Received: 25 June 2018 Revised: 2 December 2018 Accepted: 2 January 2019 DOI: 10.1002/oa.2734 RESEARCH ARTICLE Canine sex estimation and sexual dimorphism in the collection of identified skeletons of the University of Coimbra, with an application in a Roman cemetery from Faro, Portugal Leandro H. Luna Multidisciplinar Institute of History and Human Sciences (IMHICIHU)‐Council of Scientific and Technical Investigations (CONICET)/Faculty of Philosophy and Letters, University of Buenos Aires, Buenos Aires, Argentina Correspondence Leandro H. Luna, Multidisciplinar Institute of History and Human Sciences (IMHICIHU)‐ Council of Scientific and Technical Investigations (CONICET)/Faculty of Philosophy and Letters, University of Buenos Aires, Saavedra 15 (1083), Buenos Aires, Argentina. Email: lunaranda@gmail.com Abstract Sexual estimation of human remains is an aspect of great importance for the characterization of demographic profiles in bioarcheology and to identify individuals in forensic cases. The aims of this paper are threefold: to generate population‐specific formulae for sex estimation based on permanent canine metrics, to evaluate the dental sexual dimorphism, and to develop a Bayesian approach in a sample of 115 individuals from the documented human sample housed in the University of Coimbra (Portugal). Discriminant functions and logistic regression equations were developed, and posterior probabilities were calculated. Formulae offered high percentages of correct sex assignation (77.42–86.54% for the discriminant functions and 81.63–85.18% for the logistic regression), whereas posterior probabilities ranged between 0.71 and 0.85. The procedures were then applied in an archaeological sample of 32 individuals from the Roman (I‐III century AC) cemetery of Ossonoba Romana (Faro, Portugal) in order to test the relevance of their use in this geographical‐related sample. The results of correct estimation are higher than 75% for three formulae and four combinations of variables in the Bayesian approach. Although phenotypic variation may be a factor influencing the sex estimations, canine odontometrics are powerful tools when previously tested and can increase the amount of data obtained for paleodemographic and forensic purposes. In this case, some of the methods developed for the modern sample can be used in archaeological samples and in spatial and temporal‐related skeletal collections. KEY W ORDS archaeology, canine teeth, dental anthropology, odontometrics, sexual dimorphism 1 | I N T RO D U CT I O N and growth patterns, many biologic, demographic, behavioural, and pathological issues may be approached, such as sex and age‐at‐death The dentition is a very valuable tool from both bioarchaeological and estimation, diet variations, health patterns, disease affection, and forensic fields. Teeth are particularly resistant to post‐depositional biological distances. Moreover, as tooth size and shape are established deterioration because of its size, hardness, and strength, so they are in the first stages of the individuals' life and do not change throughout usually better preserved than bones and often more represented in the life, sexual information in subadults can also be obtained, which is human osteological samples (Hillson, 1996; Duckworth, 2006). In often problematic when bones are analyzed (e.g., Hillson, 1986; many cases, they are the only source of information. Considering the Kieser, minimal influence of socioeconomic factors on dental size, shape, Njemirovskij, Keros, & Brkic, 2007). Int J Osteoarchaeol. 2019;1–13. wileyonlinelibrary.com/journal/oa 2008; Luna, 2008, 2015, 2016; Vodanovic, Demo, © 2019 John Wiley & Sons, Ltd. 1 2 LUNA Sex and age‐at‐death estimation constitute the first step in any (1997), they conclude that the Y‐chromosome promotes a greater bioarchaeological or forensic investigation, and such information dentine thickness and may be the main responsible for sex differences should be as accurate as possible in order to obtain reliable identifica- in crown size. tions. Adult sex can be predicted with a high degree of accuracy when Previous research that evaluated sexual dimorphism on a Portu- the pelvis and the skull are available (Bruzek, 2002; Buikstra & guese sample analyzed a cast sample of 80 young adult students of Ubelaker, 1994; Phenice, 1969; Rosing et al., 2007; Walker, 2005), the School of Dentistry, University of Lisbon (Pereira et al., 2010). although it decreases when those elements are fragmented or deteri- Values between 3.4% and 5.7% were obtained, and population‐ orated. In these contexts, teeth are valuable and useful tools for sex specific discriminant functions for sex estimation were generated. estimation (Ling & Wong, 2007). Numerous methodologies based on For the Coimbra Identified Collection, Galera and Cunha (1993) canine metrics have been proposed for archaeological and forensic recognized an important sexual dimorphism represented by larger samples throughout the world, with satisfactory results (e.g., Acharya buccolingual dimensions than mesiodistal ones, and Bermúdez de & Mainali, 2007; Aris, Nystrom, & Craig‐Atkins, 2018; García‐Campos Castro, Sarmiento, Cunha, Rosas, and Bastir (2001) obtained values et al., 2018; Isçan & Kedici, 2003; Karaman, 2006; Khamis, Taylor, of 7.77% for males and 7.21% for females. As no research was previ- Malik, & Townsend, 2014; Kieser, 2008; Mitsea, Moraitis, Leon, ously done considering sex estimation in this collection, the aim of this 2005; paper is to develop population‐specific formulae based on direct Saunders, Chan, Kahlon, & Kluge, 2007; Thompson, 2013; Viciano, measurements of permanent canine metric data in a sample of docu- D'Anastasio, & Capasso, 2015; Viciano, López‐Lázaro, & Alemán, mented individuals housed in the University of Coimbra (Portugal). 2013; Zorba, Moraitis, Eliopoulos, & Spiliopoulou, 2012; Zorba, Univariate analyses are accomplished, and both discriminant functions Moraitis, & Manolis, 2011; Zorba, Vanna, & Moraitis, 2014). However, and logistic regression equations are generated. A Bayesian approach as the degree of sexual dimorphism varies among different samples, is also fulfilled, with the aim of obtaining posterior probabilities of their application is usually restricted to the population in which it correct sex assignations. All these procedures are then used in an was created (Cardoso, 2010; Hillson, 1986; Kondo & Townsend, archaeological sample from the Roman (I‐III century AC) cemetery of 2004; Luna, 2008, 2010, 2012, 2015; Pereira, Bernardo, Pestana, Ossonoba Romana (Faro, Portugal) in order to test the relevance of Santos, & Mendonça, 2010; Roberts & Manchester, 1999; Saunders their application in a geographically related sample. Sexual dimorphism et al., 2007; Schwartz & Dean, 2005). is also assessed in both samples with the purpose of identifying pre- Nicopoulou‐Karayianni, & Spiliopoulou, 2014; Okazaki, Sexual dimorphism (defined as the differences in size and shape liminary trends during the last two millennia. between males and females) for human teeth generally ranges between 1% and 7% (Acharya & Mainali, 2007; Garn, Lewis, Swindler, & Kerewsky, 1967; Schwartz & Dean, 2005), being male teeth larger 2 SAMPLES AND METHODS | than female ones, although in some samples, it reaches 10/15% (Frayer & Wolpoff, 1985; Luna & Flensborg, 2017; Mitsea et al., 2.1 | Sample 2014; Viciano et al., 2015); moreover, buccolingual diameters tend to show more sexual dimorphism than mesiodistal diameters (e.g., The reference sample is composed of 115 individuals (53 females and Hassett, 2011; Isçan & Kedici, 2003; Kondo & Townsend, 2004). 