Saturday , August 24 2019
Home / zimbabwe / Dental evolutionary stages and their consequences for neanderthal-modern human divergence

Dental evolutionary stages and their consequences for neanderthal-modern human divergence



The source of Neanderthal and modern human tribes is the subject of an intense debate. DNA analyzes have generally shown that the two lines between the middle period of the middle Pleistocene were different, which is a time that has greatly influenced interpretations of the hominine fossil record. However, this divergence time is incompatible with the anatomical and genetic neanderthal affinities observed in the Middle Pleistocene hominins of the Sima de los Huesos (Spain), which are up to 430,000 years old. On the basis of quantitative analyzes of dental evolution levels and Bayesian analysis of phylogenetic relationships of hominine, I prove that any time between neanderthals and modern humans under the age of 800 would lead to an unexpectedly rapid evolution of teeth in early Neanderthals from the Sima de los Huesos. These results support the previous common ancestor for Neanderthals and modern humans before the age of 800, unless unexplained mechanisms have so far accelerated the development of teeth in early Neanderthals.


The time and identity of the last common ancestor (LCA) of Homo neanderthalensis in Homo sapiens (hereinafter referred to as Neanderthals and Modern People) are issues that are discussed intensively (15). Studies of the ancient DNA (aDNA) generally showed the time of divergence ca. 400 thousand years ago (ka)6), which found support in some quantitative studies of skull changes (7). In addition, the evolutionary scenarios, which are usually considered, presuppose that at least some hominins of the middle Pleistocene were from 600 to 400 years ago or even younger part of the last common ancestors for Neanderthals and modern humans.[reviewed([reviewedin([pregledanov([reviewedin(8)]. However, several anatomical studies of fossil evidence have shown that some European Pleinocene European hominins, especially those belonging to the sample of the Sima de los Huesos (SH), show a clear affinity with the Neanderthals (911). Following some contradictory results regarding the geological age of hominins SH (12, 13), this collection is now safe 430 years ago (14), age confirmed by analysis of mitochondrial DNA (mtDNA) branch length (15). In addition, recent DNA analysis (nDNA) of these populations showed the evolutionary affinity of SH hominins with classical Neanderthals (16), making the differences between neanderthals and modern humans necessarily older than the age of SH fossils. Some recent studies reflect these new findings and favor an older age for this LCA of 550 to 765 ka (17) on the basis of recent estimates of the degree of human mutation (16). The divergence times arising from genomic data are highly dependent on the estimates of the rate of mutation and generation time that are still being discussed (18). Due to the small variations in these parameters, there can be very different estimates of the divergent time between two types. If these nuances are ignored, then accurate reading of the values ​​provided by the aDNA analysis can result in radically different interpretations of the fossil record, which may even be incompatible with the affinities derived from anatomical evidence.

Closer evolutionary affinity SH with Neanderthals than modern humans shows that hominins SH differed from the modern human line at the same point as the classic Neanderthals. Therefore, the genetic affinities, the geological age and the morphological variations of the hominine SH can be used to deduce the time of the neanderthal-modern human difference. Recent studies of the variation of hominine have shown that, in contrast to other properties, the post-cranial shape of the teeth, as described by geometric morphometric databases (Figure S1), has been developed in a highly homogeneous manner in all hominine genera (19). This finding was used in this study to conclude the time when Neanderthals and modern humans should differ in order to maintain the evolutionary stage of the tooth shape of the phylogenetic branch leading to SH hominins within the same range of variations observed in other types of hominine (tables S1 and S2). The dental form in the hominine SH is unexpectedly derived against the neanderthal state, both in the expression of neanderthal discrete characteristics (9) and at its final stage of the structural reduction of the number and size of the cubes (Figure S2) (11). The tooth shape of the SH hominins is thus derived that is not representative of other Neanderthal populations. However, this does not affect the design of this study. Even if the SH hominins do not show the average dental forms seen in later classic neanderthals, their highly developed dental habits developed from the same ancestors as classical Neanderthals and in a period separating hominine SH from a Neanderthal-modern man. LCA (see Figure 1). The homogeneity of the evolutionary stages for the shape of the teeth stands out in sharp contrast with a much more heterogeneous scenario for the size of the teeth, in which different degrees are observed in the various branches of hominine phylogeny (19).

