Navigation – Plan du site

AccueilNuméros38 (1)ArticlesHominin phylogenetics: methods an...

Articles

Hominin phylogenetics: methods and insights

Méthodes et apports des analyses phylogénétiques chez les hominines
Pierre Gousset, Florent Détroit et Jérémie Bardin

Résumés

Les analyses phylogénétiques des fossiles d’hominines sont capitales pour notre compréhension de l’évolution humaine. Cette revue de littérature détaille les bases méthodologiques sur lesquelles ces analyses ont été menées dans le but d’encourager et/ou de proposer des améliorations méthodologiques. Des tendances sont mises en lumière, comme l’utilisation habituelle de spécimens individuels comme Unités Taxonomiques Opérationnelles pour les périodes récentes ou l’application au cours des dernières années des méthodes à modèles. La rareté de l’utilisation des caractères de l’anatomie interne et du squelette postcrânien est évidente, alors que de nombreuses analyses publiées fournissent beaucoup d’informations pour intégrer ces caractères dans de futures matrices. Certaines méthodes développées et appliquées, parfois en routine, en dehors de la discipline paléoanthropologique, comme l’usage de "vrai" caractères continus, de fréquences pour tenir compte de la variabilité intra-taxonomique ou l’intégration de données de morphométrie géométrique mériteraient d’être plus largement appliquées aux hominines. Le super-arbre généré à partir des analyses décrites dans ce travail montre que les relations phylogénétiques des espèces récentes du genre Homo ne sont pas consensuelles, pointant ainsi la nécessité d’une analyse approfondie utilisant toutes les régions anatomiques et bénéficiant des récentes avancées méthodologiques.

Haut de page

Texte intégral

Introduction

Context of this review

1The aim of this review is to assess the state of the art of phylogenetics in paleoanthropology, in order to identify the methods best suited to refining our understanding of human evolution. More specifically, we will see how many of the difficulties met by paleoanthropologists when reconstructing the phylogeny, such as character integration, continuous characters, intra-taxonomic variability, etc., might be resolved through methodological advances and developments.

2Based on the authors’ field of competence and the central aim of this review, closer attention will be given to analyses focusing on recent Homo species (i.e., Homo species less than one million years old) and analyses based on morphological rather than molecular data.

A brief history of hominin phylogenetics

3Phylogenetic reconstruction is based on a number of methodological approaches, which will be divided here into two sets: parsimony analysis and model-based methods. Parsimony analysis is a phylogenetic reconstruction method based on grouping taxa with shared derived character states. The principle of parsimony makes it possible to select the tree with the fewest evolutionary steps according to the dataset being analysed. Model-based methods apply an evolutionary model defined a priori, which is used to estimate the most probable tree among the existing ones (e.g., Bayesian inference), or the probability of obtaining the data knowing a tree and the model (e.g., maximum likelihood). Whether for Bayesian inference or maximum likelihood, the evolution model used must be explicitly defined prior to the analysis (see section on "Phylogeny beyond parsimony"). A key difference between model-based and parsimony methods is that the first optimizes individual branch lengths while the second optimizes total tree length (Lewis, 2001).

4Cladistics was invented in 1950 by German entomologist Willi Hennig. He did not publish his method in English until 1965 (Hennig, 1965), at which point it began to be widely used for many taxa. The first cladistic analysis of hominins was carried out in 1975 by Nils Eldredge and Ian Tattersall (Eldredge and Tattersall, 1975; figure 1). This and subsequent studies mainly consisted of cladograms in which derived character states were identified. These analyses based on shared derived character states rather than maximum similarity represented a major shift in classification thinking, i.e., the shift from phenetics to cladistics (Strait et al., 2015). Although quantitative phylogeny developed rapidly after Hennig (1965) (see Brooks et al., 2007 for a review), the first numerical analyses on hominins were not carried out until 1987 by Chamberlain and Wood (1987) and Stringer (1987) (figure 1).

Figure 1

Figure 1

Historical timeline of major methodological steps in research on hominin phylogenetic analyses (references: © Gerd Hennig for Willi Hennig picture; Zeitoun 2000 for the matrix; unsplash.com for the lower limbs and DNA and Simon-Maciejewski et al. 2024 for the cranium with landmarks) |
Frise chronologique retraçant quelques-unes des grandes étapes des recherches portant sur l’analyse phylogénétique des hominines (références : © Gerd Hennig pour la photo de Willi Hennig ; Zeitoun 2000 pour la matrice ; unsplash.com pour les membres inférieurs et pour l’ADN et Simon-Maciejewski et al. 2024 pour le crâne avec les landmarks)

5From the late 1980s to the early 2000s, most published analyses focused on resolving relationships between ancient hominin species (Appendix A, column S). There are, however, some notable exceptions, such as Stringer (1987) who examined Homo heidelbergensis and Caparros (1997) who focused on "archaic H. sapiens", i.e., specimens from the Middle and Upper Pleistocene. More recently, Martinón-Torres et al. (2007), Mounier and Caparros (2015), Mounier et al. (2016) and Ni et al. (2021) focused on the phylogeny of Middle Pleistocene hominins. At least three analyses have focused exclusively on the phylogenetic placement of H. floresiensis (Argue et al., 2009; 2017; Zeitoun et al., 2016). Again for relatively recent periods, Zeitoun et al. (2010) analysed diversity within H. erectus, following Zeitoun (1996). In addition, other studies have focused on one or a few fossils attributed to recent Homo species (Appendix A, column S).

Criticisms of phylogenetics in paleoanthropology and outline of this review

6While phylogenetics is widely and routinely applied to study fossils (Kitching et al., 1998), paleoanthropology differs from the other sub-disciplines of paleontology due to widespread scepticism regarding the results generated by phylogenetic methods (reviewed by Strait et al., 2015). Indeed, the use of phylogenetics, and especially quantitative phylogenetics, in hominins, has been criticized since the early 1990s (e.g., Trinkaus, 1990; Collard and Wood, 2000; Curnoe, 2003; Hawks, 2004).

7Some of the frequently raised concerns, summarized by Trinkaus (1990) and Hawks (2004), apply to any method intending to study human evolution. These include incompleteness of the fossil record, small sample sizes, uncertainty about which morphological variations represent true synapomorphies rather than homoplasies. Other aspects of these criticisms, however, are specific to quantitative phylo-genetic methods. We argue in this review that all these shortcomings may find solutions within phylogenetic methods.

8More specifically, the following points have been addressed in the literature:

9- cladistic analyses based on morphological data would not be able to reconstruct phylogenetic relationships within extant taxa (Collard and Wood, 2000). This assumption is addressed in section 1 both on the scale of extant great apes and among extant H. sapiens populations;
- defining Operational Taxonomic Units (OTUs) is complicated as many disagreements occur in hominin alpha taxonomy and because, to some extent, morphological clines prevent efficient separation of groups (Trinkaus, 1990; Hawks, 2004). Carrying out analyses using individuals or very small groups of individuals as OTUs is one way to counter this issue. The choice of OTUs is more widely discussed in section 2;
- paleoanthropologists disagree strongly over the phylogenetic "quality" of different anatomical parts. While the postcranium is usually studied for its functional rather than phylogenetic meaning, the relative "quality" of craniodental elements is hotly debated. The anatomical elements used are reviewed in section 3;
- insufficient reflexion on character choice, coding and weighting is harmful, as these are the main drivers of the phylogenetic relationships proposed by the trees. For example, traits are (or have been) mainly coded as discrete while genuine variability of traits is often continuous. Furthermore, variation is often poorly taken into account in phylogenetic analyses (Trinkaus, 1990; Hawks, 2004). The first issue no longer exists since continuous characters can be treated "as such". The second issue may also be resolved using methods such as Generalized Frequency Coding (GFC; Smith and Gutberlet, 2001). Techniques used to choose, define and code characters are presented in section 4;
- the principle of parsimony does not offer a satisfactory approach to the reality of evolution (Trinkaus, 1990). This criticism has been countered (Kluge, 1997; Goloboff, 2003) and methods relying on principles other than parsimony have been developed and recently applied to hominins to reconstruct phylogenies, such as model-based methods or phylogenetic networks. These are detailed in section 5.

10The remaining sections consist of a discussion on the small number of matrices used in many published analyses (section 6) and a presentation of molecular phylogenies (section 7). A brief overview of the insights brought by phylogenetic analyses to relationships among recent Homo species is proposed in section 8, and a supertree summarizing more than three decades of phylogenetics in paleoanthropology is presented. Finally, we provide a brief overview of how phylogenetic trees can be used to understand various aspects of human evolution (section 9).

11All the published phylogenetic analyses of hominins reviewed in this study, and associated data, were used to generate the analyses, graphs and illustrations presented here. Newly published analyses based on previously published matrices were considered here as separate analyses. While this approach introduces a certain amount of redundancy in the data, (1) an investigation of the exact amount of redundancy between datasets goes well beyond the scope of this paper and (2) we think it better reflects the most common practices of paleoanthropologists doing phylogenetics. The raw data used to generate the graphs are available in Appendices A to D and Appendix E presents the raw data used to generate the supertree.

