Index Introduction Materials and Methods Results Metric dental traits Nonmetric dental traits Discussion Acknowledgments References

Introduction

The Japanese archipelago was initially occupied in prehistoric times by the Jomon people and their ancestors. Later, substantial genetic admixture occurred with Yayoi immigrants from the Asian mainland associated with the spread of rice cultivation from 2,000 years BP onward. This hypothesis, which was termed the “dual structure” model by Hanihara (1991), is now widely accepted as describing the population history of Japan. According to this scenario, the genes of the Yayoi migrants gradually spread throughout Japan and this diffusion continues even today. The morphological variety that we currently recognize is due to the admixture ratio of genetic traits of the indigenous people and the migrants, and the ratio varies according to the locality. In peripheral regions such as southernmost and northernmost Japan, where people were less affected by the genes of migrants compared to the western part of Japan, the people tend to retain the physical characteristics inherited from the indigenous Jomon people. In particular, the Ainu living on the northernmost island, Hokkaido, are considered to be direct descendants of the Jomon natives based on many morphological similarities to each other (e.g. Howells, 1966; Turner, 1976; Brace and Nagai, 1982; Yamaguchi, 1982; Hanihara, 1986; Matsumura, 1989, 1994, 1995; Dodo and Ishida, 1990; Pietrusewsky, 1994).

Thus, to reconstruct the population history of Japan, fundamental issues such as the dispersion process of the immigrants toward the northeastern mainland through time, and their level of admixture with the indigenous people need to be clarified further. A number of cranial and dental studies have demonstrated the strong genetic influence of immigrants in central Japan by the end of the protohistoric Kofun period, although the proportion of the migrants was still less than in western Japan. On the other hand, the morphological features of the people of the Tohoku district of northern mainland Honshu throughout the period from Yayoi to the early modern age were heretofore largely unknown due to the virtual lack of human remains in the area. Hanihara (1986) and Ossenberg et al. (2006), through their analysis of geographical variation of early modern and modern Japanese crania, found that Tohoku Japanese occupy an intermediate phenetic position between Hokkaido Ainu and the people living in the Kanto region of central Japan, suggesting that the genetic admixture with immigrants in this area was less than in other regions of Honshu. This view was followed by a preliminary study using dental remains from early modern Tohoku sites (Nagaoka and Hirata, 2003), although the number of samples used was quite small. Given this perspective, the genetic impact of immigrants was thought to be even less in the ancient Tohoku people. A further accumulation of early and late historic human remains from Tohoku now enables a study in greater temporal and regional scale for understanding the population history of this region. A spatial and temporal craniometric variation analysis by Kawakubo et al. (2009) reconfirms that the expansion of the immigrants gradually covered the geographic range, including Tohoku, of the previous Jomon lineages, supporting the “dual structure” model of Hanihara (1991).

Interestingly, the earliest historical record of Japan, the Nihonshoki, mentions that the proto- and early historic northern Tohoku region was occupied by people designated ‘Emishi’, who might have been regarded as distinctly different tribes by the early Japanese people living in the area dominated by the Yamato Imperial Court of Western Japan (Aston, 1972; Farris, 1996). The fact that many non-Japanese place names, which are linguistically related to the Ainu language, remain in the northern Tohoku region has given some historians the idea that the ‘Emishi’ can be ethnically identified as the Ainu themselves. There is a long-standing debate about whether the ethnicity of the ‘Emishi’ was Japanese or Ainu. Hanihara (1990) suggested that the ‘Emishi’ were non-Ainu Japanese who had descended in parallel from the Jomon population

Although skeletal remains of the ‘Emishi’ are still very scarce, discoveries of sites from later historic periods in the northern Tohoku region have produced a large number of human remains, allowing us to trace back the population affinity of the ancient inhabitants of this region. Herein we report a study based on metric and nonmetric dental morphology of the early modern and modern Tohoku peoples to address the question of the population history of the northeastern mainland of Japan.

Materials and Methods

The skeletal materials from the Tohoku district used in the present study are listed in Table 1, and the sample locations are presented in Figure 1. All samples examined comprised permanent teeth from the skeletal remains encompassing a total of 185 individuals, which are stored at the Department of Anatomy, School of Medicine, Tohoku University, Department of Anatomy, School of Medicine, Saint Marianna University, and the University Museum, the University of Tokyo.

View Details

Figure 1. Map showing locations of the Tohoku skeletal specimens used.

