G Model JCHB-25336; No. of Pages 11
ARTICLE IN PRESS
HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
HOMO - Journal of Comparative Human Biology journal homepage: www.elsevier.com/locate/jchb
Estimation of sex from the hyoid body in skeletal individuals from archeological sites R. D’Anastasio a, J. Viciano a,∗, M. Di Nicola b, D.T. Cesana a, M. Sciubba a, M. Del Cimmuto a, A. Paolucci a, A. Fazio a, L. Capasso a a
University Museum, State University “G. d’Annunzio”, Piazza Trento e Trieste 1, 66100 Chieti, Italy Laboratory of Biostatistics, Department of Biomedical Science, State University “G. d’Annunzio”, Via dei Vestini, 66100 Chieti, Italy b
a r t i c l e
i n f o
Article history: Received 8 June 2013 Accepted 29 January 2014 Available online xxx
a b s t r a c t Recent forensic studies have shown that the hyoid bone is a sexually dimorphic element of the human skeleton. Given the advanced techniques of collecting human remains in archeological and forensic contexts, the recovery of hyoid bones is now more frequent in skeletal samples. For that reason the authors propose a new method for estimating sex based on hyoid bodies from archeological sites. The study has been conducted on well-preserved hyoids of skeletal remains of 64 adult individuals (44 males and 20 females) dated from the pre-Roman to the medieval periods. The authors considered 10 linear measurements of the hyoid body. The most significant measurements showing sexual dimorphism are the body height, body length, and the maximum and minimum diameter of the articular facet for the greater horn. Discriminant function analysis achieved the allocation accuracy between 75.0% and 88.0%, depending on the measurement collected. This method represents a new, useful and easy way for increasing biological information when assessing the sex of adult human remains from an archeological sample. © 2014 Elsevier GmbH. All rights reserved.
∗ Corresponding author. Tel.: +39 0871 3553502; fax: +39 0871 410927. E-mail address: [email protected] (J. Viciano). http://dx.doi.org/10.1016/j.jchb.2014.01.002 0018-442X/© 2014 Elsevier GmbH. All rights reserved.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11 2
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Introduction Sex estimation of human skeletal remains represents a crucial stage in any palaeoanthropological, archeological or forensic study. Sexually dimorphic differences between males and females have been quantified in numerous ways in physical anthropology and osteoarcheology based on both morphological and metric point features for most of the human skeletal elements. In an archeological context, the accuracy in sex determination depends on the integrity of the skeletal elements because of the usual fragmented state of preservation of human remains. For this reason the possibility of collecting as much information as possible from every element available is very important. The recent discovery of hominin hyoid bones, including those of Australopithecus, Homo neanderthalensis, Homo heidelbergensis and other archaic humans (Alemseged et al., 2006; Arensburg et al., 1989; Capasso et al., 2008; Martínez et al., 2008; Rodríguez et al., 2003), and the increasing recovery of hyoids from archeological contexts has reawakened interest in this bony element and offers new opportunities to predict sex from both fused and unfused hyoids. The hyoid bone is a sexually dimorphic skeletal element (Jelisiejew et al., 1968; Kim et al., 2006; Miller et al., 1998; Reesink et al., 1999; Urbanová et al., 2013) that consists of a body and greater and lesser horns that articulate with the body. It is an important anchor for the tongue and other muscles that are required for swallowing and speaking, and provides a supporting structure for the larynx. The greater horns can fuse to the body, but the anatomical reason behind the fusion is still somewhat unclear: it could be related to the age (O’Halloran and Lundy, 1987), even if some hyoids remain unfused throughout old age. However, sex seems not to be correlated with the bilateral fusion of the greater horns to the body (Miller et al., 1998). The possibility of determining sex from the hyoid has been pursued for a long time through analyses of morphological and metric differences between males and females. Various authors have considered fused hyoids as a useful tool for probabilistic sex diagnosis (Jelisiejew et al., 1968; Kim et al., 2006; Miller et al., 1998; Reesink et al., 1999). In 2010 Kindschuh and co-workers presented a method for determining sex from unfused hyoids (Kindschuh et al., 2010). The hyoid body alone also displays sexual dimorphism, and thus it is useful for sex diagnosis (Reesink et al., 1999). Previous researchers investigated the utility of the hyoid as a reliable sex indicator, but they conducted metric analyses of forensic samples indirectly through X-ray images (Komenda and ˇ ´ 1990; Miller et al., 1998; Reesink et al., 1999). Kim et al. (2006) quantified sexual dimorphism Cern y, of the hyoid bone in Koreans using digital photographs and obtained a discriminant function with an accuracy of 88.2%. In a more recent study researchers attempted to obtain measurements directly from hyoid bones (Kindschuh et al., 2010) in the Terry Collection, an American skeletonized cadaver series of identified American Whites and Blacks collected since 1898 in the Anatomy Department at Washington University, St. Louis, Missouri (Hunt and Albanese, 2005). Our study represents the first attempt to identify metric dimorphism in hyoid body from an archeological sample. In particular, we employed new measurements and discriminant functions were developed for predicting sex.
Materials and methods The sample consists exclusively of complete hyoid bodies, without ossified cartilaginous tissue or fused horns. The specimens belong to 64 individuals from five archeological sites, representing the geographical region of Central Italy: Opi (VI-V century BCE), Canne della Battaglia (III century BCE), Herculaneum (I century CE), Teramo (VII-XII century CE), and Monte D’Argento (XI-XV century CE). The Herculaneum site represents a homogeneous and unique sample of a Roman town whose inhabitants died contemporaneously during the eruption of the volcano Vesuvius in the year 79 CE (Capasso, 2001). The other sites come from the Abruzzo region or neighboring areas, whose chronological contexts range from pre-Roman to the Medieval periods (Capasso et al., 1990; D’Anastasio et al., 2004; D’Anastasio and Vitullo, 2008). All these individuals are housed in the University Museum of Chieti (Italy). Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
3
Fig. 1. Hyoid body measurements (A: frontal view; B: superior view; C: lateral view). Refer to Table 1 for a description of the measurements which correspond to each acronym.
The biological profile of each individual was previously determined considering all the available bones. Sex was estimated following standard descriptive and metric criteria (Ferembach et al., 1980; Murail et al., 2005) according to cranial and pelvic features. The estimated age at death was based on dental development (Ubelaker, 1989), the degree of dental wear (Lovejoy et al., 1985; Miles, 1963), and the appearance of the pubic symphyseal surface (Katz and Suchey, 1986) and the ilium auricular surface (Buckberry and Chamberlain, 2002). We considered 10 linear measurements. All of them were taken using a standard digital caliper with an accuracy of 0.01 mm (Fig. 1 and Table 1). Data measurements were first assessed for normality using the Kolmogorov–Smirnov’s one-sample test and for homogeneity of variance using the Levene test. Afterwards, we analyzed the differences between the mean values of two measurements (FMax and FMin) to search for metric asymmetry (differences between measurements taken on the right and left side) using the paired Student’s t-test.
