Home / Science / Rates of dental evolution and its implications for the divergence between Neanderthal man and modern humans

Rates of dental evolution and its implications for the divergence between Neanderthal man and modern humans


The origin of Neanderthal and modern human lineages is the subject of intense debate. DNA analyzes generally indicated that the two lineages had diverged during the middle Pleistocene average period, an inferred period that strongly influenced interpretations of the hominin fossil record. This divergence period is however not compatible with the anatomical and genetic affinities of Neanderthal observed in the Middle Pleistocene hominins of Sima de los Huesos (Spain), dated 430 000 years ago (ka). . Based on quantitative analyzes of dental evolution rates and Bayesian analyzes of phylogenetic hominin relationships, I show that any divergence of time between Neanderthals and modern humans under 800 ka would have led to evolution. unexpected and rapid dentistry in the Neanderthals of Sima de los Huesos. These results confirm a last common ancestor before 800 ka for Neanderthals and modern humans, unless mechanisms until now unexplained accelerate the dental evolution of the primitive Neanderthals.


The timing and identity of the last common ancestor (LCA) of Homo Neanderthalensis and Homo sapiens (hereinafter referred to as Neanderthals and modern man) are intensely debated issues (15). Ancient DNA (DNA) studies have generally demonstrated a time of divergence of about 400 thousand years ago (ka) (6), which found evidence in some quantitative studies of cranial variation (7). In addition, the generally discussed evolutionary scenarios tend to assume that at least some Middle Pleistocene hominins dated between 600 and 400 ka, or even younger, were part of the older ancestral species. common to Neanderthals and modern humans.[Examinédans([Reviewedin([examinédans([reviewedin(8)]. Numerous anatomical studies on fossils, however, have shown that some Middle Pleistocene European hominins, especially those belonging to the Sima de los Huesos (SH) sample, show clear affinities with Neanderthals (911). After contradictory results concerning the geological age of SH hominins (12, 13), this collection is now well dated at 430 ka (14), age confirmed by length analysis of its mitochondrial DNA branch (mtDNA) (15). In addition, recent analyzes of nuclear DNA (nDNA) in this population have demonstrated an evolutionary affinity of SH hominins with classical Neanderthals (16), thus making the divergence between Neanderthal man and modern man necessarily older than the age of SH fossils. Some recent studies reflect these new discoveries and advocate an earlier age for this LCA of 550 to 765 ka (17) on the basis of more recent estimates of the human mutation rate (16). The divergence times deduced from the genomic data strongly depend on the mutation rate and generation time estimates, which are still under discussion (18). Small variations in these parameters can lead to very different estimates of the divergence time between two species. If these nuances are not taken into account, a strict reading of the values ​​provided by the ADNa analyzes may give rise to radically different interpretations of the fossil record, which may even be incompatible with the affinities deduced from anatomical evidence.

The greater evolutionary affinity of SH with Neanderthals compared to modern humans indicates that HS hominins diverged from the modern human lineage at the same time as classical Neanderthals. Therefore, genetic affinities, geological age, and morphological variation of SH hominins can be used to infer the moment of divergence between Neanderthals and modern humans. Recent studies on the variation of hominin have shown that, unlike other characters, the post-limestone dental form described in geometric datasets (Figure S1) has evolved neutrally and has extremely homogeneous speeds in all lineages of hominine (19). This observation was used in this study to infer when Neanderthals and modern humans should have diverged to maintain the rate of evolution of the phylogenetic branch of the dental form leading to SH hominins in the same interval. variation that observed in other hominin species (Tables S1 and S2). The dental form of the SH hominins is unexpectedly derived to the Neanderthal condition, both in the expression of the discrete characteristics of Neanderthal (9) and in its extreme degree of post-canine structural reduction in the number and size of cusps (Figure S2) (11). The dental form of SH hominins is so derived that it is not representative of other Neanderthal populations. However, this does not affect the design of this study. Although SH hominins do not show the average tooth shape observed in classical Neanderthals, their highly derived dentitions must have evolved from the same ancestral form as classical Neanderthals and during the time period between SH hominins and Neanderthals. 39, Neanderthal man – the modern man ACV (see Fig. 1). The homogeneity of the evolutionary rates for the dental form differs markedly from the much more heterogeneous scenario observed for the size of the teeth, for which different rates are observed in different branches of the phylogeny of the hominin (19).

Fig. 1 Phylogenetic scenarios and SH dental morphology.