Is it possible to reconstruct hominin phylogeny based on morphology?

12Paleoanthropologists have only skeletal and dental material to reconstruct the phylogenetic relationships of fossil hominins. While the true phylogeny of hominins is unknown, one way to better understand the potential of morphological characters to approach this unknown true phylogeny is through congruence with other data. Thus, one might check whether phylogenies based on morphology are congruent with phylogenies based on molecular data. It is important to note that molecular phylogenies should not be considered as the gold standard, as they also present shortcomings (Rodríguez-Ezpeleta et al., 2007; Philippe et al., 2011), which explains why morphological and molecular data are now frequently used in combination to reconstruct phylogenies (Lee and Palci, 2015; Pyron, 2015; Zou and Zhang, 2016; Keating et al., 2023). In any case, no molecular data exists for the majority of fossil hominins, thus preventing any comparison. The next two paragraphs focus on published analyses that have addressed this question at a higher taxonomic level compared to fossil hominins (i.e., extant great apes) or at a lower taxonomic level (i.e., between current H. sapiens populations). While such comparisons have the major drawback that the morphological characters congruent with the molecular data at a given taxonomic level might not be the same as those congruent with molecular data at another taxonomic level, we point out that these comparisons are the only empirical arguments available at present.

13In response to Collard and Wood (2000), who argued that craniodental anatomy does not permit recovery of the relationships among hominids (understood here as encompassing all extant and fossil great apes) that are known from genetic data, Strait and Grine (2004) carried out a phylogenetic analysis of hominids also including fossil hominins (understood here as encompassing all fossil taxa closer to H. sapiens than to the genus Pan). With the addition of these fossil taxa, and thanks to the authors’ work on character selection (see "Choice, definition and coding of characters"), they succeeded in recovering the relationships between hominids known thanks to genetics.

14Stringer (1993) performed a phylogenetic analysis on current H. sapiens populations. This analysis was based on non-metric dental traits. Stringer (1993) was able to extract a phylogenetic signal from his data since the tree he obtained is congruent with the tree obtained from molecular data for the same populations. Moreover, numerous studies have calculated correlations between genetic and phenotypic data and have shown the important phylogenetic signal carried by non-metric dental traits (e.g., Irish et al., 2020; Rathmann and Reyes-Centeno, 2020).

15Thus, beyond theoretical considerations, empirical evidence supports that it is reasonable to construct a hominin phylogeny based on morphological data.

Which operational taxonomic units?

16As taxonomic debates within hominins are very important, many authors have pointed out the need for a well-argued definition of the OTUs used in the analysis (e.g., Trinkaus, 1990; Hawks, 2004). One way of defining consensual OTUs is to use individual specimens or small groupings of individuals discovered in the same layer of the same locality (Appendix C). This approach has the advantage of being more agnostic and presenting greater taxonomic resolution. It also allows for discussion of the monophyly of a given group commonly used in paleoanthropology (e.g., H. neanderthalensis, H. heidelbergensis and H. sapiens in Ni et al., 2021, H. habilis and H. rudolfensis in Prat, 2000 or H. erectus and H. ergaster in Zeitoun, 1996 and Gilbert, 2008) or of the geographical origin of a taxon (Asfaw et al., 2002; Mounier and Caparros, 2015; Ni et al., 2021). This explains why studies of some taxa of which the hypodigm is hotly debated are almost systematically carried out at this taxonomic level (e.g., H. heidelbergensis, H. erectus, H. habilis). This approach also has drawbacks, however, such as the difficulty of including fragmentary specimens and a strong but disputable a priori as regards the accuracy of cladistics for low-level taxonomy (these points are discussed later in this section).

17Being aware of the difficulties inherent to the definition of paleontological species, authors have frequently used populations as OTUs, often referring to them as demes or paleo-demes (Howell, 1999; Dennell et al., 2011). These are often identified based on anatomy, but also on chronology and geography (Stringer, 1987; Asfaw et al., 2002; Cameron et al., 2004; Kimbel et al., 2006; Gilbert, 2008).

18Some authors have made in-depth investigations into the effects of changing the taxonomic level of OTUs. Gilbert (2008) performed three analyses based on different taxonomic levels for the OTUs: one with individuals, another by sites of discovery and the last by groups of demes (the latter being defined on the basis of dating and geography to avoid any circular reasoning). The three analyses produced congruent results concerning the research question of the monophyly of H. ergaster and of H. erectus sensu stricto. Caparros (1997) constructed his OTUs in two stages: after carrying out one phylogenetic analysis per geographical region with individual specimens as OTUs, he grouped individuals forming a monophyletic group in this first set of analyses into the same OTU to carry out a final analysis on the scale of the Old World.

19Still, many of the published phylogenetic analyses on hominins have used species as OTUs (figure 2). Among hominins, an increase since the 1990s in species that have been widely recognized as valid, and the resulting paradigm shift, have led to a finer taxonomic division and therefore to an increase in the number of OTUs (six hominin species in Chamberlain and Wood, 1987; eight in Martinez and Arsuaga, 1997; 20 in Dembo et al., 2015). While Ernst Mayr’s (1950) proposal of a very small number of species within the genus Homo remained fairly widespread until the early 21st century, more than 20 hominin taxa are now commonly recognized, including at least eight within the genus Homo (H. habilis, H. erectus, H. heidelbergensis, H. naledi, H. sapiens, H. neanderthalensis, H. floresiensis and H. luzonensis).

Figure 2

Figure 2

Pie chart representing the taxonomic levels of OTUs in phylogenetic analyses on hominins. On the right are analyses that focus on the phylogenetic relationships of recent Homo species (n=24); on the left are analyses that focus on earlier hominins (n=25). Recent Homo species are Homo species less than one million years old. Gilbert (2008) features in each of the three main categories as he performed three analyses, each one at a different taxonomic level. Pie charts generated from the data presented in appendix B |
Diagramme circulaire présentant le rang taxonomique des UTO par étude phylogénétique sur les hominines. À droite, les analyses phylogénétiques se concentrant sur les relations de parenté entre espèces récentes du genre Homo (n=24) et à gauche les analyses se concentrant sur des hominines plus anciens (n=25). Les espèces récentes du genre Homo sont celles ayant vécu au cours du dernier million d’années. Gilbert (2008) a été enregistré dans chacune des trois principales catégories puisqu’il a réalisé trois analyses à chaque fois à un niveau taxonomique différent. Graphique généré à partir des données présentées dans l’appendice B

20Among published analyses, a difference in what OTUs represent is clearly apparent between analyses focusing on recent Homo species and those that do not (figure 2, Appendices A and B). The former tend to use lower taxonomic levels, which allows for discussion of species hypodigms. More broadly, the scope of the studies best explains how OTUs are defined: while published analyses that focus on resolving the phylogenetic position of individual specimens more often use individual specimens or paleo-demes as OTUs (or both – in this case, the OTUs are qualified as "Mixed" in figure 2 Appendix B), published studies aiming to reconstruct the relationships between a wide range of taxa frequently use species as OTUs.

21The choice of taxonomic level(s) used for the OTUs influences many aspects of the study. Despite the numerous advantages mentioned above (more agnostic approach, higher taxonomic resolution), using individuals as OTUs also has its drawbacks (figure 3). Firstly, using individuals as OTUs generates matrices where large amounts of data are missing due to the scarcity of well-preserved and complete fossils. Therefore, fragmentary specimens cannot be readily integrated into analyses of this kind because they will show too much missing data. Lastly, this approach requires cladistics to be considered as suitable for low-level taxonomy, which is a questionable a priori (see Sneath, 1995 for a detailed argument), since, according to Darlu et al. (2019: 37), "phylogenetic reconstruction […] aims to identify the phylogenetic structure between isolated units". However, one can legitimately assume that relationships between populations (and even more so between individuals) occur in the form of networks, which some phylogenetic reconstruction methods have tried to account for (see subsection "Phylogenetic networks" in "Phylogeny beyond parsimony"). Thus, some authors (Gilbert, 2008; Mounier et al., 2009) argue that phenetics is better suited to low-level taxonomy.

Figure 3

Figure 3

Diagram summing up the advantages (in green) and drawbacks (in red) of the different taxonomic levels that can be used for OTUs. The coloured points over each taxonomic level represent individuals. The colour represents the classification for each individual depending on the taxonomic level (i.e., one colour per OTU). The "n" corresponds to the number of OTUs. Polymorphism is coloured in grey since increased polymorphism can be considered as advantageous or disadvantageous depending on the authors and/or situations (see main text) |
Schéma résumant les avantages (en vert) et inconvénients (en rouge) des différents niveaux taxonomiques pouvant être utilisés comme UTO. Les points colorés au-dessus de chaque niveau taxonomique représentent des individus. La couleur représente la façon dont les individus sont classés selon le niveau taxonomique de l’UTO (i.e., une couleur par UTO). Les "n" correspondent aux nombres d’UTO. Le polymorphisme est colorié en gris car l’augmentation du polymorphisme peut être considéré comme avantageux ou désavantageux selon les auteurs et/ou les situations (voir le texte principal)

22Increasing the taxonomic level of the OTUs also increases the amount of polymorphism. While polymorphism can be considered as an obstacle to reconstructing a phylogeny, some studies have shown the usefulness of polymorphic characters (Wiens, 1995; Smith and Gutberlet, 2001 – see the detailed discussion in "intra-taxonomic variability and phylogenetic ‘quality’ of characters depending on that variability").

