Original Articles
A comparative analysis of fetal to subadult femoral midshaft bone distribution of prehistoric Jomon hunter-gatherers and modern Japanese
2016 Volume 124 Issue 1 Pages 1-15
Details
2016 Volume 124 Issue 1 Pages 1-15
Adult Jomon femora are known to be more robust with mid-diaphyseal cross sections that are more elongated anteroposteriorly than in the modern Japanese. In the present study, we compared femoral midshaft cross-sectional geometric properties between prehistoric Jomon hunter-gatherers and modern Japanese from the seventh fetal month to 17 years of age. The cross-sectional properties reflecting mechanical strength and shape (e.g. cortical bone area, medullary cavity area, area moments of inertia, representative indices) were measured using micro-computed tomography on 58 Jomon and 73 modern Japanese femora. Results showed that the Jomon midshaft cross sections are significantly larger, more robust, and stronger under mechanical loading than those of the modern Japanese throughout the examined fetal to subadult periods. The consistently greater robusticity of the Jomon femur was caused by (1) greater bone diameter and mass established at least by late fetal life, (2) lower endosteal bone resorption rates from toddler through subadult ages, and (3) greater subperiosteal bone expansion after around puberty. In both Jomon and modern Japanese, the femoral midshaft cross-sectional shape changes through growth, on average, from near-circular to slightly anteroposteriorly elongated. This was seen to be exaggerated in the Jomon midshaft by bone distributional changes that involve enhanced posterior projection of the shaft corresponding to the well-known femoral shaft pilaster. The results also demonstrated that the fetal to infant Jomon femoral midshaft cross sections tend to be broader mediolaterally than those of the modern Japanese, possibly contributing to the adult subtrochanteric platymeric (mediolaterally broad) cross-sectional shape of the Jomon proximal femoral shaft. Our findings suggest that complex interactions of population-specific genetic background, differential response to activity level and/or mechanical load, and elevated levels of activity/load and muscle hypertrophy appear to have caused the Jomon condition of a comparatively robust femoral diaphysis.
Ever since Wolff (1892; translated into English, 1986), bone structure and morphology have been specifically related to the ability of bone to resist mechanical load (Lanyon, 1982; Endo et al., 1984; Frost, 1987, 2001; Currey, 2002; Ruff et al., 2006a). Frost (1987, 2001), for example, suggested that bone is physiologically regulated so as to keep its strain levels confined within optimal ranges. According to such ‘mechanostat’ or ‘mechanoadaptive’ hypotheses, increased strain resulting from greater load stimulates bone formation which then reduces strain back to ‘normal’ levels (within certain thresholds); conversely, decreased strain from disuse induces bone resorption without compensatory formation. The paradigm in which bones optimally change their own mass and distribution in vivo in response to external mechanical load and resulting internal strain is widely acknowledged as ‘bone functional adaptation’ (Pearson and Lieberman, 2004; Ruff et al., 2006a). We note here that the morphological consequences of such ‘mechanoadaptation’ of bone are not necessarily evolutionary adaptations, so we hereinafter refer to it in italics as bone functional adaptation.
Human limb bone structure and morphology have been extensively examined from the viewpoints of biomechanics since the mid-20th century (Aoji et al., 1959; Frankel and Burstein, 1965; Kimura, 1966, 1971; Endo and Kimura, 1970; Jernberger, 1970; Lovejoy et al., 1976; Trinkaus, 1976; Martin and Atkinson, 1977; Lovejoy and Trinkaus, 1980; Kimura and Amtmann, 1984). Some researchers who emphasize the role of bone functional adaptation consider that the diaphyseal cross-sectional geometry of human limb bones reflects to a large extent individual and populational antemortem history of mechanical stress and activity patterns (Ruff and Hayes, 1983; Ruff et al., 1984, 1993, 1994; Ruff, 1987, 1992, 2005; Bridges, 1989, 1995; Trinkaus et al., 1994; Shaw and Stock, 2009a, b). In a classic study, Ruff et al. (1984) reported that the geometric properties of the femoral subtrochanteric and midshaft cross sections differed significantly between preagricultural (more robust and elongated either mediolaterally or anteroposteriorly) and agricultural (more gracile and circular) subsistence strategy groups of the North American Atlantic coast. They interpreted these differences as due to different levels and types of activity involving the lower limb in the two populations. Ruff (1992) furthermore suggested that cross-sectional size (e.g. cortical bone area) more accurately reflects the general magnitude of mechanical load, while cross-sectional shape (e.g. ratio of maximum and minimum area moments of inertia) more accurately reflects principal load directions. Over the last decade or so, by applying ‘mechanoadaptive’ concepts to prehistoric/historic skeletal remains, archaeologists and anthropologists have attempted to interpret past human activity patterns such as discerning between settled or nomadic activity styles (Holt, 2003; Stock and Pfeiffer, 2004; Beauval et al., 2005; Rhodes and Knüsel, 2005; Marchi et al., 2006; Ruff et al., 2006b; Sládek et al., 2006a, b; Okazaki, 2007; Shackelford, 2007; Maggiano et al., 2008; Sparacello and Marchi, 2008; Nikita et al., 2011; Sparacello et al., 2011; Shaw and Stock, 2013; Hill and Durband, 2014).
Regarding prehistoric and modern Japanese skeletal remains, since the mid-1960s, Kimura and his colleagues have advanced our understanding of limb bone biomechanical characteristics (Kimura, 1966, 1971, 2006; Endo and Kimura, 1970; Kimura and Takahashi, 1982, 1984, 1992; Kimura and Amtmann, 1984). The pioneering experiment of Kimura (1966), for example, examined the load–strain relationship in a modern intact tibia by using electrical strain gauges. He suggested that bone such as the tibial diaphysis is not necessarily adapted to uniaxial body weight support, but rather to resisting a range of loading patterns that result from complex activities such as walking. Kimura and Takahashi (1982) and Kimura (2006), by means of radiographic and computed tomographic (CT) measurements, reported that the prehistoric Jomon hunter-gatherers have lower limb bone diaphyses that have relatively larger diameters, thicker cortical bone, and greater strength against bending in the anteroposterior direction than those of the modern groups. In these investigations, the marked anteroposterior elongation of the femoral shaft (pilaster) and anteroposterior elongation/mediolateral flattening of the tibial shaft (platycnemia) seen in the adult Jomon lower limb bones were interpreted as functionally adaptive morphologies that enable resistance to greater bending moments in the anteroposterior direction in relation to their lifestyle with intense and complex lower limb muscle activities (Kimura and Takahashi, 1982; Kimura, 2006). Morimoto (1971, 1981) suggested that bone functional adaptation under less than ideal nutritional conditions might have resulted in the Jomon tibial shaft platycnemia, a bone distribution pattern that emphasizes relative strength against bending in the anteroposterior direction.
