request them directly from the authors. Also, the data processing and the population genomics analyses are not well-documented in order to an independent scientist could replicate the results. This part is very quickly written, in some parts is very confusing, and in some cases the authors just mention the software tool without any detail on what parameters they used (or if they just used the default) neither are citing any particular published pipeline that they followed in order to not have to provide such details. This entire part needs revision (I provide some more details below). Moreover, in a few cases the version of the tools is omitted including but not limited to samtools, plink and treemix (different version may result in different output). Appropriate use of statistics and treatment of uncertainties: In my opinion the statistical analyses, wherever provided with details, are robust. The bar with migration plot from treemix graph in Fig.1 is missing, though. Conclusions: The conclusions are valid regarding the interpretation of the population genomics results and their association with the archaeological background of the samples and the hypotheses made. References: There are some missing citations mostly for software tools used; they are given only at the supplementary info and even there, some of them they do not have any citation e.g. bedools. Clarity and context: The main text is clear and appropriate, albeit there is erroneous link between the supplementary info (e.g. Text S2 should be Text S1 etc). The language needs to be edited a bit by a native English speaker, especially the Supplementary Info text. Attention should be given as there are a few spaces missing (maybe due to PDF conversion?). The length of the main manuscript is acceptable. Out of my expertise: I cannot evaluate the technical details for the carbon dating analyses. Suggested improvements: Other than mentioned above, here I provide some more specified details.
The manuscript "Jomon genome sheds light on East Asian population history" by Gakuhari et al. describes production and analysis of genome-wide data from a ~2.5 kya individual from Honshu, Japan, associated with the hunter-gatherer-fisher Jomon culture that is hypothesized to represent the descendants of the earliest Upper Paleolithic settlers of the Japanese archipelago.
This represents the first Jomon genome with appreciable coverage from mainland Japan, although a low coverage Jomon genome from Honshu and two higher coverage ones from Hokkaido have been published previously. Analysis of this individual, IK002, albeit at lower coverage, has been published earlier in McColl et al. 2018.
The main results of the study are that: 1) Jomon represent an early diverging lineage basal to the split of Tibetans and the ancestors of other East Asians, Eastern Siberians and Native Americans, but dating after the split of the lineage of which Tianyuan (40kya) is a part of. As split of the ancestors of East Asians and Native Americans is dated to around 26kya, the window of divergence is consistent with Jomon being the descendents of Upper Paleolithic settlers. This is consistent with the findings of Kanzawa-Kiriyama et al. 2019.
2) The ancestors of Jomon took a southern route from South East Asia to reach the archipelago (consistent with Kanzawa-Kiriyama et al. 2019) and the Jomon shows no evidence of admixture with representatives of a northern route (here represented by the Siberian MA1). Similarities in microblade technology across northern Eurasia and in Japan in the UP could have arisen through cultural diffusion or demic diffusion by a group not related to MA1.
3)Jomon show genetic affinity to modern and ancient coastal, but not inland, populations of East Asia, possibly supporting a coastal route of spread for this early ancestry. The wet lab processes and computational analyses are sound and according to best practices. The argumentation is easy to follow and supported by the results. The authors do not overstate their conclusions and present alternative hypotheses where appropriate, also taking into account archaeological and anthropological evidence.
The weakness of the paper is twofold: First, limiting the analysis to only the IK002 Jomon genome while other Jomon genomes have been published. This might be due to the very recent publication of these and lack of time to redo and add analyses. For a resubmission I suggest including the additional Jomon genomes which could also address further questions of Jomon population history, e.g. overall population size, population size differences in the different islands, contact between islands and between islands and mainland, dating/location of gene-flow into modern Japanese etc. This would help with the second weakness, the large overlap with recent publications. Additional analyses could set the paper apart from previously published ones, especially McColl et al. 2018 andKanzawa-Kiriyama et al. 2019, as many of the findings here are not exactly novel or surprising. Furthermore, the authors could address and explain differences in their findings to the previous papers, such as different admixture proportions of Jomon into Ainu and modern Japanese, or the finding in McColl that "the Jōmon individual is best-modeled as a mix between a population related to group 1/Önge and a population related to East Asians" rather than a deep diverging lineage.
-Concerning the Jomon-like ancestry in Ulchi and Nivkh, to distinguish the scenarios "This could be an influence of the Hokkaido Ainu who are likely to be direct descendants of the Jomon people. The ancestor of the Ainu people could have admixed with the Okhotsk people55 who were genetically as well as morphologically close to Ulchi and Nivkh in the the Primorye region56-60. The second explanation is that the track of the earliest-wave was erased in the inland but left over in the coastal region.", can you use ALDER to date the admixture event? Same for the Jomon-like ancestry in Devil's Gate. Are these events independent?
Suggested edits and small mistakes that should be fixed: -mention the geographic origin of the sample in the abstract -Some of the results are referenced for the first time in the discussion, such as the qpgraph modeling. These should get their own paragraph in the results. Ust'Belaya mapped in wrong location, should be near Lake Baikal -The texts especially of the supplementary material could use a round of copyediting, there are numerous typos or grammatical mistakes, e.g. in Materials and Methods: "we chose IK002, who were excavated from" ("were" > "was"). The main text looks mostly fine.
-As a visual summary of the results, a schematic such as Fig. 4 in McColl et al. 2018 (but focusing on the settlement waves of Japan and dispersal of Jomon-like ancestry) would be a nice addition, especially since this study deals with routes of dispersal.