Most of the pre-modern specimens, which were excavated from the sites in Aomori, Iwate, and Akita Prefectures, are dated to the Edo (Tokugawa) period (1603–1868 AD). In this study, these specimens were designated as the ‘Edo Northern Tohoku’ sample according to the sample localities. A few individual specimens dating to the Muromachi period (1392–1567 AD), listed in Table 1, were combined with this group. Although several sites located in the southern Tohoku district have produced human remains of early modern Edo period, this local group was not used in this study due to the small sample size.

All modern Tohoku specimens used are skeletal collections from the dissecting-room of Tohoku University whose date of death are recorded as from the period of about 1900–1950. This series comprises persons born in various districts in Tohoku region, though predominantly from Miyagi Prefecture. Therefore, these specimens were dealt with as a single sample group called ‘Modern Tohoku’.

Metric dental traits, represented by the mesiodistal and buccolingual crown diameters, were recorded for the above early modern and modern Tohoku groups. The measurements were taken according to the system of Fujita (1949). The third molars were not examined because of their small sample size.

Nonmetric dental characteristics were based on a battery of 21 traits, which were scored using protocols and criteria given in Matsumura (1995). Observations of these nonmetric traits were undertaken for teeth on the right side or antimere substitutions when necessary. All traits were scored on the basis of presence/absence to facilitate statistical comparisons.

Eight other population samples from Japan, consisting of different regional, chronological and ethnic groups, were used for comparison with above two groups from Tohoku district: neolithic Jomon, Yayoi immigrants in western Japan, protohistoric Kofun, early medieval Kanto in central Japan, early modern Edo Kanto, modern Kanto Japanese, Amami and Okinawa Islanders in southernmost Japan, and Hokkaido Ainu in northernmost Japan.

The Jomon series represented early hunter-gatherers who lived in the Japanese archipelago from c. 13000 years BP to 2300 years BP. The samples used here were from the middle to final Jomon phases (c. 4000–2300 years BP), and were excavated from various regions of Japan. The immigrant Yayoi samples, from the first rice cultivators in Japan and dated from c. 2400 years BP to 1750 years BP, were from northern Kyushu and Yamaguchi Prefecture (Nakahashi et al., 1989; Nakahashi, 1993). The protohistoric Kofun series materials ranging from 300 AD to 600 AD were excavated from northern Kyushu and the Kinki district in western Japan, and the Kanto district in central Japan. The medieval Japanese samples from the Kamakura period (1192–1331 AD) and the early modern samples from the Edo (Tokugawa) period (1603–1868 AD) were from the Kanto district.

Both the metric and nonmetric dental data for these comparative pre-modern samples, as well as the data of three modern groups, were cited from Matsumura (1989, 1990, 1994, 1995), and Matsumura and Hudson (2005).

To assess the degree of phenetic differences in these dental characteristics between the Tohoku samples and other comparison samples, multivariate statistical procedures were performed for both the metric and nonmetric tooth data.

Overall tooth sizes were compared based on crown area totals (mesiodistal diameter × buccolingual diameter). Similarities in odontometric proportions were estimated by the Q-mode correlation coefficient (Sneath and Sokal, 1973) based on the 28 mesiodistal and buccolingual crown diameters. Then, in order to transform the correlation coefficient (r) into a distance value (d), the coefficient was subtracted from 1 (d = 1 − r).

The multivariable nonmetric trait differences were explored using C.A.B. Smith’s distances (Berry and Berry, 1967), often referred to as ‘mean measure of divergence’ values, to evaluate population affinities based on the presence/absence frequencies of the 21 nonmetric traits.

To aid the interpretation of the matrix of intersample phenetic distances based on metric and nonmetric tooth traits, the neighbor-joining method (Saitou and Nei, 1987; Tamura et al., 2007) was applied to the distance (1 − r) matrix of the Q-mode correlation coefficient (r) and Smith’s distances, respectively.

Results

Metric dental traits

Table 2 gives sample sizes, means, and standard deviations for mesiodistal and buccolingual tooth crown diameters recorded in the Edo Northern Tohoku and modern Tohoku samples. For cross-reference to the tooth size, mean values of the Kanto samples (Matsumura, 1994) are presented in the adjacent columns in Table 1. In addition to these basic data, this table provides significance levels of differences resulting from t-tests comparing the Edo and modern Tohoku groups, and these Tohoku groups and contemporary samples from Kanto district. The crown diameters of the Edo Northern Tohoku were significantly smaller for several tooth types than those of the Edo Kanto Japanese in both sexes. Even in comparison between the modern samples, the Tohoku people were smaller than the Kanto Japanese for some tooth types. On the other hand, for most of the tooth kinds, the Edo Northern Tohoku sample did not differ greatly from the modern Tohoku series.