Table 1 Measurementsa of the hyoid body. Measurement
Description
BHb BLMaxb BLMinc BLAvb ThMaxc ThMedc ThBordc FMaxc FMinc LLHornb
Body height: maximum body height in the medial plane Body length: maximum length of the body in the transverse plane Minimum transverse diameter: minimum length of the body in the transverse plane Average transverse diameter: average length of the body in the transverse plane Maximum thickness of the total body Antero-posterior thickness: maximum antero-posterior thickness of the body in the medial plane Thickness of body’s inferior border Maximum diameter of the right and left articular facet for the greater horn Minimum diameter of the right and left articular facet for the greater horn Length of the lesser horn of the body
a b c
All measurements are in millimeters. Measurements applied in previous studies (Kim et al., 2006; Kindschuh et al., 2010; Reesink et al., 1999). New measurements used in the current study.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11 4
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Table 2 Distribution of the sample according to the archeological sites, age groups and sex. Century
Site of origin Opi Canne della Battaglia Herculaneum Teramo Monte D’Argento
Females
Males
VI-V BCE III BCE I CE VII-XII CE XI-XV CE
Age at death 20–40 years 40–60 years
Sexes pooled
N
%
N
%
6 2 29 6 1
100.0 66.7 59.2 100.0 50.0
– 1 18 – 1
– 33.3 40.8 – 50.0
6 3 47 6 2
25 19
71.4 65.5
10 10
28.6 34.5
35 29
The differences between the mean values in all measurements collected at two different times were also analyzed in order to assess possible intra- and interobserver error. Lin’s concordance correlation coefficient (CCC) (Lin, 1989, 2000) was computed to determine the level of agreement between repeated measurements collected by the same observer and by different observers. To assess the degree of agreement for a given set of data, the calculated CCC is compared to the strength of agreement criteria proposed by McBride (2005), which establishes four levels of qualitative assessment as follows: almost perfect, for CCC values greater than 0.99; substantial, from 0.95 to 0.99; moderate, from 0.90 to 0.95; and poor, for CCC values below 0.90. The main effects of the age at death and the archeological site of the individuals on various measurements were tested by two-way ANOVA. Next, the differences between the mean values of males and females were analyzed using the independent Student’s t-test in cases where the normality and homogeneity of variance are fulfilled. Next, a stepwise discriminant function analysis was performed to create equations to estimate sex. In order to maximize the applicability in archeological cases, when the skeletal remains are in poor condition, the functions were calculated for a maximum combination of two measurements. To test the consistency of prediction accuracy, the data were subjected to a cross-validation procedure. All statistical analyses were performed using the SPSS 15.0 software (SPSS Inc., Chicago, IL, USA). Finally, the formulae published by Kindschuh et al. (2010) for sex estimation developed in a knownsex osteological collection were applied to the present archeological samples in order to evaluate its applicability. Results The hyoid bones belong to 64 adult individuals (44 males and 20 females; Table 2). Table 3 shows the percentage of preservation of the crania and pelves of the specimens studied, on which descriptive and metric criteria were used to estimate sex. According to the estimated ages at death, the sample
Table 3 Degree of preservation of the cranium and pelvis in the specimens studied. Degree of preservation
Cranium N
Pelvis %
N
%
Optimal Good Discreet Mediocre Poor
42 15 6 1 0
65.6 23.4 9.4 1.6 0.0
42 10 4 3 5
65.6 15.6 6.3 4.7 7.8
Total
64
100.0
64
100.0
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
5
Table 4 The paired t-test evaluating metric bilateral asymmetry. Pair of measurements
N
Diff
SE
t
P
FMax(right)–FMax(left) FMin(right)–FMin(left)
22 23
0.0268 −0.1535
0.4985 0.6271
0.252 −1.174
0.803 0.253
N – number of pairs; Diff – mean difference between pairs; SE – standard error; t – Student’s t-test; P – P-value.
was divided into two groups (modified from Vallois, 1960): 20–40 years (35 young adults), and 40–60 years (29 old adults) (Table 2). Regarding the asymmetry analysis, FMin and FMax were analyzed and none of these measurements show statistically significant differences at P < 0.05 level regarding laterality (Table 4). For this reason, the average between both right and left sides of these measurements was calculated and it was used for the subsequent statistical analyses. A total number of 48 hyoid bones (representing the 75% of the total sample) were used as subsample for intra- and inter-observer error analysis. Table 5 shows the differences between mean values and the concordance correlation coefficient (CCC) for repeated measurements. In the intra-observer error analysis, the CCC value ranges from 0.985 to 0.999 (from substantial to almost perfect). With regard to the inter-observer error analysis, the value of CCC ranges from 0.955 to 0.999 (from substantial to almost perfect). In general, the results show that hyoid body measurements are less concordant between different observers than within the same observer. Considering the entire sample (sex and age groups pooled), the Kolmogorov–Smirnov’s test showed that all the measurements were normally distributed. The results of homogeneity of variance tests indicated that the sample was statistically homogeneous for all the measurements compared.
Table 5 Lin’s concordance correlation coefficient (CCC) to evaluate concordance between repeated measurements collected by the same observer (intraobserver error) and by different observers (interobserver error). N
Measurement 1
Measurement 2
Diff
CCC
Strength of agreementa
Mean
SD
Mean
SD
Same observer 46 BH 19 BLMax 19 BLMin 18 BLAv 20 ThMax ThMed 46 ThBord 46 FMax 26 28 FMin 12 LLHorn
10.5957 21.6747 17.5150 19.2117 8.5425 4.9793 1.9498 6.0487 3.7179 2.1850
1.1274 2.3927 2.3830 1.6729 1.6916 0.9065 0.4376 0.9237 0.5331 0.5234
10.6111 24.6774 17.5022 19.1872 8.5370 4.9798 1.9550 6.0715 3.7448 2.1700
1.1223 2.3998 2.3917 1.6828 1.7166 0.9112 0.4255 0.9305 0.5470 0.4860
−0.0154 −0.0027 0.0128 0.0245 0.0055 −0.0005 −0.0052 −0.0228 −0.0269 0.0150
0.999 0.999 0.998 0.996 0.992 0.996 0.991 0.995 0.985 0.997
Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Substantial Almost perfect
Different observers BH 46 19 BLMax BLMin 19 BLAv 18 20 ThMax 46 ThMed 46 ThBord 26 FMax 28 FMin 12 LLHorn
10.5628 21.7168 17.5022 19.1944 8.5222 4.9222 1.8817 6.0102 3.6632 2.1483
1.1275 2.4362 2.3917 1.6641 1.7165 0.9222 0.4630 0.9232 0.5467 0.5576
10.6034 21.6461 17.5086 19.1994 8.5398 4.9796 1.9524 6.0601 3.7313 2.1775
1.1245 2.3961 2.3871 1.6776 1.7038 0.9085 0.4306 0.9267 0.5395 0.5044
−0.0406 0.0407 −0.0064 −0.0050 −0.0178 −0.0574 −0.0707 −0.0499 −0.0681 −0.0292
0.998 0.999 0.999 0.994 0.999 0.997 0.955 0.996 0.987 0.989
Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Substantial Almost perfect Substantial Substantial
N – number of measurements; Mean – overall measurement mean; SD – standard deviation; Diff – mean difference between repeated measurements; CCC – Lin’s concordance correlation coefficient value. a See Mcbride (2005) for the strength of agreement criteria.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model
ARTICLE IN PRESS
JCHB-25336; No. of Pages 11 6
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Table 6 Two-way ANOVA for evaluating differences between archeological sites and age groups. df
MS
F
P
7.251 1.934 2.189
4 1 2
1.813 1.934 1.095
1.241 1.325 0.750
0.306 0.255 0.478
BLMax Site Age Site × age
5.543 0.130 9.262
2 1 2
2.772 0.130 4.631
0.543 0.025 0.908
0.590 0.875 0.420
BLMin Site Age Site × age
4.270 1.255 10.893
3 1 2
1.423 1.255 5.447
0.298 0.263 1.140
0.827 0.614 0.341
BLAv Site Age Site × age
0.601 0.124 12.396
2 1 2
0.301 0.124 6.198
0.101 0.042 2.090
0.904 0.840 0.151
ThMax Site Age Site × age
10.594 1.308 4.792
2 1 2
5.297 1.308 2.396
2.969 0.733 1.343
0.074 0.402 0.284
ThMed Site Age Site × age
6.207 1.166 0.885
4 1 2
1.552 1.166 0.443
1.971 1.482 0.562
0.114 0.229 0.574
ThBord Site Age Site × age
0.190 0.000 0.251
4 1 2
0.047 0.000 0.126
0.240 0.002 0.633
0.915 0.969 0.535
FMax Site Age Site × age
1.019 0.009 0.015
3 1 2
0.340 0.009 0.007
0.407 0.011 0.009
0.749 0.919 0.991
FMin Site Age Site × age
1.345 0.092 0.037
3 1 2
0.448 0.092 0.019
1.146 0.236 0.048
0.346 0.631 0.953
LLHorn Site Age Site × age
0.552 0.419 0.204
3 1 1
0.184 0.419 0.204
0.648 1.475 0.716
0.597 0.245 0.412
Measurements BH Site Age Site × age
SS
SS – sum of squares; df – degrees of freedom; MS – mean square; F – F-score statistic; P – P-value.