62 males) between 10 and 73 years of age‐at‐death (Table 1), who Canines have consistently shown the greatest sexual dimorphism in died between 1895 and 1936. They belong to the Coimbra Identified different samples (Acharya & Mainali, 2007; Garn et al., 1967; Isçan Skeletal Collections housed in the Department of Science of Life, & Kedici, 2003; Luna, 2008; Potter, Alcazaren, Herbosa, & Tomaneng, Centre of Investigation in Anthropology and Health, Faculty of 1981; Saunders et al., 2007; Schwartz & Dean, 2005); in consequence, Science and Technology, University of Coimbra, Coimbra, Portugal. they offer the most reliable information for human sex estimation. All the individuals were exhumed from the Cemitério Municipal da Variations in hormonal secretion between males and females during Conchada in Coimbra; they represent an excellent human biological subadulthood have been usually invoked as the main cause of dental sample of the modern Portuguese population. For each skeleton, sexual dimorphism (Kondo & Townsend, 2004). As levels of sexual the following reliable information is available: full name, sex and age‐ hormones are higher because of puberty (Bogin & Smith, 2000), it is at‐death, birthplace, date of birth and death, place and cause of death, stated that tooth types forming during early childhood should be less occupation, marital status, work activity, and parents' names (Rocha, dimorphic than those late forming. However, teeth do not follow this 1995; Santos, 2000). trend. Canine crowns are formed approximately between 1.9 and On the other hand, Ossonoba Romana is a I‐III AC Roman 6.2 years of age (Ubelaker, 1989; Reid & Dean, 2006; AlQahtani, cemetery located at Faro, the capital city of Algarves district, Southern Hector, & Liversidge, 2010), but the hormonal secretion is minimal Portugal. Eighty two primary adult and subadult burials were during this age range and similar for both sexes; accordingly, this is recovered in 2004 and also several multiple ossuaries (Fernandes, not the most plausible explanation about the high percentages of 2012). Adults were sexed considering morphological and metric sexual dimorphism. Guatelli‐Steinberg, Sciulli, and Betsinger (2008) methods for the coxae, skull, long and tarsal bones (Bruzek, 2002; evaluated the association between dental size and hormonal levels Silva, 1995), and aged following Lovejoy, Meindl, Pryzbeck, and for different ages and affirmed that variations in sexual hormone Mensforth (1985) for the auricular surface (see Fernandes, 2012). secretion during subadulthood are not directly related to the degrees The sex of the only subadult included in this study was estimated of sexual dimorphism for the different tooth types. Following Alvesalo considering the morphological features of the ilia and mandibles (e.g., 3 LUNA TABLE 1 Sex and age‐at‐death data (in years) for the documented and the archaeological samples (Coimbra and Ossonova Romana, respectively) <20 Sex 20.1–35 35.1–50 n % n % n M 23 20.00 27 23.47 10 F 25 21.74 19 16.52 7 M+F 48 41.74 46 39.99 50+ % Total n % % 8.70 2 1.74 62 53.91 6.09 2 1.74 53 46.09 17 14.79 4 3.48 115 n Coimbra 100 Ossonova Romana M 1 3.12 4 12.50 6 18.75 2 6.26 13 40.63 F 5 15.62 11 34.37 2 6.26 1 3.12 19 59.37 M+F 6 18.74 15 46.88 8 25.00 3 9.38 32 100 Cardoso & Saunders, 2008; Holcomb & Konigsberg, 1995; Loth & grouped together. This method is suitable for analyzing the socio‐ Henneberg, 2001; Molleson, Cruse, & Mays, 1998; Scheuer, 2002; environmental influence on dental phenotype because it evaluates Schutkowski, 1993), and age‐at‐death was estimated by dental the associations between the number of indicators of metabolic stress maturity and long‐bone maximum lengths (Scheuer & Black, 2000). and tooth dimensions (Manzi, Santandrea, & Passarello, 1997; Permanent canines from 32 primary burials were considered in the y'Edinack, 1989). present research (Table 1). For both samples, teeth with any SPSS 16.0 (2007) and PAST 3.15 programmes were used to carry pathological damage were discarded, and only left antimeres were out all the statistical procedures. The differences between the mean taken into consideration. Rights were measured only when lefts were values of males and females were analyzed using the Mann–Whitney missing. U and the Kolmogorov–Smirnov Z nonparametric tests. Means were considered to be statistically different at α = 0.05. The correlation between dental size and age‐at‐death was evaluated using Spearman 2.2 | Measuring process and errors assessment correlation coefficient (r), and sexual dimorphism was calculated for each variable considering the formula SD = (M‐F)/F*100, first The mesiodistal and buccolingual diameters of the crown, and the proposed by Garn, Lewis, and Walenga (1968) and usually chosen in cervix of permanent upper and lower canines were measured follow- the field of dental anthropology (e.g., Hamilton, 1982; Kondo & ing Mayhall (2000) and Hillson, FitzGerald, and Flinn (2005) with an Townsend, 2004; Pettenati‐Soubayroux, Signoli, & Dutour, 2002; Absolute Digimatic Caliper Mitutoyo with a precision of 0.01 mm. Zorba et al., 2011). No calculation of missing data was needed because only intact canines The percentages of correct allocation for males, females, and the were chosen. First, the dental metric intraobserver error was analyzed whole sample (P [A|B]) were obtained considering the section points using the Intraclass Correlation Coefficient. Forty teeth from the that correspond to the average of the closest values for each sex. documented sample were measured twice, with a difference of at Moreover, discriminant functions and logistic regression equations least 2 weeks, and then the results were compared. Second, the influ- were generated including the most dimorphic variables. Both approx- ence on the environment in dental size was studied in order to identify imations are multivariate techniques that evaluate the associations if it affected the sexual expression; in such a case, sex estimations may among a large number of variables to maximize the capacity of be less reliable. Individuals who suffered from more adverse socio‐ discrimination in two or more groups. These traditional techniques environmental stress situations may have lower dental size, resulting identify the variables that better differentiate those groups and create from the inability to reach their genetic potential, and generally formulae that allow assigning new cases to one of them. Different associated with earlier ages‐at‐death. Authors as Cook and Buikstra recovery scenarios were considered in order to indicate to the (1979) and Guagliardo (1982) proposed that the stressors that program which variables should be included in each case: (a) All the influenced the development of enamel formation may also cause the measurements recorded (n = 8); (b) only those of the upper canine; individual to become more susceptible to subsequent selective agents, (c) only those of the lower canine; (d) only the coronal ones; (e) only implying that both the presence of numerous stress markers and the neck ones (n = 4 in each case). earlier deaths may be the consequence of situations of severe stress A section point for each discriminant function was created from during the growth period. In that case, smaller dental sizes in younger the centroids obtained (the mean discriminant score for each sex; individuals, and an inverse association between the age‐at‐death and see Isçan & Kedici, 2003 as an example). Only the variables with the number of dental stress indicators, should be observed (Boldsen, significant differences between the sexes were included in the 2007; Cook & Buikstra, 1979; Duray, 1996; Luna, 2008, 2015; Luna analysis. The stepwise forward method was used to obtain the & Aranda, 2010). The quantity of enamel hypoplastic defects was discriminant scores, and the cross‐validation method was applied to analyzed for each canine and compared with the dimensions of each verify the predictive capacity of each function, which has the follow- tooth. Canines were divided in two groups, with and without ing structure: hypoplasiae. The different types of hypoplastic defects (such as lines, pitting, and planes; Goodman & Rose, 1990; Hillson, 1996) were Sex ¼ α þ β1 X1 þ β2 X2 þ …βn Xn ; 4 3 | RESULTS 3.1 | Intraobserver errors assessment and the effect of environmental stress on dental size The results show low‐intraobserver errors for all the variables, which renders the methodology replicable (Table 2). Moreover, neither a significant influence of environmental constraints nor age‐at‐death effects were statistically detected (Table 2). Mann–Whitney and Kolmogorov–Smirnov tests show no relation between stress indicators and tooth size. On the other, all the values of r and coefficient of determination (that indicates the proportion of the variability of one of the variables considered by the other) are very low, which 0.92 0.01 0.01 0.01 0.98 0.82 0.39 0.01 0.03 0.08 0.92 0.01 0.05 0.03 0.74 0.07 0.08 −0.05 0.23 0.16 0.88 0.03 0.01 0.01 0.23 0.36 0.50 −0.17 0.12 0.06 0.90 0.01 0.03 0.03 0.47 0.14 0.05 0.10 0.19 0.18 0.01 0.01 0.01 0.87 0.86 0.51 0.03 −0.02 0.06 0.91 0.01 0.07 0.01 0.80 0.54 0.84 −0.04 −0.08 0.02 0.89 0.01 0.01 0.01 0.03 0.04 0.09 0.80 0.77 0.32 CD p r BLNe CCI CD p r Note. BLCr: buccolingual diameter of the crown; BLNe: buccolingual diameter of the neck; CD: Coefficient of Determination; LC: lower canine; MDCr: mesiodistal diameter of the crown; MDNe: mesiodistal diameter of the neck; UC: upper canine. and to discuss the potential of canine metrics for sex estimation. 0.93 comparing the results with those obtained from the skeletal analysis 0.02 0.14 0.28 in the archaeological sample (Ossonova Romana) with the intention of 0.89 0.79 0.63 each variable. The formulae and posterior probabilities were then used 0.35 0.44 0.65 Posterior probabilities were also calculated for different dimensions of 280.00 365.00 1256.00 PðBi ÞPðAjBi Þ  
