The Baousso da Torre cave is part of the Grimaldi, or Balzi Rossi complex of caves and rock shelters, one of the most important European Paleolithic sites on the Mediterranean coast, just at the border between Italy and France (Formicola and Holt, 2015). The skeletons of two adult (BT1 and BT2) and one adolescent (BT3) individuals were discovered in 1873 in this cave (Rivière, 1887). BT1 and BT3 ¹ 1

The study of BT3, which is not presented here, is still in progress. The detailed analysis of this immature individual is problematic due to the rediscovery, in recent times, of more bones/fragments of bones belonging to this skeleton, which need to be reunited with the original material before proceeding with the study (Villotte, in prep.).

The BT1 partial skeleton preserved very limited portions of the skull and dentition, the girdles, and the axial skeleton. Remaining bones – mainly from limbs – are usually sub-complete, although their surface is not altered. The BT2 skeleton is more complete but presents some reconstruction (with plaster and wax), and the surfaces are sometimes covered by concretion, glue, consolidants, or ochre, making their evaluation difficult. Almost all the bones of BT1 and BT2 are reddish, and some display clear traces of ochre. For BT2, this ochre is sometimes represented by a thick layer of several millimeters, especially on the occipital, the left parietal, the humerus, and the scapula.

No prehistoric anthropogenic marks were found on any bone. Both femora and the left humerus of BT1 display rodent gnaw marks. Rodent and carnivore marks are also visible on the BT2 skeleton: on both femora (carnivore) and on the tibiae and right fibula (carnivore marks on the extremities, rodent marks on the diaphyses).

The comparative data is presented in detail in the Supplementary Information (SI) 1. It consists of European Late Pleistocene humans, mostly mature individuals and a few late adolescents, chronologically divided in middle Upper Paleolithic (from about 30,000 years ago to the end of the Last Glacial Maximum) and late Upper Paleolithic (from the end of the Last Glacial maximum to the end of the Pleistocene). Osteometric and cross-sectional data derive from personal research of the authors and from the literature (See SI 1). Some individuals (Oberkassel 1, Barma Grande 2, Barma del Caviglione 1, Vado all’ Arancio 1) show marked limb abnormalities or trauma, and those skeletal elements were not included in the comparative sample.

The linear measurements follow the Martin system (M-#) (Bräuer, 1988), completed by measurements defined in Sládek et al. (2000) and Murail et al. (2005). Non-metric variations were recorded following Hauser and De Stefano (1989) for the skull, Scott and Turner (2000) for the dentition, and Finnegan (1978) for the infracranial skeleton. Dental wear was recorded following Smith (1984). “Traditional” robusticity indices computed from external measurements (i.e. diaphyseal thickness standardized to bone length) were calculated following Bräuer (1988). Biomechanical robusticity (i.e. diaphyseal rigidity scaled by body size) was estimated via cross-sectional geometry (CSG, see Ruff et al., 2006a). The biomechanical competence of long bones was calculated at specified percentages of bone length (mechanical length, as indicated in Ruff, 2002), using the polar moment of area as a measure of overall bending and torsional rigidity. Results were scaled by body size using bone mechanical length and estimated body mass (Ruff, 2000). The details of the methods used for CSG are provided in the SI 2. Bilateral asymmetry for CSG properties and measurements of the upper limbs was computed as: [100 × (maximum value − minimum value)/minimum value)], and expressed as a percentage (Trinkaus et al., 1994).

The pelvis of BT1 is very badly preserved, and cannot be used for the sex assessment. However, all measurements of its appendicular skeleton are in the upper range of variation of Upper Paleolithic males (see below) and a secondary sex diagnosis based on discriminant function analysis indicates that BT1 is very likely a male (Villotte et al., 2011a). On the five morphological structures of the coxal bone used for the sex assessment in Brůžek (2002), three (preauricular surface aspects, sciatic notch aspects, and morphology of the composite arch) can be scored on the BT2 left coxal bone and all indicate a male morphology (see SI 4). The probability of BT2 being a male or female has also been calculated by comparing measurements of its left coxal bone to those from a worldwide database (Murail et al., 2005). The posterior probability of BT2 being a male is 0.99 based on seven measurements (See SI 4).