Results of the two-way ANOVA revealed no statistically significant differences among age group categories and/or archeological sites (P > 0.05) (Table 6). Table 7 shows the sample size, mean and standard deviation, t value, and the degree of significance of the differences between the male and female individual means of all measurements. Among the 10 measurements, 8 of them show a greater mean in males than in females; these differences are statistically significant at P < 0.05 level. The most sexually dimorphic measurements are BH, BLMax, FMax, and FMin (P < 0.001). Next comes BLAv, ThMed, BLMin, and ThMax (P < 0.01). There are no significant differences in ThBord and LLHorn measurements. Table 8 shows the coefficients indicating the contribution of each measurement, the classification functions of the two groups analyzed, the discriminant functions and their corresponding sectioning Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
7
Table 7 Descriptive statistics and t-test results for mean differences between the sexes. Measurement
BH BLMax BLMin BLAv ThMax ThMed ThBord FMax FMin LLHorn
Males
Females
N
Mean
SD
N
Mean
SD
40 17 18 17 19 40 40 26 26 15
10.963 23.574 18.743 20.402 9.300 5.185 1.904 6.228 3.913 2.109
1.131 1.801 2.031 1.584 1.308 0.881 0.450 0.748 0.543 0.588
18 8 9 9 7 18 18 11 12 5
9.802 20.620 16.372 18.144 7.796 4.409 1.738 5.146 3.209 1.788
0.885 1.336 1.790 1.297 1.132 0.759 0.369 0.583 0.443 0.480
t
P
3.851 4.117 2.967 3.666 2.686 3.234 1.369 4.270 3.919 1.098
0.000 0.000 0.007 0.001 0.013 0.002 0.177 0.000 0.000 0.287
N – number of individuals; Mean – overall measurement mean; SD – standard deviation; t – Student’s t-test; P – P-value.
points, and the values of F and Wilks’s lambda. The functions whose discriminant power fell below 75% were excluded as they are of little utility. Allocation accuracy of these functions is presented in Table 9. It is seen that the allocation accuracy ranges from 76.0% to 84.2% in males and 71.4% to 100% in females. Thus, females are generally classified more accurately than males by discriminant functions. For both sexes pooled, allocation accuracy ranges between 75.0% and 88.0%. Cross-validation results have not significantly changed the original accuracy (from 73.1% to 88.0%). Of the 10 functions developed, only one (Function 4) appears in the univariate analysis, with a correct allocation of sex of 80.0%. The multivariate analysis provides a greater advantage than the univariate analysis because seven multivariate functions provide highly reliable sex estimations (from 75.0% to 88.0%). Two discriminant function formulae developed by Kindschuh et al. (2010) were applied to the study sample. Table 10 shows the results of the number of individuals correctly classified. Accuracy ranges from 74.1% to 77.8% (from 58.8% to 64.7% in males and 100.0% in females). Discussion The hyoid bone has been the subject of numerous studies, especially in forensic contexts, where it can be indicative of traumatic strangulation. Furthermore, several research papers have shown that this skeletal element can be useful as a sex estimator in both forensics and bioarchelogical investigations. Researchers such as Kim et al. (2006), Kindschuh et al. (2010), Miller et al. (1998), Reesink et al. (1999), and Urbanová et al. (2013) have studied morphometric aspects of the hyoid bone, showing that male hyoids are generally larger than female ones. The majority of previous research projects have focused on studying the hyoid bone as a whole (i.e., the body fused to the great horns). However, the recent works of Kindschuh et al. (2010) and Urbanová et al. (2013), which consider the unfused hyoid body, have shown greater promise for sexing skeletal remains when the bones are poorly preserved and only measurements of the hyoid body can be collected, excluding the great horns from the analysis. The reason given for this is the high frequency of unfused or fragmented hyoid elements found in skeletal samples (Kindschuh et al., 2010). The probability of finding an intact and complete hyoid in archeological or forensic contexts significantly decrease, due to tissue decomposition and unfavorable burial conditions. This study reveals the importance of measurements of the hyoid body. Among the 10 measurements collected, 8 show statistically significant sexual dimorphism. These results in general match reports of previous studies that emphasize greater sexual dimorphism of the BH, BLMax, and ThMed measurements (Kindschuh et al., 2010; Reesink et al., 1999). It is interesting to note the significant difference observed by the same authors of the width and height of the greater horns at the fusion point with the body (inferior end), which are comparable with our measurements FMin and FMax, respectively; they are also statistically significant in our study sample. Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11 8
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Table 8 Stepwise discriminant function analysis. Functions
Function 1 BH ThMax Constant Centroid Sectioning point Function 2 BH ThMed Constant Centroid Sectioning point Function 3 BH FMin Constant Centroid Sectioning point Function 4 BLMax Constant Centroid Sectioning point Function 5 BLMax FMin Constant Centroid Sectioning point Function 6 BLAv FMax Constant Centroid Sectioning point Function 7 BLAv FMin Constant Centroid Sectioning point Function 8 FMax FMin Constant Centroid Sectioning point
Wilks’s lambda
0.630 0.769
F
Unstandardized coefficients
6.746* 7.214**
0.629 0.554 −11.623
Standardized coefficients
Classification function Male
Female
0.447
−1.212
0.402
−0.894
0.627
−1.358
0.565
−1.200
0.677
−1.438
0.613
−1.302
0.693
−1.472
0.575
−1.308
0.698 0.702
−0.383 0.791 0.729
14.831*** 10.237***
0.681 0.651 −10.440
0.723 0.550
−0.246 0.527 0.701
15.726*** 15.359***
0.773 1.526 −13.779
0.731 0.786
−0.366 0.576
16.951***
0.598 −13.522
1.000
−0.635 0.576 0.486
16.951*** 11.641***
0.457 0.986 −13.922
0.765 0.555
−0.381 0.536 0.631
9.536** 13.476**
0.397 0.896 −13.093
0.580 0.709
−0.345 0.474 0.690
12.191*** 10.333*
0.537 1.397 −15.677
0.786 0.787
−0.390 0.659 0.556
17.596*** 13.150***
0.906 1.220 −9.809
0.648 0.613
−0.367
F-values statistically significant at *P < 0.05, **P < 0.01, ***P < 0.001 level. At each step, the variable that minimizes the overall Wilks’s lambda is entered. Minimum partial F to enter is 3.84; maximum partial F to remove is 2.71.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model
ARTICLE IN PRESS
JCHB-25336; No. of Pages 11
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
9
Table 9 Allocation accuracy of the original and cross-validateda samples. Functions
Male
Total average (%)
Female
N
%
N
%
Function 1 Original Cross-validated
16/19 15/19
84.2 78.9
5/7 4/7
71.4 57.1
80.8 73.1
Function 2 Original Cross-validated
31/40 30/40
77.5 75.0
13/18 13/18
72.2 72.2
75.9 74.1
Function 3 Original Cross-validated
21/26 19/26
80.8 73.1
10/12 10/12
83.3 83.3
81.6 76.3
Function 4 Original Cross-validated
13/17 13/17
76.5 76.5
7/8 7/8
87.5 87.5
80.0 80.0
Function 5 Original Cross-validated
14/17 14/17
82.4 82.4
8/8 8/8
100.0 100.0
88.0 88.0
Function 6 Original Cross-validated
14/17 14/17
82.4 82.4
8/8 8/8
100.0 100.0
88.0 88.0
Function 7 Original Cross-validated
14/17 13/17
82.4 76.5
8/8 7/8
100.0 87.5
88.0 80.0
Function 8 Original Cross-validated
19/25 19/25
76.0 76.0
8/11 8/11
72.7 72.7
75.0 75.0
N – number of individuals correctly classified compared with the total of individuals used for the classification. a Cross-validation is done only for those cases in the analysis. In cross-validation, each case is classified by the functions derived from all cases other than that case.