Figure 1 Phylogenetic scenarios and SH tooth morphology.

(A) Hominine phylogeny is used in the analysis of evolutionary stages (phylogeny-1). SH branch is represented in teal, and the LCA branch is in orange color, which are the colors used to represent the evolutionary stages on these two branches in the images. 4 and 5 and the like. S5. The gray lines represent various time differences that have been evaluated. (B) Transformation of Neanderthal genus-Denisovan-SH into genus. (C) Densitree showing the randomly chosen sample of 100 filogenias[odcelotnegazzorca60000filogenovijihustvarilaBayesovaanalizaDemboinsodelavcevofilogenetskihodnosihhominina[ofthetotalsampleof60000phylogeniesgeneratedbyDemboandcolleagues'Bayesiananalysisofhomininphylogeneticrelationships([odcelotnegavzorca60000filogenovkijihjeustvarilaBayesovaanalizaDemboinsodelavcevofilogenetskihodnosihhominina[ofthetotalsampleof60000phylogeniesgeneratedbyDemboandcolleagues’Bayesiananalysisofhomininphylogeneticrelationships(20)]. The original Demba trees were trimmed to preserve only the species for which dental data is available. The neanderthal branch length was shortened to reflect the age of the SH branch. (D) Upper and lower postcard teeth of one representative SH individual (on the left the upper dentition is presented). Photo: A. Muela, photographs at the Institute of Health Carlos III.

Due to the lack of consensus on phylogenetic relations of hominine, the analyzes were based on two different phylogenetic frames (Fig. S3) (19, 20). The first (phylogeny-1) is the phylogenetic tree used in the previous study of evolutionary levels of hominine (19) based on the dates of the first and last appearance of those types of hominins for which data on the form for all posterior teeth were available. The second phylogeny (phylogeny-2) is the tree of maximum plausible credibility (MCC) calculated by Dembo and his colleagues (20, 21) as part of their Bayesian analysis of the phylogenetic relationships of hominine. This phylogeny was circumcised so that it included only the species for which dental data was available. In these two phylogeny, the age of the human LCA Neanderthal – modern has changed from 500 ka, which is just below the lowest interval limit proposed by the latest molecular analyzes (16, 17, 22), to the age of the subordinate node at 100 intervals (Figure 1). The uncertainty about the phylogeny of hominine and branch length was explicitly addressed by evaluating the evolution rates on a sample of 100 trees. This pattern of trees was randomly selected from a sample of 60,000 trees obtained by Bayesian hominine filogenesis analysis (20, 21) (Figure 1). Denisovci (23), which differed according to classical neanderthals after neanderthal-modern human divergence, but before fossils SH (16) were not included in these analyzes because very rare phenotypic data are available for this group. However, according to their development relations (24), Denised, as SH hominins, can be considered as part of the Neanderthal genus in the broad sense or H. neanderthalensis sensu lato (Figure 1).

The methodological approach used consists of a three-stage process, which included calculating the value of ancestors using the approach by means of multiple variance of Brown's movement (mvBM) (25), calculating the amount of change in the branch as the difference between the morphology of descendants and ancestors and comparing these values ​​with those obtained in the simulation of evolution at a constant rate in all branches of hominine phylogeny (19). The main advantage of this approach is that it specifically and quantitatively takes into account the possibility that LCA Neanderthals and modern humans (or other two species) were not intermediate in morphology between the two daughters but rather similar to the Neanderthals. This is a scenario that has recently been proposed to explain the presence of derived neanderthal features in the SH model (4) and even in previous European hominins (26), but this has not yet been officially tested.


Changing the divergence time between SH and modern human branches has strong effects on the length of the SH branch and subordinate branches, as well as on the corresponding evolutionary stages. Very late SH – modern human divergent times cause very short lengths to the SH branch, which in turn leads to very rapid evolutionary rates for this line. On the contrary, premature SH – modern human divergent times cause very short lengths to the phylogenetic branch leading to their LCA, which is reflected in a very high degree of evolution for this branch. Figure 2 shows how the evolutionary rates associated with the neanderthal-modern human block (those corresponding to the branch SH, the modern human branch and the LCA branch) are fundamentally different when changing the SH-modern human divergence time as described above. Since the neanderthal-modern human LCA schedule is the only one that can change, there is a reverse link between the evolutionary branch level leading to the hominins of SH (or neanderthals) and the branch it supports, so that the slower velocity in the SH chain is associated with faster speed in the subordinate branch (Fig. 3 and Fig. S4).

