Recent dental (
1), mandibular (
2), genetic (
3,
4), and demographic (
5) studies have predicted the existence of an as yet unidentified African or West Asian Middle Pleistocene (MP) population that contributed to the evolution of the Neanderthal clade. This contrasts with the traditional view that considered the European continent as the sole place of origin of the Neanderthals and their direct ancestors.
Here we report on the discovery of several fossils from the recently excavated MP open-air site of Nesher Ramla (NR), central Israel (
Fig. 1), in association with stone artifacts, and faunal remains (
6).
A nearly complete right parietal bone and four fragments from the left parietal bone represent the NR-1 fossil (
Fig. 2A and fig. S1). The NR-2 fossil is an almost complete mandible, missing only the left ramus, the right condylar process, and the mandibular angle of the right ramus (
Fig. 3). The lower left second molar (NR-2 M
2) and most of the dental roots are still in place (fig. S2). Both NR-1 and NR-2 were found in situ within the lowest archaeological layer (
Fig. 1D, Unit VI), together with animal bones and flint tools, and most likely represent the same individual (supplementary text A).
Unit VI is assigned an age of 140 to 120 thousand years (ka) ago, based on the electron spin resonance–uranium series (ESR-US) dates of animal teeth recovered in this unit (with a weighted mean of 125.8 ± 5.9 ka). This age was corroborated by a series of thermoluminescence (TL) dates of burnt flints from the archaeological layer immediately above the fossil (Unit V). This layer yielded a weighted mean of 127.6 ± 4.0 ka (confirmed by isochron analysis), which is in agreement with the ESR-US dates obtained for this unit, ranging between ~128 and ~120 ka (a weighted mean of 122.3 ± 3.3 ka). This chronological information is consistent with the previously published optically stimulated luminescence (OSL) dates for the entire archaeological sequence [ranging ~170 and ~78 ka; (
6)].
The preserved anatomical elements were thoroughly described and analyzed in comparison to a large number of fossils of different periods (table S1), using a combination of traditional approaches based on linear and angular measurements, as well as three-dimensional (3D) landmark-based geometric morphometric (GM) methods (supplementary text C to E).
The overall morphology of the NR-1 parietal bones (supplementary text C and tables S2 and S3) is indicative of an archaic, low cranial vault, which is typical of MP
Homo specimens and is substantially different from early and recent
H. sapiens, which instead manifest a curved parietal bone with a pronounced eminence (
7,
8).
Further support for the rather archaic morphology of the NR Homo comes from the angle formed by the coronal and sagittal sutures (c/s angle), 91.1° in the NR-1 specimen. This angle increased during the evolution of Pleistocene Homo (fig. S3): H. erectus and African MP Homo exhibit a mean angle of 92.1° ± 2.1°; the angle opens to 94.9° ± 3.4° in European MP Homo/Neanderthals and reaches 99.4° ± 4.2° in early and recent H. sapiens. The c/s angle is significantly different between these three groups (H = 22.5, p < 0.001). The c/s angle for NR-1 is similar to that of archaic Homo, particularly African MP Homo (91.1° ± 1.1°), and falls outside the range of variation of H. sapiens.
The NR-1 parietal bone is considerably thick, mainly in the parietal eminence area (figs. S4 and S5). Regarding this aspect, the NR-1 parietal is similar to that of European MP Homo specimens (e.g., Petralona, Atapuerca SH) (fig. S4). It is generally thicker than the parietal of Neanderthals (e.g., Amud 1, Guattari, and La Chapelle-aux-Saints) and most early H. sapiens (except for Laetoli H18 and Omo 2), and it is much thicker than that of recent H. sapiens.
The 3D GM analysis, used to assess NR-1 shape variation with respect to a comparative sample of Pleistocene and recent
Homo (supplementary text,
Fig. 2C, fig. S1, and table S1), confirms the archaic morphology of NR-1. The first three principal components (PCs) explain 74.5% of the total shape variance. The first PC (34.9%) differentiates early and recent
H. sapiens from all other groups, including Asian
H. erectus, European and African MP
Homo, and Neanderthals, owing to their marked curvature along both the sagittal and the coronal planes (
Fig. 2C). The second PC (21.3%) is not taxonomically informative (fig. S1). The third PC (18.3%) separates Asian
H. erectus and African MP
Homo from Neanderthals and European MP
Homo (
Fig. 2C), based on the relative development of the parietal eminence and its relative antero-posterior position. The European MP group is characterized by an antero-posteriorly and supero-inferiorly flatter parietal bone (
Fig. 2C). NR-1 is distinct from
H. sapiens; it is at an intermediate position between the Neanderthal and MP
Homo clusters (
Fig. 2C). An unrooted phylogenetic analysis, based on the mean shape of each
Homo group, placed NR-1 close to the origin of the branch leading to African MP
Homo, close to the split from the
H. erectus branch and to European MP
Homo and Neanderthals (including Atapuerca SH), and far from early and recent
H. sapiens (
Fig. 2B).
