Sexual dimorphism of the human pelvis is traditionally viewed in relation to the tight fit of the fetal head in the pelvic aperture, which is the consequence of competing selective pressures related to either bipedal locomotion or childbirth. Anteroposterior shortening of the bipedal pelvis limits pelvic canal space, which results in a much larger cephalo-pelvic proportion between the fetal head size and maternal pelvic canal size (Washburn, 1960; Rosenberg, 1992; Haeusler et al., 2021). However, the pelvis also relates to overall body size and body shape, playing a role in thermoregulation and adaptation to climate (Ruff, 1994; Holliday, 2012; Kurki, 2013; Betti, 2017; Auerbach et al., 2018), locomotor economy and efficiency (Warrener et al., 2015), load-carrying (Wall-Scheffler, 2022), posture (Been et al., 2017), and pelvic floor stability (Uy et al., 2020; Stansfield et al., 2021). These factors vary across populations living in different geographical environments and following different lifestyles, correspondingly influencing pelvic morphology.

Despite the above-mentioned factors, sexual dimorphism of the obstetrically relevant parts of the pelvis follows a similar pattern in modern humans and is shared among primates and likely even among mammals (Fischer et al., 2021; Tague, 1991). The coxal bone is thus the most reliable skeletal region for sex estimation as has been empirically demonstrated by numerous studies on various human populations (Gonzalez et al., 2009; Klales et al., 2012; Novak et al., 2012; Rmoutilová et al., 2017; Brůžek et al., 2017; Santos et al., 2019; Galeta and Brůžek, 2020; Klales, 2020a). The magnitude of pelvic sexual dimorphism varies greatly among species, which can be attributed to the degree of cephalo-pelvic proportion between the fetal head size and maternal pelvic canal size (Leutenegger, 1974; Ridley, 1995) and to a lesser extent, to sexual dimorphism in body size and steroid hormone levels that trigger both dimorphic growth and remodeling of the pelvis (Tague, 2005). Therefore, sex estimation of archaic hominins is complicated as has been previously noted (Genovés, 1954) and extensively discussed (Smith, 1980; Trinkaus, 1980; Novotný, 1983; Hager, 2005; Bonmatí and Arsuaga, 2007; 2016a; Hammond et al., 2017, 2018; Alonso-Llamazares and Pablos, 2019; Bethard and VanSickle, 2020). Despite this complexity, except for ancient DNA (Skoglund et al., 2013) or tooth enamel peptides (Lalueza-Fox et al., 2011; Stewart et al., 2017; Parker et al., 2019), the pelvis remains the most reliable means of skeletal sex estimation. The main difficulty encountered in attempting morphological assessment of archaic human pelves is the use of anatomically modern criteria on nonmodern hominins who differ in their patterns of pelvic morphology and/or degree of sexual dimorphism (Marchal, 1997). For this reason, modern methods are not guaranteed to be applicable to all hominin fossil taxa (Genovés, 1954; Hager, 1996; Bethard and VanSickle, 2020).

While sex attribution is fundamental to discussing hominin biological evolution (Frayer, 1980; Weaver and Hublin, 2009; Hora and Sladek, 2014; Estalrrich and Rosas, 2015), sex attributions of Neandertal coxal bones consist primarily of visual descriptions of morphological traits (see Supplementary Online Material [SOM] Table S1 and references therein), most of which were published before the establishment of sex estimation methods now known to be highly reliable for modern humans (Klales et al., 2012; Brůžek et al., 2017; Santos et al., 2019). Neandertal body form shows strong adaptations to cold climate (Holliday, 1997; Weaver, 2003, 2009) with markedly wide pelvic breadth (Ruff, 1994). As highly encephalized hominins with encephalization quotients falling within the modern human range (Stanyon et al., 1993; Rosenberg and Trevathan, 2002; Ponce de León et al., 2008; Holliday, 2023), Neandertals would be expected to show comparable cephalo-pelvic proportions to modern humans and thus to have comparably difficult births. However, given the fragmentary nature of fossil pelvic remains, the attribution of sex for any individual specimen remains difficult. It is therefore critical to use multiple objective methods of sex estimation for fossil specimens, and to distinguish between sex assessment, which does not provide estimable error rates, and sex estimation with associated classification accuracy (Klales, 2020b).