Figure 2 presents summations of crown areas (mesiodistal diameter × buccolingual diameter). The Yayoi immigrants and protohistoric Kofun sample were marked by the large overall tooth size for both sexes, followed by the Kanto samples. In the males the overall tooth size of the Amami-Okinawa islanders was comparable with that of the Kanto Japanese. In contrast, the Jomon and Ainu were marked by small teeth for all the samples examined. The Edo Northern Tohoku and modern Tohoku samples, as well as the Kamakura Kanto, possessed intermediate tooth sizes among the comparative samples.

View Details

Figure 2. Total of dental crown areas (mesiodistal diameter × buccolingual diameter) for the comparative samples (left, males; right, females).

Table 3 gives the results of computation of Q-mode correlation coefficients using the 28 mesiodistal and buccolingual crown diameters, as a distance (1 − r) matrix transformed from the computed correlation coefficients (r). Figure 3 represents un-rooted trees for the males and females, which were illustrated using the neighbor-joining method on the basis of the distance matrix presented in Table 3. This analysis yielded a nearly complete separation of the Jomon-Ainu from the Yayoi immigrants and Kofun sample. Among the males, the Edo Northern Tohoku, as well as the modern Tohoku samples, were clustered with the latter clumps in this tree. In contrast, the female Tohoku samples were scattered rather closer to the Jomon and Ainu than to the Yayoi and Kofun samples. The Amami-Okinawa females were adjacently branched out to the Tohoku cluster. In both the sexes, on the other hand, the Edo and modern Kanto Japanese were tied to the Yayoi and Kofun cluster.

View Details

Figure 3. Trees of the two Tohoku series and eight comparative population samples by the neighbor-joining method applied to the distance matrix of the Q-mode correlation coefficients of Table 3, based on 28 mesiodistal and buccolingual dental crown measurements (left tree, males; right tree, females).

Nonmetric dental traits

The frequencies of the 21 nonmetric dental traits for Tohoku samples are summarized in Table 4, together with the Kanto data cited from Matsumura (1995) for reference. The Edo Northern Tohoku and modern Tohoku, left first and second rows in this table, are data recorded for both sexes and combined. Because the odontmetric analysis found considerable sexual difference in the population affinities among the Tohoku samples, sex-separated comparisons will also be expected in the nonmetric dental traits. However, the sample size is insufficient for calculating frequency data when the Tohoku samples are divided into each sex group. Therefore, the frequency data divided into the males and females were provided from the composite samples of the Edo and modern Tohoku series, as given in the third and fourth columns of the table.

The significance levels of difference based on the chi-square tests, in comparisons with contemporary samples from Kanto and within Tohoku samples of different periods, and between the males and females for the Edo and modern Tohoku composite, are also given in Table 4. Large differences were detected between the Northern Tohoku and Kanto samples of Edo period for several traits. Although the significance levels were not so high between the modern Tohoku and Kanto samples, four traits differed from each other. Among these significantly differentiated traits, there was a lower level of occurrence of the shovel-shaped central incisor, and a higher frequency of the hypoconulid reduction of upper second molar was common in both the Edo and modern Tohoku groups, compared with the contemporary Kanto Japanese. On the other hand, significant sexual difference was found only in single trait frequency in the Edo and modern Tohoku composite.

Table 5 gives the Smith’s distances calculated using the 21 nonmetric dental frequency data, together with the data from the Jomon, Yayoi immigrants, protohistoric Kofun, early medieval to modern Kanto, Amami-Okinawa, and Ainu samples. For all of these comparative samples except the Tohoku males and females, sex combined data were used. The neighbor-joining method was applied to the distance matrix of Table 5 to generate unrooted trees (see Figure 4). In the sexes combined samples (the left tree), both the Edo Northern Tohoku and modern Tohoku samples, as well as the Amami-Okinawa sample, occupy an intermediate phenetic position between the Jomon-Ainu and Yayoi-Kofun clusters. On the other hand, the Kanto Japanese were more affiliated with the Yayoi and Kofun samples. The right tree represented in Figure 4 was drawn using the sex-separated samples for the Edo and modern Tohoku series. The males and females of the Tohoku group were closely joined with each other, and were branched out similarly to the left tree drawn using the sex-combined Tohoku sample.

View Details

Figure 4. Trees of the two Tohoku series and eight comparative population samples by the neighbor-joining method applied to the Smith’s distance matrix of Table 5, based on 21 nonmetric dental traits (left tree, sex-combined samples; right tree, sex-separated samples for Tohoku).