With regard to the configuration of the endoparietal surface (fig. S5), NR-1 is polygonal, i.e., the surface is clearly oriented according to three distinct planes (fig. S5). Instead, Neanderthals and
H. sapiens manifest an arched endoparietal surface. The flatness of the superior parietal lobule, seen in the NR-1 virtual endocast (fig. S6), is one of the most characteristic features of MP
Homo (
9,
10). Other important characteristics of the NR-1 endocast and that are also typical of MP
Homo are the very low position of the maximum endocranial width at the superior part of the first temporal convolution (fig. S6), the very short parietal lobe (fig. S7), and the differing lengths of the maximal endocranial width and intraparietal width as well as their posterior position on the parietal bone (table S3). These features can sometimes also be seen in Neanderthals (table S3) (
9,
10). Conversely, recent
H. sapiens specimens exhibit subequal maximal endocranial and intraparietal widths, which are located much higher and more anteriorly than in NR-1 (
9).
The vascular pattern of the middle meningeal vessels in NR-1 is simple. Only a few, short ramifications are visible and anastomoses are absent, as is the case in other MP
Homo and Neanderthals (figs. S9 and S10) (
11). The posterior branch of the middle meningeal vessel in NR-1 is as developed as the anterior one, a pattern persistent among MP
Homo. Both Neanderthals (e.g., La Quina H5 and La Chapelle-aux-Saints) and recent
H. sapiens show a dominance of the anterior branch; the latter also possesses complex vascular endocranial imprints (fig. S10).
The NR-2 specimen is a robust mandible (
Fig. 3); the corpus is medio-laterally wide, and the cortical bone is thick (
Fig. 3, fig. S11). Its most pronounced feature is the short ramus relative to the body height, with a sturdy, low, and wide coronoid process (
Fig. 3). This specimen displays several archaic features (e.g., no trigonum mentale or incurvatio mandibulae, a wide incisura submentalis, a developed planum alveolare, a strongly developed planum triangulare, and a mandibular corpus that presents fairly parallel alveolar and basal margins) commonly seen in MP
Homo (
12,
13) (supplementary text D and table S4A).
We combined taxonomically relevant mandibular features into a hierarchical clustering analysis (fig. S12). Modern and Pleistocene humans form the two main clusters: NR-2 is placed on a side branch of the latter, together with MP Homo from Atapuerca SH, Tighenif 3, Arago XIII, and one Neanderthal (fig. S12). The discrete traits underscore the mosaic nature of the NR mandible, showing archaic morphology together with some Neanderthal traits.
The metric dimensions of the NR-2 mandibular body are presented in fig. S13. The symphyseal area is considerably thick (16.6 mm), close to the values of European MP Homo mandibles (16.9 ± 2.1 mm), and moderately tall (33.7 mm), close to the Neanderthal mean (34.0 ± 4.6 mm). The body (measured between the first and the second molar) is thick (17.7 mm), within the range of European MP Homo (18.1 ± 3.1 mm), yet taller (32.7 mm) than that of European MP Homo (30.2 ± 1.6 mm) and Neanderthals (29.9 ± 3.3 mm), close to the values of early H. sapiens (33.0 ± 4.0 mm).
The results of the 3D GM analysis (fig. S14 and tables S5A and S5B) for the NR-2 mandible are illustrated in
Fig. 3C. The first two principal components explain 47.5% of the total variance. Variation along PC1 (37.9%) is driven by changes in the length of the mandibular body, the shape of the ramus (shorter and broader among archaic
Homo), and the expression of the mental area. Variation along PC2 (9.6%) reflects changes in the body height (mainly in the mental region), and the transition from a body’s parallel alveolar and basal margins to ones that converge posteriorly. In the PC1-PC2 plot, early and recent
H. sapiens separate from the other
Homo specimens, whereas European MP
Homo and Atapuerca SH are distinguished from Neanderthals (including the Levantine Amud 1) and Asian
H. erectus. NR-2 falls between Neanderthals and the European MP
Homo specimens (including Atapuerca SH), far outside the range of the variation of
H. sapiens. The phylogenetic analysis, based on the mandibular mean shape of each hominin group, placed NR-2 on a separate branch (together with Tabun C1), close to the split between MP European fossils and Neanderthals, and far from
H. erectus, African MP
Homo, and
H. sapiens (
Fig. 3B). This result, based on metrics alone, largely echoes the results of the cluster analysis based on discrete traits and confirms that NR-2 belongs to an archaic group with Neanderthal affinities.