Other studies have conversely noted that limb bone diaphyseal structure and morphology cannot necessarily be explained from response and/or correspondence to mechanical load (Bertram and Swartz, 1991; Demes et al., 1998, 2001; Hamrick et al., 2000; Pearson, 2000; Ohman and Lovejoy, 2001, 2003; Lovejoy et al., 2003; Lieberman et al., 2004; Pearson and Lieberman, 2004; Wescott, 2006; Carlson et al., 2007; Demes, 2007; Cowgill, 2010; Morimoto et al., 2011; Wallace et al., 2010, 2012, 2014). Some researchers have emphasized the strong influence of genetic factors on limb bone diaphyseal morphogenesis (Chiu and Hamrick, 2002; Lovejoy et al., 2003; Pearson and Lieberman, 2004; Wallace et al., 2010, 2012). In particular, Lovejoy and colleagues have suggested that limb bone morphology including the shape of the physis (i.e. epiphyseal cartilage) is largely determined by pattern formation effects (Lovejoy et al., 2003). According to this paradigm, although mechanical stress is necessary for bone to develop ‘normally’ and to maintain its own mass, the role in determining specific bone shapes is considered minimal. An adult bone shape is considered to largely mirror the underlying genetics, albeit overlain by loading history during growth (e.g. Shaw and Stock, 2009a, b; Warden et al., 2014), but not to optimize resistance to mechanical stress (Wallace et al., 2014, 2015; Warden et al., 2014). Lovejoy and his colleagues pointed out in a study of myostatin-knockout mice femur (Hamrick et al., 2000) that, although a marked change in bone shape and mass occurred at the lateral shaft enthesis (the connective tissue complex between bone and tendon), no clear increase in bone mass (and strength) occurred elsewhere in the diaphysis. Although limb bone entheseal morphology appears largely genetically patterned and insensitive to general physical activity (Rabey et al., 2015), and with comparatively slight age-related osteophytic and/or porotic changes being the dominant ontogenetic modifications (Milella et al., 2012), large-scale spatial adjustments of the muscle–tendon–periosteum envelope may have led to the pilaster-like bony protuberance of the myostatin-knockout mice femur (Hamrick et al., 2000).
The present study aims to further discuss these issues by examining a developmental series of the Jomon femoral diaphysis. Adult Jomon femora are known to be more robust and have mid-diaphyseal cross sections that are geometrically elongated anteroposteriorly compared to those of recent/modern Japanese. In this study, we investigate femoral midshaft cross-sectional geometric properties reflecting both mechanical strength and shape, e.g. cortical bone area, area moments of inertia, and representative indices, by means of microfocal X-ray CT imaging. We compare these variables between Jomon and modern Japanese through fetal to subadult developmental and growth periods.
In the present study, we used a total of 112 Jomon and 264 modern Japanese late fetal to subadult femora (Table 1). These skeletal remains are stored in The University Museum, The University of Tokyo; Department of Anthropology, National Museum of Nature and Science; Tohoku University School of Medicine; Laboratory of Physical Anthropology, Graduate School of Science, Kyoto University; and The Kyushu University Museum. The Jomon skeletons come from 28 archaeological shellmound/cave sites of the Japanese archipelago (Figure 1), mostly belonging to the Middle–Latest Jomon Period (c. 5300–2300 BP). One individual comes from the Earliest Jomon Period Taishaku-Kannondo cave site (c. 12000–7000 BP), and three individuals are from the Early Jomon Period Hikosaki shellmound site (c. 7000–5300 BP). The recent to modern Japanese femora come from anatomy department skeletal collections of the late 19th to early 20th centuries, and from 17th–19th century skeletal remains of the Fukagawa district of Tokyo (there were no clear morphological differences between these two groups). For simplicity, we refer to the combined recent to modern collection as ‘modern Japanese.’ The sexes of all Jomon femora are considered unknown, and the modern Japanese samples are of mixed sex consisting of 118 males, 82 females, and 64 unknowns according to archival records. The following specimens with femoral diaphyses were chosen: (1) those with the head and condyles completely unfused, (2) those with one end completely unfused and the other partially/completely fused (three Jomon and five modern Japanese), and (3) those with one end partially fused and the other partially/completely fused (seven Jomon and 17 modern Japanese). According to limb bone lengths of modern Japanese fetuses of known gestational age (Takata, 1922) and the fusion timings of the femoral epiphyses (Krogman, 1962), the ages of these femora were considered as ranging from approximately the seventh fetal month to 17 years. As a rule, the right side was investigated, and the left side was substituted when the right side was broken or missing. Only femora of individuals without apparent bone deformation or pathological features were examined.
