Coupling geometric morphometrics and machine learning for mandibular sex estimation: testing Late Pleistocene and Late Modern populations

Coupling geometric morphometrics and machine learning for mandibular sex estimation: testing Late Pleistocene and Late Modern populations | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Coupling geometric morphometrics and machine learning for mandibular sex estimation: testing Late Pleistocene and Late Modern populations Ricardo Miguel Godinho, Isabelle Crevecouer, Susana Garcia, Rebecca Whiting, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6389860/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 19 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted 4 You are reading this latest preprint version Abstract Accurate sex estimation is crucial for studying both modern and ancient human populations, yet methods are often limited to well-preserved skeletons. Here, we combine Geometric Morphometrics (GM) and Machine Learning (ML) to assess mandibular sexual dimorphism and classify sex across a wide chronological and geographic range to bracket the potential of this approach. Sixty-seven individuals from the modern, identified Luis Lopes collection (Portugal) and 18 Late Pleistocene individuals from Jebel Sahaba (Sudan) were surface scanned. Anatomical landmark coordinates were extracted and analyzed with GM, and ML models were trained on a subset of the modern sample to predict sex in both the remaining modern individuals and the Late Pleistocene specimens. GM revealed significant sexual dimorphism in all samples, and ML achieved high intrapopulation classification accuracy. However, predictions were less reliable when applied across the temporally and geographically distant Jebel Sahaba population, reflecting interpopulation differences in mandibular size and shape. These results demonstrate that while GM–ML approaches are powerful tools for sex estimation within populations, caution is required when extending models to other populations. Virtual Anthropology Skeletal remains Archaeology Morphology Palaeodemography Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 1 Introduction Sex is one of the most fundamental biological parameters assessed in forensic, biological and palaeoanthropological studies 1 , 2 . In addition to allowing the osteobiographical characterization of individuals, it enables ensuing analysis of, e.g., sex related differences in funerary behaviour 3 – 6 , weaning 7 , diet 4 , 8 – 11 , activity patterns 5 , 12 – 14 , mobility 8 , 13 , and pathology 15 – 17 . Biomolecular methods (i.e., aDNA and proteomics) have been used to establish sex and provide very reliable results 18 – 27 . Moreover, such methods typically require very reduced quantities of bone/teeth, overcoming pervasive preservation issues in archaeological collections that lead to fragmentation and incompleteness of skeletal elements, often precluding reliable morphological based sex-estimation 19 , 21 , 24 , 25 . Yet, biomolecular methods are destructive, require highly specialized (often expensive) laboratory procedures and also depend on the preservation of proteins and/or DNA 20 . Thus, morphological-based sex estimation remains the most common and feasible approach to sex classification of archaeological individuals. Previous morphological-based sex estimation studies have explored sexual dimorphism in most bones of the human skeleton. The most reliable regions for sex estimation are the os coxae and the skull, which typically provide correct classification rates above 90% 1,28–32 (but see Spradley and Jantz 33 regarding the skull). Yet, these bones are often fragmented or incomplete in archaeological and/or forensic contexts. Moreover, the complexities of funerary behaviour often involve destructive procedures (e.g., cremation and reuse of funerary spaces) and/or post-depositional manipulation of the human remains, causing truncation of individuals and/or commingling and fragmentation of bones 34 – 42 . Thus, researchers have often to estimate sex based on individual bones rather than complete skeletons, including post-cranial bones (other than the os coxae ) which typically provide lower correct sex classification rates and so are less reliable in sex assessment 32 , 43 . Sex estimation based on the os coxae and skull is frequently based on scoring along semi-quantitative ordinal scales of specific anatomical regions 44 – 47 . While conventional metric methods are also used in these regions 48 , 49 , visual-based methods readily capture morphological information not easily quantifiable with the former metric approaches 44 . Yet, visual scoring is based on somewhat subjective observer specific assessment, and so some degree of inter-observer error emerges that may lead to conflicting estimations 1 , 45 , 50 . To better represent 3D morphology objectively and quantitatively, and to reduce inter-observer error, Geometric Morphometrics (GM) has been used more recently to investigate sex related morphological differences in the pelvis 51 – 53 and skull 54 – 58 . These approaches have provided very good results, but the use of conventional landmarks (LMs) is limited in capturing 3D morphology. Hence, some studies have also used semi-sliding landmarks to enable dense coverage of the cranium and to provide better morphological representation 59 . Machine learning (ML) has also been recently used in sex estimation. Several types of data have been used to create ML models, including linear measurements from cranial 60 , 61 and post-cranial bones 62 – 67 , and cross-sectional data from long bones 68 . While ML often improves correct sex identification, it has seldom been applied together with GM methods 59 , 69 . Moreover, to the best of our knowledge, no studies have used ML models developed with identified collections and tested sex classification of archaeological specimens with sex already previously estimated based on multiple skeletal regions (including pelvises, crania and mandibles). This is particularly relevant to test the potential and limitations of applying ML models to estimate sex in archaeological specimens and so to enable examination of sex-based differences in past populations. This is the case of mandibles, which are often used in studies about the morphological impact of population history and diet on past populations 70 – 75 . Because they are often found isolated from the remaining skeleton, sex information is typically not estimated due to uncertainty of predictions. Thus, it is vital to enhance mandibular sex estimation reliability to enable further examination about past populations. Here we use 3D GM to capture the mandibular morphology of an identified skeletal collection (Luis Lopes) and examine sex differences. We then use resulting outputs to train ML models, classify the sex of held-out specimens from the same (Luis Lopes) population and quantify the reliability of the ML predictions. Further, we use those models to classify Late Pleistocene mandibles from Jebel Sahaba which have been previously sexed morphologically (based on multiple skeletal regions) to examine the reliability of the GM based ML sex classifications. The selection of such diverse testing samples is deliberate and aims to bracket the reliability of this sex estimation approach. Specifically, the Luis Lopes intra-population testing sample is expected to provide the most reliable expectable results, whereas the Jebel Sahaba testing sample the least reliable expectable results due to its extreme (intra-specific) morphological difference (see details below). 2 Results 2.1 GM based morphological analysis Our results show clear intra-population size differences between male and female mandibles. Within each population, male mandibles are clearly larger than those of females (Fig. 1 ). The archaeological specimens from Jebel Sahaba are, however, larger than those from late Modern Portugal, with no statistically significant differences between females from the former and males from the latter. The statistical analyses show morphological differences between populations and between the sexes. Specifically, the PERMANOVA including the first PCs accounting for ~ 95% of the total variance shows statistically significant differences between all groups (males and females originating from Portugal and Sudan) in form space. In shape space, all groups are significantly different, with the exception of males and females from Jebel Sahaba (Table SI 1). Such differences are apparent in the PCA plotting PC1 and PC2, in which there is a clear distinction between the sexes of both populations in PC1 (despite some overlap) in form, but not in shape space, in which PC1 separates the two populations (Fig. 2 ). Males (which display lower PC1 scores within each population) have more robust mandibles in form space in both samples, with, e.g., more vertical mandibular symphyses, broader and upright rami and wider sigmoid notches compared to females (which display higher PC1 scores within each population). PC1 in shape space, which separates the two populations, shows that the Jebel Sahaba mandibles are much more robust than those from the late modern Luis Lopes specimens. As expected, the space that includes size (form) is strongly correlated with size Table SI 2). In shape space the first (and some ensuing) PC is also tightly correlated with size when including both populations in the analysis. However, when the populations are separated, the first PCs are no longer correlated with size within each of them, suggesting the relationship between some of the shape PCs and size in this space are driven by the size differences between the populations. 2.2 Machine learning Multiple ML models were trained with different datasets to examine which would produce the best results both in shape and form space. Specifically, the models were run using eight different sets of PCs in shape and form space, including those capturing 100% of the total variance (56 PCs for shape and 57 PCs for form), 95% of the total variance (23 PCs for shape and 18 PCs for form), 90% of the total variance (16 PCs for shape and 12 PCs for form), and PCs considered significant with a value below 0.05 (Figure SI 1). Model performance varies depending on the number of PCs included in the analyses. According to our results, models generally perform best when using 90% of the total variance (Table 1 ), while they perform the worst when using the full set of PCs obtained after PCA (Table SI 3). The latter result is likely due to the inclusion of too many residuals in the analyses. Nonetheless, perfect accuracy was achieved only when using 95% of the variance in shape (Table SI 4). With 90% of the total variance, accuracy rates range from 65% to 95% (Table 1 ), though the most common accuracy rate is 90%, with analyses on form variables generally performing better than those on shape. This trend is observed across all analyses (Fig. 3 , Table SI 4 and Table SI 5), suggesting that size differences may be a significant factor in distinguishing male from female individuals in modern Homo sapiens . This is consistent with the size results above, which show significant size differences between males and females in both samples. Table 1 Results provided by ML algorithms based on PCs accounting for over 90% of the total variance in shape and form. Algorithms with accuracy above 90% are highlighted in bold. Model Data Accuracy Kappa AccLower AccUpper Sensitivity Specificity BalAccuracy kNN shape 0.75 0.5 0.509 0.9134 0.7778 0.7273 0.7525 form 0.85 0.7059 0.6211 0.9679 1 0.7273 0.8636 LGR shape 0.8 0.596 0.5634 0.9427 0.7778 0.8182 0.798 form 0.9 0.798 0.683 0.9877 0.8889 0.9091 0.899 DTC5.0 shape 0.65 0.3 0.4078 0.8461 0.6667 0.6364 0.6515 form 0.9 0.802 0.683 0.9877 1 0.8182 0.9091 RF shape 0.85 0.7 0.6211 0.9679 0.8889 0.8182 0.8535 form 0.95 0.9 0.7513 0.9987 1 0.9091 0.9545 GB shape 0.65 0.2857 0.4078 0.8461 0.5556 0.7273 0.6414 form 0.9 0.7980 0.683 0.9877 0.8889 0.9091 0.899 NB shape 0.7 0.4059 0.4572 0.8811 0.7778 0.6364 0.7071 form 0.9 0.798 0.683 0.9877 0.8889 0.9091 0.899 LDA shape 0.9 0.798 0.683 0.9877 0.8889 0.9091 0.899 form 0.9 0.798 0.683 0.9877 0.8889 0.9091 0.899 PLS shape 0.9 0.7938 0.683 0.9877 0.7778 1 0.8889 form 0.9 0.802 0.683 0.9877 1 0.8182 0.9091 SVMl shape 0.85 0.7 0.6211 0.9679 0.8889 0.8182 0.8535 form 0.9 0.7938 0.683 0.9877 0.7778 1 0.8889 SVMr shape 0.9 0.802 0.683 0.9877 1 0.8182 0.9091 form 0.85 0.7 0.6211 0.9679 0.8889 0.8182 0.8535 NNET shape 0.95 0.898 0.7513 0.9987 0.8889 1 0.9444 form 0.95 0.9 0.7513 0.9987 1 0.9091 0.9545 The inter-population reliability of the ML models predictions was tested by contrasting the ML classifications with previous sex estimations of the Late Pleistocene Jebel Sahaba archaeological sample 76 . The results for this archaeological sample do not reach the accuracy levels achieved with the modern testing sample, assuming the skeletal morphology based anthropological classifications are correct. When using 90% of the total variance, the highest agreement between anthropological and ML-based methods is 83.33%, and the lowest is 44.44%, with a typical match rate of 61.11% (Table 2 ). Across the different models, there is a general tendency to overclassify the archaeological sample as male when size is included in the analyses, while classifications based on shape alone tend to be more balanced (Table 2 , Table SI 6 – Table SI 8). Furthermore, correct female classifications tend to be assigned with higher confidence when using shape variables, whereas correct male classifications are more confidently made with form variables, though this trend is much more pronounced for females (Fig. 4 ). Table 2 Classification provided by ML algorithms for the archaeological sample using the set of PCs that account for 90% of the total variance in shape and form space. The number of samples classified consistently by both anthropological methods and ML algorithms is indicated in brackets. Model Data N male using ML methods N female using ML methods N same sex attribution (%) kNN shape 6 (4) 12 (6) 55.56% form 17 (10) 1 (1) 61.11% LGR shape 13 (7) 5 (2) 50% form 15 (9) 3 (2) 61.11% DTC5.0 shape 8 (4) 10 (4) 44.44% form 17 (10) 1 (1) 61.11% RF shape 6 (4) 12 (6) 55.56% form 17 (10) 1 (1) 61.11% GB shape 6 (4) 12 (6) 55.56% form 17 (10) 1 (1) 61.11% NB shape 3 (3) 15 (8) 61.11% form 15 (9) 3 (2) 61.11% LDA shape 9 (7) 9 (6) 72.22% form 15 (9) 3 (2) 61.11% PLS shape 8 (7) 10 (7) 77.78% form 17 (10) 1 (1) 61.11% SVMl shape 15 (9) 3 (2) 61.11% form 15 (9) 3 (2) 61.11% SVMr shape 8 (5) 10 (5) 55.56% form 9 (8) 9 (7) 83.33% NNET shape 9 (7) 9 (6) 72.22% form 17 (10) 1 (1) 61.11% However, classification probabilities do not commonly exceed 90%, even for matching classifications, and high probabilities are equally common among mismatched classifications in shape space, though less frequent in form space (Table SI 9 – Table SI 16). In other analyses using different sets of PCs, match percentages between anthropological and ML methods range from 44.44% to 88.89%. The highest match rate in the study (88.89%) is obtained in form space, where the SVMr on the significant PCs appears to avoid overestimating the number of male individuals in the sample (Table SI 8). Thus, the relationship between model accuracy on the modern sample and its performance on the archaeological sample does not appear to be straightforward. An increase in accuracy on the testing set classifications does not necessarily correspond to a direct increase in classification match for the archaeological sample. Although the accuracy of ML models trained on the modern sample is a statistically significant predictor for classifications in the archaeological sample (F < 0.05), in both shape and form, with increasing accuracy correlating with increases in the outcome variables (Figure SI 2), this association explains only a small fraction (approx. 12.5%) of the variance in the archaeological sample (Table SI 17). 3 Discussion Overall, and consistent with previous studies, our GM results show clear sexual dimorphism in mandibular morphology 77 – 98 , as well as inter-population morphological differences 70 , 71 . Consistent with previous studies using ML for sex classification 60 – 67 , the ML models were very efficient in the sex classification of the late modern identified Luis Lopes intra-population test sample, with average accuracies of 90% (see details above). However, the sex classification of the temporally and geographically distant mandibles of Late Pleistocene Jebel Sahaba was meaningfully less efficient, with average accuracies of 60–63%. Although we cannot exclude the possibility of some skeletally misclassified archaeological individuals impacting our results, these differences are most likely due to meaningful inter-population size and shape differences that resulted in frequent misclassifications (especially of females as males with form space derived data). These results are, however, predictable and consistent with previous studies highlighting inter-population morphological differences and cautioning against the use of inadequate reference samples to classify target specimens 99 – 104 . This is particularly relevant in this study because the individuals from Jebel Sahaba belong to a highly robust population characterized by plesiomorphic traits, extreme dental dimensions and complex crown morphology, as well as robust morphological features, some being related to powerful masticatory apparatus 105 – 109 . This unique phenotype has been interpreted as a consequence of population isolation during the Late Pleistocene in the Nile valley 110 , 111 , which may influence the classification efficiency. The following sections discuss in more detail the GM and ML results, along with the limitations of this study and future research prospects. 3.1 Geometric Morphometrics Our GM results show sexual dimorphism in mandibular morphology, along with inter-population differences. Males have significantly larger mandibles than females in both populations (assessed via centroid size), and shape and form sex differences were also detected. PERMANOVA also shows shape differences between males and females in the late modern identified Portuguese sample, but not in the Late Pleistocene Jebel Sahaba. In form space (which includes size) sex differences were found in both populations by the PERMANOVA. Despite these results, plotting of PC1 and PC2 in shape space shows overlapping of sexes in both populations but separation in form space. This suggests that sex differences found in lower dimensional space are mainly driven by size and that shape differences are found in higher dimensional space and in PCs which account for smaller proportions of morphological variance. This interpretation is supported by regression of PC scores against centroid size. This analysis shows that these morphological variables are (expectably) significantly related in form space in lower dimensions, and significant relationships in shape space are found only in higher dimensions. Thus, most of the sexual dimorphism we found in these samples is due to isometric size differences between sexes. 3.2 Machine Learning ML models were trained (with supervision) using 11 different algorithms, different sets of PCs derived from both shape and form space, and were first assessed with a holdout testing sample (all using the late modern identified Portuguese sample). Overall, models performed best using the PC scores accounting for 90% of the total variance and classified sex more accurately in form than in shape space. Indeed, when using 90% of the total variance, shape-based sex estimation accuracy averaged 81%, whereas form-based sex estimation averaged 90% (see details above). These intra-population accuracies are typically higher than those reported by most studies estimating sex based on mandible morphology. Most studies report accuracies ranging from ~ 60% to ~ 85% 78,79,81–85,90−92,94,97 , with only a small number reporting accuracies of ~ 90% or more 77 , 87 . Further, the use of these methods mitigates inter-observer subjectivity of morphoscopic scoring 45 , 56 , 112 – 115 and automates sex estimation, thus providing potentially less subjective and more reliable classifications. Consistent with the Luis Lopes collection test sample, accuracy of sex classification of the Jebel Sahaba mandibles was higher using form than shape space derived data. However, the difference in the performance of shape and form-based models using the archaeological specimens was small. Further, the accuracy of sex classification was also much lower, averaging only 60.10% in shape and 63.13% in form space derived models (see details above). Notwithstanding, some models provided accuracies as high as 83.33% in form (SVMr) and 77.78% in shape space (PLS). The generally low accuracies in the latter space result from generally balanced misclassifications in both sexes. In contrast, females are more frequently misclassified as males in form space. Mandibles in the Jebel Sahaba sample (which have been described previously as highly robust; see above) are significantly larger than in the identified Luis Lopes collection, with females from the former presenting comparable size to the males from the latter. Thus, these misclassifications in form space are likely driven by inter-population size differences. Consistent with previous studies, these results highlight how inter-population morphological differences impact sex estimation and may lead to potentially biased results when inadequate reference samples or methods are chosen to classify biologically distant target samples 44 , 99 – 104 , 116 , 117 . Such differences may arise due to differences in, e.g., population history 118 – 121 , climate 119 , 120 , the mechanical demands of the masticatory system 70 , 71 , 74 , 75 , 122 and nutrition 123 . These may lead to overall interpopulation differences in size and/or shape that, in turn, impact the patterns of sexual dimorphism. Indeed, the expression of sexual dimorphism varies across populations, with contrasting patterns of robusticity in both males and females potentially leading to incorrect sex estimations when inadequate reference samples are used or estimation methods are not adjusted accordingly 100 , 117 . This has been hypothesized to drive the over-representation of one sex over the other in the examination of, e.g., sex ratios in past populations 124 , 125 . 3.3 Limitations Overall, this study shows that the clear sexual dimorphism in mandibular morphology can be used to estimate sex reliably within populations. However, sex classification of specimens from other populations is more challenging with meaningfully lower accuracies. This may result from several limitations of this study that will be tackled in future studies. One of the limitations is that this study only includes one reference identified population from late modern Portugal. To classify individuals from other geographies and/or chronologies it would be very relevant to include other reference populations and so account for inter-population morphological differences. This would be particularly relevant in target samples as morphologically distinct as the Late Pleistocene Jebel Sahaba, which has been previously described as likely isolated from other populations, being particularly robust and with very large teeth (see above). Moreover, it would also be potentially relevant to increase sample size of the reference population(s) to provide a more comprehensive depiction of intra-population and sex specific morphological variance. Conversely, this study only uses two testing samples: (i) the intra-population holdout Luis Lopes and the (ii) inter-population Late Pleistocene Jebel Sahaba samples. This choice was deliberate to enable bracketing the reliability of this sex estimation approach by using (held out) specimens from the same population used to train the ML models and an extremely morphologically distinct archaeological testing sample. This results in lower reliability results in the latter sample. Notwithstanding, we predict that ensuing studies using archaeological testing samples biologically closer to the reference sample (e.g., medieval or modern age samples originating from Portugal) will result in better results than those obtained for Jebel Sahaba. Despite the very encouraging intra-population results using the LM dataset adopted in this study, no semi LMs were used. The use of the latter would provide a more detailed morphological representation of the specimens and of specific anatomical regions known to show significant sexual dimorphism (e.g., chin, gonial angle, posterior ramus). This may enable better sex classification accuracies. However, the application of such LMs to archaeological specimens will be challenging frequently. This is because archaeological specimens are often fragmented, precluding the use of dense landmarking protocols. The sex of the specimens from the Late Pleistocene Jebel Sahaba sample was previously estimated using standard multi-factorial anthropological methods, including pelvis and skull-based sex estimation 76 , and this classification was used as reference to assess the reliability of the ML predictions. Although pelvis and skull-based sex estimation typically provides very reliable results, we cannot exclude the possibility that some of the more incomplete individuals may have been misclassified also due to the inevitable absence of reference populations similar to the Late Pleistocene Jebel Sahaba. Thus, the use of modern reference data may drive misclassification in targeting biologically and morphologically distinct populations 125 . If this was the case, this would impact our results. To overcome this limitation, future studies will include archaeological samples for which sex is estimated independently using biomolecular methods (i.e., aDNA or palaeoproteomics). This powerful approach will allow sex determination of the archaeological individuals and ensuing creation of archaeological reference samples. These will account for geographical and temporal differences and enable the development of archaeological population specific morphological sexing methods. 