23It is important to note that figure 3 shows parameters increasing or decreasing linearly in relation to the taxonomic level. However, these relationships are very unlikely to be linear and depend on the taxa considered. For example, the individuals found at Sima de los Huesos, given that they come from the same layer (Aranburu et al., 2017), could together constitute an OTU that would be situated somewhere between "individual specimens" and "paleo-demes". However, due to the very large quantity of fossil hominin material found at this site, access to variability is much greater than for some paleontological species, such as H. luzonensis or H. antecessor, which are known from fewer fossils (Bermúdez de Castro et al., 2017; Détroit et al., 2019).

24To summarize, the choice of the taxonomic level of OTUs depends on the goal of the study, as any taxonomic level presents advantages and drawbacks that in turn depend on the taxa considered (figure 3). We stress the need for a detailed study of the impact that the choice of the taxonomic level of the OTUs may have on the tree topology.

Which anatomical elements?

25The vast majority of phylogenetic analyses of hominins are based on craniodental characters (reviewed by Strait et al., 2015; figures 4-5). While some focus on a particular part of the craniodental complex, and can therefore be dependent on the state of preservation of the fossil(s) studied, others attempt an exhaustive description of craniodental elements (Appendix C).

The craniodental complex

26Craniodental characters are commonly considered more informative for phylogeny because they are less plastic (Pilbeam, 1996). Within this anatomical set, some authors establish a hierarchy, in particular by emphasizing the plasticity of facial characters compared with those of the neurocranium (Lieberman et al., 2002; Harvati and Weaver, 2006) or the generally expected greater ability of dental characters to recover a phylogeny (Stringer et al., 1997; Martinón-Torres et al., 2007; Irish et al., 2018). Dental characters are considered to be very good phylogenetic indicators for several reasons. Firstly, the chemical composition of teeth increases their preservation compared to bones, thereby providing more abundant study samples. Secondly, tooth anatomy is considered to be less plastic during evolution because it is less subject to environmental variations (in the broad sense, including behaviour) than other anatomical regions, and is therefore expected to reflect the phylogenetic history of taxa more accurately (Irish et al., 2018). This consideration for dental characters explains the important part they take up within phylogenies (e.g., Skelton et al., 1986; Strait et al., 1997; Collard and Wood, 2000; Kimbel et al., 2004; Strait and Grine, 2004; Dembo et al., 2015; Mongle et al., 2019; Ni et al., 2021), and the numerous published phylogenetic studies focusing exclusively on these characters (Appendix C).

27While there is an abundant literature discussing the quality of the phylogenetic signal provided by the various anatomical regions, fewer publications have focused on making a formal study of this signal (e.g., Harvati and Weaver, 2006; Von Cramon-Taubadel and Smith, 2012; Rathmann and Reyes-Centeno, 2020), and few phylogenetic paleoanthropologists have conducted this type of study themselves, in parallel with or in preparation for a phylogenetic analysis (e.g., Strait, 1998; Collard and Wood, 2007; Dembo, 2016; Gautney, 2023).

28Similarly, many studies have addressed the question of conflict between characters in hominin phylogenetics (e.g., Chamberlain and Wood, 1985; Skelton et al., 1986; Strait et al., 1997), but few have thoroughly investigated the phylogenetic signal conveyed by different parts of the craniodental complex using published matrices. Chamberlain and Wood (1985) pioneered the individual phylogenetic analysis of "regions" of the craniodental complex, though not extending the comparison as far as Dembo (2016) and Gautney (2023) later did.

29Following the rationale of Chamberlain and Wood (1985) and each using a different dataset, Dembo (2016 - chapter 5) and Gautney (2023) divided the craniodental complex into four and six regions respectively. Each part was then studied independently, Dembo (2016) using a phylogenetic analysis and Gautney (2023) using phylogenetic networks. These different parts produced different phylogenies, which were then compared with each other, and with the phylogeny obtained from all craniodental characters.

30Both studies showed that the face and teeth independ- ently carry a phylogenetic signal that is coherent within itself with little conflict between characters compared to other anatomical parts (Dembo, 2016; Gautney, 2023). While Gautney (2023) also found this low level of conflict for the basicranium, Dembo (2016) did not. Thus, Gautney (2023) concluded that these are better candidates for phylogenetic reconstruction.

31However, for dental and facial characters, Dembo (2016) obtained unusual topologies with respect to commonly accepted relationships. She noted that evolutionary modules, which are a set of covariant characters, followed different evolutionary pathways during the course of human evolution, thus explaining the differences in the topology generated from each anatomical part (as previously advocated by Chamberlain and Wood, 1985). Comparing these topologies with the one obtained from all characters combined, Dembo (2016) found that the phylogeny obtained with neurocranial characters only is closer to the one obtained using all characters (it is also the most consistent with commonly accepted phylogeny). Furthermore, the consensus tree with the best resolution and the least distance between two trees for the same dataset (i.e., the most consistent) was the one that included all craniodental characters. As a result, and stressing that phenotypic and genetic data are not unequivocally better correlated for one anatomical region of the craniodental complex than for another, Dembo (2016) called for the integration of all anatomical features into phylogenetic analyses (a conclusion also reached by Folinsbee et al., 2015).

32The diverging conclusions reached by Dembo (2016) and Gautney (2023) reflect a broader debate in phylogenetics. While some authors are in favour of treating all data together with no a priori (e.g., Kluge, 1989), other researchers consider that data delivering a contradictory phylogenetic message should not be studied in concert (Bull et al., 1993; see also Dembo, 2016 for a more detailed discussion on this topic).

33To try to account for modularity in the evolution of hominin skulls, Gonzáles-José et al. (2008), using geometric morphometric data (see subsection on "Geometric morphometrics" in the section on "Choice, definition and coding of characters"), defined four modules prior to the phylogenetic reconstruction. Phylogenetic analysis was then performed using equal weights for the four anatomical regions. However, this study was criticized by Adams et al. (2011) for the lack of argumentation in the definition of modules and is the only one up to now that has used this approach.

34It is worth noting that many studies outside the field of phylogenetics have been carried out to understand the phylogenetic signal carried by craniodental characters, based on comparisons with molecular data (e.g., Von Cramon-Taubadel and Smith, 2012; Rathmann and Reyes-Centeno, 2020).

Internal structures

35Thanks to the advent of CT-scanning, the internal structures of fossil hominins are now frequently investigated. Furthermore, many studies have suggested that some of these structures carry a strong phylogenetic signal (e.g., Braga et al., 2017; Martin et al., 2017). Increasingly abundant data on internal structures make it possible to include them in phylogenetic analyses. For instance, the following authors have provided summaries for inner structural morphology of teeth (Macchiarelli et al., 2013), diaphyseal sections (Ruff, 2018), endocranium (Neubauer, 2015), inner ear (Braga et al., 2017) and frontal sinuses (Balzeau et al., 2022).

36However, so far, these characters have played a significant role (because they were at once numerous and useful for phylogeny reconstruction) in only one phylogenetic analysis: 34 endocranial characters were described by Mounier et al. (2016). This study highlighted the usefulness of these characters for reconstructing phylogeny by finding trees concordant with those found from classic characters and identifying endocranial traits as synapomorphies for various hominin taxa.

The postcranium

37While postcranial characters are often used to define the hominin clade with characters linked to bipedalism (e.g., vertically short and horizontally broad ilia, well-defined linea aspera, low talar angle; DeSilva, 2009; Lovejoy et al., 2009; Schwartz, 2015), they are largely absent from numerical phylogenetic analyses carried out on hominins (figure 4). One reason for this absence is that craniodental fossils (which are most often used to define species) are rarely found associated with postcranial fossils (Strait et al., 2015; Grine et al., 2022; figures 4-5). However, this issue is not a challenge in paleoanthropology only, and the phylogeny of our closest relatives, non-hominin hominoids, has been investigated using the whole skeleton (e.g., Young and Mac- latchy, 2004; Pugh, 2022 and references therein). Besides, postcranial characters are often considered as carrying a functional rather than phylogenetic signal. While the presence of a functional signal is undeniable in the postcranium, this also holds true for craniodental morphology, which depends largely on diet. Thus, the use of postcranial characters in phylogenetic analyses is also considered a major challenge by several paleoanthropologists doing phylogeny, who have stressed the need to include them in future studies (Dembo et al., 2015; Strait et al., 2015; Argue et al., 2017; Grine et al., 2022; Mongle et al., 2023).