| Group | Collecting location | n | Storage facility (number of specimens) | |
|---|---|---|---|---|
| Jomon | Hokkaido | Motowanishi | 1 | The University of Tokyo |
| Iwate | Ebishima | 2 | National Museum of Nature and Science | |
| Miyano | 2 | National Museum of Nature and Science | ||
| Nakazawahama | 6 | The University of Tokyo (3), Tohoku University (3) | ||
| Ohora | 5 | The University of Tokyo | ||
| Miyagi | Kitasakai | 1 | The University of Tokyo | |
| Numazu | 4 | The University of Tokyo | ||
| Satohama | 2 | The University of Tokyo (1), Tohoku University (1) | ||
| Fukushima | Sanganji | 7 | The University of Tokyo | |
| Chiba | Kasori | 3 | The University of Tokyo (2), National Museum of Nature and Science (1) | |
| Kusakariba | 2 | National Museum of Nature and Science | ||
| Soya | 1 | The University of Tokyo | ||
| Ubayama | 6 | The University of Tokyo | ||
| Yahagi | 2 | The University of Tokyo | ||
| Yamazaki | 1 | The University of Tokyo | ||
| Tokyo | Azusawa | 1 | The University of Tokyo | |
| Chidorikubo | 1 | The University of Tokyo | ||
| Shizuoka | Shijimizuka | 1 | The University of Tokyo | |
| Aichi | Hobi | 10 | The University of Tokyo | |
| Ikawazu | 5 | National Museum of Nature and Science | ||
| Inariyama | 1 | The University of Tokyo | ||
| Izumida | 2 | The University of Tokyo | ||
| Yoshigo | 26 | Kyoto University | ||
| Okayama | Hikosaki | 3 | National Museum of Nature and Science | |
| Tsukumo | 11 | The University of Tokyo (3), Kyoto University (8) | ||
| Hiroshima | Taishaku-Kannondo | 1 | The University of Tokyo | |
| Taishaku-Yosekura | 4 | The University of Tokyo | ||
| Fukuoka | Kojo | 1 | Kyushu University | |
| Total | 112 | |||
| Modern Japanese | Kanto region | 160 | The University of Tokyo (anatomy department, 139; Fukagawa, 21) | |
| Tohoku region | 95 | Tohoku University | ||
| Kyushu region | 9 | Kyushu University | ||
| Total | 264 | |||
Geographical locations of the Jomon sites yielding the nonadult skeletal remains used in this study. 1, Motowanishi; 2, Ebishima; 3, Miyano; 4, Nakazawahama; 5, Ohora; 6, Kitasakai; 7, Numazu; 8, Satohama; 9, Sanganji; 10, Soya; 11, Ubayama; 12, Kasori; 13, Kusakariba; 14, Yahagi; 15, Yamazaki; 16, Azusawa; 17, Chidorikubo; 18, Shijimizuka; 19, Inariyama; 20, Yoshigo; 21, Ikawazu; 22, Hobi; 23, Izumida; 24, Hikosaki; 25, Tsukumo; 26, Taishaku-Kannondo; 27, Taishaku-Yosekura; 28, Kojo.
The femoral intermetaphyseal length (IL) was measured parallel to the long axis of the diaphysis using either a digital sliding caliper or an osteometric board. According to the longitudinal radiographic data of a modern American sample (Maresh, 1970), the femoral IL is approximately 90% of total femoral length (i.e. length including proximal and distal ends). Thus, in the present study, the IL of the osteological specimens with both the head and condyles partially/completely fused was calculated as 90% of the measured total length. The IL of specimens with one end completely unfused and the other partially/completely fused was considered 94.7% (0.9 divided by 0.95) of the measured length (i.e. length including one of the epiphysis), assuming equivalent proximal and distal epiphyseal contributions. The mediolateral and anteroposterior outer diameters of the midshaft (ML and AP, respectively) were measured as follows. The former was measured perpendicular to the long axis of the diaphysis and parallel to the plane tangent to the popliteal surface. The latter was measured perpendicular to both the long axis of the diaphysis and the ML. The midshaft outer cross-sectional index was calculated as the ratio of AP to ML (AP/ML).
Midshaft cross sections of 58 Jomon and 73 modern Japanese femora were examined by means of the microfocal X-ray CT system (TXS225-ACTIS, Tesco Corp., Tokyo, Japan) at The University Museum, The University of Tokyo. All femora were imaged in reference to the following planes: the coronal plane, tangential to the posterior end of the lesser trochanteric surface and those of the medial and lateral parts of the distal metaphysis; the sagittal plane, perpendicular to the coronal plane and parallel to the long axis of the diaphysis; and the horizontal plane, perpendicular to the coronal and sagittal planes. The scanning parameters were set to a tube voltage of 130–140 kV, current of 0.20–0.24 mA, and slice thickness of 0.03–0.12 mm. Each section was captured parallel to the horizontal plane and was reconstructed in 512 × 512 matrix size with 12-bit (4096 grades) grayscale CT values. Slice interval, pixel size, and slice thickness were made equal. The images were processed using CT-Rugle 1.2 for Windows (Medic Engineering Inc., Kyoto, Japan). The bone–air boundary was determined from the median CT values of bone and air measured at arbitrarily chosen locations of the outermost compact bone and air adjacent to the diaphysis. See Figure 2 for an example of the original and binary-transformed CT images used in this study.
Example of the original (a) and binary-transformed (b) CT images used in this study. Left femur. Viewed from the distal side; anterior is up, lateral is toward the right. The length of the white bar corresponds to 1 mm. The cross line of the binary image represents the principal axes. This specimen derives from the Latest Jomon Period Hobi shellmound site, Aichi Prefecture. Registry ID: UMUT130143. IL = 70.9 mm.
We measured the following geometric properties of the femoral midshaft cross section: total subperiosteal area (TA), cortical bone area (CA), medullary cavity area (MA), maximum and minimum area moments of inertia (Imax and Imin, respectively), area moments of inertia around the coronal and sagittal axes (Ix and Iy, respectively), and polar moment of inertia (IJ). The ratios of CA to TA, TA to MA, and Ix to Iy were also calculated (CA/TA, TA/MA, and Ix/Iy, respectively). The cortical bone area indicates the relative strength of the diaphysis against compressive/tensile force in the axial direction. The area moment of inertia indicates the strength against bending loads with regard to a chosen axis. The polar moment of inertia is the sum of Imax and Imin (or Ix and Iy) and indicates the strength against torsion.
Data analysis and statisticsIn modern humans, bipedal locomotion is acquired between the ninth postnatal month to 1.5 years of age (Yaguramaki and Kimura, 2002). Thereafter, typically, a somatic growth spurt starts at around the age of 12 years in males and 10 years in females (Sinclair and Dangerfield, 1998). Considering these significant developmental patterns and events, our developmental and growth series sample was divided into the following three age groups for analysis: (1) fetal to infant periods, aged <1.5 years, referred to as FET-INF; (2) toddler to prepubertal periods, aged 1.5–10 years, referred to as CHILD; and (3) pubertal to subadult (adolescent) periods, aged >10 years, referred to as ADOLESC. Other subdivisions are possible, such as at an age of around 4–6 years, when the bicondylar angle approximates the adult condition (Tardieu and Trinkaus, 1994) signaling changes in lower limb loading patterns. However, visual examinations of the regression plots suggest that the above defined three age-group subdivisions largely capture the growth pattern changes seen in both Jomon and modern Japanese femora.
Femoral intermetaphyseal length has a very high positive correlation with age during growth (Maresh, 1970; Fazekas and Kósa, 1978). Thus, in the present study, IL was used as an indicator of the age of each individual. In order to compare Jomon and modern Japanese femora of equivalent developmental stages, IL was divided by the mean of maximum femoral lengths of adults to derive a relative age proxy. Following Takigawa (2006), the following femoral lengths were used: 403.8 mm for the Jomon, 411.2 mm for the modern Japanese male, 380.1 mm for the modern Japanese female, and 395.7 mm (average of males and females) for the modern Japanese when sex is considered unknown. In the present study, this variable, the ratio between nonadult intermetaphyseal length and adult maximum femoral lengths, is referred to as the LTratio.