In the intra- and interobserver error analyses, the mean differences between repeated measurements showed high reproducibility and concordance values ranging from substantial to almost perfect. Thus, it can be deduced that various measurements collected by the same and by different observers are completely reliable. The study has allowed us to assess the importance of the hyoid body, which was already noticed by Kim et al. (2006); significant sexual dimorphism was observed. Compared with previous studies, our percentages of accuracy of correct allocation are slightly greater than those obtained by Kindschuh et al. (2010), whose discriminant functions have a reliability between 82.8% and 85.2%, that is definitely greater than the 72.0% and the 76.0% reported by Miller et al. (1998) and Reesink et al. (1999), respectively. Although Kim et al. (2006) obtained the accuracy similar to the results of our discriminant function analysis (88.2%), the nature of their methods (based Table 10 Application of the discriminant function formulae developed by Kindschuh et al. (2010) to the study sample. Functionsa
Body only Function 3 Function 6
Male
Female
Total average (%)
N
%
N
%
11/17 10/17
64.7 58.8
10/10 10/10
100.0 100.0
77.8 74.1
N – indicates the number of individuals correctly classified compared with the total of individuals used for the classification. a
See Kindschuh et al. (2010) for the complete description of the discriminant function formulae.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11 10
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
on indirect measurements from photographic images of the complete hyoids) does not allow a reliable comparison with the present study. When applied to this archeological sample, the formulae for sex estimation proposed by Kindschuh et al. (2010) offered a low accuracy (from 74.1% to 77.8%). The discriminant function formulae developed in this study provided greater percentages of correct allocation accuracy of sex of between 75.0% and 88.0%, but these results must be interpreted with caution. Although the hyoid bone measurements used for the development of the discriminant functions indicate highly significant sexual dimorphism, the percentage of accuracy may be slightly inflated due to the small sample sizes used to obtain them. Nevertheless, the consistency of the results is an indication that the hyoid body can be valid as a predictor of sex for further studies. On the other hand, the hyoid body does not exhibit a significant asymmetry, so it is not necessary to discriminate between the right and left sides when using the discriminant functions. This consideration allows the anthropologist to use the functions when a partial hyoid body is recovered or when it is difficult to distinguish one side from another. Conclusions The present study suggests a simple method for estimating sex from the hyoid body. It introduces significant new measurements of the hyoid bone, not present in previous studies, which are easily obtainable directly from the archeological remains. This method presents a useful additional tool for collecting biological information on human remains and contributes to the reliability of sex estimation, with future prospects of application in anthropology and osteoarcheology. Acknowledgement We thank Emma Jane Booth for the English revision. References Alemseged, Z., Spoor, F., Kimbel, W.H., Bobe, R., Geraards, D., Reed, D., Wynn, J.G., 2006. A juvenile early hominin skeleton from Dikika, Ethiopia. Nature 443, 296–301. Arensburg, B., Tillier, A.M., Vandermeersch, B., Duday, H., Schepartz, L.A., Rak, Y., 1989. A Middle Palaeolithic human hyoid bone. Nature 338, 758–760. Buckberry, J.L., Chamberlain, A.T., 2002. Age estimation from the auricular surface of the ilium: a revised method. Am. J. Phys. Anthropol. 119, 231–239. Capasso, L., Di Muzio, M., Di Tota, G., Spoletini, L., 1990. Gli inumati della necropoli medievale di Teramo–Sant’Anna: studio antropologico preliminare. In: Soprintendenza Archeologica dell’Abruzzo (Ed.), Ricerche Archeologiche a S. Maria Aprutiensis (La Necropoli, La Domus), Teramo. Capasso, L., 2001. I Fuggiaschi di Ercolano. L’Erma di Bretschneider, Roma. Capasso, L., Michetti, E., D’Anastasio, R., 2008. Homo erectus hyoid bone and the origin of speech. Coll. Antropol. 32, 315–319. D’Anastasio, R., Vitullo, G., 2008. Gli inumati della necropoli sannita di Opi–Val Fondillo (VII–V sec. a.C., L’Aquila): rilievi antropologici e paleopatologici. Atti del XVII Congresso AAI, Cagliari. Int. J. Anthropol. Special issue, 148–156. D’Anastasio, R., Di Tota, G., Capasso, L., 2004. Gli inumati della necropoli di Monte D’Argento (XI–XV sec. d. C., Minturno, Latina): rilievi antropologici e paleopatologici. Archivio per l’Antropologia e la Etnologia 84, 127–150. Ferembach, D., Schwidetzky, I., Stloukal, M., 1980. Recommendations for age and sex diagnoses of skeletons. J. Hum. Evol. 9, 517–549. Hunt, D.R., Albanese, J., 2005. History and demographic composition of the Robert J. Terry Anatomical Collection. Am. J. Phys. Anthropol. 127, 406–417. Jelisiejew, T., Jaroslaw, S., Kuduk, I., 1968. Morphologic studies on the hyoid bone in man. Folia Morphol. 27, 172–182. Katz, D., Suchey, J.M., 1986. Age determination of the male os pubis. Am. J. Phys. Anthropol. 69, 427–435. Kim, D.-I., Lee, U.-Y., Park, D.-K., Kim, Y.-S., Han, K.-H., Kim, K.-H., Han, S.-H., 2006. Morphometrics of the hyoid bone for human sex determination from digital photographs. J. Forensic Sci. 51, 979–984. Kindschuh, S.C., Dupras, T.L., Cowgill, L.W., 2010. Determination of sex from the hyoid bone. Am. J. Phys. Anthropol. 143, 279–284. ˇ ´ M., 1990. Sex determination from the hyoid bone by means of discriminant analysis. Acta Univ. Palack. Komenda, S., Cern y, Olomuc. 125, 37–51. Lin, L., 1989. A concordance correlation coefficient to evaluate reproducibility. Biometrics 45, 255–268. Lin, L., 2000. Correction: a note on the concordance correlation coefficient. Biometrics 56, 67–73. Lovejoy, C.O., Meindl, R.S., Pryzbeck, T.R., Mensforth, R.P., 1985. Chronological metamorphosis of the auricular surface of the ilium: a new method for the determination of adult skeletal age at death. Am. J. Phys. Anthropol. 68, 15–28.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
11
Martínez, I., Arsuaga, J.L., Quam, R., Carretero, J.M., Gracia, A., Rodríguez, L., 2008. Human hyoid bones from the middle Pleistocene site of the Sima de los Huesos (Sierra de Atapuerca, Spain). J. Hum. Evol. 54, 118–124. McBride, G.B., 2005. A proposal for strength-of-agreement criteria for Lin’s concordance correlation coefficient. National Institute of Water and Atmospheric Client Report. HAM2005-062, Hamilton, New Zealand. Miles, A.E.W., 1963. The dentition in the assessment of individual age in skeletal material. In: Brothwell, D.R. (Ed.), Dental Anthropology. Pergamon Press, London, pp. 191–209. Miller, K.W.P., Walker, P.L., O’Halloran, R.L., 1998. Age and sex-related variation in hyoid bone morphology. J. Forensic Sci. 43, 1138–1143. Murail, P., Bruzek, J., Houët, F., Cunha, E., 2005. DSP: a tool for probabilistic sex diagnosis using worldwide variability in hip–bone measurements. Bull. Mem. Soc. Anthropol. Paris 17, 3–4. O’Halloran, R.L., Lundy, J.K., 1987. Age and ossification of the hyoid bone: forensic implications. J. Forensic Sci. 32, 1655–1659. Reesink, E.M., Van Immerseel, A.A.H., Brand, R., Bruintjes, T.J.D., 1999. Sexual dimorphism of the hyoid bone? Int. J. Osteoarchaeol. 9, 357–360. ˜ Asturias). In: Rodríguez, L., Cabo, L.L., Egocheaga, J.E., 2003. Breve nota sobre el hioides neandertalense de Sidrón (Pilona, Aluja, M.P., Malgosa, A., Nogués, R. (Eds.), Antropología y Biodiversidad. Actas XII Congreso de la SEAB. Edicions Bellaterra, Barcelona, pp. 484–493. Ubelaker, D., 1989. Human skeletal remains: excavation, analysis, interpretation. In: Aldine Manuals on Archeology. Taraxacum, Washington, DC. ˇ Urbanová, P., Hejna, P., Zátopková, L., Safr, M., 2013. The morphology of human hyoid bone in relation to sex, age and body proportions. HOMO – J. Comp. Hum. Biol. 64, 190–204. Vallois, H.V., 1960. Vital statistics in prehistoric as determined from archaeological data. In: Heizer, R.F., Cook, S.F. (Eds.), The Application of Quantitative Methods in Archaeology. Quadrable Books, Chicago, pp. 186–222.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
ARTICLE IN PRESS
HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Contents lists available at ScienceDirect
HOMO - Journal of Comparative Human Biology journal homepage: www.elsevier.com/locate/jchb
Estimation of sex from the hyoid body in skeletal individuals from archeological sites R. D’Anastasio a, J. Viciano a,∗, M. Di Nicola b, D.T. Cesana a, M. Sciubba a, M. Del Cimmuto a, A. Paolucci a, A. Fazio a, L. Capasso a a
University Museum, State University “G. d’Annunzio”, Piazza Trento e Trieste 1, 66100 Chieti, Italy Laboratory of Biostatistics, Department of Biomedical Science, State University “G. d’Annunzio”, Via dei Vestini, 66100 Chieti, Italy b
a r t i c l e
i n f o
Article history: Received 8 June 2013 Accepted 29 January 2014 Available online xxx
a b s t r a c t Recent forensic studies have shown that the hyoid bone is a sexually dimorphic element of the human skeleton. Given the advanced techniques of collecting human remains in archeological and forensic contexts, the recovery of hyoid bones is now more frequent in skeletal samples. For that reason the authors propose a new method for estimating sex based on hyoid bodies from archeological sites. The study has been conducted on well-preserved hyoids of skeletal remains of 64 adult individuals (44 males and 20 females) dated from the pre-Roman to the medieval periods. The authors considered 10 linear measurements of the hyoid body. The most significant measurements showing sexual dimorphism are the body height, body length, and the maximum and minimum diameter of the articular facet for the greater horn. Discriminant function analysis achieved the allocation accuracy between 75.0% and 88.0%, depending on the measurement collected. This method represents a new, useful and easy way for increasing biological information when assessing the sex of adult human remains from an archeological sample. © 2014 Elsevier GmbH. All rights reserved.