Figure 4

Figure 4

Bar graph representing the various anatomical regions studied by phylogenetic analyses on hominins (n=47). The graph can be read as follows: 83% of the publications considered show more than one character describing the calvarium. Graph generated from the data presented in appendix C |
Diagramme en bâton présentant les parties anatomiques étudiées par analyse phylogénétique portant sur les hominines (n=47). Le graphique peut être lu comme suit : 83 % des publications considérées présentent plus d’un caractère portant sur le calvarium. Graphique généré à partir des données présentées dans l’appendice C

Figure 5

Figure 5

Bar graph showing the association of anatomical regions studied by phylogenetic analysis on hominins (n=47). Graph generated from data presented in Appendix C. ">1 region" means that two or more of the categories defined in figure 4 are present in the analysis. Abbreviation: C=Calvarium; F=Face; M=Mandible; T=Teeth; P=Postcranium |
Diagramme en bâton présentant les associations de parties anatomiques étudiées par analyse phylogénétique portant sur les hominines (n=47). Graphique généré à partir des données présentées dans l’appendice C. ">1 region" signifie que deux ou plus des catégories définies dans la figure 4 sont présentes dans l’analyse. Abréviations : C=Calvarium ; F=Face ; M=Mandibule ; T=Dents et P=Postcrâne

38Apart from a very general character describing robustness of the whole skeleton in the matrix of Stringer (1987), no postcranial characters were included in any phylogenetic analysis of hominins before 2009. Argue et al. (2009) pioneered the inclusion of five postcranial characters, then 29 characters in a subsequent and wider analysis (Argue et al., 2017). More recently, Gousset et al. (2026) carried out a parsimony analysis focusing on the remains identified for H. luzonensis (Appendix C). DeSilva et al. (2019) carried out a parsimony analysis focusing exclusively on foot bones as part of a synthesis on the evolution of the foot in hominins. This analysis has, so far, been the only one focusing solely on postcranial characters (Appendix C).

39However, a comprehensive study of the whole skeleton is still needed. This will be helped by the increasing number of existing syntheses on postcranial remains (e.g., Larson, 2007 on the shoulder, Tocheri et al., 2008 on the hand, Churchill and Vansickle, 2017 on the pelvis, DeSilva et al., 2019 and Sekhavati et al., 2025 on the foot; Gomez-Olivencia and Been, 2019 and Meyer and Williams, 2019 on the spine).

40As mentioned earlier, the debates surrounding the specific attribution of isolated fossil specimens is a major limitation to the inclusion of postcranial characters in phylogenetic studies. Thus, Grine et al. (2022) worked on a synthesis of attributions of postcranial fossils at the species level by reviewing the African fossil record from the late Miocene to the Middle Pleistocene. This contribution is especially important as most of the postcranial fossils for which taxonomic attribution is debated were found in Africa.

Choice, definition and coding of characters

41The choice, definition and coding of characters are major aspects of phylogenetic reconstruction, and disagreements over phylogenetic hypotheses are often due to the use of different datasets (i.e., in our case different matrices; Hawks, 2004; Darlu et al., 2019). The importance of character selection was illustrated by the analysis of Strait et al. (1997) who reviewed morphological characters traditionally used in early hominin systematics and assessed their redundancy (see also Strait and Grine, 2004).

42Redundancy of characters needs to be deleted from the matrices as phylogenetic methods consider characters as independent arguments. However, identifying redundant traits is a very difficult task. First, redundancy can stem from different factors: at least descriptive, structural (including developmental) and functional types of redundancy can be listed. Functional and structural types of redundancy require in-depth investigations to be identified (see Strait, 2001 for an example in hominins and Strait et al., 1997 and Strait and Grine, 2004 for detailed discussions on this topic). Methods such as geometric morphometrics have been developed in recent decades to address character integration and to quantify shapes precisely. These methods have been extensively applied to hominins outside a phylogenetic framework (but see Strait, 2001) and many studies on character integration are likely to be used for character definition and/or weighting (e.g., Conaway and Von Cramon-Taubadel, 2022; Jung and Von Cramon-Taubadel, 2022; Komza et al., 2022).

43Descriptive redundancy seems easier to identify and is therefore usually addressed by paleoanthropologists doing phylogeny. Two opposite approaches can be identified in the paleoanthropological literature regarding ways of handling descriptive redundancy: studies focusing on a limited number of characters for which in-depth investigation is provided (e.g., Strait et al., 1997) and studies analysing large datasets with the emphasis placed on the larger quantity of data (e.g., Dembo et al., 2015; Ni et al., 2021). The second approach clearly presents drawbacks, namely that super-matrices have repeatedly shown low concern for character selection and construction criteria (Scotland et al., 2003; Simões et al., 2017; Mongle et al., 2019; 2022).

44Besides redundancy, many detailed methodological studies regarding the choice, definition and coding of characters have been carried out prior to or in parallel with phylogenetic analyses of hominins. Some of them are detailed below.

Biases linked to body size, age and sex of the individuals

45Numerous biases can obstruct the phylogenetic signal carried by a character. To counter these effects, paleoanthropologists have developed methods to account for their influence. Prat (2000; 2004) studied the effect of sex, age and intraspecific variability in all genera of present-day great apes (including H. sapiens) using Pearson’s Chi-2 test and Fisher’s test. Because they used metric characters and ratios, several authors have corrected these characters by skull size or volume (e.g., Chamberlain and Wood, 1987; Wood and Collard, 1999; Bouée and Détroit, 2010; Ni et al., 2021; see especially Gilbert et al., 2009 on papionins for a relevant methodology). While juvenile individuals are usually excluded from phylogenetic analyses, characters describing deciduous dentition can be included, as their morphology does not change once the individual is fully grown (e.g., Strait and Grine, 2004 and all analyses re-using their descriptions).

Intra-taxonomic variability and phylogenetic "quality" of characters depending on that variability

46In general, polymorphic characters within a taxon are considered unsuitable for morphological analyses (Wiens, 1995), although they are often used in phylogenetic reconstructions based on molecular data (Darlu et al., 2019). Therefore, they have frequently been excluded from morphological analyses (e.g., Caparros, 1997; Martinón-Torres et al., 2007). For example, Martinón-Torres et al. (2007) retained only nine of the non-metric dental characters they used in the phenetic analysis that is part of the same study, excluding from the parsimony analysis those that are homoplastic and polymorphic within clades. Alternatively, simple rules have been applied to the coding of these polymorphic characters: for example, Dembo et al. (2016) explicitly stated that "if less than 66% of the specimens exhibited a given character state, the species was coded as polymorphic".

47Most authors have preferred to assign to each taxon the different discrete states observed in its hypodigm (e.g., Prat, 2004). This way of operating can then be analysed in two ways: either the character state can change to accommodate the most parsimonious hypothesis, or the variable character state is considered intermediate (Argue et al., 2017). A few authors have tested both approaches (e.g., Strait et al., 1997; Argue et al.; 2017), while others have either used both characters with a "variable" state coded as intermediate and characters with multiple states coded for the same OTU (e.g., Kimbel et al., 2004; Mongle et al., 2019), or restricted their approach to the first (e.g., Prat, 2004) or the second (e.g., Dembo, 2016:14).

48Some authors have quantified and detailed the process they followed, to include polymorphic characters in phylogenetic analyses or not, or to define character states for polymorphic characters. For example, for each of his 50 initial characters, Caparros (1997) studied the retention index (RI) and the distribution of character states in the first trees he obtained, to exclude 36 characters not considered useful for reconstructing the phylogeny. His exclusions were based on both a quantitative criterion, the RI, and a qualitative one: the variability of character states in robustly defined taxa (i.e., H. neanderthalensis or present-day H. sapiens) or their informativeness for resolving the phylogeny of the "archaic H. sapiens" (i.e., Middle Pleistocene hominins) on which he focused (see subsection "How to weight characters?" for a criticism of this approach). As for the character states definition, Zeitoun (1996) proposed an original way of accounting for variation within phylogenetics (Appendix A, column Q). He provided a measure of intraspecific variability that enabled the scope of the character states to be defined by the standard deviation observed in the H. sapiens population used as a reference: if, within the H. sapiens population, a ratio or a measurement had a standard deviation of 0.1, three character states between 0 and 0.3 were defined: state 0=0 to 0.1; state 1=0.1 to 0.2 and state 2=0.2 to 0.3. Regarding the question of intra-taxonomic variability, dental characters are an interesting case study, in particular because they have been scored in many different ways. These approaches may be grouped as follows: non-normed anatomical descriptions (e.g., Kimbel et al., 2004; Strait and Grine, 2004; Martinón-Torres et al., 2007), normed using ASUDAS characters (Appendix A, column Q) and/or crown dimensions (e.g., Strait and Grine, 2004; Dembo et al., 2016; Ni et al., 2021, and see especially Irish and Grabowski, 2021 for a detailed study on teeth dimensions). ASUDAS characters are dental characters derived from a reference table called the Arizona State University Dental Anthropology System, which consists of a set of around 30 plaques (Scott and Irish, 2013), depending on whether newly described characters are incorporated in the system or not. Given that many plaques are used to describe a tooth type (i.e., Carabelli’s cusp on all three upper molars), they may generate considerably more than 30 characters. This table, initially created to identify variations between current populations of H. sapiens, has been used for some years in paleoanthropology in variably modified forms (e.g., Bailey and Wood, 2007; Martinón-Torres et al., 2012; 2019; Irish et al., 2013; 2018 – but see Kimbel et al., 2013; Carter et al., 2014; Mongle et al., 2019 – Xing et al., 2014; 2015; 2016; 2018; 2021). ASUDAS characters have the advantage of being standardized, which helps to increase the reproducibility of research work (Irish et al., 2018). Moreover, these traits are increasingly well described in all hominins, making it easier to build up a large dataset than for non-normed traits, though noise linked to inter-observer error could increase in parallel. However, they may describe only a small portion of non-metric dental variations (Carter et al., 2014), thus warranting their use in conjunction with additional non-metric dental characters (e.g., Ni et al., 2021; Gousset et al., 2026).