Based on the longitudinal radiographic data of modern Americans (Maresh, 1970), the LTratio value of the FET-INF age group was considered as <0.318 from the mean 1.5-year-old intermetaphyseal length of 154.7 mm and the mean 17-year-old total length of 485.9 mm. The LTratio value of the ADOLESC age group was considered as >0.717 from the mean 10-year-old intermetaphyseal length of 348.6 mm and the mean 17-year-old total length of 485.9 mm. The LTratio value of the CHILD age group was then considered as 0.318–0.717 according to the above. The LTratio value we used for birth is 0.212. This was derived from the mean femoral intermetaphyseal length of a neonatal modern Japanese sample (84.0 mm; Takata, 1922) and the maximum femoral length of the adult modern Japanese (395.7 mm, see above). We note here that, in the Jomon, the age–length relationship of the femur is unknown, so the timings of birth, locomotor change, and somatic growth spurt, as indicated above, are actually uncertain.
Group differences were evaluated by analysis of covariance (ANCOVA) using LTratio as the covariate and the midshaft metrics including the cross-sectional geometric properties as the dependent variables. The slope and y-intercept were determined by least-squares regression. The separate-variance t-test (Welch’s method) was performed when ANCOVA was inapplicable. The significance of correlation was evaluated using the Pearson’s or Spearman’s correlation coefficients. All analyses were performed using SYSTAT 13 for Windows (Systat Software Inc., Chicago, IL, USA). All P-values are shown, and those <0.05 are considered statistically significant.
Table 2 shows the results of ANCOVA of the two midshaft linear metrics and six cross-sectional geometric properties against relative age (variable LTratio). Figure 3 illustrates the scatter plots against LTratio. All variables had significant positive correlations with LTratio in all three age groups (FET-INF, CHILD, and ADOLESC), indicating that bone size and mass increase throughout development and growth. We found that, in all variables, the Jomon have greater values than the modern Japanese at equivalent relative age (LTratio), and that this is the case in all three age groups (FET-INF, CHILD, and ADOLESC). This indicates that the Jomon femoral midshaft is more robust and stronger against compression/tension, bending, and torsion throughout the examined fetal to subadult periods. The slopes of most variables in both the FET-INF and CHILD age groups exhibited no significant Jomon–modern Japanese differences, indicating that the growth rates of the femoral midshaft bone size, mass, and distribution are comparable in the two populations before around puberty. In contrast, the slopes of all variables of ADOLESC were significantly or near-significantly (ML, P = 0.091; AP, P = 0.092) greater in the Jomon than in the modern Japanese. This indicates that, in contrast to the modern Japanese, the robusticity of the Jomon femoral midshaft increases markedly during adolescence.
| Variable (mm) | Group | Fetal–infant period (FET-INF) | Toddler–prepubertal period (CHILD) | Pubertal–subadult period (ADOLESC) | Difference of slopes | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Mean | Slope | Y-intercept | R | n | Mean | Slope | Y-intercept | R | n | Mean | Slope | Y-intercept | R | FET-INF vs. CHILD | CHILD vs. ADOLESC | ||||||||
| ML | Jomon | 41 | 6.5 | 39.415 | P < 0.001 | −0.750 | NA | 0.94 | 50 | 13.8 | 15.550 | P = 0.988 | 5.944 | P = 0.061 | 0.89 | 21 | 21.0 | 26.091 | P = 0.091 | −1.059 | P = 0.055 | 0.79 | P < 0.001 | P = 0.009 |
| Modern Japanese | 199 | 5.7 | 28.835 | 0.560 | 0.90 | 29 | 14.0 | 15.578 | 5.527 | 0.91 | 36 | 20.3 | 15.903 | 6.789 | 0.64 | P < 0.001 | P = 0.924 | |||||||
| AP | Jomon | 41 | 6.0 | 32.770 | P = 0.171 | 0.057 | P < 0.001 | 0.91 | 50 | 13.8 | 21.670 | P = 0.818 | 2.958 | P = 0.033 | 0.93 | 21 | 23.8 | 36.636 | P = 0.092 | −7.166 | P < 0.001 | 0.83 | P = 0.008 | P = 0.001 |
| Modern Japanese | 199 | 5.5 | 28.913 | 0.379 | 0.88 | 29 | 14.2 | 21.197 | 2.686 | 0.92 | 36 | 22.0 | 25.846 | −0.019 | 0.80 | P < 0.001 | P = 0.203 | |||||||
| √CA | Jomon | 21 | 4.8 | 15.781 | P = 0.486 | 1.994 | P = 0.002 | 0.67 | 23 | 10.0 | 15.994 | P = 0.452 | 1.734 | P < 0.001 | 0.95 | 14 | 16.9 | 28.136 | P = 0.002 | −6.733 | NA | 0.85 | P = 0.967 | P = 0.005 |
| Modern Japanese | 42 | 4.8 | 12.520 | 2.188 | 0.81 | 15 | 9.6 | 14.663 | 1.487 | 0.95 | 16 | 14.9 | 9.174 | 7.287 | 0.66 | P = 0.266 | P = 0.075 | |||||||
| 4√Imax | Jomon | 21 | 2.8 | 13.052 | P = 0.789 | 0.530 | P = 0.002 | 0.83 | 23 | 6.4 | 8.544 | P = 0.908 | 1.967 | P = 0.026 | 0.95 | 14 | 10.3 | 15.791 | P = 0.016 | −2.932 | NA | 0.81 | P = 0.091 | P = 0.006 |