Reviewer #1 (Remarks to the Author):
The manuscript focuses on genomic analysis of the 2.5 kya Jomon individual, IK002, an important piece of evidence that can inform us on the demographic history of eastern Eurasia. The authors concluded that IK002 is basal to all modern and ancient eastern Eurasians, and that, based the genetic affinity to Answer to comment 2) Thank you so much for the reviewer's comments. Yes, the sentence could be misleading. We changed the sentence as below: 'This direct evidence on the link between the Jomon and Southeast Asians, thus, confirms the southern route origin of the IK002 lineage.' 3) Paragraph 2, Line 7 of Results: 'We found that IK002 clusters between present-day Southeast and East Asians and the Upper-Paleolithic human remains (40 kya) from Tiányuán Cave.' Is the position of IK002 in the plot driven by Southeast Asian + East Asian versus Hoabinhian clusters, instead of Tianyuan? This is important point to look into, especially when IK002 is regarded as genetically affiliated with Hoabinhian-related ancestry.

Answer to comment 3)
Thank you so much for the reviewer's comments. In the PCA, IK002 clusters between Southeast Asian + East Asian modern populations and Tianyuan / Hoabinhian, albeit closer to Tianyuan. The position on the PCA alone is not sufficient information to distinguish the two scenarios the reviewer mentions. To avoid he sentence being misleading we have changed it as below.  Table S4).'

'We found that IK002 clusters between present-day Southeast and East
Pairwise FST calculations using IK002 can not be carried out with pseudo-haploid genotypes as used to represent IK002. 5) Paragraph 3, Line 1. For Admixture analysis, how many iterations per K was ran? If multiple iterations was run, was CLUMPP applied after the Admixture?
How consistent is the mode produced per K? Alternatively, a calculation of cross validation error per K can be performed. What is the basis for choosing K = 10 as the best representation for the population structure? It will be useful to plot all K's in the supplementary.

Answer to comment 5)
Thank you for the reviewer's comment. We have updated the description and plots of the admixture analysis accordingly. ADMIXTURE was run in 10 replicates, for K values ranging from K=2 and K=20, and best runs were selected and aligned using pong. For this revised version, we chose to represent the structure at K=15 to highlight fine-scale structure among different modern and ancient groups from Eastern Asia, including a component maximized in the Jomon individuals We also provide all runs from K=2 to K=20 in the supplement as figure S6 for the interested reader. are consistent with those presented here, with IK002 as a basal lineage, and no apparent admixture signal from Hoabhinian hunter-gatherers. We note that with a complex admixture graph model as the one in McColl et al it is unfeasible to explore the full space of topologies, and hence equally or better fitting models might not have been explored. In particular, McColl added IK002 to an admixture graph which already included Ami, thereby not allowing for Ami to be modeled as admixed between Jomon and another Asian lineage. The Jomonlike ancestry in the Ami we demonstrate here could therefore cause the apparent Ami/Hoabhinian admixture for IK002 in their model, in order to account for the excess shared drift between IK002 and Ami. In conclusion, we are confident in our results of IK002/Jomon representing a basal East Asian lineage, but caution that more ancient samples will be needed to disentangle the more complex relationships among those early lineages. 8) Paragraph 5, Lines 9-15 of Results. 'IK002 can be modelled as a basal lineage to East Asians, Northeast Asia/East Siberians, and Native Americans, supporting a scenario in which their ancestors arrived through the southern route and migrated from Southeast Asia towards Northeast Asia.' Given the comments above in items number 2 and 7, the conclusions in this section is not supported by robust analysis.