∗ Corresponding author. Tel.: +39 0871 3553502; fax: +39 0871 410927. E-mail address: [email protected] (J. Viciano). http://dx.doi.org/10.1016/j.jchb.2014.01.002 0018-442X/© 2014 Elsevier GmbH. All rights reserved.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11 2
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Introduction Sex estimation of human skeletal remains represents a crucial stage in any palaeoanthropological, archeological or forensic study. Sexually dimorphic differences between males and females have been quantified in numerous ways in physical anthropology and osteoarcheology based on both morphological and metric point features for most of the human skeletal elements. In an archeological context, the accuracy in sex determination depends on the integrity of the skeletal elements because of the usual fragmented state of preservation of human remains. For this reason the possibility of collecting as much information as possible from every element available is very important. The recent discovery of hominin hyoid bones, including those of Australopithecus, Homo neanderthalensis, Homo heidelbergensis and other archaic humans (Alemseged et al., 2006; Arensburg et al., 1989; Capasso et al., 2008; Martínez et al., 2008; Rodríguez et al., 2003), and the increasing recovery of hyoids from archeological contexts has reawakened interest in this bony element and offers new opportunities to predict sex from both fused and unfused hyoids. The hyoid bone is a sexually dimorphic skeletal element (Jelisiejew et al., 1968; Kim et al., 2006; Miller et al., 1998; Reesink et al., 1999; Urbanová et al., 2013) that consists of a body and greater and lesser horns that articulate with the body. It is an important anchor for the tongue and other muscles that are required for swallowing and speaking, and provides a supporting structure for the larynx. The greater horns can fuse to the body, but the anatomical reason behind the fusion is still somewhat unclear: it could be related to the age (O’Halloran and Lundy, 1987), even if some hyoids remain unfused throughout old age. However, sex seems not to be correlated with the bilateral fusion of the greater horns to the body (Miller et al., 1998). The possibility of determining sex from the hyoid has been pursued for a long time through analyses of morphological and metric differences between males and females. Various authors have considered fused hyoids as a useful tool for probabilistic sex diagnosis (Jelisiejew et al., 1968; Kim et al., 2006; Miller et al., 1998; Reesink et al., 1999). In 2010 Kindschuh and co-workers presented a method for determining sex from unfused hyoids (Kindschuh et al., 2010). The hyoid body alone also displays sexual dimorphism, and thus it is useful for sex diagnosis (Reesink et al., 1999). Previous researchers investigated the utility of the hyoid as a reliable sex indicator, but they conducted metric analyses of forensic samples indirectly through X-ray images (Komenda and ˇ ´ 1990; Miller et al., 1998; Reesink et al., 1999). Kim et al. (2006) quantified sexual dimorphism Cern y, of the hyoid bone in Koreans using digital photographs and obtained a discriminant function with an accuracy of 88.2%. In a more recent study researchers attempted to obtain measurements directly from hyoid bones (Kindschuh et al., 2010) in the Terry Collection, an American skeletonized cadaver series of identified American Whites and Blacks collected since 1898 in the Anatomy Department at Washington University, St. Louis, Missouri (Hunt and Albanese, 2005). Our study represents the first attempt to identify metric dimorphism in hyoid body from an archeological sample. In particular, we employed new measurements and discriminant functions were developed for predicting sex.
Materials and methods The sample consists exclusively of complete hyoid bodies, without ossified cartilaginous tissue or fused horns. The specimens belong to 64 individuals from five archeological sites, representing the geographical region of Central Italy: Opi (VI-V century BCE), Canne della Battaglia (III century BCE), Herculaneum (I century CE), Teramo (VII-XII century CE), and Monte D’Argento (XI-XV century CE). The Herculaneum site represents a homogeneous and unique sample of a Roman town whose inhabitants died contemporaneously during the eruption of the volcano Vesuvius in the year 79 CE (Capasso, 2001). The other sites come from the Abruzzo region or neighboring areas, whose chronological contexts range from pre-Roman to the Medieval periods (Capasso et al., 1990; D’Anastasio et al., 2004; D’Anastasio and Vitullo, 2008). All these individuals are housed in the University Museum of Chieti (Italy). Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
3
Fig. 1. Hyoid body measurements (A: frontal view; B: superior view; C: lateral view). Refer to Table 1 for a description of the measurements which correspond to each acronym.
The biological profile of each individual was previously determined considering all the available bones. Sex was estimated following standard descriptive and metric criteria (Ferembach et al., 1980; Murail et al., 2005) according to cranial and pelvic features. The estimated age at death was based on dental development (Ubelaker, 1989), the degree of dental wear (Lovejoy et al., 1985; Miles, 1963), and the appearance of the pubic symphyseal surface (Katz and Suchey, 1986) and the ilium auricular surface (Buckberry and Chamberlain, 2002). We considered 10 linear measurements. All of them were taken using a standard digital caliper with an accuracy of 0.01 mm (Fig. 1 and Table 1). Data measurements were first assessed for normality using the Kolmogorov–Smirnov’s one-sample test and for homogeneity of variance using the Levene test. Afterwards, we analyzed the differences between the mean values of two measurements (FMax and FMin) to search for metric asymmetry (differences between measurements taken on the right and left side) using the paired Student’s t-test.