: ∑kj¼1 P AjBj P Bj F M F+M PðBi jAÞ ¼ LC using the traditional Bayesian theorem: 0.00 0.01 0.01 Vaupel, 2002; Irurita Olivares & Alemán Aguilera, 2016; Koch, 2007), 0.92 0.44 0.29 individual may be correctly sexed; Bocquet‐Appel, 2008; Hoppa & 0.01 0.10 0.10 the posterior probabilities (or the probability that an unknown new 0.89 probabilities derived from the reference data were used to estimate 0.10 0.22 0.19 Finally, the likelihoods of correct allocation (P [B|A]) were also calculated using a Bayesian approach. The section points and the prior 0.65 0.63 0.35 ness of fit of the logistic regressions (Nagelkerke, 1991). 0.29 0.58 0.41 showing the power of explanation of the model, to evaluate the good- 326.00 380.00 1045.00 Ingersoll, 2002). Nagelkerke's R2 was calculated from the results, F M F+M denominator of a ratio whose numerator is unity (Peng, Lee, & UC applied to the neperian logarithm, which, in turn, is included in the MDNe product between the recorded measure and an associated value, CCI calculate an exponent that includes the sum of a constant with the CD than that of the discriminant functions because it is necessary to p to create the models. The structure of the equation is more complex r Viciano et al., 2015). In this case, the backward method was chosen CCI is to 0, the greater the probability that the individual is male (see BLCr the probability that the individual is female, and the closer the value CD equal to 2.718. If P(sex) > 0.5, the most likely sex is female, and male if P(sex) < 0.5; consequently, the closer the value is to 1, the greater p where e, the Euler constant (or neperian logarithm), is approximately r 1 ; 1 þ e–Li CCI PðsexÞ ¼ MDCr probability of sex (Psex) using the function p obtained in the previous formula), named Li, is used to calculate the Z On the other hand, logistic regressions measure the significance of the model through the analysis of deviance. The logit value (result p measurement, etc. (Hair, Anderson, Tatham, & Black, 1998). U where α is the constant, β1 is the first coefficient, X1 is the first TABLE 2 Statistical evaluation of the environmental influence in canine dimensions (Mann–Whitney U and Kolmogorov–Smirnov Z; canines with and without hypoplasia are compared), intraobserver errors (intraclass correlation coefficient; n = 40), and correlation between metrics and age‐at‐death (Spearman r) for the Coimbra sample LUNA 5 LUNA means that metrics are not influenced by age‐at‐death. All these On the other hand, most of the averages are higher for the results indicate that canine metric variations are due to sex in the Ossonoba Romana sample (Table 3) than those of the Coimbra Identified Coimbra sample. Skeletal Collection. The same occurs with the standard deviations, and the means and the medians are similar. Regarding the sexual dimorphism, the values range between 6.21% and 12.70%, and those for 3.2 | Intrapopulation variability the lower canines are higher for this sample. On the contrary, upper canine sexual dimorphism is lower for three of the four variables, and Table 3 shows the descriptive statistics and the dispersion of the the same is shown in the results of the neck. In this case, U values indi- canine measurements from the Coimbra sample. Some ranges overlap cate significant statistical differences between males and females for all between males and females (e.g., MDCrLC and BLCrUC), whereas the neck measurements and the mesiodistal diameter of the crown. others show many (lower) values for females and many (higher) values for males (e.g., MDNeLC). As the means and medians are very similar 3.3 | Discriminant functions and logistic regressions for all the variables, the distributions are almost symmetrical. Sexual dimorphism ranges between 5.23% and 11.59%; neck measurements The discriminant functions obtained from the reference sample are show higher values (10.38–11.59%) than crown ones (5.23–9.35%), shown in Table 5. The variables selected and included in the functions and upper canine neck sexual dimorphism is higher (11.04–11.59%) are always those of the neck, with the exception of the functions than lower canine results (10.38–10.91%). The values of the created only through the measurements of the crown. A good predictive mesiodistal diameters are lower than those of the buccolingual diame- capacity is observed, because the percentages of correct assignation vary ters, for both upper and lower crowns, whereas the mesiodistal and between 77.42% and 86.54%, and the values from the cross validation buccolingual results for the neck are very similar. Moreover, U values range between 77.42% and 84.68%. On the other hand, the only formu- show statistical significant differences between sexes in all the lae that includes a variable of the crown shows lower percentages of cor- measurements considered. The percentages of correct assignations rect assignations (Males = 78.83%; Females = 79.00%; both for the reference sample range from 71.70% to 92.45%. Most of the sexes = 78.94%; Table 5). These results support that the canine neck variables show values higher than 75% for males and females (all dimensions are much more dimorphic than those of the crown in this except BLCrUC and MDNeUC), and when both sexes are grouped, sample. Finally, the formula for the upper canines offers better results the results vary between 78.26% and 85.21% (Tables 3 and 4). than that for the lower canines. TABLE 3 Descriptive statistics, sexual dimorphism (%), and statistical differences between sexes (Mann–Whitney U) for the Coimbra Identified Skeletal Collections (M = 62; F = 53) and the Ossonoba Romana (M = 13; F = 19) samples LC UC MDCr F BLCr M F MDNe BLNe MDCr BLCr MDNe BLNe M F M F M F M F M F M F M Coimbra Identified Skeletal Collections Min. 5.76 6.14 6.54 6.81 4.49 4.82 6.19 6.35 6.47 6.77 7.05 6.89 4.66 4.89 6.65 6.96 Max. 7.52 7.54 7.89 9.12 5.68 6.42 7.87 8.72 8.03 8.50 8.81 9.58 6.06 6.89 8.44 9.22 Mean 6.45 6.79 7.21 7.88 5.03 5.58 6.97 7.70 7.31 7.71 7.87 8.47 5.39 5.99 7.34 8.19 SD 0.35 0.33 0.4 0.49 0.28 0.36 0.44 0.59 0.36 0.41 0.47 0.64 0.33 0.43 0.44 0.59 Median 6.46 6.78 7.22 7.89 5.05 5.63 6.9 7.8 7.28 7.69 7.82 8.53 5.38 6.05 7.27 8.16 25 prcntil 6.21 6.54 6.9 7.6 4.75 5.3 6.7 7.4 7.1 7.45 7.54 8.15 5.15 5.74 7.04 7.80 75 prcntil 6.66 7.01 7.47 8.12 5.26 5.79 7.28 8.09 7.52 8.03 8.19 8.93 5.71 6.26 7.61 8.55 Dimorphism 5.23 9.35 10.91 10.38 5.47 7.58 11.04 11.59 768.5* 470.5* 385.0* 477.0* 748.0* 702.0* 444.0* 399.0* U Ossonoba Romana Min. 5.49 6.13 6.22 6.82 4.43 4.62 6.20 6.71 6.38 7.18 7.09 7.18 4.97 5.15 6.39 6.79 Max. 7.25 7.99 8.64 9.80 5.98 7.00 8.58 9.68 7.99 8.50 9.57 9.75 6.11 6.99 9.44 9.65 Mean 6.41 7.16 7.31 8.09 5.22 5.68 7.16 8.07 7.29 7.92 8.20 8.71 5.60 6.04 7.73 8.38 SD 0.47 0.53 0.56 0.92 0.42 0.76 0.56 0.87 0.43 0.49 0.67 0.77 0.37 0.61 0.78 0.86 Median 6.47 7.31 7.17 8.12 5.19 5.64 7.08 8.17 7.28 8.00 8.13 8.99 5.72 6.36 7.57 8.64 25 prcntil 6.00 6.78 6.96 7.18 5.04 5.02 6.85 7.23 7.05 7.42 7.56 8.07 5.27 5.31 7.21 7.70 75 prcntil 6.78 7.55 7.69 8.67 5.48 6.32 7.57 8.52 7.64 8.38 8.62 9.23 5.95 6.50 8.24 8.85 Dimorphism 11.70 10.67 8.81 12.70 8.64 6.21 7.85 8.41 U 39.50* 83.00 65.00** 51.00* 38.00* 68.00 64.00** 62.00** Note. BLCr: buccolingual diameter of the crown; BLNe: buccolingual diameter of the neck; LC: lower canine; MDCr: mesiodistal diameter of the crown; MDNe: mesiodistal diameter of the neck; UC: upper canine; References: see Table 2. *p < 0.01; **p < 0.05. 