(A) Phylogeny of hominin used in evolution rate analysis (phylogeny-1). The SH branch is represented in teal and the LCA branch in orange, which are the colors used to represent the rates of change on these two branches in Figs. 4 and 5 and FIG. S5. The gray lines represent the different divergence times that were evaluated. (B) Transformation of the Neanderthal-Denisovan-SH line into the SH line. (C) Densitree showing a sample of 100 randomly selected phylogeny[Surl'échantillontotalde60000phylogéniegénéréparl'analysebayésiennedesrelationsphylogénétiquesdel'hominineparDemboetsescollègues([Ofthetotalsampleof60000phylogeniesgeneratedbyDemboandcolleagues'Bayesiananalysisofhomininphylogeneticrelationships([surl’échantillontotalde60000phylogéniegénéréparl’analysebayésiennedesrelationsphylogénétiquesdel’hominineparDemboetsescollègues([ofthetotalsampleof60000phylogeniesgeneratedbyDemboandcolleagues’Bayesiananalysisofhomininphylogeneticrelationships(20)]. The original Dembo trees have been pruned to keep only those species for which dental data is available. The length of the Neanderthal branch has been shortened to reflect the age of the SH branch. (re) Upper and lower post-canine teeth of a representative SH individual (the upper dentition is shown on the left). Photo credit: A. Muela, photographs taken at the Institute of Health Carlos III.

To explain the lack of consensus on the phylogenetic relationships of hominin, the analyzes were based on two different phylogenetic frameworks (Figure S3) (19, 20). The first (phylogeny-1) is the phylogenetic tree used in a previous study on hominin evolution rates (19), which is based on the dates of first and last appearance of these hominin species for which data on the shape of all posterior teeth were available. The second phylogeny (phylogeny-2) is the tree of maximum credibility of the clade calculated by Dembo and his colleagues (20, 21) as part of their Bayesian analysis of the phylogenetic relationships of hominin. This phylogeny was pruned to include only species for which dental data were available. In these two phylogeny, the age of the Neanderthal – modern human ACV has been changed to 500 ka, which is just below the lower limit of the range suggested by the molecular analyzes the most recent (16, 17, 22) at the age of the underlying node at 100 ka intervals (Fig. 1). Uncertainty regarding phylogenetic hominin relationships and length of branches was explicitly addressed by estimating rates of change on a sample of 100 trees. This sample of trees was randomly selected from a sample of 60 000 trees generated by the Bayesian analysis of the phylogeny of hominin (20, 21) (Fig. 1). Denisovans (23), which diverged from classical Neanderthals after human divergence between Neanderthal and modern but before the age of SH fossils (16), were not included in these analyzes because very rare phenotypic data are available for this group. However, given their evolving relationships (24), Denisovans, as SH homininins, can be considered as part of the wider Neanderthal lineage or H. neanderthalensis sensu lato (Fig. 1)

The methodological approach used consisted of a three-step process that included the calculation of ancestral values ​​using a multiple variance Brownian motion (mvBM) approach (25), calculating the magnitude of the change by branch as the difference between the descending and ancestral morphologies, and comparing these values ​​with those obtained by simulating the evolution at a constant rate in all branches of phylogeny of hominin (19). The major advantage of this approach is that it gives a specific and quantitative account of the possibility that the Neanderthal LCA and modern humans (or any other species) do not have an intermediate morphology between the two species. girls but more like Neanderthals. This is a scenario that has recently been suggested to explain the presence of Neanderthal features derived in the SH sample (4) and even in earlier European hominines (26), but this has not yet been formally tested.


Changing the time of divergence between the HS and the modern human branches has strong effects on the length of the SH branch and the anterior branch, as well as on their associated rate of evolution. Very late – modern human divergence times result in very short lengths for the SH branch, resulting in very fast rates of change for this lineage. On the other hand, too early human-human divergence times lead to very short lengths for the phylogenetic branch leading to their LCA, which results in a very high rate of evolution for this branch. Figure 2 shows how the evolution rates associated with the modern Neanderthal human clade (those corresponding to the SH branch, the modern human branch and the ACL branch) differ considerably when the time of human divergence is modified. SH – modern described above. Since only the Neanderthal – modern human ACV calendar is allowed to change, there is an inverse relationship between the rate of evolution of the branch leading to SH hominins (or Neanderthals) and the branch. underlying, so that a slower rate in the SH branch is associated with faster speed in the underlying branch (Fig. 3 and Fig. S4).

Fig. 2 Branch-specific evolution rates obtained by analysis of phylogeny-1.

(A) Evolution rates obtained by setting the modern SH human divergence time at 0.5 Ma there is. (B) The rates obtained by fixing this divergence at 0.9 Ma should occur, which corresponds to the minimum SD of all the rates of the tree. (C) Rates obtained by fixing the divergence at 1.4 Ma there is. SH – modern human divergence time greater than 1.4 Ma results in even higher rates for the pre-separation branch SH – modern human separation, referred to in the following figure as the ACV branch. Evolutionary rates are shown above each branch (gray for rates that remain roughly constant in all scenarios and black for rates associated with the modern Neanderthal human clade, which are affected by changes in divergence time modern human SH).

Fig. 3 Relationship between rate of change in HS agencies and LCA agencies.

Relation observed during the analysis of the first phylogenetic scenario (phylogeny-1). The evolutionary rates in both branches show an inverse and non-linear relationship such that very high rates in the SH branch are associated with very low rates in the LCA branch and vice versa. This effect can be visualized in FIG. 2, which shows how these rates change according to the time of supposed human divergence SH-modern.