49Characters described here as "non-normed", such as "Cusp number on mandibular M2: four (0), five (1), six (2), greater than six (3)" (character 316 of Dembo et al., 2016), are frequent. However, they oversimplify the intra-taxonomic variability that is otherwise evidenced by ASUDAS descriptions.

50Different solutions have been proposed to implement ASUDAS characters in the matrix. One simple solution is to use just one or a very small number of individuals as OTUs (Ni et al., 2021), thus reducing the amount of within-OTU variability to almost zero. However, this approach may increase the noise due to excessive subsampling. Other methods take into account the variability present within an OTU. Stringer et al. (1997) and Gousset et al. (2026) calculated the frequency of individuals above a threshold value (usually called the "breakpoint" in the ASUDAS system; Scott and Irish, 2017). In hominins and for morphological characters, frequencies have mainly been used for ASUDAS characters and thus mainly for dental characters (Stringer et al., 1997; Gousset et al., 2026; but see Strait et al., 1997 for a few additional examples outside of the ASUDAS system). Conversely, Bailey (2002) calculated the average character state (the latter being ordered) by OTU (figure 6).

Figure 6

Figure 6

Diagram showing the various coding procedures used to integrate ASUDAS characters in a matrix. The upper part of the figure (over the upper horizontal black line) shows the primary data. This data may be transformed in two different ways: by using a mean value rounded to the nearest integer (method of Bailey 2002) or by counting the individuals over a breakpoint character state (method of Stringer et al. 1997). The black crosses represent individuals; the continuous and vertical black line represents the breakpoint; values in a green square are those implemented in the matrix |
Schéma présentant les différents codages utilisés pour intégrer les caractères ASUDAS dans une matrice. La partie supérieure de la figure (au-dessus de la ligne noire horizontale la plus haute) représente la donnée primaire. Elle peut être transformée de deux façons : en réalisant une moyenne puis en arrondissant cette moyenne à l’unité près (méthode de Bailey 2002) ou en codant le pourcentage d’individus au-dessus d’un état de caractère servant de seuil (méthode de Stringer et al. 1997). Les croix noires représentent des individus, la ligne continue verticale représente la valeur seuil et les valeurs dans un encadré vert sont celles implémentées dans la matrice

51The use of means or frequencies is nevertheless debated in hominin phylogenetics. The character selection made by Caparros (1997) and Martinón-Torres et al. (2007) is linked to theoretical considerations: they considered that phylogeny should be based on characters that are not polymorphic within clades but rather fixed, thus excluding the use of frequencies. This assumption (along with criticisms detailed later in this article) lead some authors to reject the use of ASUDAS characters in phylogenetics (Kimbel, 2013; Carter et al., 2014). This concern is more broadly related to the definition of character states as "mutually exclusive conditions" (Sereno, 2007), which is problematic when ASUDAS characters are implemented as frequencies or means.

52However, it is argued here that variability should be incorporated in detail into the matrix. ASUDAS characters are often described at the OTU level as frequencies (e.g., Bailey, 2002; Irish et al., 2013; 2018) precisely because their use is evidence that they are mostly variable within taxa. While Carter et al. (2014) argued that a small sample size, due to the scarcity of fossils, does not give a clear idea of intra-taxonomic variation, this limitation applies to virtually every other type of character implementation (Trinkaus, 1990; Hawks, 2004). To return to the exclusion of characters based on the observation of polymorphism, it should be noted that this approach does not test whether the variability reflects a structure within the taxon considered, in which case it may be informative from a phylogenetic point of view. Different populations can indeed be identified through anatomy in taxa that are widely recognized as monophyletic, such as H. sapiens and H. neanderthalensis (Fabre et al., 2009; Scott and Irish, 2017; Picin et al., 2020). Nor does this approach consider that variability present in one or more taxa may not be ubiquitous: if it is not, one can assume that a phylogenetic signal is nevertheless provided by the trait for those taxa in which a trait state is fixed (as advocated by de Pinna, 1991).

53Furthermore, Wiens (1995) has shown the usefulness, in reconstructing phylogenies, of such polymorphic characters when implemented as frequencies. Further use of ASUDAS characters in phylogenetics should take advantage of methods that allow more information to be integrated than only averages or frequencies over a certain grade. The Generalized Frequency Coding (GFC) proposed by Smith and Gutberlet (2001) seems to be a good option in that it implements the percentage of each character state in the matrix, using cumulative frequencies when the character is ordered.

How to transform continuous data for phylogenetic analysis?

54Most phylogenetic analyses of hominins have implemented discrete characters in their matrices (figure 7). However, genuine variability of traits is often continuous. Thus, various methods have been employed to discretize them (see in particular Zeitoun, 1996; Stringer et al., 1997; Collard and Wood, 2000; Bailey, 2002). The other analyses have discretized them explicitly but without argumentation as to the choice of character states (e.g., Cameron et al., 2004; Mounier and Caparros, 2015), or without making explicit reference to a measure (all analyses grouped in "no continuous characters" in figure 7).

Figure 7

Figure 7

Pie chart representing the type of characters used for each phylogenetic analysis on hominins (n=47). Were considered as continuous characters discretized characters for which an explicit discretization logic and method do exist, excluding for example Cameron et al. 2004 and Mounier and Caparros 2015. *GM stands for Geometric Morphometrics. Graph generated from the data presented in appendix D |
Diagramme circulaire présentant le type de caractères utilisés pour chaque analyse phylogénétique des hominines (n=47). Ont été considérés comme caractères continus discrétisés des caractères pour lesquels une argumentation et une méthode existent pour la discrétisation, excluant par exemple Cameron et al. 2004 et Mounier et Caparros 2015. Graphique généré à partir des données présentées dans l’appendice D. *GM signifie morphométrie géométrique

55While numerous discretization methods exist (Darlu et al., 2019), many authors have argued that they bias the original data (Trinkaus, 1990; Hawks, 2004). This is no longer a methodological limitation thanks to the advent of treatment of continuous characters "as such". The integration of continuous characters "as such" in a matrix was made possible by the implementation of a new algorithm in TNT software, which allows a character to be divided into 65,001 states (Goloboff et al., 2006). The number of character states that can be implemented in phylogeny softwares had previously been technically limited to 32 and then 1001 characters in PAUP (Bardin et al., 2014). The coding of continuous characters "as such" makes it possible to implement a range of variation per taxon (Goloboff et al., 2006). However, to our knowledge, and excluding "geometric morphometric characters", only two publications on hominin phylogeny have included continuous characters treated "as such" (Appendix D; figure 7), though they have been shown to carry a strong phylogenetic signal in hominoids (Pugh, 2022).

How to integrate geometric morphometric data into a phylogenetic analysis?

56Geometric morphometrics is a method designed for capturing fine morphological differences between individuals and taxa. As such, it can be used as a basis for phylogenetic analyses. Historically, the use of geometric morphometrics in phylogeny has been criticized (e.g., Adams et al., 2011), especially in paleoanthropology, although it has been shown to be effective in reconstructing phylogenies (Goloboff and Catalano, 2016; Palci and Lee, 2019). Up to now, few studies have included geometric morphometrics in phylogenetic analyses of hominins, and these have focused only on the skull (Appendix D).

57Two different approaches exist to integrate the results of a geometric morphometric analysis into the matrix. One way is to implement the landmark coordinates after alignment (Bouée and Détroit, 2010; Simon-Maciejewski et al., 2024). This procedure has the advantage of allowing the characters implemented in the matrix to fit the notion of primary and secondary homology, as defined for discrete characters (Palci and Lee, 2019). While the two published studies on hominins that have used this approach performed a procrustes alignment, Palci and Lee (2019) argued that RTFRA alignment may be better suited for phylogenetics. A second option is to perform a principal component analysis before using the values along these principal components (Appendix A, column O). This approach has been criticized by Adams et al. (2011), who noted the high susceptibility of the data included in the matrix to the addition or deletion of specimens, but it has the advantage of giving morphological independence to the characters implemented in the matrix (Simon-Maciejewski et al., 2024). This second option was the first to be used, by Gonzáles-José et al. (2008). However, these authors used only the first few principal components (accounting for 75% of the variation), while Parins-Fukuchi (2021), who took up their data, used the 13 first principal components to account for nearly 100% of the variation.