6 LUNA TABLE 4 Number of cases, percentages of correct assignations (P [A|B]), and posterior probabilities (P [B|A]) for males, females, and the whole reference sample n P (A|B) P (B|A) Tooth Variable SP (mm) F M F+M F M F+M F M F+M LC MDCr BLCr MDNe BLNe 6.83 7.35 5.25 7.11 49 40 42 40 47 55 56 53 96 95 98 92 92.45 75.47 79.24 75.47 75.80 88.71 90.32 85.48 83.48 82.61 85.21 80.00 0.83 0.77 0.78 0.78 0.77 0.75 0.85 0.82 0.83 0.76 0.82 0.80 UC MDCr BLCr MDNe BLNe 7.40 7.85 5.48 7.70 40 39 38 45 50 52 54 51 90 91 92 96 75.47 73.59 71.70 84.90 80.64 83.87 87.09 82.26 78.26 79.13 80.00 83.48 0.71 0.74 0.74 0.80 0.79 0.77 0.79 0.82 0.75 0.76 0.77 0.81 Note. F = 53; M = 62; Total = 115; References: see Table 2. TABLE 5 Discriminant functions and logistic regressions considering different recovery scenarios Cases correctly sexed Original # Variables Cross validation Discriminant functions SP M F M+F M F M+F 1 All X= ‐18.016+1.833*MDNeLC+1.055*BLNeUC ‐0.0835 83.94 86.54 85.00 80.63 84.68 82.45 2 LC X= ‐17.572+2.039*MDNeLC+0.907*BLNeLC ‐0.0705 80.71 77.42 78.94 80.71 77.42 78.94 3 UC X= ‐16.375+1.280*MDNeUC+1.161*BLNeUC ‐0.0774 84.68 80.64 82.45 84.68 80.64 82.45 4 Crowns X= ‐16.604+2.192*BLCrLC ‐0.0667 78.83 79.00 78.94 78.83 79.00 78.94 5 Necks X= ‐18.016+1.833*MDNeLC+1.055*BLNeUC ‐0.0835 83.94 86.54 85.00 80.63 84.68 82.45 # Variables 1 All 2 Logistic regressions PðsexÞ ¼ LC 3 UC 4 Crowns 5 Necks 1 1 þ e−ð−32:596þ3:793 PðsexÞ ¼ PðsexÞ ¼ PðsexÞ ¼ M F M+F Nagelkerke R2 0.5 82.75 87.10 85.18 0.76 78.86 83.97 81.63 0.70 76.95 87.10 82.50 0.70 78.86 85.55 82.45 0.69 82.75 85.55 84.20 0.75 MDNeLCþ½−1:692* BLCrUCŠ þ3:423* BLNeLCÞ 1 1 þ e−ð−29:340þ3:630 1 þ e−ð−25:554 PðsexÞ ¼ * SP * MDNeLCþ1:405* BLNeLCÞ 1 þ ½−1:479* BLCrUCŠ þ2:180* MDNeUCþ 3:284* BLNeUCSÞ 1 1 þ e−ð−32:038þ1:326 1 þ e−ð−34:515þ3:558 * * MDCrUCþ2:959* BLCrLCÞ 1 MDNeLCþ2:058* BLNeUCÞ Note. SP: section point; References: see Table 4. Regarding the logistic regressions, the cases correctly allocated The posterior probabilities were then calculated considering the for both sexes range between 81.63% and 85.18% (Table 5). The different measurements for each canine, dividing the total range in highest percentages are observed when all the measurements are 0.25‐mm sections. As shown in Table 6, it is easy to define which included to generate the formulae, as well as when those of the neck the most probable sex is, regarding the probability associated to a are selected. The same trend as in the case of the discriminant func- given measurement. Only a few exceptions are observed in the tions is seen, because the variables included are only those of the neck mesiodistal diameter of the upper canine neck (measurements: 5.50 for most of the formulae. The values of the coefficient of determina- and 5.75 mm), the buccolingual diameter of the upper canine neck tion (R2 of Nagelkerke) range between 0.70 and 0.76, which mean that (measurement: 7.75 mm), the buccolingual diameter of the lower the formulae are adequate to discriminate between sexes. canine crown (measurements: 7.50 and 7.75 mm), and the buccolingual diameter of the lower canine neck (measurements: 6.50 3.4 | Bayesian approach and 7.50 mm); in those cases, the probabilities for males and females are similar, what renders the estimation doubtful (Table 6). The likelihoods of correct allocation (P [B|A]) were first obtained con- The values offered inTable 6 were then used for sex estimation of 10 sidering the section points of each variable. The Bayesian approach females and 10 males not included in the previous analysis (Table 7). calculates the probability that an individual with a value higher than These results allow estimating the actual sex with reliability. For example, the section point, be a male, or, on the contrary, one with a lower for males, when the eight variables are included, the mean posterior value, be a female. The posterior probabilities for each variable range probability density of being male is 0.18 and of being female is much between 0.71 and 0.83 for females and between 0.75 and 0.85 for smaller (1.987E‐05). On the other hand, for females, the mean probability males. With the exception of the probabilities of female upper canines, density of being male is 0.01, whereas that of being female is 0.15. The all are higher than 0.75. Upper canines and crowns show slightly lower same trends are observed for all the combinations, but the differences values than lower canines and necks, respectively (Table 4). are much stronger when more variables are included (Table 7). This data 7 LUNA TABLE 6 Posterior probabilities (P [B|A]) for different possible measurements UC LC MDNe BLCr BLNe MDCr BLCr MDNe BLNe Mm M F M F M F M F M F M F M F M F 4.75 ‐ ‐ ‐ ‐ 0.41 0.81 ‐ ‐ ‐ ‐ ‐ ‐ 0.33 0.87 ‐ ‐ 5.00 ‐ ‐ ‐ ‐ 0.46 0.76 ‐ ‐ ‐ ‐ ‐ ‐ 0.68 0.78 ‐ ‐ 5.25 ‐ ‐ ‐ ‐ 0.49 0.74 ‐ ‐ ‐ ‐ ‐ ‐ 0.74 0.64 ‐ ‐ 5.50 ‐ ‐ ‐ ‐ 0.73 0.67 ‐ ‐ ‐ ‐ ‐ ‐ 0.76 0.54 ‐ ‐ 5.75 ‐ ‐ ‐ ‐ 0.78 0.63 ‐ ‐ ‐ ‐ ‐ ‐ 0.80 0.40 ‐ ‐ 6.00 ‐ ‐ ‐ ‐ 0.78 0.52 ‐ ‐ 0.40 0.80 ‐ ‐ 0.83 0.29 ‐ ‐ 6.25 ‐ ‐ ‐ ‐ 0.80 0.43 0.39 0.81 0.45 0.72 ‐ ‐ 0.81 0.30 0.67 0.82 0.81 0.20 0.72 0.77 ‐ 0.60 0.75 6.50 0.39 0.75 ‐ ‐ 0.82 0.34 0.39 0.81 0.54 0.66 0.39 0.80 6.75 0.40 0.75 ‐ ‐ 0.85 0.28 0.39 0.80 0.62 0.42 0.40 0.78 7.00 0.47 0.54 0.29 0.77 ‐ ‐ 0.45 0.75 0.75 0.38 0.46 0.73 ‐ ‐ 0.39 0.66 7.25 0.50 0.50 0.48 0.72 ‐ ‐ 0.48 0.70 0.79 0.31 0.50 0.69 ‐ ‐ 0.41 0.60 7.50 0.74 0.45 0.46 0.69 ‐ ‐ 0.52 0.62 0.77 0.28 0.59 0.62 ‐ ‐ 0.62 0.54 7.75 0.76 0.42 0.49 0.62 ‐ ‐ 0.57 0.56 ‐ ‐ 0.63 0.60 ‐ ‐ 0.77 0.50 0.45 8.00 0.76 0.26 0.77 0.54 ‐ ‐ 0.76 0.49 ‐ ‐ 0.79 0.48 ‐ ‐ 0.81 8.25 0.80 0.15 0.79 0.51 ‐ ‐ 0.80 0.38 ‐ ‐ 0.80 0.36 ‐ ‐ 0.81 0.36 8.50 ‐ ‐ 0.80 0.46 ‐ ‐ 0.80 0.34 ‐ ‐ 0.80 0.25 ‐ ‐ 0.81 0.26 8.75 ‐ ‐ 0.77 0.42 ‐ ‐ 0.81 0.29 ‐ ‐ 0.80 0.24 ‐ ‐ 0.81 0.26 9.00 ‐ ‐ 0.81 0.37 ‐ ‐ 0.82 0.28 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 9.25 ‐ ‐ 0.79 0.32 ‐ ‐ 0.84 0.28 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 9.50 ‐ ‐ 0.80 0.29 ‐ ‐ 0.84 0.28 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 9.75 ‐ ‐ 0.80 0.28 ‐ ‐ 0.84 0.28 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ 10.00 ‐ ‐ 0.80 0.28 ‐ ‐ 0.84 0.28 ‐ ‐ ‐ ‐ ‐ ‐ ‐ ‐ Note. Values higher than 0.75 are in bold. References: see Table 4. can be used for allocating individuals with confidence, as shown in Anyway, the highest results were obtained from the analysis of the four Table 8, where the direct measurements and probability densities for a lower canine variables, with 84.61% of correct assignations for males, single male and a single female are strongly different. 84.21% for females, and 84.37% for both sexes; the four neck variables also offer high percentages of correct allocation (76.92% for males, 81.25% for females, and 79.31% for both sexes). 3.5 | Application in the cemetery of Ossonoba Romana 4 | DISCUSSION The sexual allocation methods proposed in the present study were applied to Ossonoba Romana sample in order to discuss their potential The application of multifactorial analysis to odontometrics have emerged for sex estimation of archaeological individuals (Table 9). More than as a useful tool for sex estimation because they considerably increase the 81% of the results were adequate (>75% of the cases correctly quality and quantity of data (Viciano et al., 2015), improving the reliability assigned) for males, much better than for females (33.33%), which of paleodemographic discussions and the accuracy of forensic identifica- means that only some of the methods can be used in this sample. tions. In this research, intraobserver error is not statistically significant, One discriminant function (#1 and #5 are the same formulae) and two and factors such as age‐at‐death and the levels of stress suffered during logistic regressions show values higher than 75% for both sexes sepa- subadulthood did not affect sexual dimorphism. Moreover, the sex ratio rately. All of them include neck measurements, and one contains a var- is 1.17/1 (see Albanese, Cardoso, & Saunders, 2005), which mean that iable from the crown (BLCrUC). Logistic formula #2 also offers sex balance is acceptable and do not influence the results obtained. In satisfactory results when both sexes are grouped (75%). The most consequence, canine metrics constitute an adequate tool for sex estima- successful formulae is logistic regression #1, with 84.6% of correct allo- tion of the Portuguese population from the XIX‐XX centuries. cation for males, 82.38% for females, and 83.33% for both sexes. Con- Univariate approaches sometimes offer better results than multi- ditional probability values show that the only single variable that offers variate perspectives. For example, Potter et al. (1981) analysed the high percentages for both sexes is lower canine MDCr (76.92% for sexual dimorphism in a human sample from the Philippines and identi- males, 80.67% for females, and 78.12% for both sexes), and when con- fied high degrees in most of the variables considered in isolation, but sidering two variables, good results are given by the lower canine crown multivariate formulae offered a much lower percentage of correct (81.81% for males, 78.94% for females, and 78.12% for both sexes). allocations. The same trend was observed by Kieser, Groeneveld, 8 LUNA TABLE 7 Posterior probability densities of 20 additional individuals from the Coimbra sample, considering the section points given in Table 4 pM Variables pF Number of variables Mean SD Min. Max. Mean SD Min. Max. 1 0.78 0.07 0.50 0.76 0.43 0.06 0.26 0.50 Males MDCrUC BLCrUC 1 0.71 0.17 0.29 0.81 0.51 0.13 0.37 0.77 MDNeUC 1 0.75 0.09 0.49 0.80 0.55 0.09 0.43 0.74 BLNeUC 1 0.78 0.12 0.45 0.80 0.43 0.13 0.29 0.75 MDCrLC 1 0.65 0.14 0.45 0.79 0.489 0.14 0.31 0.75 BLCrLC 1 0.680 0.12 0.46 0.80 0.54 0.14 0.24 0.69 MDNeLC 1 0.73 0.14 0.33 0.83 0.55 0.17 0.29 0.78 BLNeLC 1 0.73 0.08 0.60 0.81 0.56 0.10 0.45 0.75 UC Neck 2 0.63 0.08 0.51 0.72 0.09 0.42 0.07 1.00 LC Neck 2 0.70 0.06 0.54 0.73 0.08 0.44 0.07 0.09 UC Crown 2 0.60 0.04 0.51 0.63 0.06 0.37 0.04 0.07 LC Crown 2 0.66 0.06 0.52 0.70 0.38 0.35 0.07 0.09 UC 4 0.37 0.09 0.29 0.53 0.01 0.02 0.01 0.01 LC 4 0.46 0.10 0.28 0.58 0.01 0.01 0.01 0.01 Neck 4 0.50 0.05 0.37 0.52 0.01 0.01 0.01 0.01 Crown 4 0.40 0.04 0.36 0.48 0.01 0.02 0.01 0.01 All 8 0.18 0.04 0.12 0.26 1.987E‐05 0.01 2.05E‐06 1.25E‐04 Females MDCrUC 1 0.51 0.13 0.40 0.74 0.55 0.12 0.50 0.75 BLCrUC 1 0.52 0.10 0.46 0.80 0.64 0.07 0.46 0.69 MDNeUC 1 0.52 0.10 0.46 0.73 0.73 0.03 0.67 0.76 BLNeUC 1 0.48 0.10 0.39 0.76 0.72 0.09 0.49 0.80 MDCrLC 1 0.46 0.06 0.40 0.62 0.73 0.04 0.66 0.80 BLCrLC 1 0.50 0.08 0.39 0.63 0.67 0.15 0.24 0.80 MDNeLC 1 0.59 0.18 0.33 0.76 0.78 0.07 0.64 0.87 BLNeLC 1 0.58 0.08 0.41 0.72 0.65 0.07 0.54 0.77 UC Neck 2 0.36 0.11 0.20 0.49 0.52 0.04 0.39 0.68 LC Neck 2 0.30 0.08 0.18 0.40 0.69 0.06 0.49 0.88 UC Crown 2 0.12 0.01 0.10 0.16 0.43 0.04 0.32 0.57 LC Crown 2 0.40 0.01 0.01 0.01 0.50 0.02 0.40 0.69 UC 4 0.04 0.01 0.03 0.06 0.43 0.07 0.30 0.55 LC 4 0.02 0.01 0.02 0.05 0.51 0.06 0.45 0.68 Neck 4 0.06 0.01 0.04 0.08 0.39 0.08 0.20 0.50 Crown 4 0.02 0.01 0.01 0.04 0.23 0.06 0.16 0.32 All 8 0.01 8.38E‐04 0.01 0.01 0.15 0.05 0.11 0.25 Note. pF: female probabilities; pM: male probabilities; SD: standard deviation; References: see Table 4. and Preston (1985), Acharya and Mainali (2008), Isçan and Kedici (2003), are similar to those shown by Galera and Cunha (1993) and show that and Falk and Corruccini (1982) in the last case for neck measurements. In sexual dimorphism is moderate for this population. The importance of the present research, all the variables showed significant differences the cervical measurements is highlighted because most of the between sexes, and except BLCrUC and MDNeUC, percentages of correct variables chosen by the statistical programme are those of the neck. assignations are higher than 75%. This trend shows the strength of single In consequence, these formulae can be used for sex estimation with measurements for sex estimation and is in concordance with previous reliance on samples phenotypically related to the identified skeletons investigations that offer high percentages of correct estimations for the of the University of Coimbra. On the other hand, the percentages Portuguese population (e.g., Bermúdez de Castro et al., 2001; Galera & of Cunha, 1993; Pereira et al., 2010). (77.42–86.54%) and the logistic regressions (81.63–85.18%) are correct allocation given by the discriminant functions The percentages of sexual dimorphism obtained here in the similar to those obtained by previous research considering different Coimbra sample range between 11.04% and 11.59% for the neck measurements in samples around the world (e.g., Isçan & Kedici, and between 5.23% and 9.35% for the crown. These latter values 2003 for Turks: 73–77%; Acharya & Mainali, 2007 for Nepalese: 9 LUNA TABLE 8 Posterior probability densities for a female and a male from the Coimbra sample, considering the section points given in Table 4 Number of variables Male individual Female individual mm pM pF mm pM pF MDCrUC 1 8.31 0.80 0.15 6.61 0.39 0.75 BLCrUC 1 9.30 0.79 0.32 7.05 0.29 0.77 MDNeUC 1 6.89 0.85 0.28 5.73 0.73 0.67 BLNeUC 1 9.21 0.82 0.29 6.24 0.39 0.81 MDCrLC 1 7.54 0.77 0.28 5.89 0.40 0.80 BLCrLC 1 8.77 0.80 0.24 6.33 0.39 0.80 MDNeLC 1 6.42 0.81 0.30 4.71 0.33 0.87 BLNeLC 1 8.90 0.81 0.26 6.30 0.67 0.82 UC Neck 2 ‐ 0.69 0.08 ‐ 0.28 0.54 LC Neck 2 ‐ 0.65 0.08 ‐ 0.23 0.67 UC Crown 2 ‐ 0.64 0.05 ‐ 0.11 0.58 LC Crown 2 ‐ 0.62 0.07 ‐ 0.01 0.64 UC 4 ‐ 0.44 0.01 ‐ 0.03 0.31 LC 4 ‐ 0.40 0.01 ‐ 0.04 0.43 Neck 4 ‐ 0.45 0.01 ‐ 0.07 0.37 Crown 4 ‐ 0.39 0.01 ‐ 0.09 0.37 All 8 ‐ 0.18 1.987E‐05 ‐ 0.01 0.13 Note. pF: female probabilities; pM: male probabilities; References: see Table 4. 77.4–83%; Viciano et al., 2013 for Spaniards: 79.4–92.6%; Khamis are different from those of the studied record must be carried out with et al., 2014 for Malaysians: 70.2–78.5%; Viciano et al., 2015 for caution, evaluating the reliability of the results. archaeological Italians: 83.7–95.9%; and García‐Campos et al., 2018 Regarding the conditional probabilities, although most of them for a modern sample from Spain, South Africa, and Sudan: cannot be used in the Ossonoba Romana sample, when previous tests 71.43–84.62%). These findings confirm the usefulness of permanent are carried out, some variables or groups of variables that offer high canine measurements for sex estimation in specific populations and probabilities of sexual estimation are identified and, in turn, can be the need to test their applicability in geographically‐related samples. applied with confidence (i.e., the mesiodistal crown diameter of the The posterior probabilities obtained considering the section lower canine, both diameters of the lower canine crown, the four points of each variable are also adequate for sex estimation variables of the lower canine, and the four of the necks; see Table 9 (pfem = 0.71–0.83; pmasc = 0.75–0.85). This approach offers a powerful ). In most of the cases, neck variables offer better results than those tool because it can be easily improved when new cases are included in of the crown for both samples, the same as for discriminant functions the calculations and addresses the problem from a probabilistic per- and logistic regressions. spective, without considering arbitrary and fixed statistical thresholds The magnitude of sexual dimorphism in tooth size is much (such as those of the discriminant functions and logistic regressions). affected by genetic differences among and within populations (Garn In fact, the data given in Tables 7 and 8 for another sample from the et al., 1967; Moss & Moss‐Salentijn, 1977; Saunders et al., 2007), same osteological collection indicate that this approach is reliable but environmental constraints may also contribute to phenotypic and can be used in human remains of similar provenance. It is diversity, such as nutrition, disease, climate, and subsistence patterns suggested that this kind of procedure be incorporated in current (Garn, Osborne, & McCabe, 1979; Harris, Hicks, & Barcroft, 2001; research for dental sex estimations because the accuracy of the results Luna, 2015; Potter, Rice, Dahlberg, & Dahlberg, 1983; Townsend & obtained are usually improved. Brown, 1978a, 1978b; Zorba et al., 2011). The differences in tooth For Ossonoba Romana, most of