The analysis of 100 phylogeny gives very few cases (3 out of 100) where the SH branch shows the highest rate in the complete tree, but a majority of cases (59 out of 100) where the anterior branch presents the highest rate. higher in the tree (Fig. 4). According to these results, scenarios with a divergence time greater than 0.75 million years (Ma) as a result of which the ACV branch exhibits the highest rate of change are more likely than scenarios with a younger divergence time (Figure 5). The fact that the branch leading to the modern SH human clade tends to display the highest rate of evolution in most phylogeny shows that the dental divergence was stronger in the later stages of gender evolution. Homo.

Fig. 4 Change in evolution rates obtained by the analysis of 100 trees.

(ADensity tree showing the sample of 100 randomly selected trees used in the calculations. (B) Box-whiskers comparing peak rate of change (gray), ACL level (orange) and SH (teal) in 100 phylogenies. (C) Evolutionary rates obtained during the analysis of each of the 100 phylogeny showing the maximum rate on the tree (in gray), the ACL level (in orange) and the rate of SH (teal). The phylogeny of (C) are sorted according to their maximum evolutionary speed. The graph shows that the ACL level is the maximum rate in a majority of phylogeny (59 out of 100), while the SH rate is the maximum rate only in three phylogenies. In all other cases, the maximum rate is found in the other branches (in most cases in P. boisei plugged).

Fig. 5 Most likely, the time of divergence between Neanderthals and modern humans is based on the analysis of phylogeny-1.

(A) Comparison of observed standard deviations of all rates of hominine phylogeny (red dots) with standard deviation distributions obtained when simulating evolution on the same tree at a constant rate. (B) Comparison of evolution rates at SH (teal), ACL branch (orange) and all other branches (gray) obtained for different times of divergence SH – modern man. (C) Comparison of SH (teal line) rate with rate interval of 95% of rates obtained for this branch by means of 100 phylogeny analysis (gray box). (re) Comparison of the ACL level (orange line) with the 95% interval obtained for this branch by the analysis of 100 phylogenies (gray box). Dashed black lines frame the most likely divergence times according to each analysis. The red dotted lines indicate the minimum and maximum values ​​obtained across all the analyzes and set the most likely divergence time when all the results are considered together. Equivalent results based on the analysis of phylogeny-2 are provided in FIG. S5.

The zero expectation that the dental form has evolved in a neutral way through the complete phylogeny of hominin is only accepted if the divergence between Neanderthal and the human being. Modern man is in the range of 0.7 to 1.2 Ma (Fig. 5A and Table S3), which strongly suggests an outside divergence time. this interval. The expectation of a neutral dental evolution is corroborated by previous studies (19) and was tested against simulated scenarios reflecting genetic drift and excluding selection (27). The standard deviation (SD) of the evolution rate on the tree reaches its minimum value there is 0.9 Ma, although the standard deviations of the tree are low and very similar for the 0.7 to 1.1 Ma interval. Figure 5B shows that the flow rates corresponding to the SH branch and the underlying branch become equal when the divergence time is set between 0.7. and 0.8 Ma before. Divergence times significantly younger or older than 0.75 Ma result in rates of change for the SH branch or for the anterior branch, which are extremely far from the range of variation observed for all other branches (Figure 5B). The rate of change at the SH branch level falls within the 95% range calculated for this branch by analysis of 100 phylogenies only when the divergence time between Neanderthals and the modern man is older than there is 0.8 Ma (Figure 5C). Since the 95% rate range for the anti-retreating branch is very wide, most divergence times are consistent with the values ​​calculated for this branch (Figure 5D). The combined result of all these analyzes gives an interval of between 0.8 and 1.2 Ma as the most likely divergence time for the SH branch and the modern human branch and, therefore, for the lineages. Neanderthal and modern human. Restart these analyzes using the My Client Center tree calculated by Dembo and his colleagues (20) provides even older divergence times, with a minimum divergence time of 0.9 Ma previously calculated from the combination of all analyzes (Fig S5 and Table S4).

Assuming a divergence between Neanderthals and modern humans there are about 600 ka, the age that the most recent molecular studies seem to indicate (16, 17, 22), would have consequences on the rates of SH dental evolution. First, the SD of all rates of hominine phylogeny would show an unusually high value (while still within the range obtained), for 1000 simulated neutral scenarios (P = 0.033 for phylogeny-1; see Figure 5A and Table S3). Second, assuming a divergence time of 600 ka there would be reason to believe that the rate of evolution at the SH branch level was the highest of the phylogeny of hominin (1.3 times greater than evolution rate at the LCA branch level). According to the analysis of 100 different phylogeny sampled in the Dembo study (20this scenario is unlikely (Fig. 4). In addition, the rate of change at the SH branch level in a divergence scenario of 600 ka would be 1.99, a value well outside the 95% rate range of observed rates for the SH branch through the analysis of 100 Dembo trees (Figure 5C). . An evolutionary rate of 1.99 at the SH branch is less than a single value observed in the analysis of 100 phylogenies (2.05), which is clearly aberrant with respect to all observed rates at this branch (Fig. 4B). The results of the various analyzes performed in this study show that SH hominins must be separated by at least 400 kA from modern human Neanderthal ACN to maintain the rate of SH hominin evolution in the range of variation observed for the other hominins. Therefore, do a ca. A divergence of 600 ka consistent with similar rates of evolution between SH hominins and other hominin species would require about 200 ka for SH hominins, which is considerably younger than all calculated values ​​for this hominin. population (1214).