| Modern Japanese | 42 | 3.0 | 12.438 | 0.448 | 0.94 | 15 | 6.5 | 8.657 | 1.656 | 0.95 | 16 | 9.5 | 7.402 | 3.396 | 0.81 | P < 0.001 | P = 0.436 | |||||||
| 4√Imin | Jomon | 21 | 2.6 | 11.248 | P = 0.864 | 0.680 | P < 0.001 | 0.79 | 23 | 6.1 | 8.380 | P = 0.931 | 1.761 | P = 0.003 | 0.93 | 14 | 9.6 | 14.678 | P = 0.021 | −2.752 | NA | 0.86 | P = 0.331 | P = 0.008 |
| Modern Japanese | 42 | 2.8 | 11.616 | 0.409 | 0.95 | 15 | 6.1 | 8.461 | 1.400 | 0.97 | 16 | 9.0 | 7.017 | 3.200 | 0.73 | P < 0.001 | P = 0.397 | |||||||
| 4√Ix | Jomon | 21 | 2.7 | 11.282 | P = 0.696 | 0.701 | P = 0.002 | 0.79 | 23 | 6.3 | 9.045 | P = 0.804 | 1.581 | P = 0.031 | 0.95 | 14 | 10.3 | 16.144 | P = 0.014 | −3.277 | NA | 0.83 | P = 0.409 | P = 0.005 |
| Modern Japanese | 42 | 2.9 | 12.140 | 0.371 | 0.95 | 15 | 6.4 | 9.298 | 1.196 | 0.95 | 16 | 9.4 | 8.067 | 2.736 | 0.85 | P = 0.007 | P = 0.439 | |||||||
| 4√Iy | Jomon | 21 | 2.8 | 13.092 | P = 0.624 | 0.503 | P < 0.001 | 0.83 | 23 | 6.3 | 7.777 | P = 0.861 | 2.225 | P = 0.004 | 0.92 | 14 | 9.6 | 14.292 | P = 0.030 | −2.367 | NA | 0.82 | P = 0.085 | P = 0.010 |
| Modern Japanese | 42 | 2.9 | 11.956 | 0.487 | 0.94 | 15 | 6.2 | 7.947 | 1.812 | 0.97 | 16 | 9.1 | 6.317 | 3.896 | 0.65 | P < 0.001 | P = 0.385 | |||||||
| 4√IJ | Jomon | 21 | 3.2 | 14.534 | P = 0.939 | 0.710 | P < 0.001 | 0.82 | 23 | 7.4 | 10.066 | P = 0.846 | 2.222 | P = 0.010 | 0.95 | 14 | 11.9 | 18.007 | P = 0.015 | −3.253 | NA | 0.85 | P = 0.168 | P = 0.005 |
| Modern Japanese | 42 | 3.4 | 14.335 | 0.510 | 0.95 | 15 | 7.5 | 10.277 | 1.784 | 0.96 | 16 | 11.0 | 8.585 | 3.927 | 0.78 | P < 0.001 | P = 0.376 | |||||||
NA, not applicable. P < 0.05 are shown in bold type.
Scatter plots of the two midshaft linear metrics and four cross-sectional geometric properties against relative age (variable LTratio). The horizontal axis corresponds to LTratio. The ○ and × represent Jomon and modern Japanese, respectively. The regression line was drawn in each age group (FET-INF, CHILD, and ADOLESC) when the correlation was statistically significant (solid line, Jomon; dashed line, modern Japanese).
Because the Jomon FET-INF sample is truncated at near birth (i.e. individuals after birth to 1.5 years old are lacking in the Jomon sample), we furthermore investigated Jomon–modern Japanese differences using only the fetal samples. Table 3 shows the ANCOVA results of this analysis, confining the specimens only to those with LTratio values <0.212. We found that the slopes of all variables except ML did not differ significantly between the Jomon and the modern Japanese, and that all variables tended to be significantly greater in the Jomon. These results indicate that the Jomon femur is more robust than that of the modern Japanese at least by the seventh fetal month.
| Variable (mm) | Group | n | Mean | Slope | Y-intercept | R | ||
|---|---|---|---|---|---|---|---|---|
| ML | Jomon | 36 | 6.1 | 37.725 | P = 0.003 | −0.432 | NA | 0.85 |
| Modern Japanese | 178 | 5.4 | 23.394 | 1.449 | 0.71 | |||
| AP | Jomon | 36 | 5.8 | 29.530 | P = 0.420 | 0.641 | P < 0.001 | 0.75 |
| Modern Japanese | 178 | 5.2 | 25.225 | 0.982 | 0.69 | |||
| √CA | Jomon | 19 | 4.7 | 19.982 | P = 0.657 | 1.305 | P < 0.001 | 0.68 |
| Modern Japanese | 29 | 4.4 | 17.135 | 1.359 | 0.67 | |||
| 4√Imax | Jomon | 19 | 2.8 | 14.046 | P = 0.663 | 0.367 | P < 0.001 | 0.78 |
| Modern Japanese | 29 | 2.6 | 12.565 | 0.411 | 0.78 | |||
| 4√Imin | Jomon | 19 | 2.6 | 13.113 | P = 0.679 | 0.374 | P < 0.001 | 0.76 |
| Modern Japanese | 29 | 2.4 | 11.778 | 0.370 | 0.79 | |||
| 4√Ix | Jomon | 19 | 2.6 | 12.914 | P = 0.836 | 0.434 | P < 0.001 | 0.75 |
| Modern Japanese | 29 | 2.5 | 12.216 | 0.349 | 0.78 | |||
| 4√Iy | Jomon | 19 | 2.7 | 14.213 | P = 0.557 | 0.319 | P < 0.001 | 0.78 |
| Modern Japanese | 29 | 2.6 | 12.230 | 0.422 | 0.78 | |||
| 4√IJ | Jomon | 19 | 3.2 | 16.175 | P = 0.669 | 0.441 | P < 0.001 | 0.77 |
| Modern Japanese | 29 | 3.0 | 14.505 | 0.465 | 0.79 | |||
NA, not applicable. P < 0.05 are shown in bold type.
Table 4 shows the results of ANCOVA of the two metrics representing the midshaft cross-sectional shape. Figure 4 illustrates the scatter plots. In both Jomon and modern Japanese, AP/ML and Ix/Iy exhibited significant positive correlations with LTratio when all three age groups were pooled for analysis. This shows that femoral midshaft cross-sectional shape and bone distribution change modally from a slight mediolaterally broad shape to an anteroposteriorly elongated shape through growth. Another observation worthy of note is the considerable shape variation that characterizes the fetal period (Figure 4). Such large degrees of variation were apparently maintained through later growth periods.