58Simon-Maciejewski et al. (2024) applied both approaches (3D coordinates after alignment and values along PC axes) to their own dataset and obtained very similar topologies for the two. Unlike the studies presented above, which used individuals as OTUs, Simon-Maciejewski et al. (2024) averaged values by OTU, after checking that specimens belonging to an OTU plotted close to each other on the first principal components.

59The integration of geometric morphometrics in phylogenetic analyses is a very promising approach that deserves to be applied to other anatomical regions.

How to weight characters?

60A priori weighting of characters is uncommon in hominin phylogenetics. It has been explicitly rejected by some authors (Zeitoun, 1996; Caparros, 1997; Dembo, 2016) on grounds of subjectivity. Character weightings have been applied when the same trait was coded for a serial element (e.g., all molars, all metatarsals), since the expression of these traits may have one single genetic origin. Consequently, these traits were weighted in inverse proportion to the number of teeth for which they were coded in Strait and Grine (2004) and Gousset et al. (2026) (although to different degrees in these two articles, see Gousset et al., 2026). In addition, Strait and Grine (2004) used the same idea to downweight characters that belong to the same complex, for example all those describing canine size.

61Following the same idea, a weighting may be given to characters depending on the number of character states they have: if the number of character states differs between characters, an artificial overweight is given to characters described with more states in the absence of such a weighting. Thus, Wiens’ (2001) between-character scaling has sometimes been used (e.g., Gousset et al., 2026) and its absence in Irish et al.’s (2013) work was criticized by Carter et al. (2014).

62The most common weighting applied in hominin phylogenetics depends on homoplasy. Homoplasy is defined as a similarity between two taxa that does not derive from a common ancestor. In the context of phylogenetic analysis, it is regarded as noise that one seeks to exclude from arguments to reconstruct evolutionary history, since it interferes with the phylogenetic signal one is trying to identify (Darlu et al., 2019). Weighting against homoplasy therefore aims to give a lower weight to the most homoplastic characters. The latter are identified as those whose states appear and/or disappear at least twice in the trees. Several methods can be used to weight characters according to their homoplasy, each aiming to increase the internal consistency (i.e., absence of conflicts between characters) of the resulting tree (Goloboff, 1993; Darlu et al., 2019).

63The first set of methods is used to weight the characters once the parsimony analysis has been carried out (Goloboff, 1993). Character weighting depends on different indices: RI, consistency index (CI) or "rescaled" CI (column R of Appendix A). While some authors have compared the results obtained before and after this reweighting (Caparros, 1997; Widianto and Zeitoun, 2003), Caparros and Prat (2021) explained that they retained the topology found after reweighting because they obtained a higher RI after reweighting than before.

64"Implied weighting", as opposed to "successive weighting", weights characters according to the rate of homoplasy during the tree search and not as a second step (Goloboff, 1993). Unlike successive weighting, implied weighting is not an iterative method that integrates the weight initially given to characters. Moreover, implied weighting can be used with variable "strength", reflecting the relationship between the homoplasy rate per character and the weight that will be given to each of them, which is not the case with successive weighting (Goloboff, 1997). Gousset et al. (2026) used different "strengths" of implied weighting in their dataset when examining the specific case of H. luzonensis, which probably evolved in a context of insular isolation (Détroit et al., 2019). In this specific case, implied weighting did not appear to be effective for reconstructing relationships, since it disfavoured reversals, which are frequently documented in insular environments (Hooijer, 1954; Sondaar, 1994; Van der Geer, 2005; van den Hoek Ostende, 2018). Furthermore, for the same analysis, the trees obtained after implied weighting had a lower RI and CI (Gousset et al., 2026).

65It is important to note that the foundations of methods for reweighting characters have been criticized (Congreve and Lamsdell, 2016): many authors have pointed out that a character which is homoplastic in one part of the tree can still support a clade in another part of the tree, and is then useful for reconstructing the phylogeny (De Pinna, 1991; Congreve and Lamsdell, 2016 and references therein). This consideration nuances the value of using methods that downweight potentially useful characters.

Phylogeny beyond parsimony

Model-based methods

66In contrast to parsimony analyses, model-based methods are based not only on the sharing of derived character states, but on an evolutionary model defined a priori that enables estimations of the most probable tree among the existing ones. Since they were conceived in 1963 by Edwards and Cavalli-Sforza, and in particular since their use in a computationally efficient context (Felsenstein, 1981), model-based methods have mainly been applied to molecular data (Folinsbee et al., 2015). Since the early 2000s, they have increasingly been used to analyse anatomical data (Folinsbee et al., 2015). Among model-based methods, it is possible to distinguish between Bayesian likelihood and maximum likelihood. Both are based on probabilities. A likelihood is a probability of observing a dataset given a model that can generate this kind of data. In its classic form, a model in phylogenetics contains a tree and an evolutionary model. The probability of data (traits), given a tree and an evolutionary model, consists of the joint probability (i.e., the product) of all the characters from the dataset (e.g., matrix taxa/characters). The probability of one of these traits is then the sum of all possible histories of the trait respecting the values of the OTUs. Finally, the probability of one of these trait histories consists of the joint probability of all branches of the tree with some trait values on internal nodes (those of the given trait history under scrutiny). The probability of each of these branches comes from the evolutionary model that provides the calculation to go from the trait value at one extremity of the branch to the other. To sum up in one sentence, the likelihood of a given tree is the probability of all characters, given that they may have different histories (with different probabilities) on that tree (Felsenstein, 1985). The maximum likelihood consists of using this metric to find the best tree. This formulation allows easy handling of the basic version of model-based phylogenetics. However, many improvements have been made over the years that complicate this scheme but allow well-identified properties of living systems to be taken into account, such as dependencies between traits or between branches (e.g., de Koning et al., 2012, Billet and Bardin, 2019). Based on conditional probabilities, the Bayes theorem can be used to link the likelihood (i.e., the probability of the data given the model) and the posterior probabilities (i.e., the probability of the model given the data). Bayesian inference is the most widely used method in hominin phylogenetics among model-based methods (Appendix A, column F) in that it allows a direct quantitative assessment of alternative hypotheses (i.e. the trees).

67Gonzáles-José et al. (2008) were the first to use model-based methods for hominins, based on geometric morphometric data (see subsection on "Geometric morphometrics"). Dembo et al. (2015; 2016) also used model-based methods, but from "classic" discrete traits. Since then, the vast majority of studies in hominin phylogeny have used model-based methods, either in parallel with a parsimony analysis or on their own (Appendix A, columns E-F). Phylogenetic analyses using geometric morphometric data tend to use model-based methods more frequently (Gonzáles-José et al., 2008; Bouée and Détroit, 2010; Parins-Fukuchi, 2021).

68Studies applying parsimony analysis or model-based methods to a matrix that has already been analysed by the other set of methods, along with published articles using both, enabled to compare the results obtained (Appendix A, columns E, F, I). No rule seems to explain the differences in the topologies obtained by one method or the other. The number of differences varied from low (Organ et al., 2011; Argue et al., 2017) to relatively high (Gonzáles-José et al., 2008; Mongle et al., 2019). Caparros and Prat (2021) rejected the most probable tree obtained by probabilistic methods and retained only those obtained by one of their parsimony analyses, because they observed far less consistency in the trees obtained by Bayesian inference than by their parsimony analysis (Caparros and Prat, 2021). Ni et al. (2021) obtained tree topologies that differed from the one obtained with maximum parsimony. They therefore decided to constrain the tree presented in the main text by using the topology obtained with maximum parsimony (Ni et al., 2021:48).

69Commonly used in association with model-based methods is the total-evidence approach (Ronquist et al., 2012). Total-evidence uses anatomical, molecular and temporal data for those taxa for which such data is available. While this approach has served to estimate divergence times and to reconstruct the evolution of key traits (e.g., Püschel et al., 2021; 2022; Mongle et al., 2022; see section "Wanted: a phylogeny to understand human evolution"), total evidence has not yet been used to reconstruct hominin phylogeny.

Phylogenetic networks

70Phylogenetic networks provide a graphic representation that allows web-like links between taxa as opposed to only diverging branches in a tree. This method is used when evolution in the form of populations exchanging genes is considered, as opposed to well-separated lineages (Trueman, 2010). It is based on congruence and conflicts between trees (Caparros and Prat, 2021). As discussed above, phylogenetic networks can also be used to explore a dataset (Gautney, 2023) but it is its application to phylogenetic relationships that is discussed here.