The evolutionary rates measured in this study are strongly influenced by the length of the branches, so that small branches accumulating strong dental changes cause high rates. Young times of divergence between Neanderthals and modern humans lead to short SH branches and, consequently, the high rates of evolution observed for SH hominins. Therefore, if SH hominins were less than 430 ka, then they would be consistent with the divergence time between Neanderthals and modern humans after 800 ka without requiring extremely high rates of evolution. Specifically, the ca. Divergence of 600 ka indicated by the most recent molecular estimates (22) would be consistent with the average change rates for the SH sample if these hominins were as young as 200 ka. This scenario is worth considering since the age of HS hominins has been the subject of controversial discussions in the past (2, 12, 13). The most recent studies, based on luminescence and paleomagnetic analyzes, safely indicate an age of 430 ka for these fossils (14). This figure is further corroborated by genetic analyzes of SH hominins dating back to about 400 ka based on the length of its mRNA branch, with the highest 95% posterior density range between 150 and 650. ka (15). This range is certainly wide enough and implies that SH hominins may be younger than 430 ka. On the basis of these data, however, they can also be considerably older, which would necessarily push the divergence between Neanderthals and modern humans to an even older date. Additional evidence to support one ca. 430 ka for the SH sample comes from other molecular studies. These studies demonstrate that SH hominins share the same lineage of tDNA as Denisovans, unlike Neanderthal and modern human (15). According to Posth and his colleagues (28), the Denisovan-SH mtDNA line is the primitive Neanderthal clade line, and the classical Neanderthal mDNA line was subsequently acquired through an introgression event of 219 at 468 ka before. If this model is correct, then the SH population must precede this introgression event, which provides additional support at an age> 400 ka for the SH sample. Therefore, on the basis of the current combined geochronological and molecular evidence, a postulate of about 430 ka for SH hominins is the most reasonable assumption, so other explanations are needed to make current results. Also related to branch lengths, it can be argued that the analytical approach presented in this study favors the times of divergence between Neanderthals and moderns, as it assumes longer branch lengths (and thus higher rates of occurrence). slower evolution) for other species of hominins. This potential bias is explained using 100 different phylogenetic scenarios based on Bayesian analyzes of the phylogenetic relationships of hominin (20), some of which show branch lengths for other species as short as the SH branch. Nevertheless, analyzes based on these 100 phylogenies also indicate that the times of divergence between Neanderthal and modern humans are very unlikely there are less than 800 ka. This means that it is unlikely that methodological artifacts will influence the observed results. Biological factors are therefore necessary to explain them.

A rapid rate of change in lower Neanderthal populations represented by SH hominins, which would be a necessary consequence of a divergence between Neanderthals and modern humans after 800 ka. , may result from a strong selection of the dental form of these hominins. Although this scenario is initially plausible, it is also very unlikely that the evolution of the initial segment of the Neanderthal line was characterized by a rapid dental course that is not observed in any other species of hominin (19) (not even those of the kind Paranthropus, which are characterized by an extreme degree of post-channel megadontia). This strong selection scenario is unlikely for two reasons. First, the differences in tooth shape seen in SH hominins compared to a hypothetical ancestral morphology (3), as well as more primitive configurations than those observed in Homo erectushave no functional significance and are considered to be selectively neutral (29). Therefore, it is highly unlikely that these dental variations were the target of rigorous selection implicated by unusually fast rates of change. Second, the dentition of SH hominins is the only skeletal region that has a highly derived state. Other features related to chewing, such as facial and mandibular anatomy, show clear affinities between Neanderthals in SH hominins, but not in the state of hyper-derived Neanderthal found in their dentitions (ten), implying lower rates of change. A strong selection scenario associated with a functional benefit would almost certainly involve other cranial regions outside the teeth. The transition state of most other characters of SH hominines most likely indicates that selection was not the main factor in SH dental evolution.

