Jomon: Paleolithic Contingent in Modern Japanese Oct 13, 2016 21:37:47 GMT
Post by Admin on Oct 13, 2016 21:37:47 GMT
FIG. 1. Three demographic models for the origins of mainland Japanese: (a) transformation, (b) replacement, and (c) hybridization. In all models, the population divergence initially occurs at TAIN-CHB between the Ainu and CHB
The dual structure model for the population history of the Japanese people is a hybridization model that Hanihara (1991) proposed concerning the demographic history of prehistoric Japanese based on morphological analyses of crania and teeth (Hanihara 1991). The dual structure model proposes two main events of (I) migration and (II) admixture: (I) the Jomon people originated somewhere in Southeast Asia and came to the Japanese archipelago around 12,000 years ago, whereas the Yayoi people came from northeast Asia around 2,000 years ago and entered the northern part of Kyushu, which is the western island of Honshu (main-island Japan), probably via the Korean peninsula. Then, (II) the Yayoi migrants admixed with the indigenous Jomon people gradually from west to east. Because the degree of Jomon or Yayoi ancestry may vary among local populations, the Japanese populations can be characterized by the dual structure of ancestral populations: people in main-island Japan have a high degree of genetic contribution from the Yayoi migrants, whereas the Ainu on Hokkaido (the northernmost island) and the Ryukyuans in Okinawa (the southernmost islands of the Japanese archipelago) have a higher degree of genetic contribution from the indigenous Jomon. Meanwhile, recent morphological studies have reported that Jomon skeletal series vary among geographic regions and that cranial variation is clearly distinct between the northeast and the southwest portions of the Japanese archipelago (Hanihara and Ishida 2009; Nakashima et al. 2010; Fukase et al. 2012a, 2012b). Therefore, the Jomon people are likely to have already diversified among the local regions before the Yayoi migration happened; thus, the population history in the Japanese archipelago may be complicated with the population structure of the Jomon.
FIG. 2. The complex hybridization model including hyper parameters of the prior for the admixture proportion (P), various ancestral population sizes and different migration rates between the Ainu and JPT.
The demographic models (transformation, replacement, and hybridization) are illustrated in figure 1, including three parameters common to the three models (NAIN: population size in Ainu; NJPT: population size in JPT; TAIN-CHB: time of population divergence between Ainu and CHB) and four parameters specific to each model (TAIN-JPT: time of population divergence between Ainu and JPT; TJPT-CHB: time of population divergence between JPT and CHB; TADMIX: time of admixture between Ainu and CHB; P: the admixture proportion from the Jomon ancestry). A migration parameter (m= 1.0E5) after the admixture event was incorporated between Ainu and JPT to evaluate this effect on the model selection. The transformation model was characterized as the modern Japanese as direct descendants of the Jomon people. Next, we considered that the Ainu split from CHB at TAIN-CHB and that JPT diverged from Ainu at TAIN-JPT. In contrast, the replacement model stated that the modern Japanese were of Yayoi ancestry and we postulated that JPT branched off from CHB at TJPT-CHB. The hybridization model was expressed as the admixture between Ainu and CHB that happened at TADMIX subsequent to the divergence between them at TAIN-CHB. Because complete isolation between Ainu and JPT seems to be an extreme assumption, we also incorporated a continuous gene flow between these two geographically proximate populations into the three models (fig. 1). Then, we tested the model selection with or without migration. We used kernel-approximate Bayesian computation (kernelABC), which is a likelihood-free approach for Bayesian inferences, to incorporate high dimensional summary statistics and improve approximation of posterior estimates given the data (Fukumizu et al. 2013; Nakagome, Fukumizu, et al. 2013; Nakagome, Mano, et al. 2013; Osada et al. 2013; Sato et al. 2014).
FIG. 3. The complex hybridization model that includes the Jomon population structure.
The simplified hybridization model chosen from our model selection can be used as a starting point for further investigation into more realistic scenarios consisting of various prior conditions. Here, we hypothesized several complex hybridization models by including different sets of priors for three parameters (fig. 2): 1) symmetric beta distributions, which is B(, ), for the prior of admixture proportion (P) (supplementary fig. S2, Supplementary Material online); 2) ancestral population sizes before the divergence between Ainu and CHB (NANC = 1,000, 2,500, 5,000, or NAIN which follows lognormal (LN; 5,000, 5,0002 )); and 3) migration rates between Ainu and JPT (m= 0.0, 1.0E5, 1.0E3, or 1.0E1). First, the hybridization model is characterized by the admixture event between Ainu and CHB; hence, the choice of hyper parameters for the prior of P can be crucial in estimating the posterior distribution. We compared MLs of six hyper parameters ( = 0.1, 0.25, 0.5, 1.0, 2.0, and 5.0) under NANC =NAIN and m= 0.0 (supplementary table S2, Supplementary Material online). Although differences in their aMLs were less signifi- cant in terms of aBFs, the U-shaped prior with = 0.5 exhibited the highest likelihood among the hyper parameters. Second, NANC is likely to greatly contribute to shaping the amount of genetic variation among the three populations. We tested four conditions in which NANC was the same as NAIN or fixed at 1,000, 2,500, and 5,000 under m= 0.0 (supplementary table S3, Supplementary Material online). The fixed NANC = 1,000 showed a significantly higher likelihood than the other conditions. Third, continuous migration between Ainu and JPT after the admixture event can be important to defining shared genetic variation (supplementary table S4, Supplementary Material online). The likelihood of m= 1.0E5 was significantly higher than m= 1.0E3 and 1.0E1. Meanwhile, the difference in the likelihoods between m= 0.0 and 1.0E5 was less significant, as indicated in table 1.