71Trueman (2010), who criticized the article by Argue et al. (2009), had already proposed the use of network representations in hominins, without using them as extensively as Caparros and Prat (2021). Caparros and Prat (2021) justified their approach by the existence of interbreeding between different fossil hominin species, which is now known to have been an important biological mechanism at least among the most recent representatives of the genus Homo (Bergström et al., 2021; Peyrégne et al., 2024). Using maximum parsimony and reweighting, Caparros and Prat (2021) obtained 3213 trees, which they implemented in the network analysis. The polytomies visible on the consensus tree of these 3213 trees served as the basis for reconstructing evolutionary lineages in the network analysis. The latter made it possible to identify "evolutionary dead ends" and anagenetic lineages in hominins. For example, H. floresiensis and H. habilis are collectively separated from the rest of the genus Homo, suggesting a clear separation between two evolutionary lineages.

Too few matrices?

72While there are over 40 papers presenting phylogenetic analyses of fossil hominin taxa, far fewer matrices have been created. Different research teams have published several times based on the same characters and character states, with each new publication often associated with the description of a new individual or taxon. The creation of a de novo matrix to describe a partial fossil (Appendix A, column I) or one or a very few specific anatomical regions (Appendix A, column J) reduces number of "complete" matrices still further (see section on "Which anatomical elements?").

73Thus, just a few matrices including multiple recent Homo species have been reused over the last 30 years, initially published in Zeitoun (1996), Gonzalez-José et al. (2008), Argue et al. (2009), Dembo et al. (2015) (that is a combination of previous matrices) and Mounier and Caparros (2015) respectively (Appendix A, column I). The same can be said about matrices focusing on early hominins, the majority of which are ultimately based on Groves (1986) and Skelton and McHenry (1992).

74The fact that some datasets have been reused multiple times should be kept in mind when trying to identify whether one phylogenetic position of a taxon is better supported than another. As an example, H. floresiensis systematically found a basal position relative to H. erectus sensu lato and more recent representatives of the genus Homo when the matrix of Dembo et al. (2016) is used (Appendix A, column I). This may give the impression that the assumption of a basal position within hominins for this taxon is well supported by different phylogenetic analyses, although only two distinct and repeatedly analysed datasets (from Dembo et al., 2016 and Argue et al., 2009 augmented in Argue et al., 2017) produce such a signal. The third dataset including H. floresiensis (that of Zeitoun, 1996 in the analysis of Zeitoun et al., 2016) produced a totally contradictory message since H. floresiensis was found within H. erectus sensu stricto (see section on "Contribution of phylogenetic analyses to resolving relationships among recent Homo").

75However, independent matrices may yield identical or close phylogenies, thus lending further support to the phylogenetic hypotheses proposed. As an example, Kimbel et al. (2004) largely reproduced the results of Strait et al. (1997) when using a new dataset (Strait et al., 2015).

76Given the significant lack of consensus on the phylogenetic relationships of late Homo (see Strait et al., 2015 and section entitled "Wanted: a phylogeny to understand human evolution"), additional analyses based on independent datasets are clearly needed for these taxa.

Molecular phylogenies

77Hominins constitute a taxon of which only one representative remains today, which for a long time prevented any interest in the phylogeny of this group based on molecular data. However, the sequencing of the mitochondrial DNA of the H. neanderthalensis holotype in 1997 opened up this field of study (Krings et al., 1997; Bocherens et al., 2022). Thirteen years later, the publication of the partial genome of H. neanderthalensis and of the Denisovans signalled a major advance in this field (Green et al., 2010; Reich et al., 2010). More recently, proteins have been extracted from Middle and Lower Pleistocene fossils, enabling taxonomic identification of three of them as well as phylogenetic analyses (Chen et al., 2019; Welker et al., 2020; 2025; Fu et al., 2025b; Madupe et al., 2025; Tsutaya et al., 2025). Often (but not universally) considered more reliable than phylogenies obtained from morphological data, molecular phylogenies have already increased our knowledge of some portions of recent human evolution and might continue to do so for other hominin taxa.

78The use of molecular data for hominin phylogeny involves multiple stages. First, stringent climato-edaphic conditions are required to be able to extract fossil DNA or proteins, particularly in the case of DNA (Willerslev et al., 2004). While the oldest hominin DNA sequenced to date is dated to around 430 ka (Meyer et al., 2016), the oldest proteomes have been identified at the sites of Dmanisi (Georgia, dated between 1.78 and 1.85 Ma – Welker et al., 2020) and Swartkrans (South Africa, dated between 1.8 and 2.2 Ma – Madupe et al., 2025). The earliest dates for any non-hominin species are 2.4 Ma for DNA (Kjær et al., 2022) and over 20 Ma for proteins (Green et al., 2025; Paterson et al., 2025). Paleo-proteome analysis therefore seems more likely to inform about the phylogenetic relationships of ancient hominins (but see below the technical limitations associated with paleo-proteomes). Proteins can be extracted from bone, enamel or dentin. For hominins, protein-based phylogenies have been derived from dentin (for the Xiahe mandible – Chen et al., 2019) and enamel proteomes (for H. antecessor – Welker et al., 2020 – and P. robustus – Madupe et al., 2025).

79Once the data has been extracted from the fossil or sediment, the sequences must be aligned for analysis. The aim is to make all the sequences comparable by analysing differences between homologous characters (in this case nucleotides or amino acids).

80Finally, molecular data can provide phylogenies via parsimony or model-based methods, both of which are commonly used for hominins (see, for example, Picin et al., 2020 for parsimony and Prüfer et al., 2014 for model-based methods). To our knowledge, hominin phylogenies from paleoproteomic data have only been generated by model-based methods (Bayesian likelihood and maximum likelihood – Chen et al., 2019; Welker et al., 2020; Madupe et al., 2025).

81As we saw above, some proteins can be preserved for longer than DNA, especially enamel proteome (Demarchi et al., 2016). On the other hand, enamel proteome is much less informative for phylogeny, due to the much smaller amount of extractable data (Welker et al., 2025). For example, the proteins extracted at the Dmanisi site did not allow to place H. georgicus within the hominin phylogeny, and those identified for H. antecessor did not allow full reconstruction of known DNA-based relationships between Denisovans, H. neanderthalensis and H. sapiens (Welker et al., 2020). Moreover, proteins found in dental enamel (which is often the only source of paleo-proteins) evolve particularly slowly, thus often preventing any identification of differences between fossil species (Fong-Zazueta et al., 2025).

82In short, while molecular data are increasingly present in phylogenetic analyses of hominins, for the time being they mainly concern the most recent fraction of human evolution. Although numerous sampling and sequencing efforts hold out hopes of obtaining molecular data from earlier times and for more taxa, certain taphonomic limitations seem to be highly restrictive.

Contribution of phylogenetic analyses to resolving relationships among recent Homo

83The various studies whose methodology is described above have led to significant advances in our understanding of hominin evolution. This section focuses on relationships between recent Homo species. The reader is invited to refer to the article by Strait et al. (2015) for an equivalent review on earlier hominins.

The supertree: is there a consensus on recent Homo phylogeny?

84To illustrate the points currently agreed upon and those still in debate in the phylogeny of recent Homo species, a supertree has been generated. This supertree is a single phylogenetic tree obtained from all those generated by the analyses mentioned in this review. To do so, one tree per article was registered in nexus format (Appendix E). Only the analyses where at least two OTUs belong to the genus Homo were considered. The supertree was generated in R (v. 4.3.1; R Core Team, 2023), using the supertree function in the "Phangorn" package (v. 2.11.1; Schliep, 2011) following the procedure described in Ragan (1992). The following packages were also used: "ape" (Paradis and Schliep, 2019), "phytools" (v. 2.1.1; Revell, 2024) and "DescTools" (Signorell, 2024).

85This procedure produced a large number of maximum parsimony trees. A majority-rule consensus tree of these most parsimonious trees was therefore computed (Margush and McMorris, 1981; Darlu et al., 2019:353). Clades found in 50% or more of the trees were retained.

86While the significant heterogeneity of the OTUs used in these analyses complicates the reading of the tree, the result obtained helps to illustrate the points discussed later in the text. Figure 8 shows the very low resolution of the supertree, reflecting the lack of consensus on the relationships of recent Homo species. The existence of numerous non-mutually exclusive OTUs (e.g., "Homo heidelbergensis sensu lato", "Homo heidelbergensis sensu stricto", "Sima de los Huesos/Petralona", "Sima de los Huesos") highlights how much the alpha-taxonomy among late hominin taxa is debated. Furthermore, clades containing few OTUs are often groups of taxa that are present in only one analysis (or in a series of analyses produced by the same authors).