Table 1 Measurementsa of the hyoid body. Measurement
Description
BHb BLMaxb BLMinc BLAvb ThMaxc ThMedc ThBordc FMaxc FMinc LLHornb
Body height: maximum body height in the medial plane Body length: maximum length of the body in the transverse plane Minimum transverse diameter: minimum length of the body in the transverse plane Average transverse diameter: average length of the body in the transverse plane Maximum thickness of the total body Antero-posterior thickness: maximum antero-posterior thickness of the body in the medial plane Thickness of body’s inferior border Maximum diameter of the right and left articular facet for the greater horn Minimum diameter of the right and left articular facet for the greater horn Length of the lesser horn of the body
a b c
All measurements are in millimeters. Measurements applied in previous studies (Kim et al., 2006; Kindschuh et al., 2010; Reesink et al., 1999). New measurements used in the current study.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11 4
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Table 2 Distribution of the sample according to the archeological sites, age groups and sex. Century
Site of origin Opi Canne della Battaglia Herculaneum Teramo Monte D’Argento
Females
Males
VI-V BCE III BCE I CE VII-XII CE XI-XV CE
Age at death 20–40 years 40–60 years
Sexes pooled
N
%
N
%
6 2 29 6 1
100.0 66.7 59.2 100.0 50.0
– 1 18 – 1
– 33.3 40.8 – 50.0
6 3 47 6 2
25 19
71.4 65.5
10 10
28.6 34.5
35 29
The differences between the mean values in all measurements collected at two different times were also analyzed in order to assess possible intra- and interobserver error. Lin’s concordance correlation coefficient (CCC) (Lin, 1989, 2000) was computed to determine the level of agreement between repeated measurements collected by the same observer and by different observers. To assess the degree of agreement for a given set of data, the calculated CCC is compared to the strength of agreement criteria proposed by McBride (2005), which establishes four levels of qualitative assessment as follows: almost perfect, for CCC values greater than 0.99; substantial, from 0.95 to 0.99; moderate, from 0.90 to 0.95; and poor, for CCC values below 0.90. The main effects of the age at death and the archeological site of the individuals on various measurements were tested by two-way ANOVA. Next, the differences between the mean values of males and females were analyzed using the independent Student’s t-test in cases where the normality and homogeneity of variance are fulfilled. Next, a stepwise discriminant function analysis was performed to create equations to estimate sex. In order to maximize the applicability in archeological cases, when the skeletal remains are in poor condition, the functions were calculated for a maximum combination of two measurements. To test the consistency of prediction accuracy, the data were subjected to a cross-validation procedure. All statistical analyses were performed using the SPSS 15.0 software (SPSS Inc., Chicago, IL, USA). Finally, the formulae published by Kindschuh et al. (2010) for sex estimation developed in a knownsex osteological collection were applied to the present archeological samples in order to evaluate its applicability. Results The hyoid bones belong to 64 adult individuals (44 males and 20 females; Table 2). Table 3 shows the percentage of preservation of the crania and pelves of the specimens studied, on which descriptive and metric criteria were used to estimate sex. According to the estimated ages at death, the sample
Table 3 Degree of preservation of the cranium and pelvis in the specimens studied. Degree of preservation
Cranium N
Pelvis %
N
%
Optimal Good Discreet Mediocre Poor
42 15 6 1 0
65.6 23.4 9.4 1.6 0.0
42 10 4 3 5
65.6 15.6 6.3 4.7 7.8
Total
64
100.0
64
100.0
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
5
Table 4 The paired t-test evaluating metric bilateral asymmetry. Pair of measurements
N
Diff
SE
t
P
FMax(right)–FMax(left) FMin(right)–FMin(left)
22 23
0.0268 −0.1535
0.4985 0.6271
0.252 −1.174
0.803 0.253
N – number of pairs; Diff – mean difference between pairs; SE – standard error; t – Student’s t-test; P – P-value.
was divided into two groups (modified from Vallois, 1960): 20–40 years (35 young adults), and 40–60 years (29 old adults) (Table 2). Regarding the asymmetry analysis, FMin and FMax were analyzed and none of these measurements show statistically significant differences at P < 0.05 level regarding laterality (Table 4). For this reason, the average between both right and left sides of these measurements was calculated and it was used for the subsequent statistical analyses. A total number of 48 hyoid bones (representing the 75% of the total sample) were used as subsample for intra- and inter-observer error analysis. Table 5 shows the differences between mean values and the concordance correlation coefficient (CCC) for repeated measurements. In the intra-observer error analysis, the CCC value ranges from 0.985 to 0.999 (from substantial to almost perfect). With regard to the inter-observer error analysis, the value of CCC ranges from 0.955 to 0.999 (from substantial to almost perfect). In general, the results show that hyoid body measurements are less concordant between different observers than within the same observer. Considering the entire sample (sex and age groups pooled), the Kolmogorov–Smirnov’s test showed that all the measurements were normally distributed. The results of homogeneity of variance tests indicated that the sample was statistically homogeneous for all the measurements compared.
Table 5 Lin’s concordance correlation coefficient (CCC) to evaluate concordance between repeated measurements collected by the same observer (intraobserver error) and by different observers (interobserver error). N
Measurement 1
Measurement 2
Diff
CCC
Strength of agreementa
Mean
SD
Mean
SD
Same observer 46 BH 19 BLMax 19 BLMin 18 BLAv 20 ThMax ThMed 46 ThBord 46 FMax 26 28 FMin 12 LLHorn
10.5957 21.6747 17.5150 19.2117 8.5425 4.9793 1.9498 6.0487 3.7179 2.1850
1.1274 2.3927 2.3830 1.6729 1.6916 0.9065 0.4376 0.9237 0.5331 0.5234
10.6111 24.6774 17.5022 19.1872 8.5370 4.9798 1.9550 6.0715 3.7448 2.1700
1.1223 2.3998 2.3917 1.6828 1.7166 0.9112 0.4255 0.9305 0.5470 0.4860
−0.0154 −0.0027 0.0128 0.0245 0.0055 −0.0005 −0.0052 −0.0228 −0.0269 0.0150
0.999 0.999 0.998 0.996 0.992 0.996 0.991 0.995 0.985 0.997
Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Substantial Almost perfect
Different observers BH 46 19 BLMax BLMin 19 BLAv 18 20 ThMax 46 ThMed 46 ThBord 26 FMax 28 FMin 12 LLHorn
10.5628 21.7168 17.5022 19.1944 8.5222 4.9222 1.8817 6.0102 3.6632 2.1483
1.1275 2.4362 2.3917 1.6641 1.7165 0.9222 0.4630 0.9232 0.5467 0.5576
10.6034 21.6461 17.5086 19.1994 8.5398 4.9796 1.9524 6.0601 3.7313 2.1775
1.1245 2.3961 2.3871 1.6776 1.7038 0.9085 0.4306 0.9267 0.5395 0.5044
−0.0406 0.0407 −0.0064 −0.0050 −0.0178 −0.0574 −0.0707 −0.0499 −0.0681 −0.0292
0.998 0.999 0.999 0.994 0.999 0.997 0.955 0.996 0.987 0.989
Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Almost perfect Substantial Almost perfect Substantial Substantial
N – number of measurements; Mean – overall measurement mean; SD – standard deviation; Diff – mean difference between repeated measurements; CCC – Lin’s concordance correlation coefficient value. a See Mcbride (2005) for the strength of agreement criteria.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model
ARTICLE IN PRESS
JCHB-25336; No. of Pages 11 6
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Table 6 Two-way ANOVA for evaluating differences between archeological sites and age groups. df
MS
F
P
7.251 1.934 2.189
4 1 2
1.813 1.934 1.095
1.241 1.325 0.750
0.306 0.255 0.478
BLMax Site Age Site × age
5.543 0.130 9.262
2 1 2
2.772 0.130 4.631
0.543 0.025 0.908
0.590 0.875 0.420
BLMin Site Age Site × age
4.270 1.255 10.893
3 1 2
1.423 1.255 5.447
0.298 0.263 1.140
0.827 0.614 0.341
BLAv Site Age Site × age
0.601 0.124 12.396
2 1 2
0.301 0.124 6.198
0.101 0.042 2.090
0.904 0.840 0.151
ThMax Site Age Site × age
10.594 1.308 4.792
2 1 2
5.297 1.308 2.396
2.969 0.733 1.343
0.074 0.402 0.284
ThMed Site Age Site × age
6.207 1.166 0.885
4 1 2
1.552 1.166 0.443
1.971 1.482 0.562
0.114 0.229 0.574
ThBord Site Age Site × age
0.190 0.000 0.251
4 1 2
0.047 0.000 0.126
0.240 0.002 0.633
0.915 0.969 0.535
FMax Site Age Site × age
1.019 0.009 0.015
3 1 2
0.340 0.009 0.007
0.407 0.011 0.009
0.749 0.919 0.991
FMin Site Age Site × age
1.345 0.092 0.037
3 1 2
0.448 0.092 0.019
1.146 0.236 0.048
0.346 0.631 0.953
LLHorn Site Age Site × age
0.552 0.419 0.204
3 1 1
0.184 0.419 0.204
0.648 1.475 0.716
0.597 0.245 0.412
Measurements BH Site Age Site × age
SS
SS – sum of squares; df – degrees of freedom; MS – mean square; F – F-score statistic; P – P-value.