As already mentioned, SH hominins show a dental anatomy that is not representative of the Neanderthal average but clearly more derived. This observation, however, does not affect the design of this study or its results. The study design does not require that SH dental anatomy be representative of the wider Neanderthal variation range. It is simply based on the fact that the SH dental features, whether representative or not, have evolved from the same ancestral condition as that of classical Neanderthals, from the period separating SH hominins for ACL. human between Neanderthal and modern. Therefore, this study does not consider SH dental anatomy as representative of classical Neanderthals, but only as the characteristic dental form of the SH population, given its evolutionary relationships and geological age. Given this unrepresentative and highly derived state of SH dentition, a plausible explanation for the rapid dental outcome implicated by a divergence posterior to 800 ka indicates that SH dental anatomy results from a strong founder effect. In this scenario, the ancestral populations of SH hominins would have had different dental morphologies, one of which would have been fixed in the SH sample as it was present in their direct ancestors. This scenario is theoretically possible and could be supported by the geographical location of SH hominins on the Iberian Peninsula, where they may have been more isolated than other Neanderthal populations in continental Europe. However, this scenario would imply that the SH dental phenotype was present, albeit to a small extent, in early Middle Pleistocene populations whose HS hominines evolved. Due to the scarcity of the fossil record, this scenario can not be ruled out at present, but current fossil hypodigms do not show these derived dental configurations in any other hominin prior to the SH population, weakening this assumption.

Hybridization is another factor likely to have affected the dental evolution of SH hominins. On the basis of the genetic analyzes, it is now confirmed that the hybridization took place between Neanderthals, modern humans and Denisovans (30, 31), probably quite often. Therefore, it is safe to assume that different Middle Pleistocene hominin lineages hybridize when in contact. The high degree of mosaicism found in the SH population, with some traits showing a totally Neanderthalian condition and others showing a much more primitive state, could potentially indicate a hybrid origin. However, SH hominins do not show the skeletal abnormalities found in first generation hybrids of live primate species, such as the presence of rotating or supernumerary teeth, and sutural anomalies in the neurocranium and the face (32). Although a hybrid origin of SH hominins is certainly possible, this hypothesis does not rely on particularly strong support, based on their anatomy or what we currently know about the phenotypic effects of hybridization.

L'explication la plus simple des résultats présentés dans cette étude est que les hommes de Néandertal et les hommes modernes ont divergé avant 0,8 Ma, ce qui rendrait les taux d'évolution de la dentition SH à peu près comparables à ceux observés chez d'autres espèces. Ce temps de divergence est nettement plus ancien que les estimations les plus récentes basées sur l'ADN ((16, 17, 22), mais pas si loin des estimations précédentes datant de cette divergence à env. Il y a 800 ka (24). Les estimations du temps de divergence entre les hommes de Néandertal et les humains modernes basées sur l'ADN diffèrent considérablement (6, 17, 22, 24), indiquant qu'une lecture stricte de ces valeurs ne peut pas conduire les interprétations du disque fossile d'hominin. En outre, le temps de divergence obtenu à partir de l'analyse des taux dentaires est étonnamment similaire au temps de divergence des lignées d'ADNmt humain SH-Denisovan et Neanderthal – modernes. La divergence entre les deux lignées d’ADNmt a été estimée à environ Il y a 1 Ma, avec un intervalle de densité postérieure le plus élevé de 95% compris entre 0,7 et 1,4 Ma (15). Comme expliqué ci-dessus, on pense que la lignée de l'ADNmt de Neandertal est le résultat d'un événement d'introgression relativement récent survenant chez l'homme moderne (28). Par conséquent, le temps de divergence des lignées d’ADMt humain et moderne de SH reflète beaucoup plus précisément la divergence population de Néandertalien – population humaine moderne que le temps de divergence entre les lignées d’ADNt humain de Néandertal et moderne, qui reflète le temps maximum imparti pour l'événement d'introgression et est sensiblement plus jeune. La divergence des ADNmt des hominines SH et des êtres humains modernes est toutefois encore plus ancienne que le temps de division de la population estimé à partir de l’ADNn entre la lignée humaine moderne et la lignée Neanderthal-SH-Denisovan, qui a récemment été calculé entre 550 et 765 ka (16, 17) ou 520 à 630 ka (22). Le temps de divergence de l'ADNm indique le moment où les deux lignées d'ADNm ont commencé à accumuler des mutations indépendamment, alors que le temps intermédiaire de la population représente le dernier moment où les deux groupes ont échangé du matériel génétique par divergence et tend donc à être plus jeune que les estimations de l'ADNm. Les résultats de la présente étude suggèrent que la différenciation phénotypique en morphologie dentaire a commencé avant que la division de la population entre les lignées de Néandertal et les lignées humaines modernes ne soit complète. Bien qu'il soit possible que ces divergences résultent des diverses approches méthodologiques utilisées dans différentes études sur la variation génétique et phénotypique (7, 16, 17, 22), il est également possible que ces différences reflètent des signaux biologiques divergents associés à des traits différents. Dans ce cas, le cadre temporel plus ancien de divergence phénotypique suggéré par la variation dentaire aura de profondes répercussions sur la façon dont nous interprétons les enregistrements d'hominine fossile et les relations entre les spécimens fossiles, en particulier pour les populations et les périodes pour lesquelles l'ADNa n'est pas disponible.