Answer to comment 8)
Thank you for the reviewer's comment. As commented in the reply to point (7) above, we disagree with the reviewer doubting the robustness of IK002 lineage as a basal position. In both McColl et al and here, IK002 consistently forms a basal lineage in the Treemix analysis, diverging prior to ancestors of modern East Asians, Northeast Asia/East Siberians, and Native Americans. We also present a formal test of the basal position using D-statistics, which are consistent with IK002 forming an outgroup to pairs of East and Northeast Asian populations and the ancient Tibetan individual from Chokhopani ( Fig. 3 and Fig.   S8). We therefore argue that our conclusions and this statement are warranted.
To clarify the dual origin of Native Americans as previously described, we have Answer to comment 10) Thank you for the reviewer's comment. We agree with the reviewer that Austronesian-related gene-flow into Japan could in principle also cause this signal. We do however find it to be a very unlikely scenario, as there is no evidence in the archaeological record for an Austronesian migration into Japan.
Furthermore, Ami are consistent with being an outgroup to IK002 and the Hokkaido Jomon when using f4(Mbuti, Ami;IK002,Hokkaido Jomon), with Z=-1.4 (Table S3). Hence any potential gene flow from Austronesians groups would have to have had a similar impact across a large area of the Japanese archipelago, as well as reaching northern Japan by at least 3,500 years ago (age of the Hokkaido Jomon samples). In conclusion we don't find support for this alternative interpretation 11) Paragraph 1, Line 2 of Discussion. As earlier presented, given the comments above in items number 2 and 7, the conclusion in this sentence is not supported by robust analysis.
Answer to comment 11) Thank you for the reviewer's comment. As we answered to your comments 2, we changed the sentence to avoid mislead as below .
"IK002 is modelled as a basal lineage to East Asians, Northeast Asians/East Siberians, and Native Americans (basal East Eurasians: bEE) after the divergence between Tianyuan and the ancestor of hunters-gatherers in Southeast Asia." 12) Paragraph 2, Line 7. See comment 9 above.

Answer to comment 12)
Thank you for the reviewer's comment. We already answered to your comment 9.
13) Paragraph 4, Last sentence of Discussion. The word can be derogatory, and does not refer to a single ethnolinguistic population. They are best referred to as the 'Kankanaey' ethnic group.
Answer to comment 13) Thank you for the reviewer's comment. However, the data are derived from Human Genome Diversity Project. If we change the name of the population, researchers cannot recognize them. So, we added some words into the sentence as below.
"though the origin of Igorot (which can be derogatory, and could be best

Answer to comment 14)
We thank the reviewer for the comment. We note that the aim of the qpGraph analysis was to quantify the range of possible Jomon-like admixture in different target populations, including Ami, and hence the same considerations with regards to the results of McColl et al discussed above apply as well. To improve clarity of the results, we have now included the worst fitting Z-score for all models in figure S10, as well as the fit of base model without the target admixed population. As can be seen in the base model, when performing the initial qpGraph fit only on ancient groups and excluding Ami, a model with Jomon as basal lineage and no contribution from an Onge/Hoabinian-like group provides a very good fit to the data (worst Z=2.1).

Reviewer #2 (Remarks to the Author):
Key results: In this manuscript the authors report the whole-genome sequencing of an ancient individual from central Honshu, the main island of Japan in order to provide insights into the origin and migration history of East Asians. The specimen dates back to ~2500 BP and it is consider a representative individual of the Jomon culture. Previous studies have been sequenced in depth other individuals (even older) of the Jomon culture, but from the northernmost island of Japan (Hokaido) and not from the main island of Japan. The main questions of the study are whether the Jomon were the direct descendants of the Upper Paleolithic first migrants into the Japanese archipelago and to determine if the Jomon culure is associated with the dispersal route from either the northern or the southern parts of Himalaya Mts. After sequencing the specimen genome at a considerable depth (~1.85X) for aDNA standards regarding the expected preservation at humid and warm climates, they performed several population genetics methods and found that, indeed, this individual forms a lineage basal to the rest of the ancient or present-day East Eurasians examined in this study and that the route of migration that this lineage followed was toward East Asia from Southeast Asia. Data & methodology: In this part I will focused on the reproducibility of the analyses taking into account the supplementary info, too. The wetlab part is properly described. However, the sequence data are not publicly available (nothing is mentioned in this manuscript about data availability) so no-one can replicate the analyses. I suggest to inform if they going to be uploaded in a public database or if can someone request them directly from the authors. Also, the data processing and the population genomics analyses are not welldocumented in order to an independent scientist could replicate the results.
This part is very quickly written, in some parts is very confusing, and in some cases the authors just mention the software tool without any detail on what parameters they used (or if they just used the default) neither are citing any particular published pipeline that they followed in order to not have to provide such details. This entire part needs revision (I provide some more details below). Moreover, in a few cases the version of the tools is omitted including but not limited to samtools, plink and treemix (different version may result in different output).
Appropriate use of statistics and treatment of uncertainties: In my opinion the statistical analyses, wherever provided with details, are robust. The bar with migration plot from treemix graph in Fig.1 is missing, though. References: There are some missing citations mostly for software tools used; they are given only at the supplementary info and even there, some of them they do not have any citation e.g. bedools.
Clarity and context: The main text is clear and appropriate, albeit there is erroneous link between the supplementary info (e.g. Text S2 should be Text S1 etc). The language needs to be edited a bit by a native English speaker, especially the Supplementary Info text. Attention should be given as there are a few spaces missing (maybe due to PDF conversion?). The length of the main manuscript is acceptable.
Out of my expertise: I cannot evaluate the technical details for the carbon dating analyses.
Suggested improvements: Other than mentioned above, here I provide some more specified details.
Note> It would have been helpful for the reviewers to provide a line numbers and speed up in this way the process of finding the text part that is mentioned here.