Figure 2 Different evolutionary stages obtained by analysis of phylogeny-1.

(A) The evolutionary levels that were achieved when setting up a SH – modern human divergent time before 0.5 Ma. (B) The steps taken to determine this difference at 0.9 Ma, which is the scenario associated with the minimum SD of all levels on the tree. (C) Degrees achieved when adjusting deviation before 1.4 ma. The time of the SH-modern human divergence, which was older than 1.4 ma, has created even higher rates for a branch that is before the SH-modern separation of a person, which is referred to as the LCA branch in the following numbers. Evolution rates are given above each branch (gray for rates that remain roughly unchanged in all scenarios, and black for rates associated with Neanderthal-modern human block, influenced by changes in times of modern human divergence).

Figure 3 Relationship between the evolutionary level in the SH branch and the LCA branch.

The relationship observed in the analysis of the first phylogenetic scenario (phylogenetics-1). Evolutionary rates on both branches show an inverse and non-linear relationship, so that very high levels in the SH chain are associated with very low rates on the LCA and vice versa. This effect can be shown in Figure 2, which shows how these velocities change according to the assumed SH – modern human divergence time.

The analysis of 100 filogeny gives very few cases (3 out of 100), where the branch SH shows the highest rate across the entire tree, but most of the cases (59 out of 100), where the antedating branch shows the highest rate across the tree (Figure 4). According to these results, divergence time scenarios that were older than 0.75 million years (Ma) are more likely to result in the LCA branch showing the highest degree of evolution as in the scenarios with the younger age of divergence (Figure 5). The fact that the branch leading to the SH – modern human clan indicates the highest degree of evolution in most phylogeny shows that the tooth divergence was the strongest in the later stages of gender development. Homo.

Figure 4 Changing the evolutionary levels obtained from the analysis of 100 trees.

(A) Densitree showing a sample of 100 randomly selected trees used in the calculations. (B) Boxplot compares the highest level of evolution (gray), the degree of LCA (orange), and the rate of SH (teal) in 100 phylogeny. (C) Evolution rates obtained in the analysis of each of the 100 phylogens showing the highest rate per tree (gray), the degree of LCA (orange), and the degree of SH (teal). The phylogeny in (C) are classified according to their highest degree of evolution. The graph shows that the LCA rate is the highest in most phylogeny (59 out of 100), while the SH is the highest level in only three phylogeny. In all other cases, the highest rate is found in other branches (in most cases in P. boisei branches).

Figure 5 Probably neanderthal-modern human time, based on the analysis of phylogeny-1.

(A) Comparison of observed SDs of all levels in hominine phylogeny (red points) with SD distributions obtained in the simulation of evolution over the same constant-speed tree. (B) Comparison of the evolutionary levels in the SH (teal) branch, the LCA branch (orange) and all other branches (gray) obtained for different SH – modern time differences between people. (C) Comparison of the SH (teal line) velocity with a 95% interval of steps obtained for this branch with the analysis of 100 filogenij (gray field). (D) Comparison of the LCA level (orange line) with a 95% interval obtained for this branch with the analysis of 100 filogenies (gray field). The black dotted lines surround the most likely time differences with respect to each analysis. The red dotted lines indicate the lowest and highest values ​​obtained with all the analyzes, and the most likely time when all the results are considered together. Equivalent results, based on the analysis of phylogeny-2, are given in Fig. S5.