4 Materials and methods This study is based on 85 adult mandibles (Table 3 ). Sixty-seven originate from the Lisbon Luís Lopes identified skeletal collection 126 and the remaining 18 from the Jebel Sahaba Late Pleistocene sample 76 . The selected specimens were surface scanned to enable GM based morphological analysis. To that end, landmark coordinates were extracted from the generated meshes for ensuing use in standard GM analysis. GM outputs were then used in supervised ML to create models that enable the mandibular-based prediction of sex in the archaeological Jebel Sahaba sample. Table 3 Sample composition of the individuals selected from the collections used in this study. Female Male Total Luis Lopes (Portugal) 30 37 67 Jebel Sahaba (Sudan) 8 10 18 Total 38 47 85 4.1 Specimen selection Only individuals no younger than ~ 18 years of estimated age were used in this study. This is because mandibular morphology diverges between males and females during puberty, and so younger individuals do not present meaningful sex-related morphological differences 127 . Further, growth and development induce major morphological changes that are not of interest in this study and that would obscure the sex-related morphological differences 127 . Age assessment of the individuals (i.e., non-adult vs. adult) from the identified collection was based on the records of the individuals and validated via scoring of dental development and eruption sequences 128 . This was particularly suitable as completion of dental development plays a role in the final shape and size of the mandible. The sex identified on the individual records of the Luis Lopes collection (curated by the co-author SG) was used for the GM analysis of sexual morphological differences and in the validation of the ML predictions. The age at death (i.e., non-adult vs. adult) of the archaeological individuals from Jebel Sahaba was estimated via direct observation of dental eruption sequences 128 . The sex of these individuals was previously estimated by Crevecoeur, et al. 76 based on the observation of the hip bones (Bruzek 129 , Murail et al. 130 and Bruzek et al. 31 ) and the skull (Buikstra and Ubelaker 46 ), with estimation reliability reported therein. These estimations by Crevecoeur, et al. 76 were used on the GM based examination of sex related morphological differences and on the ML predictions of sex. Selection was also restricted to specimens that were fully, or almost fully, complete for the landmarking protocol used in this study. This option was favoured because estimation of the original location of the anatomical LMs introduces some degree of error and, so, noise to the analysis (see details below). Thus, 53/67 (79.1%) mandibles presented no missing LMs, and only the remaining 14/67 (20.9%) presented one (9/67; 13.4%) or two (5/67; 7.5%) missing LMs in the Luís Lopes identified sample (Table SI 18). In the Late Pleistocene archaeological sample from current Sudan, 4/18 (22,2%) mandibles were complete, 5/18 (27,8%) had one missing LM, 1/18 (5,6%) 2, 5/18 (27,8%) 3, 2/18 (11,1%) 4 and 1/18 (5,6%) 5 missing LMs (Table SI 18). This strict selection criteria resulted in the exclusion of several individuals due to fragmentation or oral pathologies impacting morphology (e.g., extensive ante-mortem tooth loss). 4.2 Digitization and GM based morphological analysis After selection, specimens were digitised using an Einscan Pro 2X Plus structured light surface scanner. Points clouds were converted into meshes using the EXScan Pro software, which were then used to collect coordinates from a set of 21 anatomical landmarks per specimen (Fig. 5 and Table SI 19). The LM coordinates were collected in the open-source 3D Slicer software 131 . The coordinates were then imported into R and the packages Geomorph 132 and Morpho 133 , were used for ensuing reconstruction of incomplete specimens and GM analysis. The original location of missing anatomical landmarks was estimated using the estimate.missing Thin Plate Splines (TPS) based function of Geomorph. TPS based reconstruction provides reliable predictions of the original morphology but, nonetheless, introduces some errors. The magnitude of the errors relates to which and how many LMs are being reconstructed 134 . While we restricted the maximum number of missing LMs to 5, most of the selected specimens displayed the complete set of (or lacked up to a maximum of two) LMs (Table SI 18). Previous studies have also shown that significant estimation errors may emerge when inadequate specimens are selected as reference 135 – 137 . Although we found no significant error differences when selecting reference specimens from different modern human populations in another study 70 , we opted for population specific missing data estimation. (i.e., incomplete specimens from the Portuguese sample were reconstructed based on complete specimens from Portugal; incomplete specimens from Jebel Sahaba were reconstructed based on complete specimens from the same population). Two spaces were used in the GM analysis: shape and form. In shape space, scaling removed (isometric) size differences 138 – 140 . In form space, (isometric) size differences were first removed via scaling and were then re-introduced via inclusion of log centroid size 138 , 141 . Differences in size between the sexes were examined using Centroid Size (CS). Principal Component Analysis (PCA) ensued to examine shape and form differences between males and females, which were visualized together with TPS warpings along the relevant principal components (PCs). Shape and form PC scores were also regressed against size (i.e., centroid size) to examine the relationship between these variables for the whole pooled sample and in each individual population. A Kruskal Wallis test, ensued by post-hoc tests, was used to examine hypothetical differences in size (as assessed via centroid size) between sexes and populations. PERMANOVA was performed in Past 142 to test for shape and form statistical differences between males and females using the PC scores of the first PCs explaining ~ 95% of the total variance. 4.3 Machine Learning Principal components scores derived from shape and form analyses were subjected to ML methods. Initially, the late modern Luis Lopes identified sample was split into two sets: training (70%) and testing (30%), to assess model reliability. Analyses were conducted in shape and form space across four rounds: using the full variance, PC scores accounting for 95% of total variance, PC scores accounting for 90%, and PCs identified as significant by permutations with the GraphGMM library 143 . ML models require large sample sizes for effective training, so artificial methods to increase sample size are commonly used in archaeology, paleontology, and anatomy 144 – 146 , as these fields are often limited by access to reference collections, as is the case in the present study. However, no artificial sample-increasing methods were used in this study. Bootstrapping was avoided, as it merely duplicates known data multiple times, potentially leading to overfitting during model training 146 , 147 . Similarly, given that our sample is limited to a specific population, applying generative adversarial networks to generate artificial data within the morphological variance range was deemed unsuitable 146 . Therefore, we proceeded with only the collected sample, with the aim that additional sexed osteological collections from different geographical areas and chronologies will eventually be incorporated into the training set in future studies. Among ML approaches, supervised learning methods were preferred, requiring a pre-classified dataset with built-in learning and self-control mechanisms. To mitigate potential overfitting, self-correcting techniques such as k-fold cross-validation were employed. The original sample was partitioned into 10 sets to generate “submodels” with performance across these submodels assessing overall model efficiency. The sexed modern osteological Luis Lopes collection was used to train the models, which were then applied to classify Late Pleistocene mandibles as male or female. Eleven algorithms were used in this study (), including k-Nearest Neighbour (kNN), Logistic Regression (LGR), Decision Trees (DTC5.0), Random Forest (RF), Gradient Boosting (GB), Naïve Bayes (NB), Linear Discriminant Analysis (LDA), Partial Least Squares (PLS), Linear and Radial Support Vector Machines (SVMl and SVMr), and Neural Networks (NNET). All algorithms were trained using the ‘caret’ 148 and ‘caretEnsemble’ 149 R libraries. Model tuning, essential for classification accuracy, was conducted using hyperparameter grids to test various parameter values, with optimal values selected for each algorithm. The ‘tuneLength’ function in these libraries enabled hyperparameter configuration by generating 20 models per algorithm, with accuracy and Kappa values guiding optimal selection. Differences in prediction performance and classification rates among models were compared considering a set of factors, including kappa, sensitivity, specificity and balanced accuracy values. Kappa statistics account for chance in predictions, with values ranging from − 1 to 1 (kappa > 0.8 signals high predictive power) 150 . Sensitivity and specificity measure opposite rates: true positive versus true negative classifications 150 . These values are balanced through averaging, resulting in a model efficiency score from 0 to 1 151 . Classifications for the archaeological sample were compared across models and against classifications from traditional anthropological methods based on all available skeletal elements of each individual 76 . Additionally, posterior probability values (p) were calculated for each individual to determine group membership reliability, with values exceeding 0.9 typically regarded as strong identifications. Declarations Acknowledgements RM Godinho is funded by Fundação para a Ciência e a Tecnologia (FCT; 2023.10993.TENURE.006). This research was also funded by the FCT R&D research project “ParaFunction” (project reference 2022.07737.PTDC; https://doi.org/10.54499/2022.07737.PTDC). The British Museum for granting access to the Jebel Sahaba collection. The Portuguese Museu Nacional de História Natural e da Ciência (MUHNAC) for granting access to the Luis Lopes collection. J Aramendi is funded by the British Academy (NIF22\220310). The anthropological reassessment of the Jebel Sahaba collection by IC was supported by the International Research Project (IRP) ABASC founded by the CNRS-INEE; the French government in the framework of the University of Bordeaux's IdEx “Investments for the Future” program/GPR “Human Past” and the French National Research Agency (ANR-14-CE31, project BIG DRY) Author contributions All authors contributed substantially and reviewed the manuscript. Data availability statement The data that support this study is available from the corresponding author upon reasonable request. Permission statement Access to the Luís Lopes identified sample was granted by the housing institution, the Portuguese Museu Nacional de História Natural e da Ciência (MUHNAC). Access to the Jebel Sahaba sample was granted by the housing institution, the British Museum. Both access permissions ensued formal access requests to each of these institutions and assessment by the relevant boards. All regulations of each of these institutions were followed. References Krishan, K. et al. A review of sex estimation techniques during examination of skeletal remains in forensic anthropology casework. Forensic Science International 261 , 165.e161-165.e168 (2016). https://doi.org/https://doi.org/10.1016/j.forsciint.2016.02.007 White, T. D., Black, M. T. & Folkens, P. A. Human osteology . (Academic press, 2011). Laffranchi, Z., Beck De Lotto, M. A., Delpino, C., Lösch, S. & Milella, M. Social differentiation and well-being in the Italian Iron Age: exploring the relationship between sex, age, biological stress, and burial complexity among the Picenes of Novilara (8th–7th c. BC). Archaeological and Anthropological Sciences 13 , 182 (2021). https://doi.org/10.1007/s12520-021-01449-3 Dong, Y. et al. Shifting diets and the rise of male-biased inequality on the Central Plains of China during Eastern Zhou. 114 , 932-937 (2017). https://doi.org/10.1073/pnas.1611742114 %J Proceedings of the National Academy of Sciences Masclans, A., Duboscq, S., Achino, K. F., Morell, B. & Gibaja, J. F. Looking for sexual differences during the middle Neolithic in the northeast of the Iberian Peninsula. Journal of Archaeological Science: Reports 26 , 101858 (2019). https://doi.org/https://doi.org/10.1016/j.jasrep.2019.05.023 Rebay-Salisbury, K. et al. Gendered burial practices of early Bronze Age children align with peptide-based sex identification: A case study from Franzhausen I, Austria. Journal of Archaeological Science 139 , 105549 (2022). https://doi.org/https://doi.org/10.1016/j.jas.2022.105549 Fernández-Crespo, T. et al. Multi-isotope evidence for the emergence of cultural alterity in Late Neolithic Europe. Science Advances 6 , eaay2169 (2020). https://doi.org/10.1126/sciadv.aay2169 Rey, L., Rottier, S., Santos, F. & Goude, G. Sex and age-related social organization in the Neolithic: A promising survey from the Paris Basin. Journal of Archaeological Science: Reports 38 , 103092 (2021). https://doi.org/https://doi.org/10.1016/j.jasrep.2021.103092 Rey, L. et al. A multi-isotope analysis of Neolithic human groups in the Yonne valley, Northern France: insights into dietary patterns and social structure. Archaeological and Anthropological Sciences 11 , 5591-5616 (2019). https://doi.org/10.1007/s12520-019-00885-6 Avery, L. C., Brickley, M. B., Findlay, S., Chapelain de Seréville-Niel, C. & Prowse, T. L. Child and adolescent diet in Late Roman Gaul: An investigation of incremental dietary stable isotopes in tooth dentine. International Journal of Osteoarchaeology 31 , 1226-1236 (2021). https://doi.org/https://doi.org/10.1002/oa.3033 Miller, M. J., Dong, Y., Pechenkina, K., Fan, W. & Halcrow, S. E. Raising girls and boys in early China: Stable isotope data reveal sex differences in weaning and childhood diets during the eastern Zhou era. American Journal of Physical Anthropology 172 , 567-585 (2020). https://doi.org/https://doi.org/10.1002/ajpa.24033 Lieverse, A. R., Stock, J. T., Katzenberg, M. A. & Haverkort, C. M. The bioarchaeology of habitual activity and dietary change in the Siberian Middle Holocene. Human bioarchaeology of the transition to agriculture , 263-291 (2011). Ruff, C. B. et al. Gradual decline in mobility with the adoption of food production in Europe. Proceedings of the National Academy of Sciences 112 , 7147 (2015). Ruff, C. Sexual dimorphism in human lower limb bone structure: relationship to subsistence strategy and sexual division of labor. Journal of Human Evolution 16 , 391-416 (1987). https://doi.org/http://dx.doi.org/10.1016/0047-2484(87)90069-8 Bonsall, L. A comparison of female and male oral health in skeletal populations from late Roman Britain: Implications for diet. Archives of Oral Biology 59 , 1279-1300 (2014). https://doi.org/https://doi.org/10.1016/j.archoralbio.2014.07.019 Lukacs, J. R. Sex Differences in Dental Caries Rates With the Origin of Agriculture in South Asia. Current Anthropology 37 , 147-153 (1996). https://doi.org/doi:10.1086/204481 Milner, G. R. & Boldsen, J. L. in A Companion to Paleopathology 268-284 (2011). Lugli, F. et al. Enamel peptides reveal the sex of the Late Antique ‘Lovers of Modena’. Scientific Reports 9 , 13130 (2019). https://doi.org/10.1038/s41598-019-49562-7 Granja, R. et al. Unbalanced sex-ratio in the Neolithic individuals from the Escoural Cave (Montemor-o-Novo, Portugal) revealed by peptide analysis. Scientific Reports 13 , 19902 (2023). https://doi.org/10.1038/s41598-023-47037-4 Buonasera, T. et al. A comparison of proteomic, genomic, and osteological methods of archaeological sex estimation. Scientific Reports 10 , 11897 (2020). https://doi.org/10.1038/s41598-020-68550-w Cintas-Peña, M. et al. Amelogenin peptide analyses reveal female leadership in Copper Age Iberia (c. 2900–2650 BC). Scientific Reports 13 , 9594 (2023). https://doi.org/10.1038/s41598-023-36368-x Froment, C. et al. Analysis of 5000 year-old human teeth using optimized large-scale and targeted proteomics approaches for detection of sex-specific peptides. Journal of Proteomics 211 , 103548 (2020). https://doi.org/https://doi.org/10.1016/j.jprot.2019.103548 Gasparini, A. et al. Biological sex VS. Archaeological Gender: Enamel peptide analysis of the horsemen of the Early Middle age necropolises of Campochiaro (Molise, Italy). Journal of Archaeological Science: Reports 41 , 103337 (2022). https://doi.org/https://doi.org/10.1016/j.jasrep.2021.103337 Parker, G. J. et al. Sex estimation using sexually dimorphic amelogenin protein fragments in human enamel. Journal of Archaeological Science (2018). https://doi.org/https://doi.org/10.1016/j.jas.2018.08.011 Skoglund, P., Storå, J., Götherström, A. & Jakobsson, M. Accurate sex identification of ancient human remains using DNA shotgun sequencing. Journal of Archaeological Science 40 , 4477-4482 (2013). https://doi.org/https://doi.org/10.1016/j.jas.2013.07.004 Loreille, O. et al. Biological Sexing of a 4000-Year-Old Egyptian Mummy Head to Assess the Potential of Nuclear DNA Recovery from the Most Damaged and Limited Forensic Specimens. Genes 9 , 135 (2018). Díaz-Zorita Bonilla, M. et al. Female sex bias in Iberian megalithic societies through bioarchaeology, aDNA and proteomics. Scientific Reports 14 , 21818 (2024). https://doi.org/10.1038/s41598-024-72148-x Christensen, A. M., Passalacqua, N. V. & Bartelink, E. J. Forensic anthropology: current methods and practice . (Academic Press, 2019). Luo, Y.-C. Sex determination from the pubis by discriminant function analysis. Forensic Science International 74 , 89-98 (1995). https://doi.org/https://doi.org/10.1016/0379-0738(95)01739-6 Meindl, R. S., Lovejoy, C. O., Mensforth, R. P. & Carlos, L. D. Accuracy and direction of error in the sexing of the skeleton: Implications for paleodemography. American Journal of Physical Anthropology 68 , 79-85 (1985). https://doi.org/https://doi.org/10.1002/ajpa.1330680108 Brůžek, J., Santos, F., Dutailly, B., Murail, P. & Cunha, E. Validation and reliability of the sex estimation of the human os coxae using freely available DSP2 software for bioarchaeology and forensic anthropology. American Journal of Physical Anthropology 164 , 440-449 (2017). https://doi.org/https://doi.org/10.1002/ajpa.23282 Bruzek, J. & Murail, P. in Forensic Anthropology and Medicine: Complementary Sciences From Recovery to Cause of Death (eds Aurore Schmitt, Eugénia Cunha, & João Pinheiro) 225-242 (Humana Press, 2006). Spradley, M. K. & Jantz, R. L. Sex Estimation in Forensic Anthropology: Skull Versus Postcranial Elements. Journal of Forensic Sciences 56 , 289-296 (2011). https://doi.org/10.1111/j.1556-4029.2010.01635.x Godinho, R. M., Gonçalves, D. & Valera, A. C. The preburning condition of Chalcolithic cremated human remains from the Perdigões enclosures (Portugal). International Journal of Osteoarchaeology 29 , 706-717 (2019). https://doi.org/10.1002/oa.2768 Valera, A. C., Evangelista, L. & Godinho, R. M. in Megalithic Societies: Old Questions, New Narratives (eds G. Higginbottom et al. ) (Archaeopress, In press). Valera, A. C., Silva, A. M., Leandro, I., Godinho, R. M. & Evangelista, L. S. in Os Perdigões Neolíticos. Génese e desenvolvimento (de meados do 4º aos inícios do 3º milénio AC) (ed António Carlos Valera) 155-162 (Núcleo de Investigação Arqueológica (NIA), 2018). Evangelista, L. S. & Godinho, R. M. in O Sepulcro 4 dos Perdigões. Um Tholos da Segunda Metade do 3º Milénio A.C. (ed António Carlos Valera) 57-102 (Núcleo de Investigação Arqueológica (NIA) & Era Arqueologia, S.A., 2020). Sianto, L. et al. Evidence of contact between New and Old World: paleoparasitological and food remains study in the Tagus river population of Sarilhos Grandes (Montijo, Portugal). Archaeological and Anthropological Sciences 10 , 75-81 (2018). https://doi.org/10.1007/s12520-016-0337-9 Filipe, V., Godinho, R. M., Granja, R., Ribeiro, A. & Valera, A. Bronze age funerary spaces in Outeiro Alto 2 (Brinches, Serpa, Portugal): the hypogea cemetery. Zephyrus 71 , 107-129 (2013). Valera, A. et al. in 4º Colóquio de Arqueologia do Alqueva - O Plano de Rega (2002-2010). (eds António Carlos Silva, Frederico Tátá Regala, & Miguel Martinho) 55-73 (EDIA). Godinho, R. Deposições funerárias em fossa nos Perdigões: dados antropológicos do sector I. Apontamentos de Arqueologia e Património 3 , 29-34 (2008). Miguel, L. & Godinho, R. M. Notícia do sítio arqueológico do Monte das Covas 3 (Beja). Apontamentos de Arqueologia e Património 4 , 23-24 (2009). Introna, F., Di Vella, G. & Campobasso, C. P. Sex determination by discriminant analysis of patella measurements. Forensic Science International 95 , 39-45 (1998). https://doi.org/https://doi.org/10.1016/S0379-0738(98)00080-2 Walker, P. L. Sexing skulls using discriminant function analysis of visually assessed traits. American Journal of Physical Anthropology 136 , 39-50 (2008). https://doi.org/10.1002/ajpa.20776 Walrath, D. E., Turner, P. & Bruzek, J. Reliability test of the visual assessment of cranial traits for sex determination. American Journal Of Physical Anthropology 125 , 132-137 (2004). https://doi.org/10.1002/ajpa.10373 Buikstra, J. & Ubelaker, D. Standards for Data Collection from Human Skeletal Remains: Proceedings of a Seminar at the Field Museum of Natural History . (Arkansas Archeological Survey, 1994). Ferembach, D., Schwidetzky, I. & Stloukal, M. Recommendations for age and sex diagnoses of skeletons. Journal of Human Evolution 9 , 517-549 (1980). https://doi.org/https://doi.org/10.1016/0047-2484(80)90061-5 Santos, F., Guyomarc’h, P. & Bruzek, J. Statistical sex determination from craniometrics: Comparison of linear discriminant analysis, logistic regression, and support vector machines. Forensic Science International 245 , 204.e201-204.e208 (2014). https://doi.org/https://doi.org/10.1016/j.forsciint.2014.10.010 Techataweewan, N. et al. Metric sexual dimorphism of the skull in Thailand. Forensic Science International: Reports 4 , 100236 (2021). https://doi.org/https://doi.org/10.1016/j.fsir.2021.100236 Colman, K. L. et al. The accuracy of 3D virtual bone models of the pelvis for morphological sex estimation. International Journal of Legal Medicine 133 , 1853-1860 (2019). https://doi.org/10.1007/s00414-019-02002-7 Steyn, M., Pretorius, E. & Hutten, L. Geometric morphometric analysis of the greater sciatic notch in South Africans. HOMO - Journal of Comparative Human Biology 54 , 197-206 (2004). https://doi.org/http://dx.doi.org/10.1078/0018-442X-00076 Bytheway, J. A. & Ross, A. H. A Geometric Morphometric Approach to Sex Determination of the Human Adult Os Coxa. Journal of Forensic Sciences 55 , 859-864 (2010). https://doi.org/https://doi.org/10.1111/j.1556-4029.2010.01374.x Bilfeld, M. F. et al. Human Coxal Bone Sexual Dimorphism and Multislice Computed Tomography: Geometric Morphometric Analysis of 65 Adults. Journal of Forensic Sciences 57 , 578-588 (2012). https://doi.org/https://doi.org/10.1111/j.1556-4029.2011.02009.x Kimmerle, E. H., Ross, A. & Slice, D. Sexual Dimorphism in America: Geometric Morphometric Analysis of the Craniofacial Region. Journal of Forensic Sciences 53 , 54-57 (2008). https://doi.org/https://doi.org/10.1111/j.1556-4029.2007.00627.x Rosas, A. & Bastir, M. Thin-plate spline analysis of allometry and sexual dimorphism in the human craniofacial complex. American Journal of Physical Anthropology 117 , 236-245 (2002). https://doi.org/10.1002/ajpa.10023 Gonzalez, P. N., Bernal, V. & Perez, S. I. Analysis of sexual dimorphism of craniofacial traits using geometric morphometric techniques. International Journal of Osteoarchaeology 21 , 82-91 (2011). https://doi.org/10.1002/oa.1109 Del Bove, A., Profico, A., Riga, A., Bucchi, A. & Lorenzo, C. A geometric morphometric approach to the study of sexual dimorphism in the modern human frontal bone. American Journal of Physical Anthropology 173 , 643-654 (2020). https://doi.org/https://doi.org/10.1002/ajpa.24154 Milella, M., Franklin, D., Belcastro, M. G. & Cardini, A. Sexual differences in human cranial morphology: Is one sex more variable or one region more dimorphic? The Anatomical Record 304 , 2789-2810 (2021). https://doi.org/https://doi.org/10.1002/ar.24626 Del Bove, A. et al. Mapping sexual dimorphism signal in the human cranium. Scientific Reports 13 , 16847 (2023). https://doi.org/10.1038/s41598-023-43007-y Bertsatos, A., Chovalopoulou, M.