Si l'ACV phénotypique des hommes de Néandertal et des hommes modernes était supérieure à 800 ka, cela impliquerait que tous les hominines fossiles plus jeunes que cet âge ne sont plus des candidats valables pour occuper cette position ancestrale. Certains fossiles plus jeunes que cet âge, cependant, sont souvent considérés comme faisant partie de la dernière espèce ancestrale commune aux Néandertaliens et aux humains modernes (2, 8). Ces fossiles, généralement attribués à Homo Heidelbergensis, comprennent des spécimens européens et africains, tels que Mauer, Arago, Petralona, ​​Bodo, Kabwe, etc., et peut-être même des spécimens asiatiques. Si les Néandertaliens et les humains modernes ont divergé avant 800 ka, tous ces fossiles doivent alors être liés aux Néandertaliens ou aux humains modernes, ou ils peuvent faire partie d'une lignée soeur pour les deux. Cependant, ces fossiles ne peuvent pas être ancestraux aux Néandertaliens et aux humains modernes car ils retarderaient leur divergence évolutionnaire. Une relation évolutive entre ces fossiles et les hommes de Néandertal et les humains modernes ne serait possible que s’ils faisaient partie d’une ancienne espèce ancestrale qui a persisté dans le temps en tant qu’espèce relique après la scission réelle des deux lignées. Effectivement, ce scénario signifierait que le H. heidelbergensis les fossiles font partie d'un groupe frère des Néandertaliens et des hommes modernes, mais le changement évolutif par rapport à leurs populations supposées ancestrales ne comportait pas de spéciation.

Il a été suggéré que le processus d’acquisition d’une anatomie entièrement néandertalienne aurait peut-être commencé plus tôt et aurait été plus progressif que le processus d’acquisition d’une configuration humaine entièrement anatomiquement moderne, qui n’apparaîtra dans les archives fossiles qu’après ca. Il y a 200 ka (4). Les traits humains modernes naissants sont observés dans les archives fossiles à env. Il y a 300 ka (33), ce chiffre correspond aux estimations récentes fondées sur l'ADN de la divergence humaine moderne entre 260 et 350 ka (18). Cela contraste avec l'observation d'une dentition totalement néandertalienne (qui peut même être considérée comme hyper hyperandandhale) à 430 ka auparavant chez les hominines SH. Les divergences entre les dates auxquelles des affinités claires entre l'homme de Néandertal et l'homme moderne sont observées dans les archives d'hominine peuvent sembler indiquer des taux d'évolution différentiels dans les deux lignées, ce qui affecterait les inférences faites dans la présente étude. However, they may simply reflect the incompleteness of the fossil record, particularly for the modern human lineage, as the SH sample is the only early Neanderthal population represented in the fossil record that shows such a derived dentition. New fossil findings, as well as the reassessment of previously known ones, are essential to shed more light on the process of acquisition of a fully anatomically modern human configuration.


Experimental design

The major goal of this study was to measure evolutionary rates for dental shape in the earlier part of the evolution of the Neanderthal lineage and to compare them with the rates observed in other hominin species. The calculation of these evolutionary rates assumed different phylogenetic scenarios and different divergence times between Neanderthals and modern humans to determine the effect of these sources of uncertainty on the inferred rates. The results of these analyses have important implications regarding the mechanisms promoting dental evolution in early Neanderthals, the most likely divergence time between Neanderthals and modern humans, and, more generally, the interpretation of the Middle Pleistocene fossil record. The experimental design consisted of a three-step process including (i) the calculation of ancestral dental shapes at all the nodes of the hominin phylogeny using an mvBM approach (25), (ii) the calculation of the amount of change per branch as the difference between descendant and ancestral dental shapes, and (iii) the comparison of the observed amounts of change per branch with those expected when simulating evolution at a constant rate across all the branches of the phylogeny (19). The data and methodological approaches used in the study are explained in detail below.

The data

Species-specific dental shape was calculated for eight hominin species for which data on the variation of all postcanine teeth (upper and lower premolars and molars) were available. This sample included Australopithecus afarensis, Australopithecus africanus, Paranthropus robustus, Paranthropus boisei, Homo habilis (including H. habilis and Homo rudolfensis) H. erectus (including only Asian specimens), SH sample (as representative of H. neanderthalensis), and H. sapiens (table S1). Classic Neanderthals were not included in the analyses because their exact relationship with the SH fossils (which can be directly ancestral or a sister group within the Neanderthal lineage) is currently unknown (16). The analysis of the SH fossils is deemed substantially more relevant than the analysis of classic Neanderthals because they are closer to the divergence point between the Neanderthal and the modern human lineages, thus allowing for a finer-detailed analysis. When classic Neanderthals are used, a divergence time of 500 ka ago yields an average evolutionary rate for the Neanderthal branch and results that generally agree with the expectation of similar evolutionary rates across all the branches of the hominin phylogeny (fig. S6) (19). A 500-ka divergence, however, is younger than the youngest bound provided by the most recent molecular and anatomical estimates, which indicates that fossils that are further from the Neanderthal–modern human divergence point do not provide enough resolution to time this divergence.