The zero expectation that the tooth shape has evolved neutrally throughout the entire hominine phylogeny is only accepted if the neanderthal-modern human divergence is in the interval from 0.7 to 1.2-Ma (Fig. 5A and Table S3), which is strongly indicated against departure times outside in this interval. Expectations of neutral dental development are confirmed by previous studies (19) and has been tested according to simulated scenarios that reflect a genetic shift and no choice (27). The standard deviation (SD) of the evolutionary rates along the tree reaches its lowest value at 0.9 m, although tree SDs are low and very similar in the range from 0.7 to 1.1. Figure 5B shows that the rates corresponding to the SH branch and the support branch become the same when the divergence time is 0.7 to 0.8 Ma. Divergent times, which are significantly younger or older than 0.75 Ma, lead to evolutionary rates for the SH branch or the former branch, which are very far from the range of variations observed for all other branches (Figure 5B). The evolutionary rate on the SH branch is 95% interval calculated for this branch by the analysis of 100 filogenies only when the neanderthal-modern human divergence time is older than 0.8 ma (Fig. 5C). The 95% interval for the previous branch is very wide, so most of the deviation times are compatible with the values ​​calculated for this branch (Figure 5D). The overall result of all these analyzes gives you a time interval from 0.8 to 1.2 Ma before you as the most likely time of divergence for the branch SH and the modern human branch, and hence for the Neanderthal and modern human lines. Repeat these analyzes using the MCC tree calculated by Dembo and his co-workers (20) provides even longer time differences with a minimum deviation time of 0.9 Ma, calculated from a combination of all analyzes (figure S5 and Table S4).

Assuming neanderthal-modern human divergence about 600 years ago, age when the latest molecular studies show (16, 17, 22), it would have some consequences on SH dental development rates. First, SD of all phases of hominine phylogeny would show an unusually high value (although still within the range reached) with respect to 1000 simulated neutral scenarios (P = 0.033 for phylogeny-1; see Figure 5A and Table S3). Secondly, assuming that the divergence time is 600 years old, this would mean that the evolutionary rate in the SH branch was the highest in hominine filogenesis (1.3 times greater than the evolution rate in the LCA section). According to the analysis of 100 different phylogeny sampled from the Dembo study (20), this scenario is unlikely (Figure 4). In addition, the evolutionary rate at the SH branch in the divergence scenario 600-ka would be 1.99, a value that is well beyond the 95% interval of the rates observed for the SH branch, with the analysis of 100 trees Dembo (Figure 5C). The evolutionary level of 1.99 at the SH branch is lower than one value observed in the analysis of 100 filogenies (2.05), which is a clear exception to all the levels observed in this branch (Figure 4B). The results of the various analyzes carried out in this study show that the hominine SH should be separated from at least 400 ka from the human LCA neanderthal-modern to maintain the evolutionary rate of SH hominins in the range of variations observed for other hominins. Therefore, ca. 600 divergence, compatible with similar evolutionary rates between hominine SH and other types of hominine, would require ca. 200 ka age for hominine SH, which is substantially younger than all the values ​​calculated for this population (1214).


The evolution rates measured in this study are strongly influenced by the lengths of the branches, so that the short branches that accumulate powerful dental changes cause high rates. The young times of divergence between neanderthals and modern humans are caused by short branches of SH and consequently observed high evolutionary rates for hominine SH. If the SH hominins were less than 430 years old, they would be compatible with the time of the divergence between Neanderthals and modern humans who turned 800 years ago without the need for extremely high evolutionary rates. Specifically, ca. The tolerance of 600-fold by recent molecular evaluations (22) would be compatible with the average evolution rates for the SH sample if these hominins were already 200 years old. This scenario is worth considering, because in the past it has already been discussed about the age of hominins SH (2, 12, 13). Recent studies, based on luminescence and paleomagnetic analysis, safely indicate the age of 430 years for these fossils (14). This data is further supported by genetic analyzes derived from the SH hominins up to about 400 years ago based on the length of the mtDNA branch, with a 95% maximum posterior interval of 150 to 650 ka (15). This interval is, however, quite broad and means that the hominine SH can be less than 430 ka. On the basis of these data, they can also be much older, which would necessarily suppress neanderthal-modern human divergence to an even older date. Additional evidence supporting approx. 430, the age for the SH sample comes from other molecular studies. These studies show that SH hominins share the same line of mtDNA as Denisovans, which differs from neanderthal and modern human lines of mtDNA (15). According to Posth and co-workers (28The Denisovan-SH mtDNA line is primitive to the neanderthal block, and the classical neanderthal mtDNA line was acquired posteriorly through the introgression event of modern humans, ranging from 219 to 468 ka. If this model is correct, then the SH population must precede this introgression event, which gives additional support to the sample> 400 ka. Therefore, based on the current combined geochronological and molecular evidence, the age of approximately 430 ka is the most meaningful for the hominine SH, so other explanations are needed to present the results. Also, in relation to branch lengths, it can be argued that the analytical approach presented in this study gives priority to the older Neandertal-modern times of human divergence because it provides for longer branch branches (and hence slower evolution rates) for other types of hominins. This potential bias is taken into account using 100 different phylogenetic scenarios based on Bayesian analyzes of the phylogenetic relationships of hominine (20), some of which show the lengths of branches of other species that are as short as the SH branch. Nevertheless, analyzes based on these 100 phylogeny show that Neanderthal-modern human times, which were younger than 800 years old, are very unlikely. This means that methodological artefacts are unlikely to trigger the observed results, so they should be explained by biological factors.