Age-at-death of BT1 cannot be estimated using the pubic symphysis and the auricular surface of the ilium, since the first is missing and the latter heavily damaged. Skeletal maturity was reached (e.g., the iliac crest is fused) indicating a fully adult individual. In addition, the very limited preserved part of the auricular surface of the ilium shows no degenerative changes, no skeletal sign of degenerative processes related to senescence is present on any bone, and teeth are not heavily worn. Moreover, the metaphyseal line between the head and the diaphysis of the second right metacarpal is still macroscopically and radiologically visible. Based on those criteria, BT1 should be considered as a young adult/adult. Age-at-death estimation of BT2 is based on the state of fusion of later-fusing secondary centers of ossification, following Owings Webb and Suchey (1985), and on the morphology of the auricular surface of the ilium, following Schmitt (2005). The partial union of the epiphysis to the right medial clavicle and the morphology of the auricular surface of the ilium both identify BT2 as a young adult (20–29 years old).

In both individuals, the skull is fragmentary. The morphology of the neurocranium and the face of BT1 and BT2 is clearly modern, and none of the autapomorphic traits of Neanderthals (e.g., the suprainiac fossa, Balzeau and Rougier, 2010) is present for BT1 and BT2 partial skulls. It is however important to mention some features such as the presence of well-developed cranial structures (e.g., the external occipital protuberance) in both individuals, the partial mylohyoid bridging at the right side of the mandible of BT2, and the supernumerary ossicles of the cranial vault of BT1. The cranial capacity of BT2 can be estimated to 1502 ± 64 cm³ and to 1684 ± 76 cm³ (respectively from the third and eighth formulas of the Table 3 in Olivier et al. (1978)), in the upper part of the range of variation seen for the first part of the Upper Paleolithic [1467 ± 165 cm³, n = 29 (Trinkaus et al., 2014)].

Macroscopically, there is little to report on the tooth morphology. BT2 does not show evidence of heavy anterior tooth wear relative to the posterior teeth, contrary to some other Upper Paleolithic individuals (Clement et al., 2012). A comprehensive study of the teeth, including their internal structure, is planned in a near future.

For both individuals, the dimensions used to calculate body mass, stature and body proportions have been estimated (see SI 4). Thus, it should be kept in mind that those figures are only indicative, and were used to place these individuals in the context of the other Upper Paleolithic individuals. The body mass was estimated following the guidelines in Trinkaus and Ruff (2012), based on estimations of the supero-inferior diameter of the femoral head (see SI 4). BT1 and BT2 body masses are estimated to be 78.1 and 73.1 kg, respectively. Stature was estimated using Trotter and Gleser (1952) equation for African-Americans (based on femoral maximum length), as suggested by Formicola (2003) for European middle Upper Paleolithic specimens. The stature is of 173.0 ± 3.9 cm for BT1 and 173.4 ± 3.9 cm for BT2. All of these values fall in the range of variation seen for the MUP male sample (Table 1). Brachial and crural indices of BT1 and BT2 are slightly lower than the average values for the MUP and LUP samples (Table 1) but are still high when compared to recent Europeans (see Holliday, 1997).

Upper limbs measurements of the two BT individuals (see tables in SI 4) consistently fall in the upper portion of the variation seen for the Upper Paleolithic samples, and are even outliers for certain variables (e.g., distal radius circumference, see Fig. 1). In terms of “Traditional” robusticity computed from external measurements, the upper limb bones of BT individuals are relatively robust (Table 2, Table 3 and Table 4). When considering the mechanical robusticity via CSG properties, the humeri of the BT individuals (both sides) are consistently more robust than any middle Upper Paleolithic male in our comparative sample (Table 2, Fig. 2). As a result, BT individuals seem to fit better with the late Upper Paleolithic male sample. However, it is difficult to interpret this result from a functional adaptation point of view, given that differences in manipulatory behavior between the two periods are not well defined (see also below the discussion about upper limb asymmetry). In addition, it should be taken into account that the estimations of BT1 humeral length (see Table 11 in SI4) and of the femoral head diameter (both individuals, see Table 31 in SI4), from which body mass was calculated, suggest that the results should be interpreted with caution. CSG results for the forearm are more fragmentary, and the comparative sample size is smaller: only for BT2 was it possible to obtain CSG properties of the ulna, and only mid-proximally; the individual is one of the most robust in the whole Upper Paleolithic sample (Table 3). BT1 has the right radius preserved, while the left is preserved in BT2. Both individuals tend to show high values of robusticity (Table 4, Fig. 1 and 2 in SI 2).