-E., Brůžek, J. & Bejdová, Š. Advanced procedures for skull sex estimation using sexually dimorphic morphometric features. International Journal of Legal Medicine (2020). https://doi.org/10.1007/s00414-020-02334-9 Toneva, D. et al. Machine learning approaches for sex estimation using cranial measurements. International Journal of Legal Medicine 135 , 951-966 (2021). https://doi.org/10.1007/s00414-020-02460-4 d’Oliveira Coelho, J. & Curate, F. CADOES: An interactive machine-learning approach for sex estimation with the pelvis. Forensic Science International , 109873 (2019). https://doi.org/https://doi.org/10.1016/j.forsciint.2019.109873 Constantinou, C. & Nikita, E. SexEst: An open access web application for metric skeletal sex estimation. n/a https://doi.org/https://doi.org/10.1002/oa.3109 Mahfouz, M. et al. Patella sex determination by 3D statistical shape models and nonlinear classifiers. Forensic Science International 173 , 161-170 (2007). https://doi.org/https://doi.org/10.1016/j.forsciint.2007.02.024 Curate, F. et al. A method for sex estimation using the proximal femur. Forensic Science International 266 , 579.e571-579.e577 (2016). https://doi.org/http://dx.doi.org/10.1016/j.forsciint.2016.06.011 Curate, F. et al. Sex determination from the femur in Portuguese populations with classical and machine-learning classifiers. Journal of Forensic and Legal Medicine 52 , 75-81 (2017). https://doi.org/https://doi.org/10.1016/j.jflm.2017.08.011 du Jardin, P., Ponsaillé, J., Alunni-Perret, V. & Quatrehomme, G. A comparison between neural network and other metric methods to determine sex from the upper femur in a modern French population. Forensic Science International 192 , 127.e121-127.e126 (2009). https://doi.org/https://doi.org/10.1016/j.forsciint.2009.07.014 Bertsatos, A., Garoufi, N. & Chovalopoulou, M.-E. Advancements in sex estimation using the diaphyseal cross-sectional geometric properties of the lower and upper limbs. International Journal of Legal Medicine (2020). https://doi.org/10.1007/s00414-020-02437-3 Imaizumi, K. et al. Development of a sex estimation method for skulls using machine learning on three-dimensional shapes of skulls and skull parts. Forensic Imaging 22 , 200393 (2020). https://doi.org/https://doi.org/10.1016/j.fri.2020.200393 Godinho, R. M., Umbelino, C. & Gonçalves, C. Mesolithic and Chalcolithic Mandibular Morphology: Using Geometric Morphometrics to Reconstruct Incomplete Specimens and Analyse Morphology. Open Archaeology 8 , 536-549 (2022). https://doi.org/doi:10.1515/opar-2022-0247 Godinho, R. M. et al. Mandibular morphology and the Mesolithic–Neolithic transition in Westernmost Iberia. Scientific Reports 13 , 16648 (2023). https://doi.org/10.1038/s41598-023-42846-z Katz, D. C., Grote, M. N. & Weaver, T. D. Changes in human skull morphology across the agricultural transition are consistent with softer diets in preindustrial farming groups. Proceedings of the National Academy of Sciences (2017). https://doi.org/10.1073/pnas.1702586114 von Cramon-Taubadel, N. Global human mandibular variation reflects differences in agricultural and hunter-gatherer subsistence strategies. Proceedings of the National Academy of Sciences 108 , 19546-19551 (2011). https://doi.org/10.1073/pnas.1113050108 Galland, M., Van Gerven, D. P., Von Cramon-Taubadel, N. & Pinhasi, R. 11,000 years of craniofacial and mandibular variation in Lower Nubia. Scientific Reports 6 , 31040 (2016). https://doi.org/10.1038/srep31040 http://www.nature.com/articles/srep31040#supplementary-information Pokhojaev, A., Avni, H., Sella-Tunis, T., Sarig, R. & May, H. Changes in human mandibular shape during the Terminal Pleistocene-Holocene Levant. Scientific Reports 9 , 8799 (2019). https://doi.org/10.1038/s41598-019-45279-9 Crevecoeur, I., Dias-Meirinho, M.-H., Zazzo, A., Antoine, D. & Bon, F. New insights on interpersonal violence in the Late Pleistocene based on the Nile valley cemetery of Jebel Sahaba. Scientific Reports 11 , 9991 (2021). https://doi.org/10.1038/s41598-021-89386-y Balci, Y., Yavuz, M. F. & Cağdir, S. Predictive accuracy of sexing the mandible by ramus flexure. HOMO 55 , 229-237 (2005). https://doi.org/https://doi.org/10.1016/j.jchb.2004.07.006 Donnelly, S. M., Hens, S. M., Rogers, N. L. & Schneider, K. L. Technical note: A blind test of mandibular ramus flexure as a morphologic indicator of sexual dimorphism in the human skeleton. American Journal of Physical Anthropology 107 , 363-366 (1998). https://doi.org/https://doi.org/10.1002/(SICI)1096-8644(199811)107:33.0.CO;2-Y Franklin, D., O'Higgins, P. & Oxnard, C. E. Sexual dimorphism in the mandible of indigenous South Africans: A geometric morphometric approach. South African Journal of Science 104 , 101-106 (2008). Franklin, D., O’Higgins, P., Oxnard, C. E. & Dadour, I. Discriminant function sexing of the mandible of Indigenous South Africans. Forensic Science International 179 , 84.e81-84.e85 (2008). https://doi.org/https://doi.org/10.1016/j.forsciint.2008.03.014 Giles, E. Sex determination by discriminant function analysis of the mandible. American Journal of Physical Anthropology 22 , 129-135 (1964). https://doi.org/https://doi.org/10.1002/ajpa.1330220212 Hanihara, K. Sex Diagnosis of Japanese Skulls and Scapulae by. Journal of the Anthropological Society of Nippon 67 , 191-197 (1959). https://doi.org/10.1537/ase1911.67.191 Haun, S. J. Brief communication: A study of the predictive accuracy of mandibular ramus flexure as a singular morphologic indicator of sex in an archaeological sample. American Journal of Physical Anthropology 111 , 429-432 (2000). https://doi.org/10.1002/(sici)1096-8644(200003)111:33.0.co;2-1 Hill, C. A. Technical note: evaluating mandibular ramus flexure as a morphological indicator of sex. American Journal of Physical Anthropology 111 , 573-577 (2000). https://doi.org/10.1002/(sici)1096-8644(200004)111:43.0.co;2-i Kemkes-Grottenthaler, A., Löbig, F. & Stock, F. Mandibular ramus flexure and gonial eversion as morphologic indicators of sex. HOMO 53 , 97-111 (2002). https://doi.org/https://doi.org/10.1078/0018-442X-00039 Kuha, A., Ackermann, J., Junno, J.-A., Oettlé, A. & Oura, P. Deep learning in sex estimation from photographed human mandible using the Human Osteological Research Collection. Legal Medicine 70 , 102476 (2024). https://doi.org/https://doi.org/10.1016/j.legalmed.2024.102476 Loth, S. R. & Henneberg, M. Mandibular ramus flexure: A new morphologic indicator of sexual dimorphism in the human skeleton. 99 , 473-485 (1996). https://doi.org/10.1002/(sici)1096-8644(199603)99:33.0.Co;2-x Loth, S. R. & Henneberg, M. Mandibular ramus flexure is a good indicator of sexual dimorphism. 105 , 91-92 (1998). https://doi.org/10.1002/(sici)1096-8644(199801)105:13.0.Co;2-g Loth, S. R. & Henneberg, M. Sexually dimorphic mandibular morphology in the first few years of life. American Journal of Physical Anthropology 115 , 179-186 (2001). https://doi.org/10.1002/ajpa.1067 Oettlé, A. C., Pretorius, E. & Steyn, M. Geometric morphometric analysis of mandibular ramus flexure. 128 , 623-629 (2005). https://doi.org/10.1002/ajpa.20207 Oettlé, A. C., Pretorius, E. & Steyn, M. Geometric morphometric analysis of the use of mandibular gonial eversion in sex determination. HOMO 60 , 29-43 (2009). https://doi.org/https://doi.org/10.1016/j.jchb.2007.01.003 Ogawa, Y., Imaizumi, K., Miyasaka, S. & Yoshino, M. Discriminant functions for sex estimation of modern Japanese skulls. Journal of Forensic and Legal Medicine 20 , 234-238 (2013). https://doi.org/https://doi.org/10.1016/j.jflm.2012.09.023 Plotkin, L. I., Bruzzaniti, A. & Pianeta, R. Sexual Dimorphism in the Musculoskeletal System: Sex Hormones and Beyond. Journal of the Endocrine Society 8 (2024). https://doi.org/10.1210/jendso/bvae153 Pretorius, E., Steyn, M. & Scholtz, Y. Investigation into the usability of geometric morphometric analysis in assessment of sexual dimorphism. American Journal of Physical Anthropology 129 , 64-70 (2006). https://doi.org/https://doi.org/10.1002/ajpa.20251 Rmoutilová, R. et al. Classification performance of the Sella-Tunis et al. (2017) sex estimation method in Czech population: different posterior probability threshold approaches. International Journal of Legal Medicine (2024). https://doi.org/10.1007/s00414-024-03241-z Schmittbuhl, M., Le Minor, J.-M., Schaaf, A. & Mangin, P. The human mandible in lateral view: elliptical fourier descriptors of the outline and their morphological analysis. Annals of Anatomy - Anatomischer Anzeiger 184 , 199-207 (2002). https://doi.org/https://doi.org/10.1016/S0940-9602(02)80021-8 Steyn, M. & İşcan, M. Y. Sexual dimorphism in the crania and mandibles of South African whites. Forensic Science International 98 , 9-16 (1998). https://doi.org/https://doi.org/10.1016/S0379-0738(98)00120-0 Toneva, D. H., Nikolova, S. Y., Fileva, N. F. & Zlatareva, D. K. Size and shape of human mandible: Sex differences and influence of age on sex estimation accuracy. Legal Medicine 65 , 102322 (2023). https://doi.org/https://doi.org/10.1016/j.legalmed.2023.102322 Murail, P., Bruzek, J. & Braga, J. A new approach to sexual diagnosis in past populations. Practical adjustments from Van Vark's procedure. International Journal of Osteoarchaeology 9 , 39-53 (1999). Albanese, J., Cardoso, H. F. V. & Saunders, S. R. 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 (2005). https://doi.org/http://dx.doi.org/10.1016/j.jas.2004.08.003 Garcia, S. Is the circumference at the nutrient foramen of the tibia of value to sex determination on human osteological collections? Testing a new method. International Journal of Osteoarchaeology 22 , 361-365 (2012). https://doi.org/https://doi.org/10.1002/oa.1202 Gonçalves, D., Granja, R., Cardoso, F. A. & de Carvalho, A. F. Sample-specific sex estimation in archaeological contexts with commingled human remains: a case study from the Middle Neolithic cave of Bom Santo in Portugal. Journal of Archaeological Science 49 , 185-191 (2014). https://doi.org/http://dx.doi.org/10.1016/j.jas.2014.05.011 İşcan, M. Y., Loth, S. R., King, C. A., Shihai, D. & Yoshino, M. Sexual dimorphism in the humerus: A comparative analysis of Chinese, Japanese and Thais. Forensic Science International 98 , 17-29 (1998). https://doi.org/https://doi.org/10.1016/S0379-0738(98)00119-4 Bidmos, M. A. & Mazengenya, P. Accuracies of discriminant function equations for sex estimation using long bones of upper extremities. International Journal of Legal Medicine 135 , 1095-1102 (2021). https://doi.org/10.1007/s00414-020-02458-y Greene, D. L., Ewing, G. H. & Armelagos, G. J. Dentition of a mesolithic population from Wadi Halfa, Sudan. American Journal of Physical Anthropology 27 , 41-55 (1967). https://doi.org/10.1002/ajpa.1330270107 Irish, J. D. Population continuity vs. discontinuity revisited: Dental affinities among late Paleolithic through Christian-era Nubians. American Journal of Physical Anthropology 128 , 520-535 (2005). https://doi.org/https://doi.org/10.1002/ajpa.20109 Crevecoeur, I., Rougier, H., Grine, F. & Froment, A. Modern human cranial diversity in the Late Pleistocene of Africa and Eurasia: Evidence from Nazlet Khater, Peştera cu Oase, and Hofmeyr. 140 , 347-358 (2009). https://doi.org/https://doi.org/10.1002/ajpa.21080 Martin, N. et al. From hunter-gatherers to food producers: New dental insights into the Nile Valley population history (Late Paleolithic–Neolithic). American Journal of Biological Anthropology 184 , e24948 (2024). https://doi.org/https://doi.org/10.1002/ajpa.24948 Anderson, J. E. in The Prehistory of Nubia Vol. Vol. 2 (ed Wendorf) 996-1040 (Fort Burgwin Research Center and Southern Methodist University Press, 1968). Pagani, L. & Crevecoeur, I. in Modern Human Origins and Dispersal, Words, Bones, Genes, Tools (eds Yonatan Sahle, Hugo Reyes-Centeno, & Christian Bentz) 15-24 (Kerns Verlag (DFG Center for Advanced Studies Series), 2019). Crevecoeur, I., Matu, M., Dias-Meirinho, M.-H., Bayle, P. & Pearson, O. in Du Big dry à l’Holocène en Afrique de l’est et au-delà Vol. 20 (eds Joséphine Lesur et al. ) 9-34 (Séances de la Société Préhistorique Française, 2023). Hisham, S., Zainun, K. A. & Ibrahim, M. A. Observer error in assessing age-related morphology using digital photographs of the pubic symphyseal face. Canadian Society of Forensic Science Journal 55 , 213-220 (2022). https://doi.org/10.1080/00085030.2022.2053279 Villotte, S., Kacki, S. & Thomas, A. Sex estimation of the human os coxae in archeological contexts: An advocacy of using both Diagnose Sexuelle Probabiliste and Brůžek's morphoscopic method. International Journal of Osteoarchaeology 34 , e3334 (2024). https://doi.org/https://doi.org/10.1002/oa.3334 Pilloud, M. A., Adams, D. M. & Hefner, J. T. Observer error and its impact on ancestry estimation using dental morphology. International Journal of Legal Medicine 133 , 949-962 (2019). https://doi.org/10.1007/s00414-018-1985-3 Williams, B. A. & Rogers, T. L. Evaluating the Accuracy and Precision of Cranial Morphological Traits for Sex Determination. Journal of Forensic Sciences 51 , 729-735 (2006). https://doi.org/doi:10.1111/j.1556-4029.2006.00177.x Kemkes, A. & Göbel, T. Metric Assessment of the “Mastoid Triangle” for Sex Determination: A Validation Study. Journal of Forensic Sciences 51 , 985-989 (2006). https://doi.org/https://doi.org/10.1111/j.1556-4029.2006.00232.x Godde, K., Thompson, M. M. & Hens, S. M. Sex estimation from cranial morphological traits: use of the methods across American Indians, modern North Americans, and ancient Egyptians. Homo-Journal of Comparative Human Biology (2018). https://doi.org/https://doi.org/10.1016/j.jchb.2018.09.003 Smith, H. F. Which cranial regions reflect molecular distances reliably in humans? Evidence from three-dimensional morphology. American Journal of Human Biology 21 , 36-47 (2009). https://doi.org/https://doi.org/10.1002/ajhb.20805 Harvati, K. & Weaver, T. D. Human cranial anatomy and the differential preservation of population history and climate signatures. The Anatomical Record Part A: Discoveries in Molecular, Cellular, and Evolutionary Biology 288 , 1225-1233 (2006). Roseman, C. C. Detecting interregionally diversifying natural selection on modern human cranial form by using matched molecular and morphometric data. P Natl Acad Sci USA 101 , 12824-12829 (2004). https://doi.org/10.1073/pnas.0402637101 Roseman, C. C. & Weaver, T. D. Multivariate apportionment of global human craniometric diversity. American Journal of Physical Anthropology 125 , 257-263 (2004). https://doi.org/10.1002/ajpa.10424 González-José, R. et al. Functional-cranial approach to the influence of economic strategy on skull morphology. American Journal of Physical Anthropology 128 , 757-771 (2005). https://doi.org/10.1002/ajpa.20161 Godde, K. Secular trends in cranial morphological traits: a socioeconomic perspective of change and sexual dimorphism in North Americans 1849–1960. Annals of Human Biology 42 , 255-261 (2015). https://doi.org/10.3109/03014460.2014.941399 Weiss, K. M. On the systematic bias in skeletal sexing. American Journal of Physical Anthropology 37 , 239-249 (1972). https://doi.org/https://doi.org/10.1002/ajpa.1330370208 Cintas-Peña, M. & Herrero-Corral, A. M. Missing prehistoric women? Sex ratio as an indicator for analyzing the population of Iberia from the 8th to the 3rd millennia B.C. Archaeological and Anthropological Sciences 12 , 263 (2020). https://doi.org/10.1007/s12520-020-01215-x Cardoso, H. F. V. Brief communication: The collection of identified human skeletons housed at the Bocage Museum (National Museum of Natural History), Lisbon, Portugal. American Journal of Physical Anthropology 129 , 173-176 (2006). https://doi.org/10.1002/ajpa.20228 Coquerelle, M. et al. Sexual Dimorphism of the Human Mandible and Its Association With Dental Development. American Journal of Physical Anthropology 145 , 192-202 (2011). https://doi.org/Doi 10.1002/Ajpa.21485 AlQahtani, S. J., Hector, M. P. & Liversidge, H. M. Brief communication: The London atlas of human tooth development and eruption. American Journal of Physical Anthropology 142 , 481-490 (2010). https://doi.org/10.1002/ajpa.21258 Bruzek, J. A method for visual determination of sex, using the human hip bone. American Journal of Physical Anthropology 117 , 157-168 (2002). https://doi.org/10.1002/ajpa.10012 Murail, P., Bruzek, J., Houët, F. & Cunha, E. DSP: a tool for probabilistic sex diagnosis using worldwide variability in hip-bone measurements. Bulletins et Mémoires de la Société d’Anthropologie de Paris , 167-176 (2005). Fedorov, A. et al. 3D Slicer as an image computing platform for the Quantitative Imaging Network. Magnetic resonance imaging 30 , 1323-1341 (2012). https://doi.org/10.1016/j.mri.2012.05.001 Geomorph: Software for geometric morphometric analyses. R package version 4.0.6 (2023). Schlager, S., Jefferis, G., Ian, D. & Schlager, M. S. (Academic Press. https://doi. org/10.1016/B978–0–12–810493–4.00011–0, 2023). Godinho, R. M., O'Higgins, P. & Gonçalves, C. Assessing the reliability of virtual reconstruction of mandibles. American Journal of Physical Anthropology 172 , 723-734 (2020). https://doi.org/https://doi.org/10.1002/ajpa.24095 Gunz, P., Mitteroecker, P., Neubauer, S., Weber, G. W. & Bookstein, F. L. Principles for the virtual reconstruction of hominin crania. Journal of Human Evolution 57 , 48-62 (2009). https://doi.org/10.1016/j.jhevol.2009.04.004 Neeser, R., Ackermann, R. R. & Gain, J. Comparing the accuracy and precision of three techniques used for estimating missing landmarks when reconstructing fossil hominin crania. American Journal of Physical Anthropology 140 , 1-18 (2009). https://doi.org/10.1002/ajpa.21023 Senck, S., Bookstein, F. L., Benazzi, S., Kastner, J. & Weber, G. W. Virtual Reconstruction of Modern and Fossil Hominoid Crania: Consequences of Reference Sample Choice. The Anatomical Record 298 , 827-841 (2015). https://doi.org/10.1002/ar.23104 Mitteroecker, P., Gunz, P., Windhager, S. & Schaefer, K. A brief review of shape, form, and allometry in geometric morphometrics, with applications to human facial morphology. Hystrix-Italian Journal of Mammalogy 24 , 59-66 (2013). https://doi.org/DOI 10.4404/hystrix-24.1-6369 Zelditch, M. L., Swiderski, D. L., Sheets, H. D. & Fink, W. L. Geometric Morphometrics For Biologists: A Primer . (Elsevier, 2012). O'Higgins, P. The study of morphological variation in the hominid fossil record: biology, landmarks and geometry. Journal of Anatomy 197 , 103-120 (2000). https://doi.org/10.1046/j.1469-7580.2000.19710103.x Klingenberg, C. P. Size, shape, and form: concepts of allometry in geometric morphometrics. Development Genes and Evolution 226 , 113-137 (2016). https://doi.org/10.1007/s00427-016-0539-2 Hammer, O. PAST reference manual 4.12 . (2022). Courtenay, L. A., Aramendi, J. & González-Aguilera, D. A Graph-Based Mathematical Model for More Efficient Dimensionality Reduction of Landmark Data in Geometric Morphometrics. Evolutionary Biology (2024). https://doi.org/10.1007/s11692-024-09636-5 Abellán, N., Baquedano, E. & Domínguez-Rodrigo, M. High-accuracy in the classification of butchery cut marks and crocodile tooth marks using machine learning methods and computer vision algorithms. Geobios 72-73 , 12-21 (2022). https://doi.org/https://doi.org/10.1016/j.geobios.2022.07.001 Moclán, A. et al. Identifying the bone-breaker at the Navalmaíllo Rock Shelter (Pinilla del Valle, Madrid) using machine learning algorithms. Archaeological and Anthropological Sciences 12 , 46 (2020). https://doi.org/10.1007/s12520-020-01017-1 Courtenay, L. A. & González-Aguilera, D. Geometric Morphometric Data Augmentation Using Generative Computational Learning Algorithms. Applied Sciences 10 , 9133 (2020). McPherron, S. P., Archer, W., Otárola-Castillo, E. R., Torquato, M. G. & Keevil, T. L. Machine learning, bootstrapping, null models, and why we are still not 100% sure which bone surface modifications were made by crocodiles. Journal of Human Evolution 164 , 103071 (2022). https://doi.org/https://doi.org/10.1016/j.jhevol.2021.103071 Kuhn, M. et al. Caret: Classification and Regression Training. (2024). Deane-Mayer, Z. A., Knowles, J. E. & López, A. Package ‘caretEnsemble’. (2024). Lantz, B. Machine learning with R . (Packt Publishing, 2013). Domínguez-Rodrigo, M. Successful classification of experimental bone surface modifications (BSM) through machine learning algorithms: a solution to the controversial use of BSM in paleoanthropology? Archaeological and Anthropological Sciences 11 , 2711-2725 (2019). https://doi.org/10.1007/s12520-018-0684-9 Additional Declarations No competing interests reported. Supplementary Files Godinhoetal2025CouplingGMandML3notrackchangesnocodesSI.docx Cite Share Download PDF Status: Published Journal Publication published 19 Dec, 2025 Read the published version in Scientific Reports → Version 1 posted Editorial decision: Revision requested 01 Dec, 2025 Editor assigned by journal 01 Dec, 2025 Submission checks completed at journal 28 Nov, 2025 First submitted to journal 23 Nov, 2025 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6389860","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":550205032,"identity":"9e704a03-dbb4-409b-8596-52182bb38629","order_by":0,"name":"Ricardo Miguel Godinho","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAwklEQVRIiWNgGAWjYLACHgOGBAb2BgaGB6Rp4TnAACSBHOK0gBRLJBCpxXza8YsP3hTcyeOXfGP4IeEXQ549IS0yt3OKDecYPCuWnJ1jLJHYx1BM0BYJ6Zw0aR6Dw4kbbucYSCT2MAAR0VpunjH+QaSW9GMQLTd4zCQSfhBnCzPQL4cTZ/aklVkkNkgUgwKbkC0PH7z5czixn/3w5hsf/tjkgaKUAADGIxwwtoFihyBgf4DE+cNAjJZRMApGwSgYYQAAKm1B47YA/7wAAAAASUVORK5CYII=","orcid":"","institution":"Interdisciplinary Center for Archaeology and Evolution of Human Behaviour (ICArEHB), Faculdade das Ciências Humanas e Sociais, University of Algarve","correspondingAuthor":true,"prefix":"","firstName":"Ricardo","middleName":"Miguel","lastName":"Godinho","suffix":""},{"id":550205033,"identity":"6f46f62f-3882-473b-b1ba-170df92d4d30","order_by":1,"name":"Isabelle Crevecouer","email":"","orcid":"","institution":"UMR 5199‑PACEA, CNRS, Université de Bordeaux","correspondingAuthor":false,"prefix":"","firstName":"Isabelle","middleName":"","lastName":"Crevecouer","suffix":""},{"id":550205034,"identity":"366c1901-a7a6-4c3e-9d58-cf1d8ea8c2d1","order_by":2,"name":"Susana Garcia","email":"","orcid":"","institution":"MUHNAC, Universidade de Lisboa","correspondingAuthor":false,"prefix":"","firstName":"Susana","middleName":"","lastName":"Garcia","suffix":""},{"id":550205035,"identity":"a6982875-dcf8-40c3-9c68-8d6c4a6f538d","order_by":3,"name":"Rebecca Whiting","email":"","orcid":"","institution":"The British Museum","correspondingAuthor":false,"prefix":"","firstName":"Rebecca","middleName":"","lastName":"Whiting","suffix":""},{"id":550205036,"identity":"7a8316eb-c271-4300-a204-35c3470e1776","order_by":4,"name":"Julia Aramendi","email":"","orcid":"","institution":"University of Cambridge","correspondingAuthor":false,"prefix":"","firstName":"Julia","middleName":"","lastName":"Aramendi","suffix":""}],"badges":[],"createdAt":"2025-04-07 03:53:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6389860/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6389860/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1038/s41598-025-31365-8","type":"published","date":"2025-12-19T15:58:29+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":96917176,"identity":"6589aeb1-92f1-4f7a-a0b3-f0dcd2c9b237","added_by":"auto","created_at":"2025-11-27 14:09:20","extension":"xlsx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10549,"visible":true,"origin":"","legend":"","description":"","filename":"Table1.