Specimens with a clear taxonomic affiliation with one of these eight groups were included in the analyses. Sample size for the different species differed substantially, ranging in most cases from 3 to 53 specimens per species and tooth position, with only three cases where sample size is smaller (table S2): M2 and M3 for A. afarensis (n = 2) and P4 for A. africanus (n = 1). Considering all teeth together, sample size ranged from 5 individuals represented by at least one tooth position (P. robustus and P. boisei) to 53 individuals represented by at least one tooth position (H. sapiens), with intermediate values for the other groups (table S2). This variation in sample sizes, however, is unlikely to affect results, as previous analyses based on jackknifing (reducing all sample sizes to n = 3) and bootstrapping have demonstrated that the constant evolutionary rates for dental shape in which the present study relies are very robust to sample size and composition (19). Shape variation was described using configurations of landmarks and semilandmarks placed on occlusal photographs of premolars and molars and that have been used in previous studies of hominin dental variation (fig. S1) (3, 19). Procrustes superimposition (34) was used to remove non–shape variation corresponding to the position, size, and orientation of specimens. Procrustes superimposition was carried out for each tooth position separately, but information related to each tooth was later merged to study all postcanine variation together (19). A principal components (PC) analysis of Procrustes-superimposed coordinates was carried out, and PC scores were used in subsequent calculations. Variation in dental size was not considered because it is much more heterogenous than variation in dental shape, with some branches showing substantially faster rates than others (19). Dental traits are considered to be a good proxy for neutral genetic data because they tend to be highly heritable and selectively neutral (29).


The uncertainty about hominin phylogenetic relationships was addressed in different ways. First, two different phylogenetic scenarios were explored (fig. S3). The first one (phylogeny-1) is based on the first and last appearance dates of different hominin species (35), and it reflects the most broadly agreed hominin phylogenetic relationships (19, 36). The second one (phylogeny-2) corresponds to the MCC tree obtained as part of a previously published Bayesian analysis of hominin phylogenetic relationships (20). This phylogeny was pruned to include only those species for which data on dental shape variation were available. The major differences between the first and the second phylogenetic scenarios concern the total length of the tree measured as the patristic distance (the sum of all the branches separating two given species) between the most basal node and the H. sapiens tip (approximately 4.5 Ma for phylogeny-1 and 6.2 Ma for phylogeny-2), the lengths of the different branches, and the phylogenetic position of A. africanus, which is placed as a sister group to all Paranthropus and Homo species in the first phylogeny (19) and only to Paranthropus in the second (20). The phylogenetic position of A. africanus, however, is unstable across different studies, with previous analyses setting it as a sister group only to Homo (21). On the basis of previous studies demonstrating an evolutionary relationship between Neanderthals and SH hominins (16), the phylogenetic branch leading to Neanderthals was replaced by the branch leading to SH. This was attained by changing the length of the Neanderthal branch so that it reflects a geological age for the SH sample of 430 ka (14).

Using these two phylogenies, the age of the Neanderthal–modern human LCA was changed from 500 ka to the age of the node separating H. erectus from the Neanderthal–modern human lineage (1.7 Ma in phylogeny-1 and 2.6 Ma in phylogeny-2) at 100-ka intervals. The ages of all the other nodes—and, consequently, the other branch lengths—were kept constant. Variation in evolutionary rates across all these different divergence time scenarios was assessed and compared with results based on the analysis of different phylogenetic topologies. The use of these different phylogenetic trees explicitly addressed phylogenetic uncertainty by recalculating evolutionary rates in a sample of 100 trees that were randomly selected out of a complete sample of 60,000 phylogenies generated in Dembo’s Bayesian analysis (20). This sample excluded phylogenies in which one or more branches had lengths shorter than 70 ka, which is the shortest possible length of the SH branch, obtained when the Neanderthal–modern human LCA is dated to 500 ka ago. The use of these different phylogenies addressed the uncertainty related to unclear phylogenetic relationships and branch lengths. As for the former, different phylogenies recover different evolutionary relationships across species. As for the latter, branch lengths differ in all the different trees. Therefore, although Dembo and colleagues’ study did not specifically model the uncertainty due to the age of each fossil species, that uncertainty is implicitly included in the calculations due to the different branch lengths recovered in their sample of trees. Evolutionary rates were calculated over this sample of 100 trees, and ranges of variation were compared with results obtained when analyzing the two previously described phylogenetic contexts. Possible hybridization events between lineages were not included in these calculations.