Results of the two-way ANOVA revealed no statistically significant differences among age group categories and/or archeological sites (P > 0.05) (Table 6). Table 7 shows the sample size, mean and standard deviation, t value, and the degree of significance of the differences between the male and female individual means of all measurements. Among the 10 measurements, 8 of them show a greater mean in males than in females; these differences are statistically significant at P < 0.05 level. The most sexually dimorphic measurements are BH, BLMax, FMax, and FMin (P < 0.001). Next comes BLAv, ThMed, BLMin, and ThMax (P < 0.01). There are no significant differences in ThBord and LLHorn measurements. Table 8 shows the coefficients indicating the contribution of each measurement, the classification functions of the two groups analyzed, the discriminant functions and their corresponding sectioning Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
7
Table 7 Descriptive statistics and t-test results for mean differences between the sexes. Measurement
BH BLMax BLMin BLAv ThMax ThMed ThBord FMax FMin LLHorn
Males
Females
N
Mean
SD
N
Mean
SD
40 17 18 17 19 40 40 26 26 15
10.963 23.574 18.743 20.402 9.300 5.185 1.904 6.228 3.913 2.109
1.131 1.801 2.031 1.584 1.308 0.881 0.450 0.748 0.543 0.588
18 8 9 9 7 18 18 11 12 5
9.802 20.620 16.372 18.144 7.796 4.409 1.738 5.146 3.209 1.788
0.885 1.336 1.790 1.297 1.132 0.759 0.369 0.583 0.443 0.480
t
P
3.851 4.117 2.967 3.666 2.686 3.234 1.369 4.270 3.919 1.098
0.000 0.000 0.007 0.001 0.013 0.002 0.177 0.000 0.000 0.287
N – number of individuals; Mean – overall measurement mean; SD – standard deviation; t – Student’s t-test; P – P-value.
points, and the values of F and Wilks’s lambda. The functions whose discriminant power fell below 75% were excluded as they are of little utility. Allocation accuracy of these functions is presented in Table 9. It is seen that the allocation accuracy ranges from 76.0% to 84.2% in males and 71.4% to 100% in females. Thus, females are generally classified more accurately than males by discriminant functions. For both sexes pooled, allocation accuracy ranges between 75.0% and 88.0%. Cross-validation results have not significantly changed the original accuracy (from 73.1% to 88.0%). Of the 10 functions developed, only one (Function 4) appears in the univariate analysis, with a correct allocation of sex of 80.0%. The multivariate analysis provides a greater advantage than the univariate analysis because seven multivariate functions provide highly reliable sex estimations (from 75.0% to 88.0%). Two discriminant function formulae developed by Kindschuh et al. (2010) were applied to the study sample. Table 10 shows the results of the number of individuals correctly classified. Accuracy ranges from 74.1% to 77.8% (from 58.8% to 64.7% in males and 100.0% in females). Discussion The hyoid bone has been the subject of numerous studies, especially in forensic contexts, where it can be indicative of traumatic strangulation. Furthermore, several research papers have shown that this skeletal element can be useful as a sex estimator in both forensics and bioarchelogical investigations. Researchers such as Kim et al. (2006), Kindschuh et al. (2010), Miller et al. (1998), Reesink et al. (1999), and Urbanová et al. (2013) have studied morphometric aspects of the hyoid bone, showing that male hyoids are generally larger than female ones. The majority of previous research projects have focused on studying the hyoid bone as a whole (i.e., the body fused to the great horns). However, the recent works of Kindschuh et al. (2010) and Urbanová et al. (2013), which consider the unfused hyoid body, have shown greater promise for sexing skeletal remains when the bones are poorly preserved and only measurements of the hyoid body can be collected, excluding the great horns from the analysis. The reason given for this is the high frequency of unfused or fragmented hyoid elements found in skeletal samples (Kindschuh et al., 2010). The probability of finding an intact and complete hyoid in archeological or forensic contexts significantly decrease, due to tissue decomposition and unfavorable burial conditions. This study reveals the importance of measurements of the hyoid body. Among the 10 measurements collected, 8 show statistically significant sexual dimorphism. These results in general match reports of previous studies that emphasize greater sexual dimorphism of the BH, BLMax, and ThMed measurements (Kindschuh et al., 2010; Reesink et al., 1999). It is interesting to note the significant difference observed by the same authors of the width and height of the greater horns at the fusion point with the body (inferior end), which are comparable with our measurements FMin and FMax, respectively; they are also statistically significant in our study sample. Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11 8
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
Table 8 Stepwise discriminant function analysis. Functions
Function 1 BH ThMax Constant Centroid Sectioning point Function 2 BH ThMed Constant Centroid Sectioning point Function 3 BH FMin Constant Centroid Sectioning point Function 4 BLMax Constant Centroid Sectioning point Function 5 BLMax FMin Constant Centroid Sectioning point Function 6 BLAv FMax Constant Centroid Sectioning point Function 7 BLAv FMin Constant Centroid Sectioning point Function 8 FMax FMin Constant Centroid Sectioning point
Wilks’s lambda
0.630 0.769
F
Unstandardized coefficients
6.746* 7.214**
0.629 0.554 −11.623
Standardized coefficients
Classification function Male
Female
0.447
−1.212
0.402
−0.894
0.627
−1.358
0.565
−1.200
0.677
−1.438
0.613
−1.302
0.693
−1.472
0.575
−1.308
0.698 0.702
−0.383 0.791 0.729
14.831*** 10.237***
0.681 0.651 −10.440
0.723 0.550
−0.246 0.527 0.701
15.726*** 15.359***
0.773 1.526 −13.779
0.731 0.786
−0.366 0.576
16.951***
0.598 −13.522
1.000
−0.635 0.576 0.486
16.951*** 11.641***
0.457 0.986 −13.922
0.765 0.555
−0.381 0.536 0.631
9.536** 13.476**
0.397 0.896 −13.093
0.580 0.709
−0.345 0.474 0.690
12.191*** 10.333*
0.537 1.397 −15.677
0.786 0.787
−0.390 0.659 0.556
17.596*** 13.150***
0.906 1.220 −9.809
0.648 0.613
−0.367
F-values statistically significant at *P < 0.05, **P < 0.01, ***P < 0.001 level. At each step, the variable that minimizes the overall Wilks’s lambda is entered. Minimum partial F to enter is 3.84; maximum partial F to remove is 2.71.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model
ARTICLE IN PRESS
JCHB-25336; No. of Pages 11
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
9
Table 9 Allocation accuracy of the original and cross-validateda samples. Functions
Male
Total average (%)
Female
N
%
N
%
Function 1 Original Cross-validated
16/19 15/19
84.2 78.9
5/7 4/7
71.4 57.1
80.8 73.1
Function 2 Original Cross-validated
31/40 30/40
77.5 75.0
13/18 13/18
72.2 72.2
75.9 74.1
Function 3 Original Cross-validated
21/26 19/26
80.8 73.1
10/12 10/12
83.3 83.3
81.6 76.3
Function 4 Original Cross-validated
13/17 13/17
76.5 76.5
7/8 7/8
87.5 87.5
80.0 80.0
Function 5 Original Cross-validated
14/17 14/17
82.4 82.4
8/8 8/8
100.0 100.0
88.0 88.0
Function 6 Original Cross-validated
14/17 14/17
82.4 82.4
8/8 8/8
100.0 100.0
88.0 88.0
Function 7 Original Cross-validated
14/17 13/17
82.4 76.5
8/8 7/8
100.0 87.5
88.0 80.0
Function 8 Original Cross-validated
19/25 19/25
76.0 76.0
8/11 8/11
72.7 72.7
75.0 75.0
N – number of individuals correctly classified compared with the total of individuals used for the classification. a Cross-validation is done only for those cases in the analysis. In cross-validation, each case is classified by the functions derived from all cases other than that case.