Upper limb values for external measurements and CSG properties are mostly greater for the right side (see SI 4). As a result, BT individuals share high levels of bilateral asymmetry with Upper Paleolithic (both middle and late) males (Table 5 and Fig. 3). Previous studies suggested that the level of CSG humeral asymmetry observed in Upper Paleolithic individuals, which is comparable with modern throwing athletes such as tennis players (Ireland et al., 2013, Shaw and Stock, 2009a, Trinkaus et al., 1994 and Warden et al., 2009), may be due to a hunting technique based on projectile weapons (e.g., Sládek et al., 2016 and Sparacello et al., 2017). This activity is not only strenuous during hunting but requires continuous training during development for both strength and aim (Cattelain, 1997, Rhodes and Knüsel, 2005 and Whittaker and Kamp, 2006). The high and intermittent loading rates correspond to the pattern that best stimulate osteogenic response (Burr et al., 1996, Burr et al., 2002 and Robling et al., 2002). Experimental data on athletes also suggests that the forearm may be influenced by throwing behavior, but to a lesser extent and with greater variation in the results (Shaw and Stock, 2009a). The only element of the forearm for which bilateral CSG data were available is the mid-proximal ulna of BT 2, which is the most lateralized in the middle Upper Paleolithic sample (Fig. 3 in SI 2).

This lateralized limb use is also indicated by a lesion at the right elbow of BT2. The right common flexor origin on the medial epicondyle of BT2 was previously considered as a healthy enthesis (Villotte et al., 2010). However, a recent gentle cleaning of this area revealed an alteration of the attachment of the medial collateral ligament on the right side (Fig. 4a). The area exhibits an abnormal depression with a smooth cortical surface associated with a small isolated bone protrusion. The left medial epicondyle does not display any change. Medial epicondyle enthesopathy (also called “thrower's elbow”) is a rare condition seen mainly in sports employing the overhead throwing motion. This condition is also extremely rare in recent humans before 50 years of age (Villotte et al., 2010 and Villotte and Knüsel, 2014).

This unilateral epicondylar lesion, and the high level of asymmetry observed in both the humerus and the forearm, suggest that stressful unimanual activities shaped the upper limb properties of BT individuals. These activities can be tentatively linked with the throwing motion, as suggested in several studies (see above). However, other activities may have contributed to the differential hypertrophy of the humeri through the Late Pleistocene, since repetitive domestic activities such as scraping, flint knapping, woodworking, and others, involve asymmetric recruiting of upper limb muscles (Shaw et al., 2012).

BT individuals show large values of linear dimensions for the lower limbs (see SI 4), falling in the upper part of the variation of the Upper Paleolithic male sample, and are even outliers for some variables (as illustrated for instance in Fig. 5). The femur and the tibia of the BT individuals appear robust (both in terms of external measurements and CSG properties) with respect to the Upper Paleolithic comparative samples (Table 6 and Table 7, Fig. 6). Conversely, CSG fibular data, which was available only for BT 2, indicates a relatively less robust element, both in absolute terms and when compared with the tibia (Fig. 4 in SI 2); however, the comparative sample is small and shows high variance.

BT1 and BT2 femora exhibit very prominent gluteal buttresses leading to a low meric index (Table 6). Conversely, mid-shaft shape in both individuals is very elongated antero-posteriorly, as illustrated by both their pilastric index and CSG shape ratio (Table 6; Fig. 6), which are among the highest observed in the Upper Paleolithic sample. In contrast, in both BT individuals the midshaft tibia is eurycnemic (Table 7) and displays a low CSG shape index when compared with the rest of the Upper Paleolithic sample (Table 7, Fig. 6).