Statistical analysis: Ancestors and evolutionary rates

Ancestral values at the different nodes of the hominin phylogeny were calculated using an mvBM approach (25), which relaxes the assumption that different branches have evolved at a constant rate following a standard Brownian motion (BM) model. Biologically, this approach accounts for the fact that ancestors may have not been intermediate in shape between their descendent lineages, but more similar to one of the descendant groups (4). This situation would be reflected in different evolutionary rates across the tree, with some branches showing stasis and others showing fast evolution. Through simulations, an mvBM approach has been demonstrated to produce results equivalent to standard BM under standard BM conditions and to substantially outperform standard BM approaches when evolutionary bursts (very high evolutionary rates over short periods of time) occur (37). In addition, the results of this study indicate that standard BM approaches (38) do not accurately recover differential evolutionary rates that result from changing branch lengths, particularly for very early divergence times, as very similar SDs of rates are obtained when varying divergence times (tables S3 and S4). Short branches are expected to show fast evolutionary rates because they accumulate phenotypic change over a very short period of time. Therefore, the results obtained from standard BM approaches are counterintuitive because they yield similar evolutionary rates regardless of branch length (table S3). As inferred from these results, standard BM approaches do not accurately recover evolutionary bursts that are restricted to single branches, but they distribute change across the neighboring branches. Ancestral values were calculated using species-specific PC scores with the R package evomap (39). All PC scores were included in the calculations, and they were later transformed to ancestral landmark coordinates. Procrustes distances between descendant and estimated ancestral morphologies were compared with Procrustes distances between descendant species and ancestors obtained when simulating evolution at a constant rate across the whole hominin phylogeny 1000 times (27). For these simulations, a per-generation variance rate was calculated on the basis of available data using a generalized least squares (GLS) approach (38). These calculations were carried out using the packages Morphometrics (40) and Phylogenetics (41) for Mathematica and followed a transformation of the hominin phylogenetic tree to generations using a constant generation time of 25 years. For each branch, a ratio was calculated between the observed amount of change and the corresponding simulated amount of change in the neutral scenario where all the species were evolving at the same rate. Ratios lower than 1 indicate branches that are evolving slowly and undergoing stasis, whereas ratios greater than 1 indicate fast evolution and, when very high, are likely indicative of directional selection (19). For the sake of simplicity, this ratio of observed to simulated change per branch is referred to throughout the text as rate, but these are not rates in the strict sense because they do not represent change per unit of time.

Results obtained from the calculation of evolutionary rates when assuming different divergence times for Neanderthals and modern humans were compared in different ways. Some of these comparisons involved rates across the complete tree, whereas others focused on the branches directly related to the Neanderthal–modern human divergence. For the former, the SDs of all rates in each tree were compared with those simulated in the constant rate scenarios, and P values were calculated as the proportion of simulated SDs exceeding the observed SD for each divergence time. For the latter, the evolutionary rates of the SH branch and subtending branch (LCA branch) were compared to each other and to the rates observed in the other branches, as well as to the corresponding rates obtained when analyzing 100 different phylogenetic topologies. These diverse comparisons provided different age intervals for the Neanderthal–modern human LCA. The overlapping region of these different estimates is considered the most likely divergence time between Neanderthals and modern humans, and the lower bound of the interval is interpreted as the minimum age of their LCA.


Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/5/5/eaaw1268/DC1

Fig. S1. Configurations of landmarks and semilandmarks used to describe the shape of posterior teeth.

Fig. S2. Principal components analysis of dental shape in hominins.

Fig. S3. Comparison between the two phylogenetic scenarios used in this study.

Fig. S4. Relationship between the evolutionary rate at the SH branch and at the LCA branch in phylogeny-2.

Fig. S5. Most likely Neanderthal–modern human divergence time obtained from the analysis of Dembo and colleagues’ MCC tree (phylogeny-2).

Fig. S6. Rate analysis based on classic Neanderthals and phylogeny-1.

Table S1. List of specimens used in this study.

Table S2. Sample size per species and tooth position.

Table S3. Comparison of observed and simulated SDs of rates across the tree for the different SH–modern human divergence times.

Table S4. Comparison of observed and simulated SDs of rates across the tree for the different SH–modern human divergence times calculated when using the Dembo et al. phylogenetic tree (phylogeny-2).

References (4246)

This is an open-access article distributed under the terms of the Creative Commons Attribution license, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.


  1. D. Strait, F. E. Grine, J. G. Fleagle, in Handbook of Paleoanthropology, W. Henke, I. Tattersall, Eds. (Springer Berlin Heidelberg, Berlin, Heidelberg, 2015), pp. 1989–2014.

Acknowledgments: I am grateful to J. Smaers and D. Polly for the earlier methodological help; B. Wood for the input on hominin phylogenetic relationships; M. Dembo for sharing the hominin phylogenies generated through the Bayesian analysis of the hominin phylogeny; A. Andrés and C. Posth for the discussions on nDNA, mtDNA, and dental divergence times; and A. Goswami and C. Soligo for the logistic support. I thank the following people for facilitating access to materials: J. M. Bermúdez de Castro, J. L. Arsuaga, E. Carbonell, all the other members of the Atapuerca Research Team, O. Kullmer, B. Denkel, F. Schrenk, M. A. de Lumley, A. Vialet, I. Tattersall, G. Sawyer, G. García, Y. Haile-Selassie, L. Jellema, and M. Botella. Funding: Research was supported by a UCL-Excellence Fellowship. Author contributions: A.G.-R. designed the research, collected the data, analyzed the data, interpreted the results, and wrote the paper. Competing interests: The author declares that she has no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the author.

Source link