| Variable | Group | Fetal–infant period (FET-INF) | Toddler–prepubertal period (CHILD) | Pubertal–subadult period (ADOLESC) | Difference of slopes | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Mean | Slope | Y-intercept | R | n | Mean | Slope | Y-intercept | R | n | Mean | Slope | Y-intercept | R | FET-INF vs. CHILD | CHILD vs. ADOLESC | ||||||||
| AP/ML | Jomon | 41 | 0.94 | NS | NA | NS | P = 0.007* | −0.20 | 50 | 1.00 | 0.451 | P = 0.684 | 0.776 | P = 0.749 | 0.60 | 21 | 1.14 | NS | NA | NS | P = 0.042* | 0.26 | NA | NA |
| Modern Japanese | 199 | 0.97 | NS | NS | 0.07 | 29 | 1.01 | 0.383 | 0.805 | 0.44 | 36 | 1.08 | 0.427 | 0.719 | 0.38 | NA | P = 0.854 | |||||||
| Ix/Iy | Jomon | 21 | 0.85 | NS | NA | NS | P = 0.084* | −0.19 | 23 | 1.01 | 0.864 | P = 0.929 | 0.565 | P = 0.478 | 0.64 | 14 | 1.32 | NS | NA | NS | P = 0.078* | 0.17 | NA | NA |
| Modern Japanese | 42 | 0.90 | NS | NS | 0.20 | 15 | 1.09 | 0.913 | 0.581 | 0.44 | 16 | 1.15 | NS | NS | 0.30 | NA | NA | |||||||
NS, not significant; NA, not applicable. The slope and y-intercept are shown when the correlation was statistically significant or near-significant (P < 0.1). The near-significant R-value is underlined. P < 0.1 are shown in bold type.
Scatter plots of the two metrics representing midshaft cross-sectional shape against relative age (variable LTratio). The horizontal axis corresponds to LTratio. The ○ and × represent Jomon and modern Japanese, respectively. The regression line was drawn in each age group (FET-INF, CHILD, and ADOLESC) when the correlation was statistically significant or near-significant (P < 0.1). Solid line, Jomon; dashed line, modern Japanese.
When the developmental periods were examined separately, we found that, in the FET-INF age group, AP/ML and Ix/Iy did not correlate significantly with LTratio. This was the case in both Jomon and modern Japanese, indicating that cross-sectional shape did not change during the FET-INF period. Both indices were significantly or near-significantly (P = 0.084) smaller in Jomon than in modern Japanese, indicating that the fetal to infant Jomon femoral midshaft tends to be broader mediolaterally than that of the modern Japanese. In the CHILD age group, AP/ML and Ix/Iy exhibited significant or near-significant (P = 0.098) positive correlations with LTratio in both Jomon and modern Japanese. The slopes and intercepts of these indices did not differ significantly between Jomon and modern Japanese, indicating similar cross-sectional shape throughout childhood. In the ADOLESC age group, AP/ML and Ix/Iy exhibited weak positive correlations with LTratio, although this was significant at the P < 0.05 level only in the AP/ML of the modern Japanese. These indices were significantly or near-significantly (P = 0.078) greater in Jomon than in modern Japanese, corresponding to the Jomon femoral midshaft cross-sectional shape that tends to be markedly anteroposteriorly elongated in the ADOLESC period.
Bone modeling patternTable 5 shows the results of ANCOVA of the variables reflecting bone modeling patterns against LTratio. Figure 5 illustrates the scatter plots. For TA, which reflects the amount of subperiosteal bone apposition, the slopes of both the FET-INF and CHILD age groups did not differ significantly between Jomon and modern Japanese, while the ADOLESC age group slope was significantly greater in the Jomon. These results indicate that the rates of subperiosteal bone formation are similar in Jomon and modern Japanese before around puberty, but that during the ADOLESC period an acceleration of bone formation occurs in the Jomon. For MA, which reflects the amount of endosteal bone resorption, the slope of the FET-INF age group did not differ significantly between Jomon and modern Japanese. The slope of the CHILD age group was near-significantly (P = 0.074) smaller in Jomon than in modern Japanese, and there were no significant differences between the slopes of the CHILD and ADOLESC age groups in either Jomon or modern Japanese. These results indicate that the rates of endosteal bone resorption are similar in Jomon and modern Japanese during fetal and infant ages, while the Jomon medullary cavity expands at a slower rate than that of the modern Japanese throughout subsequent growth. As a consequence, the Jomon femoral midshaft medullary cavity tends to be consistently smaller than that of the modern Japanese after early childhood.
| Variable | Group | Fetal–infant period (FET-INF) | Toddler–prepubertal period (CHILD) | Pubertal–subadult period (ADOLESC) | Difference of slopes | |||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| n | Mean | Slope | Y-intercept | R | n | Mean | Slope | Y-intercept | R | n | Mean | Slope | Y-intercept | R | FET-INF vs. CHILD | CHILD vs. ADOLESC | ||||||||
| √TA (mm) | Jomon | 21 | 5.20 | 24.764 | P = 0.686 | 0.870 | P < 0.001 | 0.85 | 23 | 12.10 | 15.141 | P = 0.580 | 4.262 | P = 0.113 | 0.93 | 14 | 18.87 | 28.410 | P = 0.023 | −4.982 | NA | 0.86 | P = 0.082 | P = 0.003 |
| Modern Japanese | 42 | 5.63 | 26.457 | 0.202 | 0.96 | 15 | 12.36 | 16.143 | 3.401 | 0.97 | 16 | 17.88 | 14.548 | 5.849 | 0.78 | P < 0.001 | P = 0.611 | |||||||
| √MA (mm) | Jomon | 21 | 2.06 | 24.644 | P = 0.483 | −2.242 | P = 0.630 | 0.92 | 23 | 6.73 | 3.606 | P = 0.074 | 4.863 | P = 0.003 | 0.43 | 14 | 8.37 | 7.544 | P = 0.352 | 2.038 | P < 0.001 | 0.48 | P = 0.002 | P = 0.347 |
| Modern Japanese | 42 | 2.85 | 28.337 | −2.962 | 0.93 | 15 | 7.72 | 7.516 | 3.549 | 0.86 | 16 | 9.87 | 12.655 | −0.590 | 0.71 | P < 0.001 | P = 0.127 | |||||||
| CA/TA | Jomon | 21 | 0.84 | −2.312 | P = 0.846 | 1.245 | P = 0.407 | −0.68 | 23 | 0.68 | 0.501 | P = 0.245 | 0.422 | P < 0.001 | 0.60 | 14 | 0.80 | NS | NA | NS | P < 0.001* | 0.31 | P < 0.001 | NA |
| Modern Japanese | 42 | 0.75 | −2.457 | 1.255 | −0.76 | 15 | 0.60 | 0.309 | 0.430 | 0.66 | 16 | 0.69 | NS | NS | −0.32 | P < 0.001 | NA | |||||||
| TA/MA | Jomon | 21 | 7.16 | −101.430 | P = 0.173 | 24.888 | P = 0.893 | −0.68 | 23 | 3.31 | 4.668 | P = 0.118 | 0.893 | P < 0.001 | 0.60 | 14 | 5.18 | NS | NA | NS | P < 0.001* | 0.31 | P < 0.001 | NA |
| Modern Japanese | 42 | 5.74 | −46.936 | 15.363 | −0.76 | 15 | 2.56 | 1.954 | 1.478 | 0.66 | 16 | 3.38 | NS | NS | −0.32 | P < 0.001 | NA | |||||||
NS, not significant; NA, not applicable. The slope and y-intercept are shown when the correlation was statistically significant or near-significant (P < 0.1). The near-significant R-value is underlined. P < 0.1 are shown in bold type.