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/019cde9a90b9023b7d9f398d.xlsx"},{"id":96822584,"identity":"64e5f14d-b18d-4cf3-a46a-dd604cd06cc6","added_by":"auto","created_at":"2025-11-26 12:22:04","extension":"xlsx","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10190,"visible":true,"origin":"","legend":"","description":"","filename":"Table2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/adba7155a11819711d0fb1ba.xlsx"},{"id":96822587,"identity":"8e58fa77-8991-4c27-beb7-95dad0bbed6f","added_by":"auto","created_at":"2025-11-26 12:22:04","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8933,"visible":true,"origin":"","legend":"","description":"","filename":"Table3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/0e6ec93a42e0fbb0bc6a7e3c.xlsx"},{"id":96917780,"identity":"c0597c9e-29e2-416e-96ec-8fb2e9299c76","added_by":"auto","created_at":"2025-11-27 14:10:33","extension":"docx","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":979244,"visible":true,"origin":"","legend":"","description":"","filename":"Godinhoetal2025CouplingGMandML3notrackchanges.docx","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/25c0748592f9bb510466608e.docx"},{"id":96822586,"identity":"31083d1e-5488-4cae-a821-c5aa58d90620","added_by":"auto","created_at":"2025-11-26 12:22:04","extension":"json","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":7150,"visible":true,"origin":"","legend":"","description":"","filename":"cfec6dfd857c44ada1d581e2e3e6ef75.json","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/31a2a1e21822cb27fc38ef29.json"},{"id":96822600,"identity":"725d28fc-65f9-41db-9f14-69512890a420","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"docx","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":414850,"visible":true,"origin":"","legend":"","description":"","filename":"Godinhoetal2025CouplingGMandML3notrackchangesnocodesSI.docx","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/8504581ab47aa588de8c8f45.docx"},{"id":96917145,"identity":"477bd7e5-437a-488b-87a0-31bafcf8aa2c","added_by":"auto","created_at":"2025-11-27 14:09:18","extension":"xml","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":301166,"visible":true,"origin":"","legend":"","description":"","filename":"cfec6dfd857c44ada1d581e2e3e6ef751enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/f291d96ff3128f9aa9f6e218.xml"},{"id":96822589,"identity":"a0ddd3fa-febb-4c55-83cc-7a47c9e0276d","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"pdf","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":11121,"visible":true,"origin":"","legend":"","description":"","filename":"Fig1.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/e93c4ee4c0fb3b9dbabc4aab.pdf"},{"id":96822602,"identity":"0fec0e03-f55b-4b42-a556-328ff737805d","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"pdf","order_by":13,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":378074,"visible":true,"origin":"","legend":"","description":"","filename":"Fig2.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/52e23cfe40ca0cb3f40a6c91.pdf"},{"id":96822596,"identity":"498f2879-d563-4711-ac3a-8cffdf06d651","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"pdf","order_by":14,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":1089961,"visible":true,"origin":"","legend":"","description":"","filename":"Fig3.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/2e1c06827bd496e908bf731b.pdf"},{"id":96822608,"identity":"4459bade-2753-4bf2-bba4-e69d9f693920","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"pdf","order_by":15,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":2456482,"visible":true,"origin":"","legend":"","description":"","filename":"Fig4.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/30ea42bb747426c26841cad9.pdf"},{"id":96822599,"identity":"669307c2-b471-4b8b-a175-67ada6e479d9","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"pdf","order_by":16,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":160312,"visible":true,"origin":"","legend":"","description":"","filename":"Fig5.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/79fb4ab66574b1639438c312.pdf"},{"id":96822593,"identity":"368a715f-b8fe-4a69-b1f0-6ac71988556b","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"png","order_by":17,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9485,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/8032ae513d6768c6e1a13def.png"},{"id":96918743,"identity":"3f34e35d-7cad-43fb-868b-594fbe571466","added_by":"auto","created_at":"2025-11-27 14:12:30","extension":"png","order_by":18,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":287931,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/a8e8bdba7cc6ff56910f09d7.png"},{"id":96917858,"identity":"62fa021f-fecc-4107-902b-d645456e4007","added_by":"auto","created_at":"2025-11-27 14:10:38","extension":"png","order_by":19,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9189,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/619dc030fe65267c8f457963.png"},{"id":96822591,"identity":"eeac8bd7-b39b-4170-9044-62dc86ae0186","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"png","order_by":20,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":111468,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/e494c08266679a7328736fb7.png"},{"id":96822603,"identity":"1afbf881-8d5a-4cc5-9f8e-c9d10322bd9f","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"jpeg","order_by":21,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":361653,"visible":true,"origin":"","legend":"","description":"","filename":"floatimage5.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/cfabf14b6ee298ce9fac69f3.jpeg"},{"id":96917580,"identity":"1713ee5e-2afe-447e-9ab1-11b2a4affc8d","added_by":"auto","created_at":"2025-11-27 14:10:09","extension":"png","order_by":22,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":9907,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/f9dc3851648f38e1b373e259.png"},{"id":96918068,"identity":"a0222807-06db-4c43-86e5-11ac208b13a3","added_by":"auto","created_at":"2025-11-27 14:11:08","extension":"png","order_by":23,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":63211,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/5db4fac3ef228769ea6ea34e.png"},{"id":96917693,"identity":"87e7b20c-945e-4f64-9e62-0e8b7af8a9ca","added_by":"auto","created_at":"2025-11-27 14:10:25","extension":"png","order_by":24,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":10347,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/2c412a814eeb0371c3f91443.png"},{"id":96916866,"identity":"0b7b5cbf-6a52-44f4-a246-715a05d6ca8d","added_by":"auto","created_at":"2025-11-27 14:08:59","extension":"png","order_by":25,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":21298,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/79397bd0e025178c368acc21.png"},{"id":96917707,"identity":"0e0edfa5-5235-4a2a-a7d5-20334c61ec8b","added_by":"auto","created_at":"2025-11-27 14:10:26","extension":"png","order_by":26,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":119437,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/6067fe45d0aac3ed134368bb.png"},{"id":96918556,"identity":"cbc54cae-bf75-45f9-9d0b-14d885570858","added_by":"auto","created_at":"2025-11-27 14:12:07","extension":"xml","order_by":27,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":299265,"visible":true,"origin":"","legend":"","description":"","filename":"cfec6dfd857c44ada1d581e2e3e6ef751structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/47186db223521a2f3fa319ca.xml"},{"id":96822605,"identity":"5c5cffb7-16b3-4304-8600-b25f66a5af1b","added_by":"auto","created_at":"2025-11-26 12:22:05","extension":"html","order_by":28,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":325909,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/60565f825f7cd31d6d524643.html"},{"id":96822578,"identity":"ab36d3bb-c266-4830-8ec6-5614b0937fbf","added_by":"auto","created_at":"2025-11-26 12:22:04","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":98687,"visible":true,"origin":"","legend":"\u003cp\u003eCentroid size of the specimens selected from the Luis Lopes and Jebel Sahaba collection. Results are grouped by sex and origin, with corresponding p-values of non-parametric pairwise post-hoc statistical testing. Note that all groups are significantly different from each other (except for late modern males from Portugal and Late Pleistocene females from present Sudan).\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/9d8b36a5bda94640bf586dba.png"},{"id":96917769,"identity":"57ea9970-0d3c-4ee4-908f-aa277af6c783","added_by":"auto","created_at":"2025-11-27 14:10:32","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":155238,"visible":true,"origin":"","legend":"\u003cp\u003e(A) Shape and (B) form PCAs. Results are colour coded and grouped by sex and origin. Note that form space shows apparent intra-population sex differences. In shape space there are no apparent intra-population sex differences, but clear inter-population differences.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/3ec9202d03551f62482b9cb9.png"},{"id":96822580,"identity":"0a05b656-3be5-4490-8a99-fd0b728d00b8","added_by":"auto","created_at":"2025-11-26 12:22:04","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":108532,"visible":true,"origin":"","legend":"\u003cp\u003eAccuracy achieved by the best-performing ML models trained with the different sets of PCs (including 100%, 95%, 90% of the total variance and the PCs established based on p values) in shape and form created for this particular study.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/1483574222c53f81f8346cc5.png"},{"id":96822582,"identity":"2003efcc-e677-49b0-a353-a652b90460fd","added_by":"auto","created_at":"2025-11-26 12:22:04","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":61775,"visible":true,"origin":"","legend":"\u003cp\u003eProbability percentages assigned to each archaeological sample following ML model training on the PCs capturing 90% of the total variance in both shape and form space. Classifications under \u003cem\u003eMm\u003c/em\u003e and \u003cem\u003eFf\u003c/em\u003e represent samples consistently assigned to male and female groups by both anthropological and ML methods. \u003cem\u003eFm\u003c/em\u003e denotes samples classified as female by ML models but as male by anthropological methods, while \u003cem\u003eMf\u003c/em\u003e indicates cases where ML models classify as male but anthropological methods classify as female.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/7b0c3b8ecebde57fe70ffb71.png"},{"id":96917408,"identity":"c2887e43-4724-4a6d-8eb8-46290246df93","added_by":"auto","created_at":"2025-11-27 14:09:42","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":165228,"visible":true,"origin":"","legend":"\u003cp\u003eLandmarks used for extraction of coordinates. Specimen in (A) anterior, (B) lateral and (C) superior view.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/f301acce1bce227251024afa.png"},{"id":98814257,"identity":"637f1f9c-6463-41cd-91d7-4d27e03c4198","added_by":"auto","created_at":"2025-12-22 16:12:09","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1967313,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/61b5a562-43c7-4616-91e3-d3dc0cf292d0.pdf"},{"id":96918655,"identity":"5c425fb9-cbde-486c-8fef-0886941bceaf","added_by":"auto","created_at":"2025-11-27 14:12:17","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":414850,"visible":true,"origin":"","legend":"","description":"","filename":"Godinhoetal2025CouplingGMandML3notrackchangesnocodesSI.docx","url":"https://assets-eu.researchsquare.com/files/rs-6389860/v1/73608152f37678add9238a51.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"\u003cp\u003eCoupling geometric morphometrics and machine learning for mandibular sex estimation: testing Late Pleistocene and Late Modern populations\u003c/p\u003e","fulltext":[{"header":"1 Introduction","content":"\u003cp\u003eSex is one of the most fundamental biological parameters assessed in forensic, biological and palaeoanthropological studies\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. In addition to allowing the osteobiographical characterization of individuals, it enables ensuing analysis of, e.g., sex related differences in funerary behaviour\u003csup\u003e\u003cspan additionalcitationids=\"CR4 CR5\" citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e, weaning\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e, diet\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan additionalcitationids=\"CR9 CR10\" citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e, activity patterns\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan additionalcitationids=\"CR13\" citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e, mobility\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e, and pathology\u003csup\u003e\u003cspan additionalcitationids=\"CR16\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eBiomolecular methods (i.e., aDNA and proteomics) have been used to establish sex and provide very reliable results\u003csup\u003e\u003cspan additionalcitationids=\"CR19 CR20 CR21 CR22 CR23 CR24 CR25 CR26\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. Moreover, such methods typically require very reduced quantities of bone/teeth, overcoming pervasive preservation issues in archaeological collections that lead to fragmentation and incompleteness of skeletal elements, often precluding reliable morphological based sex-estimation\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e,\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e,\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. Yet, biomolecular methods are destructive, require highly specialized (often expensive) laboratory procedures and also depend on the preservation of proteins and/or DNA\u003csup\u003e\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u003c/sup\u003e. Thus, morphological-based sex estimation remains the most common and feasible approach to sex classification of archaeological individuals.\u003c/p\u003e\u003cp\u003ePrevious morphological-based sex estimation studies have explored sexual dimorphism in most bones of the human skeleton. The most reliable regions for sex estimation are the \u003cem\u003eos coxae\u003c/em\u003e and the skull, which typically provide correct classification rates above 90%\u003csup\u003e1,28\u0026ndash;32\u003c/sup\u003e (but see Spradley and Jantz \u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e regarding the skull). Yet, these bones are often fragmented or incomplete in archaeological and/or forensic contexts. Moreover, the complexities of funerary behaviour often involve destructive procedures (e.g., cremation and reuse of funerary spaces) and/or post-depositional manipulation of the human remains, causing truncation of individuals and/or commingling and fragmentation of bones\u003csup\u003e\u003cspan additionalcitationids=\"CR35 CR36 CR37 CR38 CR39 CR40 CR41\" citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. Thus, researchers have often to estimate sex based on individual bones rather than complete skeletons, including post-cranial bones (other than the \u003cem\u003eos coxae\u003c/em\u003e) which typically provide lower correct sex classification rates and so are less reliable in sex assessment\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e,\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eSex estimation based on the \u003cem\u003eos coxae\u003c/em\u003e and skull is frequently based on scoring along semi-quantitative ordinal scales of specific anatomical regions\u003csup\u003e\u003cspan additionalcitationids=\"CR45 CR46\" citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e. While conventional metric methods are also used in these regions\u003csup\u003e\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e,\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e, visual-based methods readily capture morphological information not easily quantifiable with the former metric approaches\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e. Yet, visual scoring is based on somewhat subjective observer specific assessment, and so some degree of inter-observer error emerges that may lead to conflicting estimations\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e,\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e. To better represent 3D morphology objectively and quantitatively, and to reduce inter-observer error, Geometric Morphometrics (GM) has been used more recently to investigate sex related morphological differences in the pelvis\u003csup\u003e\u003cspan additionalcitationids=\"CR52\" citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e and skull\u003csup\u003e\u003cspan additionalcitationids=\"CR55 CR56 CR57\" citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e\u003c/sup\u003e. These approaches have provided very good results, but the use of conventional landmarks (LMs) is limited in capturing 3D morphology. Hence, some studies have also used semi-sliding landmarks to enable dense coverage of the cranium and to provide better morphological representation\u003csup\u003e\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eMachine learning (ML) has also been recently used in sex estimation. Several types of data have been used to create ML models, including linear measurements from cranial\u003csup\u003e\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e,\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e\u003c/sup\u003e and post-cranial bones\u003csup\u003e\u003cspan additionalcitationids=\"CR63 CR64 CR65 CR66\" citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e\u003c/sup\u003e, and cross-sectional data from long bones\u003csup\u003e\u003cspan citationid=\"CR68\" class=\"CitationRef\"\u003e68\u003c/span\u003e\u003c/sup\u003e. While ML often improves correct sex identification, it has seldom been applied together with GM methods\u003csup\u003e\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e,\u003cspan citationid=\"CR69\" class=\"CitationRef\"\u003e69\u003c/span\u003e\u003c/sup\u003e. Moreover, to the best of our knowledge, no studies have used ML models developed with identified collections and tested sex classification of archaeological specimens with sex already previously estimated based on multiple skeletal regions (including pelvises, crania and mandibles). This is particularly relevant to test the potential and limitations of applying ML models to estimate sex in archaeological specimens and so to enable examination of sex-based differences in past populations. This is the case of mandibles, which are often used in studies about the morphological impact of population history and diet on past populations\u003csup\u003e\u003cspan additionalcitationids=\"CR71 CR72 CR73 CR74\" citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e\u003c/sup\u003e. Because they are often found isolated from the remaining skeleton, sex information is typically not estimated due to uncertainty of predictions. Thus, it is vital to enhance mandibular sex estimation reliability to enable further examination about past populations.\u003c/p\u003e\u003cp\u003eHere we use 3D GM to capture the mandibular morphology of an identified skeletal collection (Luis Lopes) and examine sex differences. We then use resulting outputs to train ML models, classify the sex of held-out specimens from the same (Luis Lopes) population and quantify the reliability of the ML predictions. Further, we use those models to classify Late Pleistocene mandibles from Jebel Sahaba which have been previously sexed morphologically (based on multiple skeletal regions) to examine the reliability of the GM based ML sex classifications. The selection of such diverse testing samples is deliberate and aims to bracket the reliability of this sex estimation approach. Specifically, the Luis Lopes intra-population testing sample is expected to provide the most reliable expectable results, whereas the Jebel Sahaba testing sample the least reliable expectable results due to its extreme (intra-specific) morphological difference (see details below).\u003c/p\u003e"},{"header":"2 Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003e2.1 GM based morphological analysis\u003c/h2\u003e\u003cp\u003eOur results show clear intra-population size differences between male and female mandibles. Within each population, male mandibles are clearly larger than those of females (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The archaeological specimens from Jebel Sahaba are, however, larger than those from late Modern Portugal, with no statistically significant differences between females from the former and males from the latter.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe statistical analyses show morphological differences between populations and between the sexes. Specifically, the PERMANOVA including the first PCs accounting for ~\u0026thinsp;95% of the total variance shows statistically significant differences between all groups (males and females originating from Portugal and Sudan) in form space. In shape space, all groups are significantly different, with the exception of males and females from Jebel Sahaba (Table SI 1). Such differences are apparent in the PCA plotting PC1 and PC2, in which there is a clear distinction between the sexes of both populations in PC1 (despite some overlap) in form, but not in shape space, in which PC1 separates the two populations (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Males (which display lower PC1 scores within each population) have more robust mandibles in form space in both samples, with, e.g., more vertical mandibular symphyses, broader and upright rami and wider sigmoid notches compared to females (which display higher PC1 scores within each population). PC1 in shape space, which separates the two populations, shows that the Jebel Sahaba mandibles are much more robust than those from the late modern Luis Lopes specimens.\u003c/p\u003e\u003cp\u003eAs expected, the space that includes size (form) is strongly correlated with size Table SI 2). In shape space the first (and some ensuing) PC is also tightly correlated with size when including both populations in the analysis. However, when the populations are separated, the first PCs are no longer correlated with size within each of them, suggesting the relationship between some of the shape PCs and size in this space are driven by the size differences between the populations.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e\u003ch2\u003e2.2 Machine learning\u003c/h2\u003e\u003cp\u003eMultiple ML models were trained with different datasets to examine which would produce the best results both in shape and form space. Specifically, the models were run using eight different sets of PCs in shape and form space, including those capturing 100% of the total variance (56 PCs for shape and 57 PCs for form), 95% of the total variance (23 PCs for shape and 18 PCs for form), 90% of the total variance (16 PCs for shape and 12 PCs for form), and PCs considered significant with a value below 0.05 (Figure SI 1).\u003c/p\u003e\u003cp\u003eModel performance varies depending on the number of PCs included in the analyses. According to our results, models generally perform best when using 90% of the total variance (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), while they perform the worst when using the full set of PCs obtained after PCA (Table SI 3). The latter result is likely due to the inclusion of too many residuals in the analyses. Nonetheless, perfect accuracy was achieved only when using 95% of the variance in shape (Table SI 4). With 90% of the total variance, accuracy rates range from 65% to 95% (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), though the most common accuracy rate is 90%, with analyses on form variables generally performing better than those on shape. This trend is observed across all analyses (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, Table SI 4 and Table SI 5), suggesting that size differences may be a significant factor in distinguishing male from female individuals in modern \u003cem\u003eHomo sapiens\u003c/em\u003e. This is consistent with the size results above, which show significant size differences between males and females in both samples.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eResults provided by ML algorithms based on PCs accounting for over 90% of the total variance in shape and form. Algorithms with accuracy above 90% are highlighted in bold.