Discussion

During the aeneolithic Yayoi and protohistoric Kofun periods, Japan underwent a major new population influx from the eastern part of the Asian continent into the Japanese archipelago (Hanihara, 1991; Nakahashi, 2005). These new immigrants, with their advanced technology and culture, were probably responsible for establishing the imperial court in the Kinki district in western Japan, the influence and power of which would eventually expand eastward. The initial immigrant lineage, which originally entered part of western Japan, can also be found in Yayoi specimens from central Japan, including the Kanto district (Matsumura, 2001). Several studies of Kofun specimens from the Kanto region suggest that the genetic impact of the immigrants was not small even in eastern Japan of this period, where people still retained the physical features of the pre-existing Jomon people to a certain extent (Yamaguchi, 1985a, 1985b, 1987; Hanihara, 1987; Mizoguchi, 1988; Matsumura, 1990). Furthermore, a few researchers have detected the gene flow of immigrants in the northern districts on which this study focused. Regarding the protohistoric specimens from Miyagi Prefecture in southern Tohoku district such as the Kumanodo, Goshozan and Suzu series, skeletal morphology resembling that of the Kanto and Western Kofun people and Yayoi immigrants was demonstrated by Yamaguchi (1988), Matsumura and Ishida (1995), and Mizoguchi (1995), despite them also partially exhibiting some archaic features akin to the Jomon. In terms of the dental morphology, many of these specimens were classified as the immigrant type through discriminant function analysis using tooth measurements (Matsumura, 2001). Mosaic features in a combination of archaic Jomon and new immigrant morphologies were also found in human remains of the early historic Heian period (794–1191 AD) excavated from the Tekiana Cave site in Yamagata Prefecture (Ishida, 1992; Yamaguchi and Ishida, 2000). More recently, the immigrant features were partially traced back to a Yayoi child from the Abakuchi Cave site in Miyagi Prefecture (Nara and Suzuki, 2003). These findings suggest the ancient existence of a significant immigrant source from the west expanding into this region.

Our current study examined the genetic impact of immigrants among the more recent Tohoku people, in various respects of the dental morphology, represented by the overall tooth size, odontometric proportion, and nonmetric traits. As far as the odontometric proportion is concerned, the population affinities of the Tohoku people were somewhat inconsistent between the sexes. In the males, the metric dental proportion of the early modern Edo and modern Tohoku people were quite similar to those of the contemporary Kanto Japanese and even to those of the Kofun and Yayoi immigrants. On the other hand, the females were closer to the Jomon and Ainu rather than to the Yayoi immigrants and Kofun samples. These odontmetric proportions imply the genetic influence of the immigrants was stronger in the males than in the females among the Edo and modern Tohoku residents. The reason why the females exhibited more archaic Jomon features in some aspects of dental morphology than the males in Tohoku district is not clear. If the gene flow from the west to the north was still due mainly to males even in recent times, the females should retain more Jomon features than males. Suzuki (2007) examined dental morphology for the early modern samples from two local sites located in Iwate Prefecture, and demonstrated remarkable immigrant features not only in the males but also in females. Our current analysis of dental morphology did not deal the data with site by site, whereas Suzuki’s study implies a large influence of immigrants in certain local sites of northern Tohoku even in the females.

There is much debate on differences in migration rates between males and females in human populations, especially from views of genetic studies. According to Burton et al. (1996), patrilocality, in which males stay in their birth-place and women move, occurs in about 70% of human societies. This widespread practice of patrilocality should increase variation in maternally inheritated mitochondrial (mt) DNA and decrease variation in paternally inheritated Y-chromosomes. Such patterns of genetic diversities have in fact been detected among some local population samples (Oota et al., 2001). However, continental and global level analysis by Wilder et al. (2004) demonstrated that genetic differentiation of mtDNA and Y-chromosomes between populations is much more similar than commonly thought. Shriver (2005) inferred that female migration rate might not be greater than male rate on a broader scale. Our current study of Tohoku people deals neither with small local-scale nor global-scale movements of modern human populations but rather considers the mid-island scale, and hence is not camparable to the genetic studies debating sex-specific migrate rate. Mizoguchi (1988), based on a craniometric study, speculated that intensive hybridization occurred predominantly as a result of male immigration since the Yayoi period, when the initial immigrants arrived at western Japan. In this case, the reason for the dominance of male migration may be explained by the expansion of ancient migrants being more aggressive than the patterns in modern society discussed by Burton et al. (1996). Our study of Tohoku residents implies that the dominance of male migration toward northern Japan continued even until more recent times although the aggressive approach might have decreased. Analyzing the variability of mtDNA and Y-chromosome of Japanese samples is recommended to ascertain the existence of sex-specific migration.