The lower second molar (NR-2 M
2) is complete and shows some occlusal wear causing a slight exposure of the dentine horns (
Fig. 4A and supplementary text E). The occlusal surface of the NR-2 M
2 reveals four well-developed cusps and a hypoconulid. The presence of five main cusps is typical for most (70%) of the Atapuerca SH
Homo (
14) and Neanderthals (
15). The NR-2 M
2 has a clear continuous mid-trigonid crest and a discontinuous distal trigonid crest on the dentine surface, corresponding to grade 3 of Bailey
et al. (
16) (fig. S15). A mid-trigonid crest is present in more than 90% of Neanderthals and MP
Homo from Atapuerca SH (
14,
15). A grade 3 expression of the mid-trigonid crest, as in the NR-2 M
2, is present in nearly 60% of the Neanderthal specimens, but it is absent in
H. sapiens (
16). The Qesem Cave M
2 specimen (QC-J15) (
17) shows a similar pattern of a continuous mid-trigonid crest and a discontinuous distal trigonid crest (fig. S15). The Ehringsdorf G specimen presents only a mid-trigonid crest (but no distal crest) (fig. S15), whereas the Mauer specimen does not manifest a mid-trigonid crest at all.
The NR-2 M
2 has a single, pyramidal root bifurcating at the apical fourth of the root (
Fig. 4, C and D). The large pulp cavity extends to the middle of the root and branches out into short root canals that extend into the apices, a configuration of the roots known as taurodontism (
Fig. 4). This pyramidal root, with a taurodontic pulp cavity, is frequent in Neanderthals (
18). In modern humans, the second lower molars possess separate mesial and distal roots with some variation in the canals. The root of the NR-2 M
2 (
Fig. 4 and fig. S15) is relatively long (16.4 mm), falling toward the higher end of the range of the variation of both Upper Paleolithic
H. sapiens (11.3 to 16.8 mm) and Neanderthals (14.3 to 16.5 mm).
The 3D GM analysis for the dentinal crown shape (landmark configuration combining the information from the enamel-dentin junction or EDJ, and that from the cemento-enamel junction or CEJ: fig. S16 for the measurement template, table S6 for the landmark definitions, and fig. S17 for the PC1-PC2 plot and the PC1-PC3 plot) showed that the NR-2 M2 falls at the upper distant margin of the Neanderthal range, close to the Krapina specimens and Ehringsdorf G.
Shape variation along PC1 (30.6% of the total variance) is driven by the relative height of the crown and by the bucco-lingual expansion of the EDJ relative to the dentine outline. Like the M
2 of Neanderthals and
H. sapiens, NR-2 M
2 exhibits a relatively high crown and a bucco-lingually expanded EDJ. Along PC2 (14.7%), the NR-2 M
2 plots toward the most extreme range of the distribution, opposite to the
H. sapiens, Atapuerca SH, and African MP specimens. The associated shape is characterized by the expansion of the distal aspect of the dentine crown, a feature that NR-2 M
2 shares with some Neanderthal specimens (Krapina and El Sidrón) and the European MP
Homo Ehringsdorf G (supplementary text E). Differently from the parietal and mandible, the unrooted phylogenetic trees’ construction, based on the combined CEJ-EDJ data (
Fig. 4B), resulted in a clear affiliation with Neanderthals, whereas Qesem QC-J15 associated with Atapuerca SH. Concerning crown size, NR2 M
2 is outside the modern human range (fig. S18).
The cumulative evidence from the three analyzed anatomical elements (parietal bone, mandible, and M2) reveal a unique combination of archaic and Neanderthal features, supporting the existence of a local, Levantine population at the final MP. The results of the quadratic discriminant analyses (QDAs) (table S7) reinforce this observation, showing that an affiliation of the NR fossils with early and recent H. sapiens is highly unlikely, but that it is impossible to establish whether NR fossils are more likely to be classified as MP Homo, Neanderthal, or H. erectus (the latter for the parietal only). Consequently, the discriminant function plot (fig. S1) shows that the NR-1 parietal falls between the H. erectus/African MP Homo group and the European MP Homo/Neanderthals, with a similar likelihood of belonging to either cluster (H. erectus = 0.41, MP Homo = 0.34, Neanderthal = 0.25, based on the first three PCs).
The earliest that Neanderthal features in Levantine fossils have been discernible in the MP was around 400 ka ago at Qesem Cave (
19), the earliest modern humans were present in the Levant around 180 ka ago (
20), and unequivocal Neanderthals did not appear in the Middle East before ~70 ka ago. NR bridges a gap in this record, by displaying a highly heterogeneous, yet archaic morphology. The parietal documents a rather archaic shape of the braincase; the mandible is similar to that of MP
Homo; the molar is quite Neanderthal-like, similar to Ehringsdorf G.