Results
Ancient DNA prescreening, dating, and sequencing by high-throughput sequencers 1. Line 6: It is not clear why Text S2 is given here. I cannot find a connection.
Thank you for the reviewer's comment. Because this is a simple mistake, we removed "Text S2" in Line 6. We noticed Text S2 in Line 3 should also be Text S1.
2. The method/algorithm the authors used for degerming the genetic sex is not given in the M&M.
Thank you for the reviewer's comment. We added the sentence as below. Thank you for the reviewer's comment. We wanted to provide some more detail on the merged dataset for the results section here, but, if the reviewer and editor strongly recommend removing the explanation, we would be ok with moving it to the methods section 2. The ALDER analysis is not mentioned at all in the M&M section.
Thank you for the reviewer's comment. We added the ALDER in the M&M section. Materials 1. In Line 7 the link to Table S1 is not provide any information associated to the text. Maybe the authors mean Text S1? Thank you for the reviewer's comment. Yes, it is not Table S1 but Text S1.
We changed it. 3. Since the HGDP is mentioned for the first time please provide a reference.
Thank you for the reviewer's comment. Yes, we added the reference in the text. 4. Please provide here or in the supplementary info a rationale of selecting 1 to 6 possible migration events.
Thank you for the reviewer's comment. When using m=8 migration events, the worst residuals of the treemix fit are within 3.3 standard errors. We chose to not include more migration events to avoid overfitting and difficulties with interpretations of the resulting complex graphs. We updated Fig. S9 to now also include the matrix of residuals. Thank you for the reviewer's comment. We changed the sentence as below.
"The shape represents statistical significances of genetic affinities based on Z score. Triangle label means strong/high statistical significance with |Z|>3, inverted triangle means weak significance with |Z|=2~3 and circlemeans nonsignificance with |Z|<2." 2. Please ensure that the format is the same for every reference. For example there a few ones that the journal abbreviation does not have dots.
Thank you for the reviewer's suggestion. We revised it. Table S2 1. Please inform the reader what do the terms endogenous and efficiency are expressing. How can someone calculate them?

Supplementary Info
Thank you for the reviewer's suggestion. We changed as below.
Thank you for the reviewer's comment. We changed all the mistakes in Sup Info involving Text S2. This suggest this part to be written in order to be less confusing.
Thank you for the reviewer's comment. Yes, we agree that the parts were described in a confusing way. We changed the two paragraphs which start from "The petrous bone was cut......." 3. The part from "To remove large DNA fragment…" to "…for DNA solution" is very confusing. Please rephrase.
Thank you for the reviewer's comment. Yes, we removed "for DNA fragment solution." Text S3 1. Where can someone find this customizable NGS pipeline that is mentioned?
Is it described below or in another paper perhaps? Are any in-house scripts used and not given to the reader in a way that he will not be able to replicate the results? Please elaborate.
Thank you for the reviewer's comment. We added the details of NGS pipeline on the Sup Text.
2. Some references are missing e.g. BWA and others.
Thank you for the reviewer's suggestion. We put the references on the manuscript of Text S3.
3. Please ensure that in all the cases that you are mention the version of a software tool (here and in the main text) are following the same format (e.g. ver. 1.0.0 vs v1.0.0 vs -1.0.0).
We have now used consistent format for version names across the manuscript Text S4 1. I think you should mention in every case after you merged IK002 with the different genome databases how many SNPs had in common. Are all the analyses using the same number of SNPs? If not please inform the reader when the number changes.
Thank you for the reviewer's suggestion. We ddws the Sup Tables including   the number of SNPs. 2. Please provide a rationale to the reader to help understand why did you select the Ainu people in present-day populations and the Chokhopani 1 and Tiányuán individuals in ancient populations in ADMIXTURE analysis.
Thank you for the reviewer's suggestion. All modern and ancient individuals in the merged dataset were used in the analysis, and we changed the description accordingly." 3. The first sentence in the TreeMix paragraph is confusing. Please rephrase.
Thank you for the reviewer's suggestion. We separated the sentence into two sentences.