Nonetheless, such difference in migration rates between males and females can be inferred from dental metric proportions alone. Both sexes indicate the absolute overall tooth size intermediate between the smallest Jomon-Ainu and the largest Kofun-Yayoi samples. The absolute overall tooth size itself does not always suggest the population lineage, because this value might vary to a certain extent depending on contemporary nutrient conditions (Suzuki, 1993). The modern Tohoku specimens used in this study were sampled from persons who lived just before and after World War II. There is a possibility that their relatively small tooth size related to poor nutrition available in wartime.

It is noteworthy that such an intermediate position was likewise drawn from the nonmetric trait comparisons, which might be less affected by nutrition; furthermore, a considerable sexual difference in the population affinity was not detected in the nonmetric dental data from the composite samples of the Edo and modern Tohoku specimens. The inconsistency of the population affinities between the odontmetric proportions and nonmetric dental traits in the males is a complex issue when it comes to interpreting their genetic structures. It is entirely not clear whether the differentiation is due to some statistical bias, or to a different manner of inheritance between metric and nonmetric traits, or some other factors. Regardless, some sexual discordance in population affinity was found in the odontmetric proportions, from the complete perspective of dental morphology including the overall tooth size and nonmetric traits, and this suggests that the people in the Tohoku region had not yet received so large a genetic influence of immigrants from the west even in recent times, as compared with the Kanto region.

According to the Nihonshoki, which is the earliest historical record for the 4th–7th centuries in Japan, the ancient northern Tohoku region was occupied by the so-called ‘Emishi’ (Aston, 1972). The early historic people in the Yamato Imperial Court in western Japan regarded this remote tribe as an ethnically quite different population. Takahashi (1978) stated that the political influence of the imperial court in the Kinki district reached Miyagi Prefecture by the 7th century and expanded northward to Iwate Prefecture by the 10th century. Further documentary evidence records that the northern Fujiwara family, who ruled over northern Honshu in the 12th century, followed earlier chieftains in the region in referring to themselves as ‘Emishi’ or non-Japanese ‘Eastern Barbarians’ (Batten, 2003). Analyses of the mummies of northern Fujiwara chiefs have shown that they were biologically similar to the Kinki Japanese (Hanihara, 1998). The interpretation of such results is, of course, by no means simple. Using the definition of ethnicity used by most cultural anthropologists and sociologists, the northern Fujiwara chiefs would probably be classified as ‘Emishi’ on the grounds that that was how they perceived themselves, though in this case the precise meaning of the term ‘Emishi’ is unclear. Hanihara (1990) stated these people were certainly not related to the ethnic Ainu but were like other contemporary ethnic Japanese who lived in northeastern Japan, outside of imperial Yamato rule.

When addressing the ethnicity of northern Tohoku people, more complexity arises from other historical records on the settlements of Ainu. According to Emori (2003), Ainu people living in Hokkaido had expanded their settlements to the northernmost end of Honshu. Although discoveries of human remains are still limited in this region, some skeletal specimens possessing Ainu features have been reported. Suzuki (1951) and Suzuki et al. (1952) studied late historic skeletal remains from the city of Hachinohe and the Iwaya cave site in the Shimokita Peninsula, respectively, and concluded that both skeletal series were Ainu. Although our study of the early modern northern Tohoku specimens found dental traits like the Jomon and Ainu in a certain extent, we do not conclude that these dental characteristics were derived from the past southward movement of the Ainu. As was mentioned by Suzuki et al. (2004), who found dental traits akin to the Ainu and Jomon in a medieval skeleton from the Hamashiriya site in Shimokita, it is hard to identify the ethnicity not only of these northern Tohoku people, but also of other ancient inhabitants of the presumed Emishi territory. The question of whether the historic northern Tohoku people were Ainu or local descendants from Emishi cannot be addressed without consideration of their cultural background, which lacking in our current anthropological study. Although such problems concerning the northern Tohoku people still remain to be solved, the presence of dental features akin to the Jomon and Ainu, which are partially found in the early modern northern Tohoku people, supports Hanihara’s (1990) perspective that the ‘Emishi’ descended in parallel from the Jomon population with the Hokkaido Ainu.

Acknowledgments

We express our sincere gratitude to Kazuaki Hirata, Department of Anatomy, School of Medicine, Saint Mariana University, and Gen Suwa, the University Museum of the University of Tokyo, for permission to study skeletal materials from Tohoku. This study was supported in part by a Grant-in-Aid in 2005–2007 (No. 17370089) from the Japan Society for the Promotion of Science.