the direct measurements and dimensions may also be related to demographic dynamics and to the values of sexual dimorphism are higher than for the Coimbra sample. degree of admixture and genetic drift (Hartl & Jones, 2010). In this In consequence, a trend towards reducing dental size is seen, as case, genetic admixture may have played a central role in dental size previously shown for other populations (e.g., Brace & Ryan, 1980; and shape variation, taking into account the succession of invasions, Pereira et al., 2010), in this case during the last two millennia. Only retractions, and population extinctions that occurred in the Iberian three formulae offered reliable sex estimations, which points that Peninsula during the last 2,000 years. Not only the imprint of the the application in samples from other sources must be done with cau- Muslim populations that settled in the region since the VIII century tion. As Roberts and Manchester (1999) stated almost two decades should be considered as a crucial factor in the phenotypic variation ago, when different human populations are compared, the dimorphic of the Portuguese populations, but also previous entrances of patterns can be very variable for which the application of some populations, such as the Germanic Kingdoms of the Suevis and the techniques developed with reference samples whose characteristics Visigoths, significantly contributed to the phenotypic variability 10 LUNA TABLE 9 Percentages of correct sex assignations for the Ossonoba Romana sample (13 males, 19 males) using the formulae and the Bayesian approach generated with the Coimbra sample M F et al., 2014; Pettenati‐Soubayroux et al., 2002; Thompson, 2013). Finally, it is stressed than the application of the posterior probabilities provided in this paper may improve the quality of information M+F obtained in such cases. This approximation to sex estimation is an Discriminant function 1 76.96 76.43 76.66 overarching approach that offers additional strategies to deal with Discriminant function 2 61.54 68.45 65.63 the problem from both forensic and archaeological perspectives. Discriminant function 3 69.25 50.00 58.00 Discriminant function 4 61.54 63.14 62.55 ACKNOWLEDGEMENTS Discriminant function 5 76.96 76.43 76.66 Thanks are given to Claudia Aranda and Ana Luisa Santos for their Logistic regression 1 84.60 82.38 83.33 comments, help, and advisements regarding the recording stage and Logistic regression 2 76.96 73.63 75.00 the content of this paper and to Sofía Luna Aranda for her language Logistic regression 3 78.58 64.70 70.96 revision. The author also thanks the Department of Life Sciences Logistic regression 4 76.96 64.70 70.00 and the Centre of Investigation in Anthropology and Health (CIAS; Logistic regression 5 76.96 76.43 76.66 Pest‐UID/ANT/0283/2013), University of Coimbra, for allowing MDCrUC 76.92 35.29 50.00 carrying out this investigation. It was partially developed with a BLCrUC 84.61 29.41 56.66 postdoctoral grant awarded by the National Council of Scientific and MDNeUC 76.92 35.29 50.00 Technical Investigations (CONICET), Argentina. I also thank the BLNeUC 76.92 58.82 64.51 reviewers, who substantially improved the content of this paper MDCrLC 76.92 80.67 78.12 through their useful comments. BLCrLC 61.53 76.47 65.62 MDNeLC 61.53 41.17 46.87 BLNeLC 76.92 70.58 78.75 UC Neck 76.92 47.06 61.29 ORCID Leandro H. Luna https://orcid.org/0000-0002-5454-5570 LC Neck 76.92 73.68 71.87 RE FE RE NC ES UC Crown 92.30 47.06 67.74 LC Crown 81.81 78.94 78.12 UC 84.61 52.94 67.74 LC 84.61 84.21 84.37 Acharya, A., & Mainali, S. (2007). Univariate sex dimorphism in the Nepalese dentition and the use of discriminant functions in gender assessment. Forensic Science International, 173, 47–56. https://doi. org/10.1016/j.forsciint.2007.01.024 Neck 76.92 81.25 79.31 Crown 76.92 52.94 63.33 All 76.92 52.94 63.33 Note. Percentages higher than 75% are shown in italics. References: see Table 2 subsequently observed (see for example, Disney, 2009). Of course, microevolutionary processes and the action of local environmental effects must also be taken into account to explain this process of diminishing tooth size and sexual dimorphism. The relevance of considering some metric variables of the canine for sexual estimation in human samples of similar provenance, although temporarily different, is highlighted as a conclusion of this research. As clearly showed in previous articles (e.g., Isçan & Kedici, 2003; Karaman, 2006; Mitsea et al., 2014; Rosing, 1983; Viciano et al., 2015; Vodanovic et al., 2007; Zorba et al., 2011), tooth size and sexual dimorphism may vary in different populations, and that is why specific population data are needed, and previous testing is mandatory before using it in samples different from which the method was created. In this research, reliable information is offered for sex estimation from traditional and Bayesian statistics, which proved to be adequate for application in spatial‐related samples. This data aid in bioarchaeological and forensic cases, mostly when the human Acharya, A., & Mainali, S. (2008). Sex discrimination potential of buccolingual and mesiodistal tooth dimensions. Journal of Forensic Sciences, 53(4), 790–792. https://doi.org/10.1111/j.1556‐4029.2008.00778.x Albanese, J., Cardoso, H., & Saunders, S. (2005). Universal methodology for developing univariate sample‐specific sex determination methods: An example using the epicondylar breadth of the humerus. Journal of Archaeological Science, 32, 143–152. https://doi.org/10.1016/j. jas.2004.08.003 AlQahtani, S., Hector, M., & Liversidge, H. (2010). Brief communication: The London Atlas of human tooth development and eruption. American Journal of Physical Anthropology, 142(3), 481–190. https://doi.org/ 10.1002/ajpa.21258 Alvesalo, L. (1997). Sex chromosomes and human growth: A dental approach. Human Genetics, 101, 1–5. https://doi.org/10.1007/ s004390050575 Aris, C., Nystrom, P., & Craig‐Atkins, E. (2018). A new multivariate method for determining sex of immature human remains using the maxillary first molar. American Journal of Physical Anthropology, 167(3), 672–683. https://doi.org/10.1002/ajpa.23695 Bermúdez de Castro, J. M., Sarmiento, S., Cunha, E., Rosas, A., & Bastir, M. (2001). Dental size variation in the Atapuerca‐SH Middle Pleistocene hominids. Journal of Human Evolution, 41, 195–209. https://doi.org/ 10.1006/jhev.2001.0491 Bocquet‐Appel, J. (2008). Recent advances in palaeodemography. In Data, Techniques, Patterns. London: Springer. https://doi.org/10.1007/978‐ 1‐4020‐6424‐1 remains are fragmented or commingled and especially for subadult sex estimation, a very important aspect for the identification of missing individuals and the development of complete mortality profiles (e.g., Khamis et al., 2014; Luna, 2008, 2010, 2012, 2016; Mitsea Bogin, B., & Smith, H. (2000). Evolution of the human life cycle. In S. Stinson, B. Bogin, R. Huss‐Ashmore, & D. O'Rourke (Eds.), Human biology: An evolutionary and biocultural perspective (pp. 377–424). New York: Wiley Liss. 11 LUNA Boldsen, J. (2007). Early childhood stress and adult age mortality. A study of dental enamel hypoplasia in the Medieval Danish village of Tirup. American Journal of Physical Anthropology, 132, 59–66. https://doi. org/10.1002/ajpa.20467 Brace, C., & Ryan, A. (1980). Sexual dimorphism and human tooth size differences. Journal of Human Evolution, 9, 417–435. https://doi.org/ 10.1016/0047‐2484(80)90051‐2 Bruzek, J. (2002). A method for visual determination of sex, using the human hip bones. American Journal of Physical Anthropology, 117(2), 157–168. https://doi.org/10.1002/ajpa.10012 Buikstra, J., & Ubelaker, D. (1994). Standards for data collection from human skeletal remains. Arkansas Archaeological Survey Research Series No. 44. Arkansas Archeological Survey: Faytteville, Arkansas. Cardoso, H. (2010). Testing discriminant functions for sex determination from deciduous teeth. Journal of Forensic Science, 55(6), 1557–1560. https://doi.org/10.1111/j.1556‐4029.2010.01530.x Cardoso, H., & Saunders, S. (2008). Two arch criteria of the ilium for sex determination of immature skeletal remains: A test of their accuracy and an assessment of intra‐ and inter‐observer error. Forensic Science International, 178(1), 