Arsuaga
et al. (
21) advocated an earlier evolutionary development of the masticatory apparatus, compared with the braincase in Neanderthals. Similarly, the Jebel Irhoud fossils from North Africa possess a more primitive neurocranium but a more
H. sapiens–like face and dentition (
22). Archaic populations carrying Neanderthal-like features were also present across much of the Eurasian continent during the MP, revealed by the Chinese findings of Maba, Xujiayao, and Xuchang (
23–
27). The existence of MP Asian populations deviating markedly from the
H. erectus paradigm has been repeatedly proposed, for instance, for the Tongzi teeth (
28) or the Sambungmacan 3 cranium (
29); the latter (together with Ngandong 6 and 7) shows strong morphological affinities with the NR-1 parietals.
The NR fossils could represent late-surviving examples (140 to 120 ka) of a distinctive Southwest Asian MP
Homo group, predating Levantine Neanderthals from Amud, Kebara, and Ein Qashish (70 to 50 ka). On the basis of their mosaic morphology showing a different degree of Neanderthal features, other MP Levantine fossils, whose taxonomic affinities have long been debated, from the sites of Qesem Cave (
19), Zuttiyeh Cave (
30), and probably Tabun Cave (
31), might also be attributed to this group (supplementary text F). Adopting the cautious approach advocated by Mayr (
32), we suggest addressing this Levantine MP paleodeme as the “Nesher Ramla
Homo.” Its presence from ~420 to 120 ka ago in a geographically restricted area may have allowed for repeated interbreeding with modern human populations such as the people from Misliya Cave (
20), a notion also supported by their shared technological tradition [(
6); supplementary text F]. This scenario is compatible with evidence of an early (200 to 400 ka ago) gene flow between modern humans and Neanderthals (
3,
4) and helps explain the variable expression of the dental and skeletal features of later Levantine fossils from the Skhul and Qafzeh populations, a phenomenon noted by anthropologists since the 1930s (
31,
33). Moreover, a recent study of the Atapuerca SH and Arago dental remains (
1) suggested the existence of more than one
Homo lineage in MP Europe [see also (
34)] and hypothesized the contribution of Levantine
Homo groups carrying Neanderthal-like traits to European
Homo lineages. The NR
Homo, carrying Neanderthal-like traits, could thus represent the “source” population postulated in the demographic “sources and sinks” model (
5), according to which Western Europe was repopulated through a series of successive migrations.
Acknowledgments
M. Prévost made the drawings in
Fig. 1. P. Hervé made the drawings in fig. S10; A. Ehrenreich photographed the parietal bone in figs. S1D and S9A. E. Santos performed the virtual reconstruction of the SH mandibles. F. L. Bookstein supplied the R script for the quadratic discriminant analysis. J. J. Hublin, Department of Human Evolution, Max Planck Institute for Evolutionary Anthropology, Leipzig, Germany, supplied some of the specimens for the study. Micro–computed tomography scans of the NR fossils were done by S. Ellenbogen at the Shmunis Family Anthropology Institute, Dan David Center for Human Evolution and Biohistory Research, Tel Aviv University.
Funding: This work was funded by grants from the Dan David Foundation; the Shmunis Family Anthropology Institute; the Leakey Foundation; the Care Archaeological Foundation; the LabEx Sciences Archéologiques de Bordeaux (LaScArBx ANR-10-LABX-52); the Dirección General de Investigación of the Ministerio de Ciencia, Innovación y Universidades, grant nos. PGC2018-093925-B-C31 and C33 (MCI/AEI/FEDER, UE);and the Israel Science Foundation (1936/18, 1773/15). C.F. and V.A.K. were financially supported by the Swiss National Science Foundation (grant nos. 31003A_156299/1 and 31003A_176319). V.S. acknowledges funding from the Alon Fellowship.
Author contributions: I.H., H.M., R.S., G.W.W., and Y.Z. conceived the project, analyzed the data, and wrote the manuscript. C.F., V.A.K., and Ar.P. performed the investigation of the 3D datasets, from segmentation to data collection and morphometric analysis. C.F. and G.W.W. performed the QDA analysis. D.G.H., L.A.-B., and E.B. performed the endocast analysis. R.Q., J.-L.A., C.F., and V.S. helped interpret the work and supervised writing the manuscript. Y.Z. provided data on the lithic industry, site formation, environmental conditions, and subsistence. M.M.-T., J.-M.BdC., L.M.-F., A.V., T.S., G.M., A.Po., and F.D.V. provided crucial data on their fossils and thoroughly discussed the manuscript. All authors drafted the manuscript. Ar.P. prepared the figures.
Competing interests: The authors declare no competing interests.
Data and materials availability: Data related to the new fossils are available from the Shmunis Family Anthropology Institute website (
https://sfai.tau.ac.il/virtual_fossils_archive). Formal applications to access the fossils should follow the regulations listed at
https://en-med.tau.ac.il/dan_david_center.