A rapid evolutionary rate in the early neanderthal populations represented by hominine SH, which would be a necessary consequence of neanderthal-modern human diversity after 800 years, may result from a strong selection of teeth in these hominins. Although this scenario is initially probable, it is also very unlikely that the development of the early Neanderthal segment was characterized by rapid dental development, which is not observed in any other type of hominine (19) (even in genera Paranthropus, which is characterized by the extreme degree of postcard megadontia). This strong choice is unlikely for two reasons. First, the differences in the shape of the teeth were observed in the SH hominins in relation to the hypothetical morphology of the ancestors (3), and also in relation to more primitive configurations, such as those seen in. t Homo erectus, are of no functional significance and are considered selectively neutral (i.e.29). Therefore, it is very unlikely that these dental changes have been the target of a large selection of unusually rapid evolutionary stages. Second, the SH hominine's teeth are the only skeletal region that shows a highly derived state. Other properties associated with chewing, such as facial anatomy and mandible, show a clear neanderthal affinity in the SH hominins, but not the hyper-binding neanderthal state found in their teeth (10), which means lower levels of evolution. A strong elective scenario that is associated with a certain functional advantage would almost certainly include other cranial regions in addition to teeth. The transition state of most other properties of SH hominins most likely shows that the selection was not a major factor that promotes SH dental development.

As already mentioned, SH hominins show anatomy of teeth, which is not representative of the Neanderthal average, but is substantially more derived. However, this conclusion does not affect the design of this study nor its results. The study plan does not require SH's dental anatomy to be representative of a wider range of neanderthal variations. On the contrary, it is simply based on the fact that, irrespective of whether they are representative or not, SH dental properties are derived from the same ancestor, as classical neanderthals developed during a period that separates the hominine SH from Neanderthal-modern human LCA. Therefore, this study does not treat SH-anatomy as representative of classical neanderthals, but only as a form of teeth that was characteristic of the SH population with respect to its evolutionary relationships and geological age. Considering this unrepresentative and highly derived state of SH dentition, a plausible explanation of rapid dental development, which is the consequence of the divergence of postdatation before 800 ka, revealed to SH dental anatomy as a result of the founder's strong effect. In this scenario, the ancestors of the SH hominine population would have different dental morphologies, one of which would have been determined in the SH pattern because they were present in their immediate ancestors. This scenario is theoretically possible and can be supported by the geographic position of hominins SH on the Iberian Peninsula, where they may have been more isolated than other Neanderthal populations from continental Europe. However, this scenario would mean that SH tooth phenotype was present, although in a small proportion, in early middle Pleistocene populations, from which SH hominins were developed. Due to the lack of fossil records, this scenario can not be excluded at this time, but the current fossil hypodiges do not show these derived dental configurations in any other hominine that preceded the SH population, which undermines this hypothesis.

Another factor that could affect dental development in hominine SH is hybridization. Based on genetic analyzes, it has now been confirmed that hybridization has occurred among Neanderthals, modern humans and the Denisans (30, 31), probably very often. Therefore, it is possible to assume, without a doubt, that the various tribes of the middle Pleistocene hominine were hybridized in contact with the body. The high degree of mosaicism in the SH population, with some characteristics that show a completely Neanderthal state and others that show a much more primitive state, could potentially point to a hybrid source. However, the hominine SH does not exhibit skeletal anomalies found in hybrid groups of living primates that occurred in the early generation, such as the presence of rotated or excess teeth, and seam abnormalities in neurocrania and face (32). While the hybrid origin of SH hominins is definitely possible, this hypothesis does not have particularly strong support based on their anatomy and what we currently know about the phenotypic effects of hybridization.