24–29. https://doi.org/ 10.1016/j.forsciint.2008.01.012 Cook, D., & Buikstra, J. (1979). Health and differential survival in prehistoric populations: Prenatal dental defects. American Journal of Physical Anthropology, 51, 649–664. https://doi.org/10.1002/ ajpa.1330510415 Disney, A. (2009). A history of Portugal and the Portuguese empire. From beginnings to 1807. Victoria: La Trobe University. Duckworth, R. (2006). The teeth and their environment. Physical, chemical and biochemical influences. Basel: Karger. Duray, S. (1996). Dental indicators of stress and reduced age at death in prehistoric Native Americans. American Journal of Physical Anthropology, 99, 275–286. https://doi.org/10.1002/(SICI)1096‐8644(199602) 99:2<275::AID‐AJPA5>3.0.CO;2‐Y Falk, D., & Corruccini, R. (1982). Efficacy of cranial versus dental measurements for separating human populations. American Journal of Physical Anthropology, 57, 123–128. https://doi.org/10.1002/ ajpa.1330570113 Fernandes, H. (2012). Ossonoba Romana. Análise paleopatológica e paleobiológica de indivíduos adultos da amostra da necrópole do sec. I‐III d.C. exumada no Largo 25 de Abril (Master Thesis). University of Coimbra, Coimbra, Portugal. Frayer, D., & Wolpoff, M. (1985). Sexual dimorphism. Annual Review of Anthropology, 14, 429–473. https://doi.org/10.1146/annurev. an.14.100185.002241 Galera, V., & Cunha, E. (1993). Dental patterns of Coimbra population. L'Anthropologie, 31(1/2), 35–44. García‐Campos, C., Martiñón‐Torres, M., Martín‐Francés, L., Martínez de Pinillos, M., Modesto‐Mata, M., Perea‐Pérez, B., … Bermúdez de Castro, J. M. (2018). Contribution of dental tissues to sex determination in modern human populations. American Journal of Physical Anthropology, 166(2), 459–472. https://doi.org/10.1002/ajpa.23447 Garn, S., Lewis, A., Swindler, D., & Kerewsky, R. (1967). Genetic control of sexual dimorphism in tooth size. Journal of Dental Research, 46(5), 963–972. https://doi.org/10.1177/00220345670460055801 Garn, S., Lewis, A., & Walenga, A. (1968). Crown size profile pattern comparisons of 14 human populations. Archives of Oral Biology, 13, 1235–1242. https://doi.org/10.1016/0003‐9969(68)90079‐4 Garn, S., Osborne, R., & McCabe, K. (1979). The effect of prenatal factors on crown dimensions. American Journal of Physical Anthropology, 51, 665–678. https://doi.org/10.1002/ajpa.1330510416 Goodman, A., & Rose, J. (1990). Assessment of systemic physiological perturbations from dental enamel hypoplasias and associated histological structures. Yearbook of Physical Anthropology, 33, 59–110. https://doi. org/10.1002/ajpa.1330330506 Guagliardo, M. (1982). Tooth crown size differences between age groups: A possible new indicator of stress in skeletal samples. American Journal of Physical Anthropology, 58, 383–389. https://doi.org/10.1002/ ajpa.1330580405 Guatelli‐Steinberg, D., Sciulli, P., & Betsinger, T. (2008). Dental crown size and sex hormone concentrations: Another look at the development of sexual dimorphism. American Journal of Physical Anthropology, 137, 324–333. https://doi.org/10.1002/ajpa.20878 Hair, J., Anderson, R., Tatham, R., & Black, W. (1998). Multivariate data analysis. Prentice Hall: Upper Saddle River, NJ. Hamilton, M. (1982). Sexual dimorphism in skeletal samples. In R. Hall (Ed.), Sexual dimorphism in Homo sapiens (pp. 107–163). New York: Praeger. Harris, E., Hicks, J., & Barcroft, B. (2001). Tissue contributions to sex and race: Differences in tooth crown size of deciduous molars. American Journal of Physical Anthropology, 115, 223–237. https://doi.org/ 10.1002/ajpa.1077 Hartl, D., & Jones, E. (2010). Genetics: Principles and analysis. London: Jones and Bartlett Publishers. Hassett, B. (2011). Technical note: Estimating sex using cervical canine odontometrics: a test using a known sex sample. American Journal of Physical Anthropology, 146, 486–489. https://doi.org/10.1002/ ajpa.21584 Hillson, S. (1986). Teeth. Cambridge manuals in archaeology. Cambridge University Press: Cambridge. Hillson, S. (1996). Dental anthropology. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9781139170697 Hillson, S., FitzGerald, C., & Flinn, H. (2005). Alternative dental measurements: Proposals and relationships with other measurements. American Journal of Physical Anthropology, 126, 413–426. https://doi. org/10.1002/ajpa.10430 Holcomb, S., & Konigsberg, L. (1995). Statistical study of sexual dimorphism in the human fetal sciatic notch. American Journal of Physical Anthropology, 97, 113–125. https://doi.org/10.1002/ajpa.1330970204 Hoppa, R., & Vaupel, J. (2002). Paleodemography. Age distributions from skeletal samples. Cambridge: Cambridge University Press. https://doi. org/10.1017/CBO9780511542428 Irurita Olivares, J., & Alemán Aguilera, I. (2016). Validation of the sex estimation method elaborated by Schutkowski in the Granada Osteological Collection of identified infant and young children: analysis of the controversy between the different ways of analyzing and interpreting the results. International Journal of Legal Medicine, 130, 1623–1632. https://doi.org/10.1007/s00414‐016‐1354‐z Isçan, S., & Kedici, P. (2003). Sexual variation in buccolingual dimensions in Turkish dentition. Forensic Science International, 137, 160–164. https:// doi.org/10.1016/S0379‐0738(03)00349‐9 Karaman, F. (2006). Use of diagonal teeth measurements in predicting gender in a Turkish population. Journal of Forensic Science, 51(3), 630–635. https://doi.org/10.1111/j.1556‐4029.2006.00133.x Khamis, M., Taylor, J., Malik, S., & Townsend, G. (2014). Odontometric sex variation in Malaysians with application to sex prediction. Forensic Science International, 234, 183.e1–183.e7. Kieser, J. (2008). Human adult Odontometrics. The study of variation in adult tooth size. Cambridge University Press: Cambridge. Kieser, J., Groeneveld, H., & Preston, C. (1985). An odontometric analysis of the Lengua Indians dentition. Human Biology, 57, 611–620. Koch, K. (2007). Introduction to Bayesian statistics. Berlin: Springer. Kondo, S., & Townsend, G. (2004). Sexual dimorphism in crown units of mandibular deciduous and permanent molars in Australian Aborigines. Homo‐ Journal of Comparative Human Biology, 55, 53–64. https://doi. org/10.1016/j.jchb.2003.10.001 12 Ling, J., & Wong, R. (2007). Tooth dimensions in Southern Chinese. Homo‐ Journal of Comparative Human Biology, 58, 67–73. https://doi.org/ 10.1016/j.jchb.2006.08.003 Loth, S., & Henneberg, M. (2001). Sexually dimorphic mandibular morphology in the first few years of life. American Journal of Physical Anthropology, 115, 179–186. https://doi.org/10.1002/ajpa.1067 Lovejoy, C., Meindl, R., Pryzbeck, T., & Mensforth, R. (1985). Chronological metamorphosis of the auricular surface of the ilium: A new method for the determination of adult skeletal age at death. American Journal of Physical Anthropology, 68, 15–28. https://doi.org/10.1002/ ajpa.1330680103 Luna, L. (2008). Estructura Demográfica, Estilo de Vida y Relaciones Biológicas de Cazadores‐recolectores en un Ambiente de Desierto. Sitio Chenque I (Parque Nacional Lihué Calel, Provincia de La Pampa). BAR International Series, vol. 1886. Archaeopress: Oxford. Luna, L. (2010). Alternative methodological procedures in sex determination of commingled and fragmentary human remains: An example from Argentine Pampean Region. In K. Weiss (Ed.), Trends in physical anthropology (pp. 1–48). New York: Nova Science Publishers. Luna, L. (2012). Validación de métodos para la generación de perfiles de mortalidad a través de la dentición. Su Importancia Para la caracterización paleodemográfica. Revista Argentina de Antropología Biológica, 14, 33–51. Luna, L. (2015). Interpretative potential of dental metrics for biodistance analyses in hunter‐gatherers from Central Argentina. A theoretical‐ methodological approach. Homo‐ Journal of Comparative Human Biology, 66, 432–447. https://doi.org/10.1016/j.jchb.2015.03.001 Luna, L. (2016). Some achievements and challenges of dental anthropology. ARC Journal of Dental Science, 1(1), 5–9. LUNA Okazaki, K. (2005). Sex assessment of subadult skeletons based on tooth crown measurements: an examination on the interpopulational variation of sex differences and an application to excavated skeletons. Anthropological Science (Japanese Series), 113, 