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"9\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModel\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eData\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eAccuracy\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eKappa\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eAccLower\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eAccUpper\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c7\"\u003e\u003cp\u003eSensitivity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c8\"\u003e\u003cp\u003eSpecificity\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c9\"\u003e\u003cp\u003eBalAccuracy\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003ekNN\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.75\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.5\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.509\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.9134\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.7778\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.7273\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.7525\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.85\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.7059\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.6211\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.9679\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.7273\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.8636\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eLGR\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.596\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.5634\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.9427\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.7778\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.8182\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.798\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.798\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.8889\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.899\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eDTC5.0\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.65\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.4078\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.8461\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.6667\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.6364\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.6515\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.802\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.8182\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eRF\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.85\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.6211\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.9679\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.8889\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.8182\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.8535\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.95\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.7513\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9987\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.9545\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eGB\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.65\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.2857\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.4078\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.8461\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.5556\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.7273\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.6414\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.7980\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.8889\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.899\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eNB\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.4059\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.4572\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.8811\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.7778\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.6364\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.7071\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.798\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.8889\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.899\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eLDA\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.798\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.8889\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.899\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.798\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.8889\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.899\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003ePLS\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.7938\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.7778\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.8889\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.802\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.8182\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eSVMl\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.85\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.6211\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.9679\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.8889\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.8182\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.8535\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.7938\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.7778\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.8889\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eSVMr\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.802\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.683\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9877\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.8182\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e0.85\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.7\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.6211\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.9679\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e0.8889\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e0.8182\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e0.8535\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e\u003cp\u003e\u003cb\u003eNNET\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.95\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.898\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.7513\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9987\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e0.8889\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.9444\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e\u003cb\u003e0.95\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e\u003cb\u003e0.9\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e\u003cb\u003e0.7513\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e\u003cb\u003e0.9987\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c7\"\u003e\u003cp\u003e\u003cb\u003e1\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c8\"\u003e\u003cp\u003e\u003cb\u003e0.9091\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e\u003cp\u003e\u003cb\u003e0.9545\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe inter-population reliability of the ML models predictions was tested by contrasting the ML classifications with previous sex estimations of the Late Pleistocene Jebel Sahaba archaeological sample\u003csup\u003e\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e76\u003c/span\u003e\u003c/sup\u003e. The results for this archaeological sample do not reach the accuracy levels achieved with the modern testing sample, assuming the skeletal morphology based anthropological classifications are correct. When using 90% of the total variance, the highest agreement between anthropological and ML-based methods is 83.33%, and the lowest is 44.44%, with a typical match rate of 61.11% (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Across the different models, there is a general tendency to overclassify the archaeological sample as male when size is included in the analyses, while classifications based on shape alone tend to be more balanced (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Table SI 6 \u0026ndash; Table SI 8). Furthermore, correct female classifications tend to be assigned with higher confidence when using shape variables, whereas correct male classifications are more confidently made with form variables, though this trend is much more pronounced for females (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eClassification provided by ML algorithms for the archaeological sample using the set of PCs that account for 90% of the total variance in shape and form space. The number of samples classified consistently by both anthropological methods and ML algorithms is indicated in brackets.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModel\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eData\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eN male using ML methods\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eN female using ML methods\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eN same sex attribution (%)\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003ekNN\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6 (4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12 (6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e55.56%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17 (10)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLGR\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e13 (7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e5 (2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e50%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15 (9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3 (2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eDTC5.0\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8 (4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10 (4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e44.44%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17 (10)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eRF\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6 (4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12 (6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e55.56%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17 (10)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eGB\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e6 (4)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e12 (6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e55.56%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17 (10)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eNB\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e3 (3)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e15 (8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15 (9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3 (2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLDA\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e9 (7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9 (6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e72.22%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15 (9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3 (2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003ePLS\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8 (7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10 (7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e77.78%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17 (10)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eSVMl\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15 (9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3 (2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e15 (9)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e3 (2)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eSVMr\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e8 (5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e10 (5)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e55.56%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e9 (8)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9 (7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e83.33%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003e\u003cb\u003eNNET\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eshape\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e9 (7)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e9 (6)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e72.22%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eform\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e17 (10)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e1 (1)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e61.11%\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eHowever, classification probabilities do not commonly exceed 90%, even for matching classifications, and high probabilities are equally common among mismatched classifications in shape space, though less frequent in form space (Table SI 9 \u0026ndash; Table SI 16). In other analyses using different sets of PCs, match percentages between anthropological and ML methods range from 44.44% to 88.89%. The highest match rate in the study (88.89%) is obtained in form space, where the SVMr on the significant PCs appears to avoid overestimating the number of male individuals in the sample (Table SI 8).\u003c/p\u003e\u003cp\u003eThus, the relationship between model accuracy on the modern sample and its performance on the archaeological sample does not appear to be straightforward. An increase in accuracy on the testing set classifications does not necessarily correspond to a direct increase in classification match for the archaeological sample. Although the accuracy of ML models trained on the modern sample is a statistically significant predictor for classifications in the archaeological sample (F\u0026thinsp;\u0026lt;\u0026thinsp;0.05), in both shape and form, with increasing accuracy correlating with increases in the outcome variables (Figure SI 2), this association explains only a small fraction (approx. 12.5%) of the variance in the archaeological sample (Table SI 17).\u003c/p\u003e\u003c/div\u003e"},{"header":"3 Discussion","content":"\u003cp\u003eOverall, and consistent with previous studies, our GM results show clear sexual dimorphism in mandibular morphology\u003csup\u003e\u003cspan additionalcitationids=\"CR78 CR79 CR80 CR81 CR82 CR83 CR84 CR85 CR86 CR87 CR88 CR89 CR90 CR91 CR92 CR93 CR94 CR95 CR96 CR97\" citationid=\"CR77\" class=\"CitationRef\"\u003e77\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR98\" class=\"CitationRef\"\u003e98\u003c/span\u003e\u003c/sup\u003e, as well as inter-population morphological differences\u003csup\u003e\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e,\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e\u003c/sup\u003e. Consistent with previous studies using ML for sex classification\u003csup\u003e\u003cspan additionalcitationids=\"CR61 CR62 CR63 CR64 CR65 CR66\" citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR67\" class=\"CitationRef\"\u003e67\u003c/span\u003e\u003c/sup\u003e, the ML models were very efficient in the sex classification of the late modern identified Luis Lopes intra-population test sample, with average accuracies of 90% (see details above). However, the sex classification of the temporally and geographically distant mandibles of Late Pleistocene Jebel Sahaba was meaningfully less efficient, with average accuracies of 60\u0026ndash;63%. Although we cannot exclude the possibility of some skeletally misclassified archaeological individuals impacting our results, these differences are most likely due to meaningful inter-population size and shape differences that resulted in frequent misclassifications (especially of females as males with form space derived data). These results are, however, predictable and consistent with previous studies highlighting inter-population morphological differences and cautioning against the use of inadequate reference samples to classify target specimens\u003csup\u003e\u003cspan additionalcitationids=\"CR100 CR101 CR102 CR103\" citationid=\"CR99\" class=\"CitationRef\"\u003e99\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR104\" class=\"CitationRef\"\u003e104\u003c/span\u003e\u003c/sup\u003e. This is particularly relevant in this study because the individuals from Jebel Sahaba belong to a highly robust population characterized by plesiomorphic traits, extreme dental dimensions and complex crown morphology, as well as robust morphological features, some being related to powerful masticatory apparatus\u003csup\u003e\u003cspan additionalcitationids=\"CR106 CR107 CR108\" citationid=\"CR105\" class=\"CitationRef\"\u003e105\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR109\" class=\"CitationRef\"\u003e109\u003c/span\u003e\u003c/sup\u003e. This unique phenotype has been interpreted as a consequence of population isolation during the Late Pleistocene in the Nile valley\u003csup\u003e\u003cspan citationid=\"CR110\" class=\"CitationRef\"\u003e110\u003c/span\u003e,\u003cspan citationid=\"CR111\" class=\"CitationRef\"\u003e111\u003c/span\u003e\u003c/sup\u003e, which may influence the classification efficiency.\u003c/p\u003e\u003cp\u003eThe following sections discuss in more detail the GM and ML results, along with the limitations of this study and future research prospects.\u003c/p\u003e\u003cdiv id=\"Sec6\" class=\"Section2\"\u003e\u003ch2\u003e3.1 Geometric Morphometrics\u003c/h2\u003e\u003cp\u003eOur GM results show sexual dimorphism in mandibular morphology, along with inter-population differences. Males have significantly larger mandibles than females in both populations (assessed via centroid size), and shape and form sex differences were also detected. PERMANOVA also shows shape differences between males and females in the late modern identified Portuguese sample, but not in the Late Pleistocene Jebel Sahaba. In form space (which includes size) sex differences were found in both populations by the PERMANOVA. Despite these results, plotting of PC1 and PC2 in shape space shows overlapping of sexes in both populations but separation in form space. This suggests that sex differences found in lower dimensional space are mainly driven by size and that shape differences are found in higher dimensional space and in PCs which account for smaller proportions of morphological variance. This interpretation is supported by regression of PC scores against centroid size. This analysis shows that these morphological variables are (expectably) significantly related in form space in lower dimensions, and significant relationships in shape space are found only in higher dimensions. Thus, most of the sexual dimorphism we found in these samples is due to isometric size differences between sexes.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003e3.2 Machine Learning\u003c/h2\u003e\u003cp\u003eML models were trained (with supervision) using 11 different algorithms, different sets of PCs derived from both shape and form space, and were first assessed with a holdout testing sample (all using the late modern identified Portuguese sample). Overall, models performed best using the PC scores accounting for 90% of the total variance and classified sex more accurately in form than in shape space. Indeed, when using 90% of the total variance, shape-based sex estimation accuracy averaged 81%, whereas form-based sex estimation averaged 90% (see details above). These intra-population accuracies are typically higher than those reported by most studies estimating sex based on mandible morphology. Most studies report accuracies ranging from ~\u0026thinsp;60% to ~\u0026thinsp;85%\u003csup\u003e78,79,81\u0026ndash;85,90\u0026minus;92,94,97\u003c/sup\u003e, with only a small number reporting accuracies of ~\u0026thinsp;90% or more\u003csup\u003e\u003cspan citationid=\"CR77\" class=\"CitationRef\"\u003e77\u003c/span\u003e,\u003cspan citationid=\"CR87\" class=\"CitationRef\"\u003e87\u003c/span\u003e\u003c/sup\u003e. Further, the use of these methods mitigates inter-observer subjectivity of morphoscopic scoring\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e,\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e,\u003cspan additionalcitationids=\"CR113 CR114\" citationid=\"CR112\" class=\"CitationRef\"\u003e112\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR115\" class=\"CitationRef\"\u003e115\u003c/span\u003e\u003c/sup\u003e and automates sex estimation, thus providing potentially less subjective and more reliable classifications.\u003c/p\u003e\u003cp\u003eConsistent with the Luis Lopes collection test sample, accuracy of sex classification of the Jebel Sahaba mandibles was higher using form than shape space derived data. However, the difference in the performance of shape and form-based models using the archaeological specimens was small. Further, the accuracy of sex classification was also much lower, averaging only 60.10% in shape and 63.13% in form space derived models (see details above). Notwithstanding, some models provided accuracies as high as 83.33% in form (SVMr) and 77.78% in shape space (PLS). The generally low accuracies in the latter space result from generally balanced misclassifications in both sexes. In contrast, females are more frequently misclassified as males in form space. Mandibles in the Jebel Sahaba sample (which have been described previously as highly robust; see above) are significantly larger than in the identified Luis Lopes collection, with females from the former presenting comparable size to the males from the latter. Thus, these misclassifications in form space are likely driven by inter-population size differences.\u003c/p\u003e\u003cp\u003eConsistent with previous studies, these results highlight how inter-population morphological differences impact sex estimation and may lead to potentially biased results when inadequate reference samples or methods are chosen to classify biologically distant target samples\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e,\u003cspan additionalcitationids=\"CR100 CR101 CR102 CR103\" citationid=\"CR99\" class=\"CitationRef\"\u003e99\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR104\" class=\"CitationRef\"\u003e104\u003c/span\u003e,\u003cspan citationid=\"CR116\" class=\"CitationRef\"\u003e116\u003c/span\u003e,\u003cspan citationid=\"CR117\" class=\"CitationRef\"\u003e117\u003c/span\u003e\u003c/sup\u003e. Such differences may arise due to differences in, e.g., population history\u003csup\u003e\u003cspan additionalcitationids=\"CR119 CR120\" citationid=\"CR118\" class=\"CitationRef\"\u003e118\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR121\" class=\"CitationRef\"\u003e121\u003c/span\u003e\u003c/sup\u003e, climate\u003csup\u003e\u003cspan citationid=\"CR119\" class=\"CitationRef\"\u003e119\u003c/span\u003e,\u003cspan citationid=\"CR120\" class=\"CitationRef\"\u003e120\u003c/span\u003e\u003c/sup\u003e, the mechanical demands of the masticatory system\u003csup\u003e\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e,\u003cspan citationid=\"CR71\" class=\"CitationRef\"\u003e71\u003c/span\u003e,\u003cspan citationid=\"CR74\" class=\"CitationRef\"\u003e74\u003c/span\u003e,\u003cspan citationid=\"CR75\" class=\"CitationRef\"\u003e75\u003c/span\u003e,\u003cspan citationid=\"CR122\" class=\"CitationRef\"\u003e122\u003c/span\u003e\u003c/sup\u003e and nutrition\u003csup\u003e\u003cspan citationid=\"CR123\" class=\"CitationRef\"\u003e123\u003c/span\u003e\u003c/sup\u003e. These may lead to overall interpopulation differences in size and/or shape that, in turn, impact the patterns of sexual dimorphism. Indeed, the expression of sexual dimorphism varies across populations, with contrasting patterns of robusticity in both males and females potentially leading to incorrect sex estimations when inadequate reference samples are used or estimation methods are not adjusted accordingly\u003csup\u003e\u003cspan citationid=\"CR100\" class=\"CitationRef\"\u003e100\u003c/span\u003e,\u003cspan citationid=\"CR117\" class=\"CitationRef\"\u003e117\u003c/span\u003e\u003c/sup\u003e. This has been hypothesized to drive the over-representation of one sex over the other in the examination of, e.g., sex ratios in past populations\u003csup\u003e\u003cspan citationid=\"CR124\" class=\"CitationRef\"\u003e124\u003c/span\u003e,\u003cspan citationid=\"CR125\" class=\"CitationRef\"\u003e125\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec8\" class=\"Section2\"\u003e\u003ch2\u003e3.3 Limitations\u003c/h2\u003e\u003cp\u003eOverall, this study shows that the clear sexual dimorphism in mandibular morphology can be used to estimate sex reliably within populations. However, sex classification of specimens from other populations is more challenging with meaningfully lower accuracies. This may result from several limitations of this study that will be tackled in future studies.