Scatter plots of the variables reflecting bone modeling patterns against relative age (variable LTratio). The horizontal axis corresponds to LTratio. The ○ and × represent Jomon and modern Japanese, respectively. The regression line was drawn in each age group (FET-INF, CHILD, and ADOLESC) when the correlation was statistically significant or near-significant (P < 0.1). Solid line, Jomon; dashed line, modern Japanese. One individual of the modern Japanese is not shown in the TA/MA scatter plot due to its outlier position (LTratio = 0.164, TA/MA = 30.8).
Variable CA/TA reflects the proportion of bone within the midshaft cross section. CA/TA did not differ significantly between Jomon and modern Japanese in the FET-INF age group, while bone proportion was significantly greater in the Jomon in both the CHILD and ADOLESC age groups. Variable TA/MA expresses the proportion of medullary cavity within the midshaft cross section. The results show a distinctive change of bone formation pattern at the FET-INF to CHILD transitional period. Until around birth or toddler age, relative medullary cavity expansion outpaced relative subperiosteal bone apposition, so that medullary proportion increased. Thereafter, in both Jomon and modern Japanese, subperiosteal bone deposition considerably outpaced medullary bone resorption, so that relative medullary proportion decreased. The TA/MA values of the modern Japanese appeared to reach a plateau at about the end of the CHILD period, while those of the Jomon continued to increase in ADOLESC, corresponding to the increasingly robust Jomon femoral midshaft in the ADOLESC age group.
In the present study, we confirmed that the Jomon femoral midshaft cross sections are significantly larger, more robust, and stronger under mechanical loading than those of the modern Japanese, and found that this is consistent throughout the examined fetal to subadult developmental and growth periods. Similar differences in femoral midshaft geometric properties have been reported between adult Jomon and modern Japanese (Kimura and Takahashi, 1982; Kimura, 2006). Thus, the robust Jomon morphology (larger subperiosteal diameter, greater bone mass, larger area and polar moments of inertia) can be considered to be a characteristic of the Jomon femur that is consistent throughout fetal to adult life.
Importantly, we found that robusticity of the Jomon femoral midshaft is greater than that of the modern Japanese from as early as late fetal life. At the same time, subperiosteal deposition and endosteal resorption rates in the fetal period were found to be comparable between the two populations. These observations suggest the presence of population-specific genetic signals that influence early fetal bone mass and shape, which then continue to be expressed through later growth periods via common growth trajectories underlain by conserved regulatory mechanisms shared among populations (Chiu and Hamrick, 2002; Lovejoy et al., 2003; Pearson and Lieberman, 2004; Ruff et al., 2006a). One hypothesis that can be tested by future studies would be that populational differences in long-bone diaphyseal robusticity derive from subtle differences in mesenchymal and/or cartilaginous long-bone template size and shape, the latter established early in development (Scheuer and Black, 2000).
In the toddler to childhood growth periods, increased cortical bone thickness and area were attained by a considerably greater rate of subperiosteal bone apposition relative to that of medullary cavity expansion. This general pattern was common to both Jomon and modern Japanese, and the two populations exhibited comparable subperiosteal diameter growth rates. However, we found that medullary expansion rate is significantly lower in the Jomon than in the modern Japanese femora, accentuating femoral diaphyseal bone mass and robusticity in the toddler to childhood period of the Jomon (see, for example, the elevated regression line of CA against LTratio in Figure 3). Recent field work on the Cameroon Baka pygmy (Hagino and Yamauchi, 2014) reported that the physical activity levels of hunting-gathering subsistence children are probably considerably greater than in urbanized children. It is possible that greater physical activity results in lower rates of medullary cavity bone resorption, which may have resulted in the comparatively smaller endosteal dimensions of femora of the Jomon toddler to childhood age group. However, because outer subperiosteal growth trajectories (ML, AP, TA) did not differ significantly between Jomon and modern Japanese, alternatively, endosteal resorptive response of the two populations may have differed from factors other than activity and/or loading levels (e.g. genetic, endocrine, and/or metabolic).
Finally, while the rate of subperiosteal bone formation in the modern Japanese femoral midshaft remained largely constant throughout the toddler to subadult growth periods, that of the Jomon became significantly enhanced after around puberty. This effect is clearly seen in the various scatter plots of Figure 3, where the variables representing bone mass and mechanical strength show a clear slope difference when plotted against LTratio. These observations suggest that there is a post-adolescent difference in modeling response of subperiosteal bone apposition that is enhanced in the Jomon compared to the modern Japanese. This could have been caused either by greater activity level and/or mechanical load in the adolescent period Jomon, or from enhanced genetic/endocrine-based subperiosteal responsiveness in the adolescent Jomon femora.
In order to further examine the Jomon diaphyseal bone growth pattern of the adolescent period, the total subperiosteal area was divided into anterior and posterior halves (AntTA and PostTA, respectively), and the anteroposterior outer diameter was divided into its anterior and posterior portions (AntAP and PostAP, respectively). This was done in relation to the mediolateral axis passing through the centroid of the medullary cavity (Figure 6, Figure 7). In both Jomon and modern Japanese, there were no significant growth pattern differences between the AntTA and PostTA throughout development; an increase in the outer diameter of the Jomon femoral midshaft occurs comparably in the anterior and posterior shafts. However, with regards to AntAP and PostAP, although growth pattern differences were insignificant between Jomon and modern Japanese in AntAP, the Jomon PostAP of the ADOLESC period was particularly enhanced (Figure 7). This difference can be quantified more explicitly as a shape difference, in the form of a ratio between the mediolateral widths and anteroposterior lengths of the anterior and posterior shafts (Figure 8). In the ADOLESC period, although there were no significant differences between Jomon and modern Japanese in the anterior shaft, the Jomon posterior shaft tended to be posteriorly elongated by a significant degree (P < 0.05). These observations indicate that the strong anteroposterior elongation of the Jomon femoral midshaft cross section is caused primarily by bone mass distributional change of the posterior shaft, from a semielliptical to a posteriorly elongated and acute-angled cross-sectional shape during adolescence. This pattern of posterior shaft bone modeling corresponds to the well-known femoral shaft pilaster, morphologically expressed as a posteriorly projecting linea aspera region.