Lower limbs morphology is influenced by mobility patterns, load carrying, and the terrain on which those activities are performed (review in Carlson and Marchi, 2014). In CSG research, the rationale is that high terrestrial mobility will increase the frequency, stressfulness, and amount of antero-posteriorly directed loading, and therefore bending, of bones in the lower limb, which will lead to increased strength and high “shape indices” (ratios of antero-posterior and medio-lateral bending moments) at midshaft diaphysis. Accordingly, highly mobile hunter-gatherers, including Upper Paleolithic humans, generally show higher shape ratios (femoral I _x /I _y and tibial I _max/I _min) than sedentary agriculturalists (Carlson and Marchi, 2014). However, the scenario is complicated by the fact that body proportions and bone length might influence femoral shape (Ruff, 1995, Ruff et al., 2006b and Sparacello et al., n.d.). Also, different types of loading, for example running in a straight line versus frequent swerving, can have an impact on the functional adaptations of the tibio-fibular complex (Marchi and Shaw, 2011 and Shaw and Stock, 2009b). Overall, the results obtained for BT individuals suggest high mobility levels, with high femoral shape indices and robust tibiae (Holt, 2003 and Shaw and Stock, 2009b). However, the very low CSG shape index of the tibia, at least for an Upper Paleolithic individual, observed in both BT1 and BT2 suggests a high degree of medio-lateral stress in the tibio-fibular complex (Shaw and Stock, 2009b). Interestingly, this does not result in a particularly robust fibula, as expected based on previous research (Hagihara and Nara, 2016, Marchi and Shaw, 2011 and Sparacello et al., 2014). However, in a multivariate setting, BT2 conforms to the field hockey profile (Sparacello et al., n.d.), i.e. an individual performing high mobility with frequent swerving. This could be interpreted as indicating long-distance traversing of an uneven terrain, like the extremely mountainous inland of Liguria (Sparacello et al., 2014).

Further support to the scenario of high mobility is Fig. 7, which plots midshaft tibial rigidity (or polar moment of area, J) against midshaft humeral rigidity (cf. Shaw and Stock, 2013) for the Upper Paleolithic individuals, including BT 2. Samples with increased mobility appear to show a steeper linear fit, as suggested in Shaw and Stock (2013). In a pairwise comparison, the slope of the UP sample (b = 3.573; standard error of b = 0.351) is significantly steeper (P < 0.01 or smaller) than all the comparative samples, except for the Late Stone Age southern Africans (Shaw and Stock, 2013). When a Bonferroni correction for multiple comparisons is applied (for the five samples being compared), the slope for the UP individuals is still steeper when compared to all the comparative samples (P < 0.01 or smaller), except for runners and Late Stone Age southern Africans. BT 2 falls close to the linear fit of the data, indicating that his values were typical of a group with presumably high mobility levels.

The anterior process of the left calcaneus of BT1 is abnormally elongated. The protuberance is oriented towards where the navicular bone would have been during life, although this bone has not survived. This individual also displays abnormal bossing of the dorsal part of the base of the right second metacarpal. A carpal boss, formed by an abnormal dorsal bossing of the left trapezoid and the base of the left second metacarpal was also found in BT2. These osseous modifications are likely related to congenital tarsal and carpometacarpal coalitions. This suggests that these two adult males may have been genetically related (Villotte et al., 2011b).

Other minor abnormalities were also noticed for these individuals. The left olecranon fossa of BT1 (not preserved on the right side) is characterized by a circular depression (with a diameter of 7.5 mm and a depth of about 4 mm) in its upper part, undercutting the fossa (Fig. 4b). This change is likely due to a space-occupying lesion (i.e. a lesion that has impinged on the bone), but we were not able to identify a possible cause. However, it is worth remarking that this change appears completely different from an incomplete humeral aperture. BT1 also exhibits, on both fifth metatarsals, a bone lacuna distal to the articular surface for the fourth metatarsal, the floors of these lacunae resembling remodeled spongy bone (Fig. 4c). The diaphysis of BT2 fifth metacarpal is slightly oblique, maybe in relation with a healed fracture. BT2 exhibits some possible traces of a healed periosteal reaction on the right femur, whereas BT1 displays similar changes on the right tibia.