\u003c/p\u003e\u003cp\u003eOne of the limitations is that this study only includes one reference identified population from late modern Portugal. To classify individuals from other geographies and/or chronologies it would be very relevant to include other reference populations and so account for inter-population morphological differences. This would be particularly relevant in target samples as morphologically distinct as the Late Pleistocene Jebel Sahaba, which has been previously described as likely isolated from other populations, being particularly robust and with very large teeth (see above). Moreover, it would also be potentially relevant to increase sample size of the reference population(s) to provide a more comprehensive depiction of intra-population and sex specific morphological variance.\u003c/p\u003e\u003cp\u003eConversely, this study only uses two testing samples: (i) the intra-population holdout Luis Lopes and the (ii) inter-population Late Pleistocene Jebel Sahaba samples. This choice was deliberate to enable bracketing the reliability of this sex estimation approach by using (held out) specimens from the same population used to train the ML models and an extremely morphologically distinct archaeological testing sample. This results in lower reliability results in the latter sample. Notwithstanding, we predict that ensuing studies using archaeological testing samples biologically closer to the reference sample (e.g., medieval or modern age samples originating from Portugal) will result in better results than those obtained for Jebel Sahaba.\u003c/p\u003e\u003cp\u003eDespite the very encouraging intra-population results using the LM dataset adopted in this study, no semi LMs were used. The use of the latter would provide a more detailed morphological representation of the specimens and of specific anatomical regions known to show significant sexual dimorphism (e.g., chin, gonial angle, posterior ramus). This may enable better sex classification accuracies. However, the application of such LMs to archaeological specimens will be challenging frequently. This is because archaeological specimens are often fragmented, precluding the use of dense landmarking protocols.\u003c/p\u003e\u003cp\u003eThe sex of the specimens from the Late Pleistocene Jebel Sahaba sample was previously estimated using standard multi-factorial anthropological methods, including pelvis and skull-based sex estimation\u003csup\u003e\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e76\u003c/span\u003e\u003c/sup\u003e, and this classification was used as reference to assess the reliability of the ML predictions. Although pelvis and skull-based sex estimation typically provides very reliable results, we cannot exclude the possibility that some of the more incomplete individuals may have been misclassified also due to the inevitable absence of reference populations similar to the Late Pleistocene Jebel Sahaba. Thus, the use of modern reference data may drive misclassification in targeting biologically and morphologically distinct populations\u003csup\u003e\u003cspan citationid=\"CR125\" class=\"CitationRef\"\u003e125\u003c/span\u003e\u003c/sup\u003e. If this was the case, this would impact our results. To overcome this limitation, future studies will include archaeological samples for which sex is estimated independently using biomolecular methods (i.e., aDNA or palaeoproteomics). This powerful approach will allow sex determination of the archaeological individuals and ensuing creation of archaeological reference samples. These will account for geographical and temporal differences and enable the development of archaeological population specific morphological sexing methods.\u003c/p\u003e\u003c/div\u003e"},{"header":"4 Materials and methods","content":"\u003cp\u003eThis study is based on 85 adult mandibles (Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Sixty-seven originate from the Lisbon Lu\u0026iacute;s Lopes identified skeletal collection\u003csup\u003e\u003cspan citationid=\"CR126\" class=\"CitationRef\"\u003e126\u003c/span\u003e\u003c/sup\u003e and the remaining 18 from the Jebel Sahaba Late Pleistocene sample\u003csup\u003e\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e76\u003c/span\u003e\u003c/sup\u003e. The selected specimens were surface scanned to enable GM based morphological analysis. To that end, landmark coordinates were extracted from the generated meshes for ensuing use in standard GM analysis. GM outputs were then used in supervised ML to create models that enable the mandibular-based prediction of sex in the archaeological Jebel Sahaba sample.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eSample composition of the individuals selected from the collections used in this study.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u0026nbsp;\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eFemale\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eMale\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eTotal\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLuis Lopes (Portugal)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e30\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e37\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e67\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eJebel Sahaba (Sudan)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e8\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e10\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e18\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eTotal\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e\u003cp\u003e38\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e47\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e85\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cdiv id=\"Sec10\" class=\"Section2\"\u003e\u003ch2\u003e4.1 Specimen selection\u003c/h2\u003e\u003cp\u003eOnly individuals no younger than ~\u0026thinsp;18 years of estimated age were used in this study. This is because mandibular morphology diverges between males and females during puberty, and so younger individuals do not present meaningful sex-related morphological differences\u003csup\u003e\u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e127\u003c/span\u003e\u003c/sup\u003e. Further, growth and development induce major morphological changes that are not of interest in this study and that would obscure the sex-related morphological differences\u003csup\u003e\u003cspan citationid=\"CR127\" class=\"CitationRef\"\u003e127\u003c/span\u003e\u003c/sup\u003e. Age assessment of the individuals (i.e., non-adult vs. adult) from the identified collection was based on the records of the individuals and validated via scoring of dental development and eruption sequences\u003csup\u003e\u003cspan citationid=\"CR128\" class=\"CitationRef\"\u003e128\u003c/span\u003e\u003c/sup\u003e. This was particularly suitable as completion of dental development plays a role in the final shape and size of the mandible. The sex identified on the individual records of the Luis Lopes collection (curated by the co-author SG) was used for the GM analysis of sexual morphological differences and in the validation of the ML predictions. The age at death (i.e., non-adult vs. adult) of the archaeological individuals from Jebel Sahaba was estimated via direct observation of dental eruption sequences\u003csup\u003e\u003cspan citationid=\"CR128\" class=\"CitationRef\"\u003e128\u003c/span\u003e\u003c/sup\u003e. The sex of these individuals was previously estimated by Crevecoeur, et al. \u003csup\u003e76\u003c/sup\u003e based on the observation of the hip bones (Bruzek\u003csup\u003e\u003cspan citationid=\"CR129\" class=\"CitationRef\"\u003e129\u003c/span\u003e\u003c/sup\u003e, Murail et al.\u003csup\u003e130\u003c/sup\u003e and Bruzek et al.\u003csup\u003e31\u003c/sup\u003e) and the skull (Buikstra and Ubelaker\u003csup\u003e\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e), with estimation reliability reported therein. These estimations by Crevecoeur, et al. \u003csup\u003e76\u003c/sup\u003e were used on the GM based examination of sex related morphological differences and on the ML predictions of sex.\u003c/p\u003e\u003cp\u003eSelection was also restricted to specimens that were fully, or almost fully, complete for the landmarking protocol used in this study. This option was favoured because estimation of the original location of the anatomical LMs introduces some degree of error and, so, noise to the analysis (see details below). Thus, 53/67 (79.1%) mandibles presented no missing LMs, and only the remaining 14/67 (20.9%) presented one (9/67; 13.4%) or two (5/67; 7.5%) missing LMs in the Lu\u0026iacute;s Lopes identified sample (Table SI 18). In the Late Pleistocene archaeological sample from current Sudan, 4/18 (22,2%) mandibles were complete, 5/18 (27,8%) had one missing LM, 1/18 (5,6%) 2, 5/18 (27,8%) 3, 2/18 (11,1%) 4 and 1/18 (5,6%) 5 missing LMs (Table SI 18). This strict selection criteria resulted in the exclusion of several individuals due to fragmentation or oral pathologies impacting morphology (e.g., extensive ante-mortem tooth loss).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003e4.2 Digitization and GM based morphological analysis\u003c/h2\u003e\u003cp\u003eAfter selection, specimens were digitised using an Einscan Pro 2X Plus structured light surface scanner. Points clouds were converted into meshes using the EXScan Pro software, which were then used to collect coordinates from a set of 21 anatomical landmarks per specimen (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e and Table SI 19). The LM coordinates were collected in the open-source 3D Slicer software\u003csup\u003e\u003cspan citationid=\"CR131\" class=\"CitationRef\"\u003e131\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe coordinates were then imported into R and the packages Geomorph\u003csup\u003e\u003cspan citationid=\"CR132\" class=\"CitationRef\"\u003e132\u003c/span\u003e\u003c/sup\u003e and Morpho\u003csup\u003e\u003cspan citationid=\"CR133\" class=\"CitationRef\"\u003e133\u003c/span\u003e\u003c/sup\u003e, were used for ensuing reconstruction of incomplete specimens and GM analysis. The original location of missing anatomical landmarks was estimated using the \u003cem\u003eestimate.missing\u003c/em\u003e Thin Plate Splines (TPS) based function of Geomorph. TPS based reconstruction provides reliable predictions of the original morphology but, nonetheless, introduces some errors. The magnitude of the errors relates to which and how many LMs are being reconstructed\u003csup\u003e\u003cspan citationid=\"CR134\" class=\"CitationRef\"\u003e134\u003c/span\u003e\u003c/sup\u003e. While we restricted the maximum number of missing LMs to 5, most of the selected specimens displayed the complete set of (or lacked up to a maximum of two) LMs (Table SI 18). Previous studies have also shown that significant estimation errors may emerge when inadequate specimens are selected as reference\u003csup\u003e\u003cspan additionalcitationids=\"CR136\" citationid=\"CR135\" class=\"CitationRef\"\u003e135\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR137\" class=\"CitationRef\"\u003e137\u003c/span\u003e\u003c/sup\u003e. Although we found no significant error differences when selecting reference specimens from different modern human populations in another study\u003csup\u003e\u003cspan citationid=\"CR70\" class=\"CitationRef\"\u003e70\u003c/span\u003e\u003c/sup\u003e, we opted for population specific missing data estimation. (i.e., incomplete specimens from the Portuguese sample were reconstructed based on complete specimens from Portugal; incomplete specimens from Jebel Sahaba were reconstructed based on complete specimens from the same population).\u003c/p\u003e\u003cp\u003eTwo spaces were used in the GM analysis: shape and form. In shape space, scaling removed (isometric) size differences\u003csup\u003e\u003cspan additionalcitationids=\"CR139\" citationid=\"CR138\" class=\"CitationRef\"\u003e138\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR140\" class=\"CitationRef\"\u003e140\u003c/span\u003e\u003c/sup\u003e. In form space, (isometric) size differences were first removed via scaling and were then re-introduced via inclusion of log centroid size\u003csup\u003e\u003cspan citationid=\"CR138\" class=\"CitationRef\"\u003e138\u003c/span\u003e,\u003cspan citationid=\"CR141\" class=\"CitationRef\"\u003e141\u003c/span\u003e\u003c/sup\u003e. Differences in size between the sexes were examined using Centroid Size (CS). Principal Component Analysis (PCA) ensued to examine shape and form differences between males and females, which were visualized together with TPS warpings along the relevant principal components (PCs). Shape and form PC scores were also regressed against size (i.e., centroid size) to examine the relationship between these variables for the whole pooled sample and in each individual population.\u003c/p\u003e\u003cp\u003eA Kruskal Wallis test, ensued by post-hoc tests, was used to examine hypothetical differences in size (as assessed via centroid size) between sexes and populations. PERMANOVA was performed in Past\u003csup\u003e\u003cspan citationid=\"CR142\" class=\"CitationRef\"\u003e142\u003c/span\u003e\u003c/sup\u003e to test for shape and form statistical differences between males and females using the PC scores of the first PCs explaining\u0026thinsp;~\u0026thinsp;95% of the total variance.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003e4.3 Machine Learning\u003c/h2\u003e\u003cp\u003ePrincipal components scores derived from shape and form analyses were subjected to ML methods. Initially, the late modern Luis Lopes identified sample was split into two sets: training (70%) and testing (30%), to assess model reliability. Analyses were conducted in shape and form space across four rounds: using the full variance, PC scores accounting for 95% of total variance, PC scores accounting for 90%, and PCs identified as significant by permutations with the GraphGMM library\u003csup\u003e\u003cspan citationid=\"CR143\" class=\"CitationRef\"\u003e143\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eML models require large sample sizes for effective training, so artificial methods to increase sample size are commonly used in archaeology, paleontology, and anatomy\u003csup\u003e\u003cspan additionalcitationids=\"CR145\" citationid=\"CR144\" class=\"CitationRef\"\u003e144\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR146\" class=\"CitationRef\"\u003e146\u003c/span\u003e\u003c/sup\u003e, as these fields are often limited by access to reference collections, as is the case in the present study. However, no artificial sample-increasing methods were used in this study. Bootstrapping was avoided, as it merely duplicates known data multiple times, potentially leading to overfitting during model training\u003csup\u003e\u003cspan citationid=\"CR146\" class=\"CitationRef\"\u003e146\u003c/span\u003e,\u003cspan citationid=\"CR147\" class=\"CitationRef\"\u003e147\u003c/span\u003e\u003c/sup\u003e. Similarly, given that our sample is limited to a specific population, applying generative adversarial networks to generate artificial data within the morphological variance range was deemed unsuitable\u003csup\u003e\u003cspan citationid=\"CR146\" class=\"CitationRef\"\u003e146\u003c/span\u003e\u003c/sup\u003e. Therefore, we proceeded with only the collected sample, with the aim that additional sexed osteological collections from different geographical areas and chronologies will eventually be incorporated into the training set in future studies.\u003c/p\u003e\u003cp\u003eAmong ML approaches, supervised learning methods were preferred, requiring a pre-classified dataset with built-in learning and self-control mechanisms. To mitigate potential overfitting, self-correcting techniques such as k-fold cross-validation were employed. The original sample was partitioned into 10 sets to generate \u0026ldquo;submodels\u0026rdquo; with performance across these submodels assessing overall model efficiency.\u003c/p\u003e\u003cp\u003eThe sexed modern osteological Luis Lopes collection was used to train the models, which were then applied to classify Late Pleistocene mandibles as male or female. Eleven algorithms were used in this study (), including k-Nearest Neighbour (kNN), Logistic Regression (LGR), Decision Trees (DTC5.0), Random Forest (RF), Gradient Boosting (GB), Na\u0026iuml;ve Bayes (NB), Linear Discriminant Analysis (LDA), Partial Least Squares (PLS), Linear and Radial Support Vector Machines (SVMl and SVMr), and Neural Networks (NNET). All algorithms were trained using the \u0026lsquo;caret\u0026rsquo;\u003csup\u003e148\u003c/sup\u003e and \u0026lsquo;caretEnsemble\u0026rsquo; \u003csup\u003e149\u003c/sup\u003e R libraries. Model tuning, essential for classification accuracy, was conducted using hyperparameter grids to test various parameter values, with optimal values selected for each algorithm. The \u0026lsquo;tuneLength\u0026rsquo; function in these libraries enabled hyperparameter configuration by generating 20 models per algorithm, with accuracy and Kappa values guiding optimal selection.\u003c/p\u003e\u003cp\u003eDifferences in prediction performance and classification rates among models were compared considering a set of factors, including kappa, sensitivity, specificity and balanced accuracy values. Kappa statistics account for chance in predictions, with values ranging from \u0026minus;\u0026thinsp;1 to 1 (kappa\u0026thinsp;\u0026gt;\u0026thinsp;0.8 signals high predictive power)\u003csup\u003e\u003cspan citationid=\"CR150\" class=\"CitationRef\"\u003e150\u003c/span\u003e\u003c/sup\u003e. Sensitivity and specificity measure opposite rates: true positive versus true negative classifications\u003csup\u003e\u003cspan citationid=\"CR150\" class=\"CitationRef\"\u003e150\u003c/span\u003e\u003c/sup\u003e. These values are balanced through averaging, resulting in a model efficiency score from 0 to 1\u003csup\u003e151\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eClassifications for the archaeological sample were compared across models and against classifications from traditional anthropological methods based on all available skeletal elements of each individual\u003csup\u003e\u003cspan citationid=\"CR76\" class=\"CitationRef\"\u003e76\u003c/span\u003e\u003c/sup\u003e. Additionally, posterior probability values (p) were calculated for each individual to determine group membership reliability, with values exceeding 0.9 typically regarded as strong identifications.\u003c/p\u003e\u003c/div\u003e"},{"header":"Declarations","content":"\u003cp\u003eAcknowledgements\u003c/p\u003e\n\u003cp\u003eRM Godinho is funded by Funda\u0026ccedil;\u0026atilde;o para a Ci\u0026ecirc;ncia e a Tecnologia (FCT; 2023.10993.TENURE.006). This research was also funded by the FCT R\u0026amp;D research project \u0026ldquo;ParaFunction\u0026rdquo; (project reference 2022.07737.PTDC; https://doi.org/10.54499/2022.07737.PTDC). The British Museum for granting access to the Jebel Sahaba collection. The Portuguese Museu Nacional de Hist\u0026oacute;ria Natural e da Ci\u0026ecirc;ncia (MUHNAC) for granting access to the Luis Lopes collection. J Aramendi is funded by the British Academy (NIF22\\220310). The anthropological reassessment of the Jebel Sahaba collection by IC was supported by the International Research Project (IRP) ABASC founded by the CNRS-INEE; the French government in the framework of the University of Bordeaux\u0026apos;s IdEx \u0026ldquo;Investments for the Future\u0026rdquo; program/GPR \u0026ldquo;Human Past\u0026rdquo; and the French National Research Agency (ANR-14-CE31, project BIG DRY)\u003c/p\u003e\n\u003cp\u003eAuthor contributions\u003c/p\u003e\n\u003cp\u003eAll authors contributed substantially and reviewed the manuscript.\u003c/p\u003e\n\u003cp\u003eData availability statement\u003c/p\u003e\n\u003cp\u003eThe data that support this study is available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003cp\u003ePermission statement\u003c/p\u003e\n\u003cp\u003eAccess to the Lu\u0026iacute;s Lopes identified sample was granted by the housing institution, the Portuguese Museu Nacional de Hist\u0026oacute;ria Natural e da Ci\u0026ecirc;ncia (MUHNAC). Access to the Jebel Sahaba sample was granted by the housing institution, the British Museum. Both access permissions ensued formal access requests to each of these institutions and assessment by the relevant boards. All regulations of each of these institutions were followed.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eKrishan, K.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e A review of sex estimation techniques during examination of skeletal remains in forensic anthropology casework. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e261\u003c/strong\u003e, 165.e161-165.e168 (2016). https://doi.org/https://doi.org/10.1016/j.forsciint.2016.02.007\u003c/li\u003e\n \u003cli\u003eWhite, T. D., Black, M. T. \u0026amp; Folkens, P. A. \u003cem\u003eHuman osteology\u003c/em\u003e. (Academic press, 2011).\u003c/li\u003e\n \u003cli\u003eLaffranchi, Z., Beck De Lotto, M. A., Delpino, C., L\u0026ouml;sch, S. \u0026amp; Milella, M. Social differentiation and well-being in the Italian Iron Age: exploring the relationship between sex, age, biological stress, and burial complexity among the Picenes of Novilara (8th\u0026ndash;7th c. BC). \u003cem\u003eArchaeological and Anthropological Sciences\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 182 (2021). https://doi.org/10.1007/s12520-021-01449-3\u003c/li\u003e\n \u003cli\u003eDong, Y.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Shifting diets and the rise of male-biased inequality on the Central Plains of China during Eastern Zhou. \u003cstrong\u003e114\u003c/strong\u003e, 932-937 (2017). https://doi.org/10.1073/pnas.1611742114 %J Proceedings of the National Academy of Sciences\u003c/li\u003e\n \u003cli\u003eMasclans, A., Duboscq, S., Achino, K. F., Morell, B. \u0026amp; Gibaja, J. F. Looking for sexual differences during the middle Neolithic in the northeast of the Iberian Peninsula. \u003cem\u003eJournal of Archaeological Science: Reports\u003c/em\u003e \u003cstrong\u003e26\u003c/strong\u003e, 101858 (2019). https://doi.org/https://doi.org/10.1016/j.jasrep.2019.05.023\u003c/li\u003e\n \u003cli\u003eRebay-Salisbury, K.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Gendered burial practices of early Bronze Age children align with peptide-based sex identification: A case study from Franzhausen I, Austria. \u003cem\u003eJournal of Archaeological Science\u003c/em\u003e \u003cstrong\u003e139\u003c/strong\u003e, 105549 (2022). https://doi.org/https://doi.org/10.1016/j.jas.2022.105549\u003c/li\u003e\n \u003cli\u003eFern\u0026aacute;ndez-Crespo, T.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Multi-isotope evidence for the emergence of cultural alterity in Late Neolithic Europe. \u003cem\u003eScience Advances\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, eaay2169 (2020). https://doi.org/10.1126/sciadv.aay2169\u003c/li\u003e\n \u003cli\u003eRey, L., Rottier, S., Santos, F. \u0026amp; Goude, G. Sex and age-related social organization in the Neolithic: A promising survey from the Paris Basin. \u003cem\u003eJournal of Archaeological Science: Reports\u003c/em\u003e \u003cstrong\u003e38\u003c/strong\u003e, 103092 (2021). https://doi.org/https://doi.org/10.1016/j.jasrep.2021.103092\u003c/li\u003e\n \u003cli\u003eRey, L.