The aim of this study is to estimate the sex of the Regourdou 1 (R1) Neandertal, whose right coxal bone was reconstructed to provide new information on the individual's sex. Many large fragments of the R1 coxal bones have been recently identified among faunal remains from the site (Madelaine et al., 2008; Maureille et al., 2015), so it currently represents the only Western Eurasian Neandertal pelvis (Marine Isotope Stage [MIS] 5 to 3) with no taphonomic distortion and no prior physical restoration, as well as the only non-deformed sacroiliac articulation. To sex R1, we used morphological and metrical sex estimation methods previously standardized on modern human populations. Applied sex estimation methods were then cross validated with other Neandertal specimens to evaluate the reliability of methods developed on modern humans for sex estimation of these fossil hominins.

The site of Regourdou is located at the top of the Lascaux hill near Montignac in southwestern France. It is a partially collapsed karstic cave discovered by R. Constant, the Regourdou landowner, in 1954. The R1 fossil was accidently discovered in 1957 during amateur underground excavations by friends of R. Constant and a Swiss citizen (Piveteau, 1959, 1963). Despite being authorized to excavate the site, Constant had no background in paleontology or prehistory and thus did not record provenience of his discoveries. Between 1961 and 1964, part of the site was scientifically excavated by E. Bonifay (Bonifay, 1964; Bonifay et al., 2007). A project to re-excavate the site and to study the collections was subsequently directed by one of us (B.M.) between 2013 and 2018 (Maureille, 2013; Discamps et al., 2016).

According to Bonifay's interpretation of the sedimentological filling of the cave (SOM Fig. S1), the R1 skeleton was discovered in layer 4 (Bonifay, 1964), which yielded a few Discoid Mousterian lithics (Turq and Royer, 2016), and was associated with a temperate faunal assemblage dominated by Ursus arctos (Bonifay, 1964; Delpech, 1996; Cavanhié, 2009). During the formation of layer 4, the site functioned as a brown bear den as shown by a natural mortality profile of young and adult bears (Cavanhié, 2009). Based on the sedimentological and paleontological evidence, layer 4 was attributed to MIS 5 by Bonifay (1964) and Bonifay et al. (2007). This chronological attribution was confirmed during fieldwork at the Regourdou site between 2014 and 2016 (Maureille, 2016, unpublished data). Moreover, red sandy sediments washed from the plateau above the Regourdou karstic cave roof were then water-deposited within the cave below resulting in a 50-cm-thick brown to red sandy laminated layer that is very easily identified. This sedimentological formation corresponds to Bonifay's layer 3 (Maureille, 2014). Layer 3 overlies Bonifay's layer 4 and has yielded preliminary optically stimulated luminescence dates of ca. 80 ka (Lebrun and Lahaye, 2014). Given its antiquity, R1 is now the earliest, best-preserved Western European Neandertal skeleton (Peyrégne et al., 2019). We have also demonstrated the existence of specific post-depositional taphonomic processes related to MIS 3 European rabbits (Oryctolagus cuniculus) digging burrows within Regourdou's sedimentological filling—burrows that had an impact on the integrity of the site's archaeological assemblages (Pelletier et al., 2017).

From 1957 to 2008, the R1 skeleton was limited to 100 bones or bony fragments with especially well-preserved, fully complete, upper limbs. From 2009 to 2018 several new pieces of the skeleton were recovered. Some were recovered from the faunal collections associated with the 1961 to 1964 excavations that have been curated at the Musée national de Préhistoire at Les Eyzies-de-Tayac since 2002. These collections were checked piece by piece in 2008 due to a French law mandating the taking of inventories of national museum collections (Madelaine et al., 2008). These, plus investigations of the collection of Musée d'art et d'archéologie du Périgord and the Regourdou private site museum, led to the rediscovery and identification of 72 new pieces belonging to the R1 skeleton (Madelaine et al., 2008; Maureille et al., 2015), most of which are elements of the trunk, pelvis, and lower limbs. This has resulted in new studies of the skeletal anatomy of R1 (Gómez-Olivencia et al., 2013a, 2019; Pablos et al., 2019), including the sternum (Gómez-Olivencia et al., 2012) and sacrum (Rmoutilová et al., 2020) as well as the calcium isotope signatures of the hominin and the faunal taxa, which allowed the first evaluation of the diet and principal ungulate prey of R1 (Dodat et al., 2021). Moreover, it has also been confirmed that there was a second individual (Regourdou 2) discovered during Bonifay's fieldwork. Regourdou 2 is only represented by an isolated right calcaneus with Neandertal morphological affinities and it is larger and longer than those of R1 (Coutinho Nogueira et al., 2017). In contrast to most of the skeletal remains of R1, the Regourdou 2 calcaneus was not recorded by Bonifay, and thus its exact provenience within the site remains uncertain.