In the intra- and interobserver error analyses, the mean differences between repeated measurements showed high reproducibility and concordance values ranging from substantial to almost perfect. Thus, it can be deduced that various measurements collected by the same and by different observers are completely reliable. The study has allowed us to assess the importance of the hyoid body, which was already noticed by Kim et al. (2006); significant sexual dimorphism was observed. Compared with previous studies, our percentages of accuracy of correct allocation are slightly greater than those obtained by Kindschuh et al. (2010), whose discriminant functions have a reliability between 82.8% and 85.2%, that is definitely greater than the 72.0% and the 76.0% reported by Miller et al. (1998) and Reesink et al. (1999), respectively. Although Kim et al. (2006) obtained the accuracy similar to the results of our discriminant function analysis (88.2%), the nature of their methods (based Table 10 Application of the discriminant function formulae developed by Kindschuh et al. (2010) to the study sample. Functionsa
Body only Function 3 Function 6
Male
Female
Total average (%)
N
%
N
%
11/17 10/17
64.7 58.8
10/10 10/10
100.0 100.0
77.8 74.1
N – indicates the number of individuals correctly classified compared with the total of individuals used for the classification. a
See Kindschuh et al. (2010) for the complete description of the discriminant function formulae.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11 10
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
on indirect measurements from photographic images of the complete hyoids) does not allow a reliable comparison with the present study. When applied to this archeological sample, the formulae for sex estimation proposed by Kindschuh et al. (2010) offered a low accuracy (from 74.1% to 77.8%). The discriminant function formulae developed in this study provided greater percentages of correct allocation accuracy of sex of between 75.0% and 88.0%, but these results must be interpreted with caution. Although the hyoid bone measurements used for the development of the discriminant functions indicate highly significant sexual dimorphism, the percentage of accuracy may be slightly inflated due to the small sample sizes used to obtain them. Nevertheless, the consistency of the results is an indication that the hyoid body can be valid as a predictor of sex for further studies. On the other hand, the hyoid body does not exhibit a significant asymmetry, so it is not necessary to discriminate between the right and left sides when using the discriminant functions. This consideration allows the anthropologist to use the functions when a partial hyoid body is recovered or when it is difficult to distinguish one side from another. Conclusions The present study suggests a simple method for estimating sex from the hyoid body. It introduces significant new measurements of the hyoid bone, not present in previous studies, which are easily obtainable directly from the archeological remains. This method presents a useful additional tool for collecting biological information on human remains and contributes to the reliability of sex estimation, with future prospects of application in anthropology and osteoarcheology. Acknowledgement We thank Emma Jane Booth for the English revision. References Alemseged, Z., Spoor, F., Kimbel, W.H., Bobe, R., Geraards, D., Reed, D., Wynn, J.G., 2006. A juvenile early hominin skeleton from Dikika, Ethiopia. Nature 443, 296–301. Arensburg, B., Tillier, A.M., Vandermeersch, B., Duday, H., Schepartz, L.A., Rak, Y., 1989. A Middle Palaeolithic human hyoid bone. Nature 338, 758–760. Buckberry, J.L., Chamberlain, A.T., 2002. Age estimation from the auricular surface of the ilium: a revised method. Am. J. Phys. Anthropol. 119, 231–239. Capasso, L., Di Muzio, M., Di Tota, G., Spoletini, L., 1990. Gli inumati della necropoli medievale di Teramo–Sant’Anna: studio antropologico preliminare. In: Soprintendenza Archeologica dell’Abruzzo (Ed.), Ricerche Archeologiche a S. Maria Aprutiensis (La Necropoli, La Domus), Teramo. Capasso, L., 2001. I Fuggiaschi di Ercolano. L’Erma di Bretschneider, Roma. Capasso, L., Michetti, E., D’Anastasio, R., 2008. Homo erectus hyoid bone and the origin of speech. Coll. Antropol. 32, 315–319. D’Anastasio, R., Vitullo, G., 2008. Gli inumati della necropoli sannita di Opi–Val Fondillo (VII–V sec. a.C., L’Aquila): rilievi antropologici e paleopatologici. Atti del XVII Congresso AAI, Cagliari. Int. J. Anthropol. Special issue, 148–156. D’Anastasio, R., Di Tota, G., Capasso, L., 2004. Gli inumati della necropoli di Monte D’Argento (XI–XV sec. d. C., Minturno, Latina): rilievi antropologici e paleopatologici. Archivio per l’Antropologia e la Etnologia 84, 127–150. Ferembach, D., Schwidetzky, I., Stloukal, M., 1980. Recommendations for age and sex diagnoses of skeletons. J. Hum. Evol. 9, 517–549. Hunt, D.R., Albanese, J., 2005. History and demographic composition of the Robert J. Terry Anatomical Collection. Am. J. Phys. Anthropol. 127, 406–417. Jelisiejew, T., Jaroslaw, S., Kuduk, I., 1968. Morphologic studies on the hyoid bone in man. Folia Morphol. 27, 172–182. Katz, D., Suchey, J.M., 1986. Age determination of the male os pubis. Am. J. Phys. Anthropol. 69, 427–435. Kim, D.-I., Lee, U.-Y., Park, D.-K., Kim, Y.-S., Han, K.-H., Kim, K.-H., Han, S.-H., 2006. Morphometrics of the hyoid bone for human sex determination from digital photographs. J. Forensic Sci. 51, 979–984. Kindschuh, S.C., Dupras, T.L., Cowgill, L.W., 2010. Determination of sex from the hyoid bone. Am. J. Phys. Anthropol. 143, 279–284. ˇ ´ M., 1990. Sex determination from the hyoid bone by means of discriminant analysis. Acta Univ. Palack. Komenda, S., Cern y, Olomuc. 125, 37–51. Lin, L., 1989. A concordance correlation coefficient to evaluate reproducibility. Biometrics 45, 255–268. Lin, L., 2000. Correction: a note on the concordance correlation coefficient. Biometrics 56, 67–73. Lovejoy, C.O., Meindl, R.S., Pryzbeck, T.R., Mensforth, R.P., 1985. Chronological metamorphosis of the auricular surface of the ilium: a new method for the determination of adult skeletal age at death. Am. J. Phys. Anthropol. 68, 15–28.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
G Model JCHB-25336; No. of Pages 11
ARTICLE IN PRESS
R. D’Anastasio et al. / HOMO - Journal of Comparative Human Biology xxx (2014) xxx–xxx
11
Martínez, I., Arsuaga, J.L., Quam, R., Carretero, J.M., Gracia, A., Rodríguez, L., 2008. Human hyoid bones from the middle Pleistocene site of the Sima de los Huesos (Sierra de Atapuerca, Spain). J. Hum. Evol. 54, 118–124. McBride, G.B., 2005. A proposal for strength-of-agreement criteria for Lin’s concordance correlation coefficient. National Institute of Water and Atmospheric Client Report. HAM2005-062, Hamilton, New Zealand. Miles, A.E.W., 1963. The dentition in the assessment of individual age in skeletal material. In: Brothwell, D.R. (Ed.), Dental Anthropology. Pergamon Press, London, pp. 191–209. Miller, K.W.P., Walker, P.L., O’Halloran, R.L., 1998. Age and sex-related variation in hyoid bone morphology. J. Forensic Sci. 43, 1138–1143. Murail, P., Bruzek, J., Houët, F., Cunha, E., 2005. DSP: a tool for probabilistic sex diagnosis using worldwide variability in hip–bone measurements. Bull. Mem. Soc. Anthropol. Paris 17, 3–4. O’Halloran, R.L., Lundy, J.K., 1987. Age and ossification of the hyoid bone: forensic implications. J. Forensic Sci. 32, 1655–1659. Reesink, E.M., Van Immerseel, A.A.H., Brand, R., Bruintjes, T.J.D., 1999. Sexual dimorphism of the hyoid bone? Int. J. Osteoarchaeol. 9, 357–360. ˜ Asturias). In: Rodríguez, L., Cabo, L.L., Egocheaga, J.E., 2003. Breve nota sobre el hioides neandertalense de Sidrón (Pilona, Aluja, M.P., Malgosa, A., Nogués, R. (Eds.), Antropología y Biodiversidad. Actas XII Congreso de la SEAB. Edicions Bellaterra, Barcelona, pp. 484–493. Ubelaker, D., 1989. Human skeletal remains: excavation, analysis, interpretation. In: Aldine Manuals on Archeology. Taraxacum, Washington, DC. ˇ Urbanová, P., Hejna, P., Zátopková, L., Safr, M., 2013. The morphology of human hyoid bone in relation to sex, age and body proportions. HOMO – J. Comp. Hum. Biol. 64, 190–204. Vallois, H.V., 1960. Vital statistics in prehistoric as determined from archaeological data. In: Heizer, R.F., Cook, S.F. (Eds.), The Application of Quantitative Methods in Archaeology. Quadrable Books, Chicago, pp. 186–222.
Please cite this article in press as: D’Anastasio, R., et al., Estimation of sex from the hyoid body in skeletal individuals from archeological sites. HOMO - J. Comp. Hum. Biol. (2014), http://dx.doi.org/10.1016/j.jchb.2014.01.002