Scatter plots of the anterior and posterior halves of total subperiosteal area (AntTA and PostTA, respectively) against relative age (variable LTratio). The horizontal axis corresponds to LTratio. The ○ and × represent Jomon and modern Japanese, respectively. The regression line was drawn in each age group (CHILD and ADOLESC) when the correlation was statistically significant (solid line, Jomon; dashed line, modern Japanese). The cross section was divided in relation to the mediolateral axis passing through the centroid of the medullary cavity.
Scatter plots of the anterior and posterior portions of anteroposterior outer diameter (AntAP and PostAP, respectively) against relative age (variable LTratio). The horizontal axis corresponds to LTratio. The ○ and × represent Jomon and modern Japanese, respectively. The regression line was drawn in each age group (CHILD and ADOLESC) when the correlation was statistically significant (solid line, Jomon; dashed line, modern Japanese). The cross section was divided in relation to the mediolateral axis passing through the centroid of the medullary cavity.
Scatter plots of the outer shapes of anterior and posterior shafts against relative age (variable LTratio). The horizontal axis corresponds to LTratio. The ○ and × represent Jomon and modern Japanese, respectively. In the CHILD age group, the regression line was drawn when the correlation was statistically significant (solid line, Jomon). The horizontal line of the ADOLESC age group represents the mean value (solid line, Jomon; dashed line, modern Japanese). The cross section was divided in relation to the mediolateral axis passing through the centroid of the medullary cavity. NS, not significant.
In an experimental study of myostatin-knockout mice femur (Hamrick et al., 2000), lateral projection of the dominant enthesis (third trochanter) increased dramatically, and this was interpreted as being related to muscle and tendinous tissue hypertrophy (~90% increase of quadriceps muscle mass). On the contrary, Rabey et al.’s (2015) experiment on mice humeri showed that the comparatively smaller degree of muscle mass enhancement (10–20%) accompanying volunteer exercise did not significantly affect enthesis shape or dimensions. Hamrick et al. (2000) suggested that, in the extreme case of myostatin-knockout mice, stretching of collagen fibers and periosteum accompanying muscle fiber hypertrophy might have resulted in the enthesis size and shape changes. In the case of modern humans, the thigh muscle group (quadriceps femoris, biceps femoris, and several adductors) inserts at the linea aspera, and their hypertrophy peaks during adolescence (Ruff, 2003). Thus, it is tempting to suggest that the anteroposteriorly elongated mid-diaphyseal cross section of the Jomon femur (i.e. the pilaster) is the result of bone apposition at the linea aspera region in response to extreme thigh muscle growth after around puberty, perhaps the degree of such periosteal response in part depending on genetic background. These observations suggest a complex underlying cause for the enhanced Jomon femoral midshaft robusticity after around puberty. It appears that at least a part of this bone mass increase and geometric distribution relates to muscle mass and attachment patterns, but not to load. At the same time, an increased muscle mass would associate with elevated activity levels and acting forces, and this would potentially enhance subperiosteal bone apposition along the entire diaphysis during growth via increased load. The latter mechanical factor is the conventionally preferred explanation (bone functional adaptation), although experimental (Wallace et al., 2014, 2015) and empirical (Warden et al., 2014) studies suggest lack of correspondence between load/strain levels and location of bone apposition.
In the present study, we also found that the Jomon femoral midshaft cross sections tend to be broader mediolaterally than those of the modern Japanese during late fetal life and infancy, while this difference disappears soon after toddler age. According to previous studies (Digby, 1916; Pritchett, 1992), the growth rate (in the axial direction) of the distal metaphyseal cartilage plate is significantly greater than that of the proximal metaphysis, by a ratio of approximately 7:3. For this reason, the fetal midshaft position ‘moves’ relatively proximally during growth and eventually lies at the subtrochanteric level in adults. It has been widely acknowledged that the subtrochanteric cross section of the adult Jomon femur is mediolaterally more elongated than that of the modern Japanese (Sato, 1918; Kiyono and Hirai, 1928; Takahashi, 1982; Kohara et al., 2011). This morphology, the platymeric proximal femoral shaft (mediolaterally broad and anteroposteriorly flattened subtrochanteric cross-sectional shape), often seen in prehistoric hunter-gatherers (e.g. Trinkaus and Ruff, 2012), has been considered a result of bone functional adaptation to mechanical load (bending in the mediolateral direction) (Trinkaus, 1976; Ruff and Hayes, 1983; Ruff et al., 1984; Ruff, 1987; Bridges, 1995; Ruff, 2005). However, the shape correspondence that we observed between equivalent fetal and adult section positions (midshaft in fetuses and subtrochanteric in adults) suggests that initial femoral template (mesenchymal or cartilaginous femur) shape may in part underlie these adult subtrochanteric shape characteristics. Further investigations are needed to test this hypothesis, for example by using the nutrient foramen to determine the position of the primary ossification center (cf. Digby, 1916; Lee, 1968; Pritchett, 1991, 1992).
To summarize, the greater robusticity of the Jomon femoral midshaft cross section compared to the modern Japanese condition can be considered to be caused by multiple factors: (1) greater bone diameter and mass established probably in early fetal life, (2) lower endosteal bone resorption rates from toddler through subadult ages, (3) greater subperiosteal bone expansion after around puberty, and (4) bone apposition possibly related to muscle hypertrophy at the linea aspera region during adolescence. Thus, complex interactions of population-specific genetic background, differential response to activity level and/or mechanical load, and activity-induced factors such as mechanical load and/or muscle hypertrophy appear to have caused the Jomon condition of a comparatively robust femoral diaphysis and strong tendency for pilastric shape.
We thank R.T. Kono (National Museum of Nature and Science), Y. Dodo (Tohoku University School of Medicine), K. Katayama (Graduate School of Science, Kyoto University), T. Nakahashi (Graduate School of Social and Cultural Studies, Kyushu University), and S. Iwanaga (The Kyushu University Museum) for access to the collections in their care. We also thank two anonymous reviewers, and Professor T. Kimura and Professor C.O. Lovejoy for comments and suggestions on a previous draft of the manuscript.