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e A multi-isotope analysis of Neolithic human groups in the Yonne valley, Northern France: insights into dietary patterns and social structure. \u003cem\u003eArchaeological and Anthropological Sciences\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 5591-5616 (2019). https://doi.org/10.1007/s12520-019-00885-6\u003c/li\u003e\n \u003cli\u003eAvery, L. C., Brickley, M. B., Findlay, S., Chapelain de Ser\u0026eacute;ville-Niel, C. \u0026amp; Prowse, T. L. Child and adolescent diet in Late Roman Gaul: An investigation of incremental dietary stable isotopes in tooth dentine. \u003cem\u003eInternational Journal of Osteoarchaeology\u003c/em\u003e \u003cstrong\u003e31\u003c/strong\u003e, 1226-1236 (2021). https://doi.org/https://doi.org/10.1002/oa.3033\u003c/li\u003e\n \u003cli\u003eMiller, M. J., Dong, Y., Pechenkina, K., Fan, W. \u0026amp; Halcrow, S. E. Raising girls and boys in early China: Stable isotope data reveal sex differences in weaning and childhood diets during the eastern Zhou era. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e172\u003c/strong\u003e, 567-585 (2020). https://doi.org/https://doi.org/10.1002/ajpa.24033\u003c/li\u003e\n \u003cli\u003eLieverse, A. R., Stock, J. T., Katzenberg, M. A. \u0026amp; Haverkort, C. M. The bioarchaeology of habitual activity and dietary change in the Siberian Middle Holocene. \u003cem\u003eHuman bioarchaeology of the transition to agriculture\u003c/em\u003e, 263-291 (2011).\u003c/li\u003e\n \u003cli\u003eRuff, C. B.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Gradual decline in mobility with the adoption of food production in Europe. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e \u003cstrong\u003e112\u003c/strong\u003e, 7147 (2015).\u003c/li\u003e\n \u003cli\u003eRuff, C. Sexual dimorphism in human lower limb bone structure: relationship to subsistence strategy and sexual division of labor. \u003cem\u003eJournal of Human Evolution\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 391-416 (1987). https://doi.org/http://dx.doi.org/10.1016/0047-2484(87)90069-8\u003c/li\u003e\n \u003cli\u003eBonsall, L. A comparison of female and male oral health in skeletal populations from late Roman Britain: Implications for diet. \u003cem\u003eArchives of Oral Biology\u003c/em\u003e \u003cstrong\u003e59\u003c/strong\u003e, 1279-1300 (2014). https://doi.org/https://doi.org/10.1016/j.archoralbio.2014.07.019\u003c/li\u003e\n \u003cli\u003eLukacs, J. R. Sex Differences in Dental Caries Rates With the Origin of Agriculture in South Asia. \u003cem\u003eCurrent Anthropology\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 147-153 (1996). https://doi.org/doi:10.1086/204481\u003c/li\u003e\n \u003cli\u003eMilner, G. R. \u0026amp; Boldsen, J. L. in \u003cem\u003eA Companion to Paleopathology\u003c/em\u003e 268-284 (2011).\u003c/li\u003e\n \u003cli\u003eLugli, F.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Enamel peptides reveal the sex of the Late Antique \u0026lsquo;Lovers of Modena\u0026rsquo;. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 13130 (2019). https://doi.org/10.1038/s41598-019-49562-7\u003c/li\u003e\n \u003cli\u003eGranja, R.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Unbalanced sex-ratio in the Neolithic individuals from the Escoural Cave (Montemor-o-Novo, Portugal) revealed by peptide analysis. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 19902 (2023). https://doi.org/10.1038/s41598-023-47037-4\u003c/li\u003e\n \u003cli\u003eBuonasera, T.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e A comparison of proteomic, genomic, and osteological methods of archaeological sex estimation. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 11897 (2020). https://doi.org/10.1038/s41598-020-68550-w\u003c/li\u003e\n \u003cli\u003eCintas-Pe\u0026ntilde;a, M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Amelogenin peptide analyses reveal female leadership in Copper Age Iberia (c. 2900\u0026ndash;2650 BC). \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 9594 (2023). https://doi.org/10.1038/s41598-023-36368-x\u003c/li\u003e\n \u003cli\u003eFroment, C.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Analysis of 5000 year-old human teeth using optimized large-scale and targeted proteomics approaches for detection of sex-specific peptides. \u003cem\u003eJournal of Proteomics\u003c/em\u003e \u003cstrong\u003e211\u003c/strong\u003e, 103548 (2020). https://doi.org/https://doi.org/10.1016/j.jprot.2019.103548\u003c/li\u003e\n \u003cli\u003eGasparini, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Biological sex VS. Archaeological Gender: Enamel peptide analysis of the horsemen of the Early Middle age necropolises of Campochiaro (Molise, Italy). \u003cem\u003eJournal of Archaeological Science: Reports\u003c/em\u003e \u003cstrong\u003e41\u003c/strong\u003e, 103337 (2022). https://doi.org/https://doi.org/10.1016/j.jasrep.2021.103337\u003c/li\u003e\n \u003cli\u003eParker, G. J.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Sex estimation using sexually dimorphic amelogenin protein fragments in human enamel. \u003cem\u003eJournal of Archaeological Science\u003c/em\u003e (2018). https://doi.org/https://doi.org/10.1016/j.jas.2018.08.011\u003c/li\u003e\n \u003cli\u003eSkoglund, P., Stor\u0026aring;, J., G\u0026ouml;therstr\u0026ouml;m, A. \u0026amp; Jakobsson, M. Accurate sex identification of ancient human remains using DNA shotgun sequencing. \u003cem\u003eJournal of Archaeological Science\u003c/em\u003e \u003cstrong\u003e40\u003c/strong\u003e, 4477-4482 (2013). https://doi.org/https://doi.org/10.1016/j.jas.2013.07.004\u003c/li\u003e\n \u003cli\u003eLoreille, O.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Biological Sexing of a 4000-Year-Old Egyptian Mummy Head to Assess the Potential of Nuclear DNA Recovery from the Most Damaged and Limited Forensic Specimens. \u003cem\u003eGenes\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 135 (2018).\u003c/li\u003e\n \u003cli\u003eD\u0026iacute;az-Zorita Bonilla, M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Female sex bias in Iberian megalithic societies through bioarchaeology, aDNA and proteomics. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 21818 (2024). https://doi.org/10.1038/s41598-024-72148-x\u003c/li\u003e\n \u003cli\u003eChristensen, A. M., Passalacqua, N. V. \u0026amp; Bartelink, E. J. \u003cem\u003eForensic anthropology: current methods and practice\u003c/em\u003e. (Academic Press, 2019).\u003c/li\u003e\n \u003cli\u003eLuo, Y.-C. Sex determination from the pubis by discriminant function analysis. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e74\u003c/strong\u003e, 89-98 (1995). https://doi.org/https://doi.org/10.1016/0379-0738(95)01739-6\u003c/li\u003e\n \u003cli\u003eMeindl, R. S., Lovejoy, C. O., Mensforth, R. P. \u0026amp; Carlos, L. D. Accuracy and direction of error in the sexing of the skeleton: Implications for paleodemography. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e68\u003c/strong\u003e, 79-85 (1985). https://doi.org/https://doi.org/10.1002/ajpa.1330680108\u003c/li\u003e\n \u003cli\u003eBrůžek, J., Santos, F., Dutailly, B., Murail, P. \u0026amp; Cunha, E. Validation and reliability of the sex estimation of the human os coxae using freely available DSP2 software for bioarchaeology and forensic anthropology. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e164\u003c/strong\u003e, 440-449 (2017). https://doi.org/https://doi.org/10.1002/ajpa.23282\u003c/li\u003e\n \u003cli\u003eBruzek, J. \u0026amp; Murail, P. in \u003cem\u003eForensic Anthropology and Medicine: Complementary Sciences From Recovery to Cause of Death\u003c/em\u003e (eds Aurore Schmitt, Eug\u0026eacute;nia Cunha, \u0026amp; Jo\u0026atilde;o Pinheiro) 225-242 (Humana Press, 2006).\u003c/li\u003e\n \u003cli\u003eSpradley, M. K. \u0026amp; Jantz, R. L. Sex Estimation in Forensic Anthropology: Skull Versus Postcranial Elements. \u003cem\u003eJournal of Forensic Sciences\u003c/em\u003e \u003cstrong\u003e56\u003c/strong\u003e, 289-296 (2011). https://doi.org/10.1111/j.1556-4029.2010.01635.x\u003c/li\u003e\n \u003cli\u003eGodinho, R. M., Gon\u0026ccedil;alves, D. \u0026amp; Valera, A. C. The preburning condition of Chalcolithic cremated human remains from the Perdig\u0026otilde;es enclosures (Portugal). \u003cem\u003eInternational Journal of Osteoarchaeology\u003c/em\u003e \u003cstrong\u003e29\u003c/strong\u003e, 706-717 (2019). https://doi.org/10.1002/oa.2768\u003c/li\u003e\n \u003cli\u003eValera, A. C., Evangelista, L. \u0026amp; Godinho, R. M. in \u003cem\u003eMegalithic Societies: Old Questions, New Narratives\u003c/em\u003e (eds G. Higginbottom\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e) (Archaeopress, In press).\u003c/li\u003e\n \u003cli\u003eValera, A. C., Silva, A. M., Leandro, I., Godinho, R. M. \u0026amp; Evangelista, L. S. in \u003cem\u003eOs Perdig\u0026otilde;es Neol\u0026iacute;ticos. G\u0026eacute;nese e desenvolvimento (de meados do 4\u0026ordm; aos in\u0026iacute;cios do 3\u0026ordm; mil\u0026eacute;nio AC)\u003c/em\u003e (ed Ant\u0026oacute;nio Carlos Valera) 155-162 (N\u0026uacute;cleo de Investiga\u0026ccedil;\u0026atilde;o Arqueol\u0026oacute;gica (NIA), 2018).\u003c/li\u003e\n \u003cli\u003eEvangelista, L. S. \u0026amp; Godinho, R. M. in \u003cem\u003eO Sepulcro 4 dos Perdig\u0026otilde;es. Um Tholos da Segunda Metade do 3\u0026ordm; Mil\u0026eacute;nio A.C.\u003c/em\u003e (ed Ant\u0026oacute;nio Carlos Valera) 57-102 (N\u0026uacute;cleo de Investiga\u0026ccedil;\u0026atilde;o Arqueol\u0026oacute;gica (NIA) \u0026amp; Era Arqueologia, S.A., 2020).\u003c/li\u003e\n \u003cli\u003eSianto, L.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Evidence of contact between New and Old World: paleoparasitological and food remains study in the Tagus river population of Sarilhos Grandes (Montijo, Portugal). \u003cem\u003eArchaeological and Anthropological Sciences\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 75-81 (2018). https://doi.org/10.1007/s12520-016-0337-9\u003c/li\u003e\n \u003cli\u003eFilipe, V., Godinho, R. M., Granja, R., Ribeiro, A. \u0026amp; Valera, A. Bronze age funerary spaces in Outeiro Alto 2 (Brinches, Serpa, Portugal): the hypogea cemetery. \u003cem\u003eZephyrus\u003c/em\u003e \u003cstrong\u003e71\u003c/strong\u003e, 107-129 (2013).\u003c/li\u003e\n \u003cli\u003eValera, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e in \u003cem\u003e4\u0026ordm; Col\u0026oacute;quio de Arqueologia do Alqueva - O Plano de Rega (2002-2010).\u003c/em\u003e (eds Ant\u0026oacute;nio Carlos Silva, Frederico T\u0026aacute;t\u0026aacute; Regala, \u0026amp; Miguel Martinho) 55-73 (EDIA).\u003c/li\u003e\n \u003cli\u003eGodinho, R. Deposi\u0026ccedil;\u0026otilde;es funer\u0026aacute;rias em fossa nos Perdig\u0026otilde;es: dados antropol\u0026oacute;gicos do sector I. \u003cem\u003eApontamentos de Arqueologia e Patrim\u0026oacute;nio\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 29-34 (2008).\u003c/li\u003e\n \u003cli\u003eMiguel, L. \u0026amp; Godinho, R. M. Not\u0026iacute;cia do s\u0026iacute;tio arqueol\u0026oacute;gico do Monte das Covas 3 (Beja). \u003cem\u003eApontamentos de Arqueologia e Patrim\u0026oacute;nio\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 23-24 (2009).\u003c/li\u003e\n \u003cli\u003eIntrona, F., Di Vella, G. \u0026amp; Campobasso, C. P. Sex determination by discriminant analysis of patella measurements. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e95\u003c/strong\u003e, 39-45 (1998). https://doi.org/https://doi.org/10.1016/S0379-0738(98)00080-2\u003c/li\u003e\n \u003cli\u003eWalker, P. L. Sexing skulls using discriminant function analysis of visually assessed traits. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e136\u003c/strong\u003e, 39-50 (2008). https://doi.org/10.1002/ajpa.20776\u003c/li\u003e\n \u003cli\u003eWalrath, D. E., Turner, P. \u0026amp; Bruzek, J. Reliability test of the visual assessment of cranial traits for sex determination. \u003cem\u003eAmerican Journal Of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e125\u003c/strong\u003e, 132-137 (2004). https://doi.org/10.1002/ajpa.10373\u003c/li\u003e\n \u003cli\u003eBuikstra, J. \u0026amp; Ubelaker, D. \u003cem\u003eStandards for Data Collection from Human Skeletal Remains: Proceedings of a Seminar at the Field Museum of Natural History\u003c/em\u003e. (Arkansas Archeological Survey, 1994).\u003c/li\u003e\n \u003cli\u003eFerembach, D., Schwidetzky, I. \u0026amp; Stloukal, M. Recommendations for age and sex diagnoses of skeletons. \u003cem\u003eJournal of Human Evolution\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 517-549 (1980). https://doi.org/https://doi.org/10.1016/0047-2484(80)90061-5\u003c/li\u003e\n \u003cli\u003eSantos, F., Guyomarc\u0026rsquo;h, P. \u0026amp; Bruzek, J. Statistical sex determination from craniometrics: Comparison of linear discriminant analysis, logistic regression, and support vector machines. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e245\u003c/strong\u003e, 204.e201-204.e208 (2014). https://doi.org/https://doi.org/10.1016/j.forsciint.2014.10.010\u003c/li\u003e\n \u003cli\u003eTechataweewan, N.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Metric sexual dimorphism of the skull in Thailand. \u003cem\u003eForensic Science International: Reports\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, 100236 (2021). https://doi.org/https://doi.org/10.1016/j.fsir.2021.100236\u003c/li\u003e\n \u003cli\u003eColman, K. L.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e The accuracy of 3D virtual bone models of the pelvis for morphological sex estimation. \u003cem\u003eInternational Journal of Legal Medicine\u003c/em\u003e \u003cstrong\u003e133\u003c/strong\u003e, 1853-1860 (2019). https://doi.org/10.1007/s00414-019-02002-7\u003c/li\u003e\n \u003cli\u003eSteyn, M., Pretorius, E. \u0026amp; Hutten, L. Geometric morphometric analysis of the greater sciatic notch in South Africans. \u003cem\u003eHOMO - Journal of Comparative Human Biology\u003c/em\u003e \u003cstrong\u003e54\u003c/strong\u003e, 197-206 (2004). https://doi.org/http://dx.doi.org/10.1078/0018-442X-00076\u003c/li\u003e\n \u003cli\u003eBytheway, J. A. \u0026amp; Ross, A. H. A Geometric Morphometric Approach to Sex Determination of the Human Adult Os Coxa. \u003cem\u003eJournal of Forensic Sciences\u003c/em\u003e \u003cstrong\u003e55\u003c/strong\u003e, 859-864 (2010). https://doi.org/https://doi.org/10.1111/j.1556-4029.2010.01374.x\u003c/li\u003e\n \u003cli\u003eBilfeld, M. F.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Human Coxal Bone Sexual Dimorphism and Multislice Computed Tomography: Geometric Morphometric Analysis of 65 Adults. \u003cem\u003eJournal of Forensic Sciences\u003c/em\u003e \u003cstrong\u003e57\u003c/strong\u003e, 578-588 (2012). https://doi.org/https://doi.org/10.1111/j.1556-4029.2011.02009.x\u003c/li\u003e\n \u003cli\u003eKimmerle, E. H., Ross, A. \u0026amp; Slice, D. Sexual Dimorphism in America: Geometric Morphometric Analysis of the Craniofacial Region. \u003cem\u003eJournal of Forensic Sciences\u003c/em\u003e \u003cstrong\u003e53\u003c/strong\u003e, 54-57 (2008). https://doi.org/https://doi.org/10.1111/j.1556-4029.2007.00627.x\u003c/li\u003e\n \u003cli\u003eRosas, A. \u0026amp; Bastir, M. Thin-plate spline analysis of allometry and sexual dimorphism in the human craniofacial complex. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e117\u003c/strong\u003e, 236-245 (2002). https://doi.org/10.1002/ajpa.10023\u003c/li\u003e\n \u003cli\u003eGonzalez, P. N., Bernal, V. \u0026amp; Perez, S. I. Analysis of sexual dimorphism of craniofacial traits using geometric morphometric techniques. \u003cem\u003eInternational Journal of Osteoarchaeology\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 82-91 (2011). https://doi.org/10.1002/oa.1109\u003c/li\u003e\n \u003cli\u003eDel Bove, A., Profico, A., Riga, A., Bucchi, A. \u0026amp; Lorenzo, C. A geometric morphometric approach to the study of sexual dimorphism in the modern human frontal bone. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e173\u003c/strong\u003e, 643-654 (2020). https://doi.org/https://doi.org/10.1002/ajpa.24154\u003c/li\u003e\n \u003cli\u003eMilella, M., Franklin, D., Belcastro, M. G. \u0026amp; Cardini, A. Sexual differences in human cranial morphology: Is one sex more variable or one region more dimorphic? \u003cem\u003eThe Anatomical Record\u003c/em\u003e \u003cstrong\u003e304\u003c/strong\u003e, 2789-2810 (2021). https://doi.org/https://doi.org/10.1002/ar.24626\u003c/li\u003e\n \u003cli\u003eDel Bove, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Mapping sexual dimorphism signal in the human cranium. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 16847 (2023). https://doi.org/10.1038/s41598-023-43007-y\u003c/li\u003e\n \u003cli\u003eBertsatos, A., Chovalopoulou, M.-E., Brůžek, J. \u0026amp; Bejdov\u0026aacute;, \u0026Scaron;. Advanced procedures for skull sex estimation using sexually dimorphic morphometric features. \u003cem\u003eInternational Journal of Legal Medicine\u003c/em\u003e (2020). https://doi.org/10.1007/s00414-020-02334-9\u003c/li\u003e\n \u003cli\u003eToneva, D.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Machine learning approaches for sex estimation using cranial measurements. \u003cem\u003eInternational Journal of Legal Medicine\u003c/em\u003e \u003cstrong\u003e135\u003c/strong\u003e, 951-966 (2021). https://doi.org/10.1007/s00414-020-02460-4\u003c/li\u003e\n \u003cli\u003ed\u0026rsquo;Oliveira Coelho, J. \u0026amp; Curate, F. CADOES: An interactive machine-learning approach for sex estimation with the pelvis. \u003cem\u003eForensic Science International\u003c/em\u003e, 109873 (2019). https://doi.org/https://doi.org/10.1016/j.forsciint.2019.109873\u003c/li\u003e\n \u003cli\u003eConstantinou, C. \u0026amp; Nikita, E. SexEst: An open access web application for metric skeletal sex estimation. \u003cstrong\u003en/a\u003c/strong\u003e https://doi.org/https://doi.org/10.1002/oa.3109\u003c/li\u003e\n \u003cli\u003eMahfouz, M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Patella sex determination by 3D statistical shape models and nonlinear classifiers. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e173\u003c/strong\u003e, 161-170 (2007). https://doi.org/https://doi.org/10.1016/j.forsciint.2007.02.024\u003c/li\u003e\n \u003cli\u003eCurate, F.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e A method for sex estimation using the proximal femur. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e266\u003c/strong\u003e, 579.e571-579.e577 (2016). https://doi.org/http://dx.doi.org/10.1016/j.forsciint.2016.06.011\u003c/li\u003e\n \u003cli\u003eCurate, F.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Sex determination from the femur in Portuguese populations with classical and machine-learning classifiers. \u003cem\u003eJournal of Forensic and Legal Medicine\u003c/em\u003e \u003cstrong\u003e52\u003c/strong\u003e, 75-81 (2017). https://doi.org/https://doi.org/10.1016/j.jflm.2017.08.011\u003c/li\u003e\n \u003cli\u003edu Jardin, P., Ponsaill\u0026eacute;, J., Alunni-Perret, V. \u0026amp; Quatrehomme, G. A comparison between neural network and other metric methods to determine sex from the upper femur in a modern French population. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e192\u003c/strong\u003e, 127.e121-127.e126 (2009). https://doi.org/https://doi.org/10.1016/j.forsciint.2009.07.014\u003c/li\u003e\n \u003cli\u003eBertsatos, A., Garoufi, N. \u0026amp; Chovalopoulou, M.-E. Advancements in sex estimation using the diaphyseal cross-sectional geometric properties of the lower and upper limbs. \u003cem\u003eInternational Journal of Legal Medicine\u003c/em\u003e (2020). https://doi.org/10.1007/s00414-020-02437-3\u003c/li\u003e\n \u003cli\u003eImaizumi, K.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Development of a sex estimation method for skulls using machine learning on three-dimensional shapes of skulls and skull parts. \u003cem\u003eForensic Imaging\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 200393 (2020). https://doi.org/https://doi.org/10.1016/j.fri.2020.200393\u003c/li\u003e\n \u003cli\u003eGodinho, R. M., Umbelino, C. \u0026amp; Gon\u0026ccedil;alves, C. Mesolithic and Chalcolithic Mandibular Morphology: Using Geometric Morphometrics to Reconstruct Incomplete Specimens and Analyse Morphology. \u003cem\u003eOpen Archaeology\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e, 536-549 (2022). https://doi.org/doi:10.1515/opar-2022-0247\u003c/li\u003e\n \u003cli\u003eGodinho, R. M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Mandibular morphology and the Mesolithic\u0026ndash;Neolithic transition in Westernmost Iberia. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e13\u003c/strong\u003e, 16648 (2023). https://doi.org/10.1038/s41598-023-42846-z\u003c/li\u003e\n \u003cli\u003eKatz, D. C., Grote, M. N. \u0026amp; Weaver, T. D. Changes in human skull morphology across the agricultural transition are consistent with softer diets in preindustrial farming groups. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e (2017). https://doi.org/10.1073/pnas.1702586114\u003c/li\u003e\n \u003cli\u003evon Cramon-Taubadel, N. Global human mandibular variation reflects differences in agricultural and hunter-gatherer subsistence strategies. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e \u003cstrong\u003e108\u003c/strong\u003e, 19546-19551 (2011). https://doi.org/10.1073/pnas.1113050108\u003c/li\u003e\n \u003cli\u003eGalland, M., Van Gerven, D. P., Von Cramon-Taubadel, N. \u0026amp; Pinhasi, R. 11,000 years of craniofacial and mandibular variation in Lower Nubia. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e6\u003c/strong\u003e, 31040 (2016). https://doi.org/10.1038/srep31040 http://www.nature.com/articles/srep31040#supplementary-information\u003c/li\u003e\n \u003cli\u003ePokhojaev, A., Avni, H., Sella-Tunis, T., Sarig, R. \u0026amp; May, H. Changes in human mandibular shape during the Terminal Pleistocene-Holocene Levant. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 8799 (2019). https://doi.org/10.1038/s41598-019-45279-9\u003c/li\u003e\n \u003cli\u003eCrevecoeur, I., Dias-Meirinho, M.-H., Zazzo, A., Antoine, D. \u0026amp; Bon, F. New insights on interpersonal violence in the Late Pleistocene based on the Nile valley cemetery of Jebel Sahaba. \u003cem\u003eScientific Reports\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 9991 (2021). https://doi.org/10.1038/s41598-021-89386-y\u003c/li\u003e\n \u003cli\u003eBalci, Y., Yavuz, M. F. \u0026amp; Cağdir, S. Predictive accuracy of sexing the mandible by ramus flexure. \u003cem\u003eHOMO\u003c/em\u003e \u003cstrong\u003e55\u003c/strong\u003e, 229-237 (2005). https://doi.org/https://doi.org/10.1016/j.jchb.2004.07.006\u003c/li\u003e\n \u003cli\u003eDonnelly, S. M., Hens, S. M., Rogers, N. L. \u0026amp; Schneider, K. L. Technical note: A blind test of mandibular ramus flexure as a morphologic indicator of sexual dimorphism in the human skeleton. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e107\u003c/strong\u003e, 363-366 (1998). https://doi.org/https://doi.org/10.1002/(SICI)1096-8644(199811)107:3\u0026lt;363::AID-AJPA11\u0026gt;3.0.CO;2-Y\u003c/li\u003e\n \u003cli\u003eFranklin, D., O\u0026apos;Higgins, P. \u0026amp; Oxnard, C. E. Sexual dimorphism in the mandible of indigenous South Africans: A geometric morphometric approach. \u003cem\u003eSouth African Journal of Science\u003c/em\u003e \u003cstrong\u003e104\u003c/strong\u003e, 101-106 (2008).\u003c/li\u003e\n \u003cli\u003eFranklin, D., O\u0026rsquo;Higgins, P., Oxnard, C. E. \u0026amp; Dadour, I. Discriminant function sexing of the mandible of Indigenous South Africans. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e179\u003c/strong\u003e, 84.e81-84.e85 (2008). https://doi.org/https://doi.org/10.1016/j.forsciint.2008.03.014\u003c/li\u003e\n \u003cli\u003eGiles, E. Sex determination by discriminant function analysis of the mandible. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 129-135 (1964). https://doi.org/https://doi.org/10.1002/ajpa.1330220212\u003c/li\u003e\n \u003cli\u003eHanihara, K. Sex Diagnosis of Japanese Skulls and Scapulae by. \u003cem\u003eJournal of the Anthropological Society of Nippon\u003c/em\u003e \u003cstrong\u003e67\u003c/strong\u003e, 191-197 (1959). https://doi.org/10.1537/ase1911.67.191\u003c/li\u003e\n \u003cli\u003eHaun, S. J. Brief communication: A study of the predictive accuracy of mandibular ramus flexure as a singular morphologic indicator of sex in an archaeological sample. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e111\u003c/strong\u003e, 429-432 (2000). https://doi.org/10.1002/(sici)1096-8644(200003)111:3\u0026lt;429::aid-ajpa9\u0026gt;3.0.co;2-1\u003c/li\u003e\n \u003cli\u003eHill, C. A. Technical note: evaluating mandibular ramus flexure as a morphological indicator of sex. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e111\u003c/strong\u003e, 573-577 (2000). https://doi.org/10.1002/(sici)1096-8644(200004)111:4\u0026lt;573::aid-ajpa11\u0026gt;3.0.co;2-i\u003c/li\u003e\n \u003cli\u003eKemkes-Grottenthaler, A., L\u0026ouml;big, F. \u0026amp; Stock, F. Mandibular ramus flexure and gonial eversion as morphologic indicators of sex. \u003cem\u003eHOMO\u003c/em\u003e \u003cstrong\u003e53\u003c/strong\u003e, 97-111 (2002). https://doi.org/https://doi.org/10.1078/0018-442X-00039\u003c/li\u003e\n \u003cli\u003eKuha, A., Ackermann, J., Junno, J.-A., Oettl\u0026eacute;, A. \u0026amp; Oura, P. Deep learning in sex estimation from photographed human mandible using the Human Osteological Research Collection. \u003cem\u003eLegal Medicine\u003c/em\u003e \u003cstrong\u003e70\u003c/strong\u003e, 102476 (2024). https://doi.org/https://doi.org/10.1016/j.legalmed.2024.102476\u003c/li\u003e\n \u003cli\u003eLoth, S. R. \u0026amp; Henneberg, M. Mandibular ramus flexure: A new morphologic indicator of sexual dimorphism in the human skeleton. \u003cstrong\u003e99\u003c/strong\u003e, 473-485 (1996). https://doi.org/10.1002/(sici)1096-8644(199603)99:3\u0026lt;473::Aid-ajpa8\u0026gt;3.0.Co;2-x\u003c/li\u003e\n \u003cli\u003eLoth, S. R. \u0026amp; Henneberg, M. Mandibular ramus flexure is a good indicator of sexual dimorphism. \u003cstrong\u003e105\u003c/strong\u003e, 91-92 (1998). https://doi.org/10.1002/(sici)1096-8644(199801)105:1\u0026lt;91::Aid-ajpa9\u0026gt;3.0.Co;2-g\u003c/li\u003e\n \u003cli\u003eLoth, S. R. \u0026amp; Henneberg, M. Sexually dimorphic mandibular morphology in the first few years of life. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e115\u003c/strong\u003e, 179-186 (2001). https://doi.org/10.1002/ajpa.1067\u003c/li\u003e\n \u003cli\u003eOettl\u0026eacute;, A. C., Pretorius, E. \u0026amp; Steyn, M. Geometric morphometric analysis of mandibular ramus flexure. \u003cstrong\u003e128\u003c/strong\u003e, 623-629 (2005). https://doi.org/10.1002/ajpa.20207\u003c/li\u003e\n \u003cli\u003eOettl\u0026eacute;, A. C., Pretorius, E. \u0026amp; Steyn, M. Geometric morphometric analysis of the use of mandibular gonial eversion in sex determination. \u003cem\u003eHOMO\u003c/em\u003e \u003cstrong\u003e60\u003c/strong\u003e, 29-43 (2009). https://doi.org/https://doi.org/10.1016/j.jchb.2007.01.003\u003c/li\u003e\n \u003cli\u003eOgawa, Y., Imaizumi, K., Miyasaka, S. \u0026amp; Yoshino, M. Discriminant functions for sex estimation of modern Japanese skulls. \u003cem\u003eJournal of Forensic and Legal Medicine\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 234-238 (2013). https://doi.org/https://doi.org/10.1016/j.jflm.2012.09.023\u003c/li\u003e\n \u003cli\u003ePlotkin, L. I., Bruzzaniti, A. \u0026amp; Pianeta, R. Sexual Dimorphism in the Musculoskeletal System: Sex Hormones and Beyond. \u003cem\u003eJournal of the Endocrine Society\u003c/em\u003e \u003cstrong\u003e8\u003c/strong\u003e (2024). https://doi.org/10.1210/jendso/bvae153\u003c/li\u003e\n \u003cli\u003ePretorius, E., Steyn, M. \u0026amp; Scholtz, Y. Investigation into the usability of geometric morphometric analysis in assessment of sexual dimorphism. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e129\u003c/strong\u003e, 64-70 (2006). https://doi.org/https://doi.org/10.1002/ajpa.20251\u003c/li\u003e\n \u003cli\u003eRmoutilov\u0026aacute;, R.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Classification performance of the Sella-Tunis et al. (2017) sex estimation method in Czech population: different posterior probability threshold approaches. \u003cem\u003eInternational Journal of Legal Medicine\u003c/em\u003e (2024). https://doi.org/10.1007/s00414-024-03241-z\u003c/li\u003e\n \u003cli\u003eSchmittbuhl, M., Le Minor, J.-M., Schaaf, A. \u0026amp; Mangin, P. The human mandible in lateral view: elliptical fourier descriptors of the outline and their morphological analysis. \u003cem\u003eAnnals of Anatomy - Anatomischer Anzeiger\u003c/em\u003e \u003cstrong\u003e184\u003c/strong\u003e, 199-207 (2002). https://doi.org/https://doi.org/10.1016/S0940-9602(02)80021-8\u003c/li\u003e\n \u003cli\u003eSteyn, M. \u0026amp; İşcan, M. Y. Sexual dimorphism in the crania and mandibles of South African whites. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e98\u003c/strong\u003e, 9-16 (1998). https://doi.org/https://doi.org/10.1016/S0379-0738(98)00120-0\u003c/li\u003e\n \u003cli\u003eToneva, D. H., Nikolova, S. Y., Fileva, N. F. \u0026amp; Zlatareva, D. K. Size and shape of human mandible: Sex differences and influence of age on sex estimation accuracy. \u003cem\u003eLegal Medicine\u003c/em\u003e \u003cstrong\u003e65\u003c/strong\u003e, 102322 (2023). https://doi.org/https://doi.org/10.1016/j.legalmed.2023.102322\u003c/li\u003e\n \u003cli\u003eMurail, P., Bruzek, J. \u0026amp; Braga, J. A new approach to sexual diagnosis in past populations. Practical adjustments from Van Vark\u0026apos;s procedure. \u003cem\u003eInternational Journal of Osteoarchaeology\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 39-53 (1999).\u003c/li\u003e\n \u003cli\u003eAlbanese, J., Cardoso, H. F. V. \u0026amp; Saunders, S. R. Universal methodology for developing univariate sample-specific sex determination methods: an example using the epicondylar breadth of the humerus. \u003cem\u003eJournal of Archaeological Science\u003c/em\u003e \u003cstrong\u003e32\u003c/strong\u003e, 143-152 (2005). https://doi.org/http://dx.doi.org/10.1016/j.jas.2004.08.003\u003c/li\u003e\n \u003cli\u003eGarcia, S. Is the circumference at the nutrient foramen of the tibia of value to sex determination on human osteological collections? Testing a new method. \u003cem\u003eInternational Journal of Osteoarchaeology\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 361-365 (2012). https://doi.org/https://doi.org/10.1002/oa.1202\u003c/li\u003e\n \u003cli\u003eGon\u0026ccedil;alves, D., Granja, R., Cardoso, F. A. \u0026amp; de Carvalho, A. F. Sample-specific sex estimation in archaeological contexts with commingled human remains: a case study from the Middle Neolithic cave of Bom Santo in Portugal. \u003cem\u003eJournal of Archaeological Science\u003c/em\u003e \u003cstrong\u003e49\u003c/strong\u003e, 185-191 (2014). https://doi.org/http://dx.doi.org/10.1016/j.jas.2014.05.011\u003c/li\u003e\n \u003cli\u003eİşcan, M. Y., Loth, S. R., King, C. A., Shihai, D. \u0026amp; Yoshino, M. Sexual dimorphism in the humerus: A comparative analysis of Chinese, Japanese and Thais. \u003cem\u003eForensic Science International\u003c/em\u003e \u003cstrong\u003e98\u003c/strong\u003e, 17-29 (1998). https://doi.org/https://doi.org/10.1016/S0379-0738(98)00119-4\u003c/li\u003e\n \u003cli\u003eBidmos, M. A. \u0026amp; Mazengenya, P. Accuracies of discriminant function equations for sex estimation using long bones of upper extremities. \u003cem\u003eInternational Journal of Legal Medicine\u003c/em\u003e \u003cstrong\u003e135\u003c/strong\u003e, 1095-1102 (2021). https://doi.org/10.1007/s00414-020-02458-y\u003c/li\u003e\n \u003cli\u003eGreene, D. L., Ewing, G. H. \u0026amp; Armelagos, G. J. Dentition of a mesolithic population from Wadi Halfa, Sudan. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e27\u003c/strong\u003e, 41-55 (1967). https://doi.org/10.1002/ajpa.1330270107\u003c/li\u003e\n \u003cli\u003eIrish, J. D. Population continuity vs. discontinuity revisited: Dental affinities among late Paleolithic through Christian-era Nubians. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e128\u003c/strong\u003e, 520-535 (2005). https://doi.org/https://doi.org/10.1002/ajpa.20109\u003c/li\u003e\n \u003cli\u003eCrevecoeur, I., Rougier, H., Grine, F. \u0026amp; Froment, A. Modern human cranial diversity in the Late Pleistocene of Africa and Eurasia: Evidence from Nazlet Khater, Peştera cu Oase, and Hofmeyr. \u003cstrong\u003e140\u003c/strong\u003e, 347-358 (2009). https://doi.org/https://doi.org/10.1002/ajpa.21080\u003c/li\u003e\n \u003cli\u003eMartin, N.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e From hunter-gatherers to food producers: New dental insights into the Nile Valley population history (Late Paleolithic\u0026ndash;Neolithic). \u003cem\u003eAmerican Journal of Biological Anthropology\u003c/em\u003e \u003cstrong\u003e184\u003c/strong\u003e, e24948 (2024). https://doi.org/https://doi.org/10.1002/ajpa.24948\u003c/li\u003e\n \u003cli\u003eAnderson, J. E. in \u003cem\u003eThe Prehistory of Nubia\u003c/em\u003e Vol. Vol. 2 (ed Wendorf) 996-1040 (Fort Burgwin Research Center and Southern Methodist University Press, 1968).\u003c/li\u003e\n \u003cli\u003ePagani, L. \u0026amp; Crevecoeur, I. in \u003cem\u003eModern Human Origins and Dispersal, Words, Bones, Genes, Tools\u003c/em\u003e (eds Yonatan Sahle, Hugo Reyes-Centeno, \u0026amp; Christian Bentz) 15-24 (Kerns Verlag (DFG Center for Advanced Studies Series), 2019).\u003c/li\u003e\n \u003cli\u003eCrevecoeur, I., Matu, M., Dias-Meirinho, M.-H., Bayle, P. \u0026amp; Pearson, O. in \u003cem\u003eDu Big dry \u0026agrave; l\u0026rsquo;Holoc\u0026egrave;ne en Afrique de l\u0026rsquo;est et au-del\u0026agrave;\u003c/em\u003e Vol. 20 (eds Jos\u0026eacute;phine Lesur\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e) 9-34 (S\u0026eacute;ances de la Soci\u0026eacute;t\u0026eacute; Pr\u0026eacute;historique Fran\u0026ccedil;aise, 2023).\u003c/li\u003e\n \u003cli\u003eHisham, S., Zainun, K. A. \u0026amp; Ibrahim, M. A. Observer error in assessing age-related morphology using digital photographs of the pubic symphyseal face. \u003cem\u003eCanadian Society of Forensic Science Journal\u003c/em\u003e \u003cstrong\u003e55\u003c/strong\u003e, 213-220 (2022). https://doi.org/10.1080/00085030.2022.2053279\u003c/li\u003e\n \u003cli\u003eVillotte, S., Kacki, S. \u0026amp; Thomas, A. Sex estimation of the human os coxae in archeological contexts: An advocacy of using both Diagnose Sexuelle Probabiliste and Brůžek\u0026apos;s morphoscopic method. \u003cem\u003eInternational Journal of Osteoarchaeology\u003c/em\u003e \u003cstrong\u003e34\u003c/strong\u003e, e3334 (2024). https://doi.org/https://doi.org/10.1002/oa.3334\u003c/li\u003e\n \u003cli\u003ePilloud, M. A., Adams, D. M. \u0026amp; Hefner, J. T. Observer error and its impact on ancestry estimation using dental morphology. \u003cem\u003eInternational Journal of Legal Medicine\u003c/em\u003e \u003cstrong\u003e133\u003c/strong\u003e, 949-962 (2019). https://doi.org/10.1007/s00414-018-1985-3\u003c/li\u003e\n \u003cli\u003eWilliams, B. A. \u0026amp; Rogers, T. L. Evaluating the Accuracy and Precision of Cranial Morphological Traits for Sex Determination. \u003cem\u003eJournal of Forensic Sciences\u003c/em\u003e \u003cstrong\u003e51\u003c/strong\u003e, 729-735 (2006). https://doi.org/doi:10.1111/j.1556-4029.2006.00177.x\u003c/li\u003e\n \u003cli\u003eKemkes, A. \u0026amp; G\u0026ouml;bel, T. Metric Assessment of the \u0026ldquo;Mastoid Triangle\u0026rdquo; for Sex Determination: A Validation Study. \u003cem\u003eJournal of Forensic Sciences\u003c/em\u003e \u003cstrong\u003e51\u003c/strong\u003e, 985-989 (2006). https://doi.org/https://doi.org/10.1111/j.1556-4029.2006.00232.x\u003c/li\u003e\n \u003cli\u003eGodde, K., Thompson, M. M. \u0026amp; Hens, S. M. Sex estimation from cranial morphological traits: use of the methods across American Indians, modern North Americans, and ancient Egyptians. \u003cem\u003eHomo-Journal of Comparative Human Biology\u003c/em\u003e (2018). https://doi.org/https://doi.org/10.1016/j.jchb.2018.09.003\u003c/li\u003e\n \u003cli\u003eSmith, H. F. Which cranial regions reflect molecular distances reliably in humans? Evidence from three-dimensional morphology. \u003cem\u003eAmerican Journal of Human Biology\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 36-47 (2009). https://doi.org/https://doi.org/10.1002/ajhb.20805\u003c/li\u003e\n \u003cli\u003eHarvati, K. \u0026amp; Weaver, T. D. Human cranial anatomy and the differential preservation of population history and climate signatures. \u003cem\u003eThe Anatomical Record Part A: Discoveries in Molecular, Cellular, and Evolutionary Biology\u003c/em\u003e \u003cstrong\u003e288\u003c/strong\u003e, 1225-1233 (2006).\u003c/li\u003e\n \u003cli\u003eRoseman, C. C. Detecting interregionally diversifying natural selection on modern human cranial form by using matched molecular and morphometric data. \u003cem\u003eP Natl Acad Sci USA\u003c/em\u003e \u003cstrong\u003e101\u003c/strong\u003e, 12824-12829 (2004). https://doi.org/10.1073/pnas.0402637101\u003c/li\u003e\n \u003cli\u003eRoseman, C. C. \u0026amp; Weaver, T. D. Multivariate apportionment of global human craniometric diversity. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e125\u003c/strong\u003e, 257-263 (2004). https://doi.org/10.1002/ajpa.10424\u003c/li\u003e\n \u003cli\u003eGonz\u0026aacute;lez-Jos\u0026eacute;, R.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Functional-cranial approach to the influence of economic strategy on skull morphology. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e128\u003c/strong\u003e, 757-771 (2005). https://doi.org/10.1002/ajpa.20161\u003c/li\u003e\n \u003cli\u003eGodde, K. Secular trends in cranial morphological traits: a socioeconomic perspective of change and sexual dimorphism in North Americans 1849\u0026ndash;1960. \u003cem\u003eAnnals of Human Biology\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, 255-261 (2015). https://doi.org/10.3109/03014460.2014.941399\u003c/li\u003e\n \u003cli\u003eWeiss, K. M. On the systematic bias in skeletal sexing. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e37\u003c/strong\u003e, 239-249 (1972). https://doi.org/https://doi.org/10.1002/ajpa.1330370208\u003c/li\u003e\n \u003cli\u003eCintas-Pe\u0026ntilde;a, M. \u0026amp; Herrero-Corral, A. M. Missing prehistoric women? Sex ratio as an indicator for analyzing the population of Iberia from the 8th to the 3rd millennia B.C. \u003cem\u003eArchaeological and Anthropological Sciences\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 263 (2020). https://doi.org/10.1007/s12520-020-01215-x\u003c/li\u003e\n \u003cli\u003eCardoso, H. F. V. Brief communication: The collection of identified human skeletons housed at the Bocage Museum (National Museum of Natural History), Lisbon, Portugal. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e129\u003c/strong\u003e, 173-176 (2006). https://doi.org/10.1002/ajpa.20228\u003c/li\u003e\n \u003cli\u003eCoquerelle, M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Sexual Dimorphism of the Human Mandible and Its Association With Dental Development. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e145\u003c/strong\u003e, 192-202 (2011). https://doi.org/Doi 10.1002/Ajpa.21485\u003c/li\u003e\n \u003cli\u003eAlQahtani, S. J., Hector, M. P. \u0026amp; Liversidge, H. M. Brief communication: The London atlas of human tooth development and eruption. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e142\u003c/strong\u003e, 481-490 (2010). https://doi.org/10.1002/ajpa.21258\u003c/li\u003e\n \u003cli\u003eBruzek, J. A method for visual determination of sex, using the human hip bone. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e117\u003c/strong\u003e, 157-168 (2002). https://doi.org/10.1002/ajpa.10012\u003c/li\u003e\n \u003cli\u003eMurail, P., Bruzek, J., Hou\u0026euml;t, F. \u0026amp; Cunha, E. DSP: a tool for probabilistic sex diagnosis using worldwide variability in hip-bone measurements. \u003cem\u003eBulletins et M\u0026eacute;moires de la Soci\u0026eacute;t\u0026eacute; d\u0026rsquo;Anthropologie de Paris\u003c/em\u003e, 167-176 (2005).\u003c/li\u003e\n \u003cli\u003eFedorov, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e 3D Slicer as an image computing platform for the Quantitative Imaging Network. \u003cem\u003eMagnetic resonance imaging\u003c/em\u003e \u003cstrong\u003e30\u003c/strong\u003e, 1323-1341 (2012). https://doi.org/10.1016/j.mri.2012.05.001\u003c/li\u003e\n \u003cli\u003eGeomorph: Software for geometric morphometric analyses. R package version 4.0.6 (2023).\u003c/li\u003e\n \u003cli\u003eSchlager, S., Jefferis, G., Ian, D. \u0026amp; Schlager, M. S. (Academic Press. https://doi. org/10.1016/B978\u0026ndash;0\u0026ndash;12\u0026ndash;810493\u0026ndash;4.00011\u0026ndash;0, 2023).\u003c/li\u003e\n \u003cli\u003eGodinho, R. M., O\u0026apos;Higgins, P. \u0026amp; Gon\u0026ccedil;alves, C. Assessing the reliability of virtual reconstruction of mandibles. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e172\u003c/strong\u003e, 723-734 (2020). https://doi.org/https://doi.org/10.1002/ajpa.24095\u003c/li\u003e\n \u003cli\u003eGunz, P., Mitteroecker, P., Neubauer, S., Weber, G. W. \u0026amp; Bookstein, F. L. Principles for the virtual reconstruction of hominin crania. \u003cem\u003eJournal of Human Evolution\u003c/em\u003e \u003cstrong\u003e57\u003c/strong\u003e, 48-62 (2009). https://doi.org/10.1016/j.jhevol.2009.04.004\u003c/li\u003e\n \u003cli\u003eNeeser, R., Ackermann, R. R. \u0026amp; Gain, J. Comparing the accuracy and precision of three techniques used for estimating missing landmarks when reconstructing fossil hominin crania. \u003cem\u003eAmerican Journal of Physical Anthropology\u003c/em\u003e \u003cstrong\u003e140\u003c/strong\u003e, 1-18 (2009). https://doi.org/10.1002/ajpa.21023\u003c/li\u003e\n \u003cli\u003eSenck, S., Bookstein, F. L., Benazzi, S., Kastner, J. \u0026amp; Weber, G. W. Virtual Reconstruction of Modern and Fossil Hominoid Crania: Consequences of Reference Sample Choice. \u003cem\u003eThe Anatomical Record\u003c/em\u003e \u003cstrong\u003e298\u003c/strong\u003e, 827-841 (2015). https://doi.org/10.1002/ar.23104\u003c/li\u003e\n \u003cli\u003eMitteroecker, P., Gunz, P., Windhager, S. \u0026amp; Schaefer, K. A brief review of shape, form, and allometry in geometric morphometrics, with applications to human facial morphology. \u003cem\u003eHystrix-Italian Journal of Mammalogy\u003c/em\u003e \u003cstrong\u003e24\u003c/strong\u003e, 59-66 (2013). https://doi.org/DOI 10.4404/hystrix-24.1-6369\u003c/li\u003e\n \u003cli\u003eZelditch, M. L., Swiderski, D. L., Sheets, H. D. \u0026amp; Fink, W. L. \u003cem\u003eGeometric Morphometrics For Biologists: A Primer\u003c/em\u003e. (Elsevier, 2012).\u003c/li\u003e\n \u003cli\u003eO\u0026apos;Higgins, P. The study of morphological variation in the hominid fossil record: biology, landmarks and geometry. \u003cem\u003eJournal of Anatomy\u003c/em\u003e \u003cstrong\u003e197\u003c/strong\u003e, 103-120 (2000). https://doi.org/10.1046/j.1469-7580.2000.19710103.x\u003c/li\u003e\n \u003cli\u003eKlingenberg, C. P. Size, shape, and form: concepts of allometry in geometric morphometrics. \u003cem\u003eDevelopment Genes and Evolution\u003c/em\u003e \u003cstrong\u003e226\u003c/strong\u003e, 113-137 (2016). https://doi.org/10.1007/s00427-016-0539-2\u003c/li\u003e\n \u003cli\u003eHammer, O. \u003cem\u003ePAST reference manual 4.12\u003c/em\u003e. (2022).\u003c/li\u003e\n \u003cli\u003eCourtenay, L. A., Aramendi, J. \u0026amp; Gonz\u0026aacute;lez-Aguilera, D. A Graph-Based Mathematical Model for More Efficient Dimensionality Reduction of Landmark Data in Geometric Morphometrics. \u003cem\u003eEvolutionary Biology\u003c/em\u003e (2024). https://doi.org/10.1007/s11692-024-09636-5\u003c/li\u003e\n \u003cli\u003eAbell\u0026aacute;n, N., Baquedano, E. \u0026amp; Dom\u0026iacute;nguez-Rodrigo, M. High-accuracy in the classification of butchery cut marks and crocodile tooth marks using machine learning methods and computer vision algorithms. \u003cem\u003eGeobios\u003c/em\u003e \u003cstrong\u003e72-73\u003c/strong\u003e, 12-21 (2022). https://doi.org/https://doi.org/10.1016/j.geobios.2022.07.001\u003c/li\u003e\n \u003cli\u003eMocl\u0026aacute;n, A.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Identifying the bone-breaker at the Navalma\u0026iacute;llo Rock Shelter (Pinilla del Valle, Madrid) using machine learning algorithms. \u003cem\u003eArchaeological and Anthropological Sciences\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, 46 (2020). https://doi.org/10.1007/s12520-020-01017-1\u003c/li\u003e\n \u003cli\u003eCourtenay, L. A. \u0026amp; Gonz\u0026aacute;lez-Aguilera, D. Geometric Morphometric Data Augmentation Using Generative Computational Learning Algorithms. \u003cem\u003eApplied Sciences\u003c/em\u003e \u003cstrong\u003e10\u003c/strong\u003e, 9133 (2020).\u003c/li\u003e\n \u003cli\u003eMcPherron, S. P., Archer, W., Ot\u0026aacute;rola-Castillo, E. R., Torquato, M. G. \u0026amp; Keevil, T. L. Machine learning, bootstrapping, null models, and why we are still not 100% sure which bone surface modifications were made by crocodiles. \u003cem\u003eJournal of Human Evolution\u003c/em\u003e \u003cstrong\u003e164\u003c/strong\u003e, 103071 (2022). https://doi.org/https://doi.org/10.1016/j.jhevol.2021.103071\u003c/li\u003e\n \u003cli\u003eKuhn, M.\u003cem\u003e\u0026nbsp;et al.\u003c/em\u003e Caret: Classification and Regression Training. (2024).\u003c/li\u003e\n \u003cli\u003eDeane-Mayer, Z. A., Knowles, J. E. \u0026amp; L\u0026oacute;pez, A. Package \u0026lsquo;caretEnsemble\u0026rsquo;. (2024).\u003c/li\u003e\n \u003cli\u003eLantz, B. \u003cem\u003eMachine learning with R\u003c/em\u003e. (Packt Publishing, 2013).\u003c/li\u003e\n \u003cli\u003eDom\u0026iacute;nguez-Rodrigo, M. Successful classification of experimental bone surface modifications (BSM) through machine learning algorithms: a solution to the controversial use of BSM in paleoanthropology? \u003cem\u003eArchaeological and Anthropological Sciences\u003c/em\u003e \u003cstrong\u003e11\u003c/strong\u003e, 2711-2725 (2019). https://doi.org/10.1007/s12520-018-0684-9\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"Virtual Anthropology, Skeletal remains, Archaeology, Morphology, Palaeodemography","lastPublishedDoi":"10.21203/rs.3.rs-6389860/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6389860/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAccurate sex estimation is crucial for studying both modern and ancient human populations, yet methods are often limited to well-preserved skeletons. Here, we combine Geometric Morphometrics (GM) and Machine Learning (ML) to assess mandibular sexual dimorphism and classify sex across a wide chronological and geographic range to bracket the potential of this approach.\u003c/p\u003e\u003cp\u003eSixty-seven individuals from the modern, identified Luis Lopes collection (Portugal) and 18 Late Pleistocene individuals from Jebel Sahaba (Sudan) were surface scanned. Anatomical landmark coordinates were extracted and analyzed with GM, and ML models were trained on a subset of the modern sample to predict sex in both the remaining modern individuals and the Late Pleistocene specimens.\u003c/p\u003e\u003cp\u003eGM revealed significant sexual dimorphism in all samples, and ML achieved high intrapopulation classification accuracy. However, predictions were less reliable when applied across the temporally and geographically distant Jebel Sahaba population, reflecting interpopulation differences in mandibular size and shape. These results demonstrate that while GM\u0026ndash;ML approaches are powerful tools for sex estimation within populations, caution is required when extending models to other populations.\u003c/p\u003e","manuscriptTitle":"Coupling geometric morphometrics and machine learning for mandibular sex estimation: testing Late Pleistocene and Late Modern populations","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-11-26 12:22:00","doi":"10.21203/rs.3.rs-6389860/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2025-12-01T09:17:41+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2025-12-01T09:16:51+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-11-29T01:42:07+00:00","index":"","fulltext":""},{"type":"submitted","content":"Scientific Reports","date":"2025-11-23T23:27:18+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"scientific-reports","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"scirep","sideBox":"Learn more about [Scientific Reports](http://www.nature.com/srep/)","snPcode":"","submissionUrl":"","title":"Scientific Reports","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Scientific Reports","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"f184f1bf-043b-4650-9d7f-52fa093a04d7","owner":[],"postedDate":"November 26th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2025-12-22T16:06:20+00:00","versionOfRecord":{"articleIdentity":"rs-6389860","link":"https://doi.org/10.1038/s41598-025-31365-8","journal":{"identity":"scientific-reports","isVorOnly":false,"title":"Scientific Reports"},"publishedOn":"2025-12-19 15:58:29","publishedOnDateReadable":"December 19th, 2025"},"versionCreatedAt":"2025-11-26 12:22:00","video":"","vorDoi":"10.1038/s41598-025-31365-8","vorDoiUrl":"https://doi.org/10.1038/s41598-025-31365-8","workflowStages":[]},"version":"v1","identity":"rs-6389860","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6389860","identity":"rs-6389860","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}