Based on the available skeletal material, Regourdou 1 has been regarded as either a male (Vallois, 1965; Gómez-Olivencia et al., 2007; Volpato et al., 2012) or as sex-indeterminate (Trinkaus, 1980; Vandermeersch and Trinkaus, 1995; Meyer et al., 2011; Plavcan et al., 2014). Until 2008, only the sacrum, right ischium, and superior pubic ramus were recognized from the pelvis, but they lacked reliable features for sex attribution. For this reason, the sex of R1 was historically assessed based on the size and robusticity of its different skeletal regions (see below).

Male sex was initially proposed without any specific justification by Vallois (1965), who referred to previous papers by Piveteau (1959, 1963, 1964, 1966) in which no sex attribution was given. This initial attribution was later contradicted (Trinkaus, 1980; Vandermeersch and Trinkaus, 1995) on the basis of upper limb articulations and pubic ramus thickness that were reported to be closer to Neandertals considered females or to lie between the European Neandertals considered males and females. On the other hand, male sex was supported by Volpato et al. (2012) based on male proportions of the sacral body and alae.

With regard to the overall body size, several studies estimated the sex of R1 based on extra-pelvic measurements. In contrast to the above-mentioned gracile upper limb articulations, humeral, radial, and talar lengths were reported to be in the male Neandertal range of variation but lie within its lower portion (Trinkaus, 1980; Vandermeersch and Trinkaus, 1995), while canine breadth was reported to be within the upper range of European Neandertal variability (Volpato et al., 2012). Concordantly, R1 was classified as male using clavicular dimensions (maximum length and mid-shaft vertical and anteroposterior diameters; Carretero, 1994) or based on modern human formulae (Wescott, 2000) using second cervical vertebra (Gómez-Olivencia et al., 2007). Subsequent analyses led to small body mass estimates for R1 (64.4–65.6 kg; Table 1) compared to presumed male Neandertals (males: 65–85 kg, females: 60–75 kg; Plavcan et al., 2014) based on femoral head diameter and talar trochlea breadth (Plavcan et al., 2014; Pablos et al., 2019). Stature estimates of R1 based on various skeletal regions and equations yielded more variable results (Table 1). Estimates based on the humerus and the radius (162.2–167.5 cm) were lower than or close to the mean for Neandertals considered males (males: 167 cm, females: 155 cm; Carretero et al., 2012), while stature was higher when predicted from the lengths of the talus, calcaneus, and third and fourth metatarsals, or from vertebral body heights (170.3–173.3 cm). These differences in stature estimates likely reflect different body proportions between Neandertals and extant reference samples (Holliday, 1997; Pablos et al., 2019).

Since 2008, new pelvic remains belonging to R1 have been identified in the faunal collections from the Regourdou site. Given the recorded location of skeletal remains, their articulation with the sacrum, estimated age, and taphonomic state, their assignment to R1 is the most parsimonious interpretation (Madelaine et al., 2008; Meyer et al., 2011; Maureille et al., 2015). Although both ilia show a rather deep and asymmetric morphology of the greater sciatic notch corresponding to male sex, both notches are incompletely preserved, resulting in uncertainty about their exact shape (Meyer et al., 2011; Plavcan et al., 2014). Here, we undertake a new sex attribution of R1 based on the original left and reconstructed right coxal bones using modern sex estimation methods.