Text S5
1. This entire text section is not linked to the main article. Maybe somewhere in the Discussion?
Thank you for the reviewer's suggestion. But, we want to keep this in Text S5, because this information is important to interpret the genome data and too long to insert in the Discussion session. And also, we noticed that in the Discussion session of the main text, we cited Text S1 instead of Text S5. Therefore, probably the reviewer thought that Text S5 is not linked to the main article. We revised the Discussion session of the main text.
Thank you for the reviewer's suggestion. We changed the format.
2. Baba et al. 1991: Can the journal name be abbreviated?
Thank you for the reviewer's suggestion. But, we cannot do it, because this is a Japanese journal (but, in English). Furthermore, the authors could address and explain differences in their findings to the previous papers, such as different admixture proportions of Jomon into Ainu and modern Japanese, or the finding in McColl that "the Jōmon individual is best-modeled as a mix between a population related to group 1/Önge and a population related to East Asians" rather than a deep diverging lineage.
-Concerning the Jomon-like ancestry in Ulchi and Nivkh, to distinguish the scenarios "This could be an influence of the Hokkaido Ainu who are likely to be direct descendants of the Jomon people. The ancestor of the Ainu people could have admixed with the Okhotsk people55 who were genetically as well as morphologically close to Ulchi and Nivkh in the the Primorye region56-60. The second explanation is that the track of the earliest-wave was erased in the inland but left over in the coastal region.", can you use ALDER to date the admixture event? Same for the Jomon-like ancestry in Devil's Gate. Are these events independent?
Thank you for the reviewer's suggestion. We have now included an ALDER analysis using Devil's Gate Cave and Jomon as source groups for dating the admixture in the Ulchi, and find evidence for recent admixture date of 26 generations ago, thereby supporting the first scenario. We also ran ALDER on Devil's Gate cave as a target, but find no significant LD curve. Together this suggests that Jomon-like ancestry in Devil's Gate cave is due to older admixture Jomon-related peoples, and hence different from the recent admixture observed in Ulchi.
Suggested edits and small mistakes that should be fixed: -mention the geographic origin of the sample in the abstract Thank you for the reviewer's suggestion. We changed as below.
Here, we analyze the whole-genome sequence of a 2.5 kya individual (IK002) characterized with a typical Jomon culture.

-->
Here, we analyze the whole-genome sequence of a 2.5 kya individual (IK002) from the main island of Japanese archipelago characterized with a typical Jomon culture.
-Some of the results are referenced for the first time in the discussion, such as the qpgraph modeling. These should get their own paragraph in the results.
Thank you for the reviewer's suggestion. We have moved this paragraph to the Thank you for the reviewer's comment. We switched ancient and modern in the caption, and updated the plot with the correct location for Ust'Belaya -The texts especially of the supplementary material could use a round of copyediting, there are numerous typos or grammatical mistakes, e.g. in Materials and Methods: "we chose IK002, who were excavated from" ("were" > "was"). The main text looks mostly fine.
Thank you for the reviewer's comment. We revised the mistake. Onge-related ancestries, which disputes the core claim of this paper.

Answer:
We thank the reviewer's comment. We have now added qpGraph analyses for the alternative model as a new supplementary figure (Fig. S12). The new analyses show that both alternatives have equivalent support in qpGraph, as there is a lack of additional samples in that part of the graph to further constrain the topologies. We have therefore added the following discussion on these observations to the final paragraph of Results section in the manuscript: "We note that we fit the backbone graph assuming an unadmixed Jomon lineage, as opposed to a previously suggested dual-ancestry model where Jomon is admixed between Önge-and Ami-related ancestry. This alternative base model provides an equivalent admixture graph fit, however we find no evidence for shared genetic drift between the Önge and the ancestral Jomon lineage in qpGraph ( Fig.S12a &  Fig.S12b), or using direct f4 statistics (Table S9 & Table S10). Additional sampling of early East Asian human remains will be needed to further resolve the relationships among these deep lineages, but nevertheless either model supports the deep origins of Jomon."