The simplest explanation of the results presented in this study is that Neanderthals and modern humans were diverging before 0.8 m, which would cause evolution rates for dentic SH to be comparable to those of other species. This deviation time is significantly older than the latest aDNA estimates (16, 17, 22), but not so far from previous estimates that show this deviation at ca. 800 years ago (24). estimates of the differences between neandertals and modern aDNA-based people differ significantly (6, 17, 22, 24), which shows that strict reading of these values ​​can not lead to the interpretation of the hominine fossil record. In addition, the time of discrepancy obtained by analyzing dental levels is strikingly similar to the time of divergence of SH-Denisovan and neanderthal-modern human lines of mtDNA. The divergence between the two mtDNA lines was estimated at ca. 1m back, with 95% highest posterior interval of density from 0.7 to 1.4m before15). As explained above, it is assumed that the neanderthal mtDNA is subordinated due to a relatively new event of introgression of modern humans (28). Therefore, the timing of SH and modern human lines of mtDNA far more accurately reflected the divergence of the neanderthal-modern human population than the time of discrepancy between neanderthal and modern human lines of mtDNA, which reflects the greatest time for the introgression event and is significantly younger. The diversity of mtDNA SH hominins and modern humans is still older than the time estimated by the population, broken down by the nDNA between the modern human line and the Neanderthal-SH-Denisovan line, which was recently calculated to 550 to 765 ka (16, 17) or 520 to 630 years ago (22). The divergence time of mtDNA indicates the moment when both mtDNA lines began to accumulate mutations independently, while the time sharing period represents the last time that both groups exchanged genetic material with each other with a divergence and thus became younger than mtDNA estimates. The results of this study show that the phenotypic differentiation in dental morphology began before completing the division of the population between Neanderthal and modern human tribes. Although it is possible that these differences are due to different methodological approaches used in various genetic and phenotypic variation studies (7, 16, 17, 22), it is also possible that these differences reflect different biological signals associated with different properties. In this case, older timeframes for phenotypic divergence, as evidenced by changes in dentistry, will have profound implications for how we interpret fossil data on hominine and the relationship between fossil patterns, especially for those populations and time periods for which aDNA is not available .

If the phenotypic LCA of Neanderthals and modern humans were older than 800 years, this would mean that all fossil hominins younger than this age were no longer valid candidates to occupy this position of ancestors. Some fossils, younger than this age, are often regarded as part of the last common ancestors for Neanderthals and modern humans (2, 8). These fossils are usually attributed Homo heidelbergensis, includes European and African patterns such as Mauer, Arago, Petralona, ​​Bodo, Kabwe, etc., and possibly even some Asian patterns. If Neanderthals and modern humans were abolished earlier than 800 years ago, then all these fossils must be associated with Neanderthals or with modern humans, or they may be part of a sister's relatives for both. These fossils can not be the ancestors of Neanderthals and modern humans, because they would follow their evolutionary divergence. The evolutionary relationship between these fossils and Neanderthals and modern humans would only be possible if they were part of the older species of ancestors who existed in time as relict species after the actual split of the two genera. This scenario would in fact mean that H. heidelbergensis fossils are part of the Neanderthal and modern humans sister group, but the evolutionary change from the supposed ancestor populations did not include speciation.

It was suggested that the process of obtaining a completely Neanderthal anatomy might have begun earlier and could have been more progressive than the process of obtaining a completely anatomically modern human configuration that does not appear in the fossil record to ca. 200 years ago (4). The initial modern human characteristics are observed in the fossil record at approx. 300 years ago (33), a figure that is in line with recent DNA estimates of modern human disparities at 260 to 350 years ago (18). This is contrary to the observation of a completely Neanderthal (which can even be considered hyper-neanderthal) teeth 430 years ago in SH hominins. The discrepancies between the dates at which clear Neanderthal and modern human affinities are observed in the hominine fossil record may seem to indicate a differential evolution rate in both the lineages, which would affect the conclusions made through the present study. However, they can simply reflect the incompleteness of the fossil record, especially for the modern human lineage, as the SH sample is the only early Neanderthal population represented in the fossil record that shows such derived dentition. New fossil findings, as well as the reassessment of previously known ones, are essential to shed more light on the process of acquiring a completely anatomically modern human configuration.


Experimental design

The main goal of this study was to measure the evolutionary rates for the dental shape in the earlier part of the evolution of the Neanderthal lineage and to compare them with 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 analyzes 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.


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

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