Table 2. Estimated Demographic Parameters under the Complex Hybridization Models.
The highest likelihood model indicated an old divergence of Ainu and CHB populations an estimated 736 generations ago (95% confidence interval: 607–873), which was translated into 18,400 years ago based on the assumption of 25 years per generation. This model predicts the divergence of Jomon lineages occurred 598 generations ago (438–708) (14,950 years ago) with admixture occurring 229 generations ago (130– 291) (5,725 years ago) with 36.0% (22.2–53.3%) of the admixture proportion from Jomon ancestry to the mainland Japanese. NAIN was estimated at 6,219 (4,967–7,395) which is approximately a half of NJPT= 10,207 (7,366–12,036). In addition, the other models, either with or without assuming the population structure of the Jomon, indicated that TAIN-CHB was old and NAIN was a half or a third of NJPT. Meanwhile, the posterior estimates of TADMIX are still older than 120 generations ago (3,000 years ago), which archaeological and anthropological studies have traditionally proposed. Although estimating the parameters by fixing TADMIX at 120 revealed that the other parameters of NAIN, NJPT, and P were scaled down to half of those estimates from the models with unfixed TADMIX, the likelihoods were similar between the models with unfixed and fixed TADMIX (table 2 and supplementary table S6, Supplementary Material online). These results suggest that the estimates of TADMIX strongly depend on the other parameters, implying the difficulty in estimating TADMIX under the complex hybridization models. Our combined approaches of parameter estimation with model selection support the new model that the Jomon people were already diversified in the Japanese archipelago before the admixture event with the Yayoi occurred, and one of the Jomon lineages is a direct ancestor of the modern Japanese, yet the other lineage leads to the contemporary Ainu.
Our inference shows that the Ainu and the CHB share an ancestral population that dates to the late Pleistocene; archaeological evidence shows the peopling of the Japanese archipelago began during this time frame. From these results we conclude that the ancestral people of the Ainu (or some of their ancestors) established the Jomon culture sometime at the end of Pleistocene, while a part of the ancestral population of CHB migrated to Japan and started admixing with the Jomon people about 5,000– 6,000 years ago. Our results strongly support the dual structure model for the peopling in the Japanese archipelago produced through hybridization between the Jomon and Yayoi people. Furthermore, we propose a more complicated model in which the Jomon people have diversified on the Japanese archipelago because they were isolated from the continent around 18,000 years ago. Admixture did occur between the Jomon and Yayoi people, yet some of the Jomon people still kept their own lineages without being strongly affected by the gene flow from the Yayoi people. The older divergence time of the Jomon and Yayoi lineages is compatible with the Hanihara’s hypothesis (1991), which claims the Jomon people are direct descendants of late-Paleolithic populations somewhere in Southeast Asia, whereas the Yayoi people were a population adapted to a cold climate in Northeast Asia. As the first step for resolving the peopling of Japan statistically and quantitatively, we simplified the models to make it possible to test the three hypotheses established from the previous studies. Then, we tested complex scenarios under the hybridization model to fit them with the gwSNP data. However, there should be other possible scenarios that we have to consider, and further studies taking account of realistic scenarios should be necessary to untangle a true evolutionary history of Japanese. To this end, our results can be used as a starting point to test more complicated hypotheses using a genomic-scaled data. In particular, the origins of the Jomon and Yayoi people and their demographic history, including the source of gene flow to Ainu, still remain unresolved. Future studies, including ancient genome analysis of Jomon and Yayoi skeletal remains as well as populationscaled gwSNP and genomic sequence analysis for east and northeast Asians will bring new insights into the complex demography in Japanese populations.
Nakagome, Shigeki, et al. "Model-based verification of hypotheses on the origin of modern Japanese revisited by Bayesian inference based on genome-wide SNP data." Molecular biology and evolution 32.6 (2015): 1533-1543.