139–159. https://doi. org/10.1537/asj.113.139 Peng, C., Lee, K., & Ingersoll, G. (2002). An introduction to logistic regression. Analysis and reporting. The Journal of Educational Research, 96, 3–14. https://doi.org/10.1080/00220670209598786 Pereira, C., Bernardo, M., Pestana, D., Santos, J. C., & Mendonça, M. C. (2010). Contribution of teeth in human forensic identification—Discriminant function sexing odontometrical techniques in Portuguese population. Journal of Forensic and Legal Medicine, 17(2), 105–110. https://doi.org/10.1016/j.jflm.2009.09.001 Pettenati‐Soubayroux, I., Signoli, M., & Dutour, O. (2002). Sexual dimorphism in teeth: discriminatory effectiveness of permanent lower canine size observed in a XVIIIth century osteological series. Forensic Science International, 126, 227–232. https://doi.org/10.1016/S0379‐ 0738(02)00080‐4 Phenice, T. (1969). A newly developed visual method of sexing in the os pubis. American Journal of Physical Anthropology, 30, 297–301. https://doi.org/10.1002/ajpa.1330300214 Potter, R., Alcazaren, A., Herbosa, F., & Tomaneng, J. (1981). Dimensional characteristics of the Filipino dentition. American Journal of Physical Anthropology, 55, 33–42. https://doi.org/10.1002/ajpa.1330550106 Potter, R., Rice, J., Dahlberg, A., & Dahlberg, T. (1983). Dental size traits within families: path analysis for first molar and lateral incisor. American Journal of Physical Anthropology, 61, 283–289. https://doi.org/ 10.1002/ajpa.1330610302 Reid, D., & Dean, M. (2006). Variation in modern human enamel formation times. Journal of Human Evolution, 50, 329–346. https://doi.org/ 10.1016/j.jhevol.2005.09.003 Luna, L., & Aranda, C. (2010). Asociación entre cantidad de indicadores dentales de estrés metabólico y edad de muerte en el Sitio Chenque I: su variación por sexo y patrones de inhumación. In M. Berón, L. Luna, M. Bonomo, C. Montalvo, C. Aranda, & M. Carrera Aizpitarte (Eds.), Mamül Mapu: Pasado y Presente desde la Arqueología Pampeana (pp. 211–226). Ayacucho, Buenos Aires, Argentina: Editorial Libros del Espinillo. Roberts, C., & Manchester, K. (1999). The archaeology of disease. New York: Cornell University Press. Luna, L., & Flensborg, G. (2017). Determinación del sexo a través de la métrica dental en cazadores recolectores de la transición pampeano‐ patagónica oriental. Revista del Museo de Antropología de Córdoba, 10(1), 53–60. https://doi.org/10.31048/1852.4826.v10.n1.15309 Rosing, F. (1983). Sexing immature human skeletons. Journal of Human Evolution, 12, 149–155. Manzi, G., Santandrea, E., & Passarello, P. (1997). Dental size and shape in the Roman Imperial Age: Two samples from the area of Rome. American Journal of Physical Anthropology, 102, 469–479. https://doi. org/10.1002/(SICI)1096‐8644(199704)102:4<469::AID‐AJPA4>3.0. CO;2‐N Mayhall, J. (2000). Dental morphology: Techniques and strategies. In M. Katzemberg, & S. Saunders (Eds.), Biological anthropology of the human skeleton (pp. 103–134). New York: Wiley‐Liss. Mitsea, A., Moraitis, K., Leon, G., Nicopoulou‐Karayianni, K., & Spiliopoulou, C. (2014). Sex determination by tooth size in a sample of Greek population. Homo‐ Journal of Comparative Human Biology, 65, 322–329. https://doi.org/10.1016/j.jchb.2014.05.002 Molleson, T., Cruse, K., & Mays, S. (1998). Some sexually dimorphic features of the human juvenile skull and their value in sex determination in immature skeletal remains. Journal of Archaeological Science, 25, 719–728. https://doi.org/10.1006/jasc.1997.0215 Rocha, M. (1995). Les collections ostéologiques humaines identifiées du Musée Anthropologique de l' Université de Coimbra. Antropologia Portuguesa, 13, 7–38. Rosing, F., Graw, M., Marré, B., Ritz‐Timme, S., Rothschild, M., Rotzscher, K., … Geserick, G. (2007). Recommendations for the forensic diagnosis of sex and age from skeletons. Homo‐ Journal of Comparative Human Biology, 58, 75–89. https://doi.org/10.1016/j.jchb.2005.07.002 Santos, AL. (2000). A skeletal picture of tuberculosis. Macroscopic, radiological, biomolecular and historical evidence from the Coimbra Identified Skeletal Collection (Unpublished Phd. Thesis). University of Coimbra, Coimbra, Portugal. Saunders, S., Chan, A., Kahlon, B., & Kluge, H. (2007). Sexual dimorphism of the dental tissues in human permanent mandibular canines and third premolars. American Journal of Physical Anthropology, 133, 735–740. https://doi.org/10.1002/ajpa.20553 Scheuer, L. (2002). A blind test of mandibular morphology for sexing mandibles in the first few years of life. American Journal of Physical Anthropology, 119(2), 189–191. https://doi.org/10.1002/ajpa.10098 Scheuer, L., & Black, S. (2000). Developmental Juvenile Osteology. London: Academic Press. Moss, M., & Moss‐Salentijn, L. (1977). Analysis of developmental processes possibly related to human dental sexual dimorphism in permanent and deciduous canines. American Journal of Physical Anthropology, 46, 407–414. https://doi.org/10.1002/ajpa.1330460305 Schutkowski, H. (1993). Sex determination of infant and juvenile skeletons: I. Morphognostic features. American Journal of Physical Anthropology, 90, 199–205. https://doi.org/10.1002/ajpa.1330900206 Nagelkerke, N. (1991). A note on a general definition of the coefficient of determination. Biometrika, 78, 691–692. https://doi.org/10.1093/ biomet/78.3.691 Schwartz, G., & Dean, M. (2005). Sexual dimorphism in modern human permanent teeth. American Journal of Physical Anthropology, 128, 312–317. https://doi.org/10.1002/ajpa.20211 13 LUNA Silva, A. (1995). Sex assessment using the calcaneus and talus. Antropologia Portuguesa, 13, 107–119. Thompson, A. (2013). Odontometric determination of sex at Mound 72 Cahokia. American Journal of Physical Anthropology, 151, 408–419. https://doi.org/10.1002/ajpa.22282 Townsend, G., & Brown, T. (1978a). Heritabilities of permanent tooth sizes. American Journal of Physical Anthropology, 49, 497–502. https://doi. org/10.1002/ajpa.1330490409 Townsend, G., & Brown, T. (1978b). Inheritance of tooth size in Australian Aborigines. American Journal of Physical Anthropology, 48, 305–314. https://doi.org/10.1002/ajpa.1330480306 Ubelaker, D. (1989). The estimation of age at death from immature human bone. In M. Isçan (Ed.), Age markers in the human skeleton (pp. 55–70). Springfield, Illinois: Ch. C. Thomas Publisher. Viciano, J., D'Anastasio, R., & Capasso, L. (2015). Odontometric sex estimation on three populations of the Iron Age from Abruzzo region (central‐southern Italy). Archives of Oral Biology, 60, 100–115. https:// doi.org/10.1016/j.archoralbio.2014.09.003 Viciano, J., López‐Lázaro, S., & Alemán, I. (2013). Sex estimation based on deciduous and permanent dentition in a contemporary Spanish population. American Journal of Physical Anthropology, 152, 31–43. https://doi.org/10.1002/ajpa.22324 Vodanovic, M., Demo, Z., Njemirovskij, V., Keros, J., & Brkic, H. (2007). Odontometrics: a useful method for sex determination in an archaeological skeletal population? Journal of Archaeological Science, 34, 905–913. https://doi.org/10.1016/j.jas.2006.09.004 Walker, P. (2005). Greater sciatic notch morphology: Sex, age, and population differences. American Journal of Physical Anthropology, 127, 385–391. https://doi.org/10.1002/ajpa.10422 y'Edinack, G. (1989). Yugoslav Mesolithic dental reduction. American Journal of Physical Anthropology, 78, 17–36. https://doi.org/10.1002/ ajpa.1330780105 Zorba, E., Moraitis, K., Eliopoulos, C., & Spiliopoulou, C. (2012). Sex determination in modern Greeks using diagonal measurements of molar teeth. Forensic Science International, 217, 19–26. https://doi. org/10.1016/j.forsciint.2011.09.020 Zorba, E., Moraitis, K., & Manolis, S. (2011). Sexual dimorphism in permanent teeth of modern Greeks. Forensic Science International, 210, 74–81. https://doi.org/10.1016/j.forsciint.2011.02.001 Zorba, E., Vanna, V., & Moraitis, K. (2014). Sexual dimorphism of root length on a Greek population sample. HOMO ‐ Journal of Comparative Human Biology, 65, 143–154. https://doi.org/10.1016/j.jchb.2013.09.005 How to cite this article: Luna LH. Canine sex estimation and sexual dimorphism in the collection of identified skeletons of the University of Coimbra, with an application in a Roman cemetery from Faro, Portugal. Int J 2019;1–13. https://doi.org/10.1002/oa.2734 Osteoarchaeol.