A brief history by taxon

87Chronologically, the first major contribution of phylogenetic analyses was the invalidation of the multiregional model. Though other (and perhaps stronger) arguments were brought by other methods than phylogenetics to refute the multiregional model (e.g., Wainscoat et al., 1986; Cann et al., 1987; reviewed in Stringer, 2022), the earliest phylogenetic analyses of late Homo focused on this topic and contributed important arguments based on the fossil record. The multiregional model stipulates that today’s populations of H. sapiens descend from locally established populations in the same regions over a long period of time. In the framework of this model, Asian H. erectus, for example, are considered ancestors of today’s Asian and Oceanic populations (Wolpoff et al., 2000). The content of the multiregional model has evolved over time, with older hypotheses proposing very strong local continuity between extinct species of the genus Homo and present-day H. sapiens (Coon, 1962), while more recent proposals emphasize, above all, the major role of gene flows between the populations occupying the different regions during the Pleistocene (Wolpoff et al., 2000). However, phylogenetic analyses do not support any of these proposals. Analyses carried out with specimens as OTUs clearly showed the monophyly of H. sapiens, with older representatives of this taxon, identified in Africa, in a more basal position (Caparros, 1997; Bouée and Détroit, 2010; Mounier and Caparros, 2015; Mounier et al., 2016; Ni et al., 2021; Appendix E, column E). The latter analyses therefore clearly favour the out-of-Africa model, which is classically opposed to the multiregional model (Scerri et al., 2019). Another argument against the multiregional model is the identification of one or more lineages of H. erectus sensu lato, as distinct from Asian H. sapiens, highlighted by Zeitoun (1996), Zeitoun et al. (2010) and Bouée and Détroit (2010). Indeed, if an anagenetic Asian H. erectus lineage had led to the appearance of Asian H. sapiens, one would have expected the fossils belonging to Asian H. erectus to branch out successively as sister taxa to current Asian H. sapiens. The more recent the Asian H. erectus, the closer to H. sapiens they would have been. The multiregional model is invalidated in the supertree by the absence of Asian H. erectus in clade 3, where specimens attributed to H. sapiens, or which could belong to their lineage (H. rhodesiensis), are found (figure 8).

Figure 8

Figure 8

Supertree generated from 37 phylogenetic analyses on hominins. Green points represent clades found in each tree considered, orange points represent clades found in more than 75% of the trees but less than 100% and red points represent clades found between 50% and 75% of the trees. Taxa of which the position is discussed in the text are signalled by a blue star. Purple stars point to the fossils frequently included in H. heidelbergensis sensu lato |
Superarbre généré à partir de 37 analyses phylogénétiques portant sur les hominines. Les points verts représentent des clades trouvés dans tous les arbres considérés, les points orange des clades dont la fréquence est supérieure à 75 % et strictement inférieure à 100% et les points rouges des clades dont la fréquence est comprise entre 50 % et 75 %. Les taxons dont la position est discutée dans le texte sont signalés par une étoile bleue. Les étoiles violettes désignent les fossiles communément inclus dans H. heidelbergensis sensu lato

88More recently, there has been a shift in paradigm to overcome the binary division between these two models (recent out-of-Africa vs. multiregional) (Scerri et al., 2019). Since hybridization between multiple hominin taxa is now widely documented (Green et al., 2010; Reich et al., 2010; Slon et al., 2018), more comprehensive models have been proposed for the emergence of our species (Scerri et al., 2019). However, phylogenetics from morphological data has brought little to this debate, owing to the difficulty of identifying hybridization from morphology alone (Ackermann et al., 2019; Harvati and Ackermann, 2022), although phylogenetic networks could be a useful approach to address this issue (Caparros and Prat, 2021; Gautney, 2023 but see Meneganzin and Bernardi, 2023 for counterarguments).

89Phylogenetic analyses of recent taxa of the genus Homo have also made it possible to validate several hypotheses concerning the taxonomy and phylogeny of these taxa, and to propose new ones.

90Most of the phylogenetic analyses used to test the monophyly of H. neanderthalensis highlight the coherence of this group on a phylogenetic level (Zeitoun, 2001; González-José et al., 2008; Mounier and Caparros, 2015; Mounier et al., 2016; Ni et al., 2021; Profico et al., 2023; figure 8; Appendix E, column F). These same analyses have made it possible to argue for or against the inclusion in the H. neanderthalensis lineage of fossils whose taxonomic attribution is in debate. Thus, the Sima de los Huesos fossils belong to the Neanderthal lineage according to all recent phylogenetic analyses (González-José et al., 2008; Mounier and Caparros, 2015; Ni et al., 2021; contra Caparros, 1997; Cameron et al., 2004; Appendix E, column G). Analyses by Mounier and Caparros (2015), Mounier et al. (2016) and Ni et al. (2021) have also made it possible to include numerous late Middle Pleistocene European fossils, such as those from Ehringsdorf or Saccopastore, in the Neanderthal lineage. All these analyses corroborate the proposal that a single evolutionary lineage can be identified in the European fossil record from the very late Middle Pleistocene onwards. This proposal is also compatible with the accretion model (Hublin, 2009). The supertree finds clade 1 composed exclusively of specimens attributed to H. neanderthalensis (figure 8). However, other specimens usually attributed to this taxon are absent from this group. Most of these specimens are Middle Pleistocene fossils (figure 8 – e.g., Steinheim, Swanscombe, Ehringsdorf, Sima de los Huesos).

91In contrast, the taxonomic attribution and relationships of early to mid-Middle Pleistocene European and African fossils, which are sometimes grouped under H. heidelbergensis, are much less clear. The monophyly and the phylogenetic relationships of fossils attributed to this taxon vary widely from one study to another (Stringer, 1987; Martinez and Arsuaga, 1997; Zeitoun, 2001; Cameron et al., 2004; Martinón-Torres et al., 2007; Gilbert, 2008; Argue, 2015; Mounier and Caparros, 2015; Mounier et al., 2016; Ni et al., 2021), with H. heidelbergensis mostly appearing as paraphyletic. Note that in cases where different H. heidelbergensis specimens branch successively into a predominantly pectinate tree structure, this paraphyly does not necessarily invalidate the notion of H. heidelbergensis as a chronospecies, ancestral to monophyletic taxa (H. neanderthalensis and perhaps H. sapiens). In the supertree, they do not form a monophyletic group (figure 8).

92Conversely, the phylogenetic relationships of H. antecessor appear to be better established. When describing this species, Bermúdez de Castro et al. (1997) proposed that this new taxon was a common ancestor of H. neanderthalensis and H. sapiens. Phylogenetic analyses generally confirm this hypothesis, since they mostly confer a basal position to H. antecessor relatively to H. heidelbergensis, H. neanderthalensis and H. sapiens and a derived position compared to all other hominins (Cameron et al., 2004; Martinón-Torres et al., 2007; Dembo et al., 2015; 2016; Parins-Fukuchi et al., 2019; Caparros and Prat, 2021, see however Ni et al., 2021; Appendix E, column H and discussion below). Since then, phylogenetic analyses based on paleoproteomic data have partially confirmed this position in the hominin tree (Welker et al., 2020).

93More recently, three hominin taxa with numerous plesiomorphic characters have been described (Brown et al., 2004; Berger et al., 2015; Détroit et al., 2019), uncovering previously unsuspected phyletic diversity for recent periods of human evolution (Middle and Upper Pleistocene). As soon as these studies were published, the ensuing debates focused on the relationships of these hominins, with phylogenetic analyses soon making contributions. For the two island species, two main hypotheses concerning their relationships with hominins have been formulated (Brown et al., 2004; Détroit et al., 2019):

94- hypothesis 1: H. floresiensis/H. luzonensis evolved from a population of H. erectus, probably Asian. Once they arrived on their respective islands, these species evolved in relative isolation, which explains the small body and brain size of H. floresiensis (which can be considered as island dwarfism) and the surprising postcranial anatomy of both taxa;
- hypothesis 2: H. floresiensis/H. luzonensis originate from a population that already possessed the small dimensions and postcranial anatomy mentioned above. This ancestral population is currently unknown in the Asian fossil record.

95H. floresiensis has been studied through numerous phylogenetic analyses (Dembo et al., 2016; Parins-Fukuchi et al., 2019; Caparros and Prat, 2021), three of which looked specifically into its relationships (Argue et al., 2009; 2017; Zeitoun et al., 2016). No consensus has emerged from these analyses, since Zeitoun et al. (2016) found H. floresiensis among the H. erectus, while Argue et al. (2009; 2017) and the matrix of Dembo et al. (2015) invariably produced a basal position for H. floresiensis within the genus Homo Appendix E, column I). Uncertainty about the phylogenetic position of H. floresiensis is reflected in the supertree by its position in a multifurcation comprising taxa as diverse as members of the genus Australopithecus, early Homo (i.e., H. habilis and H. rudolfensis) and H. erectus (figure 8).

96The only phylogenetic analysis of H. luzonensis supported a close relationship with Asian H. erectus (Gousset et al., 2026), which corroborates the detailed analysis of this species’ teeth (Zanolli et al., 2022). The studies by Zeitoun et al. (2016) and Gousset et al. (2026) invalidated the hypothesis that character reversals specific to island contexts render phylogenetic analyses inapplicable to island taxa (Van der Geer, 2005; van den Hoek Ostende, 2018). If this were the case, H. floresiensis and H. luzonensis would never appear in derived positions in cladistic analyses.

97The numerous phylogenetic analyses that include H. naledi have proposed various hypotheses for this taxon, with no consensus emerging (Appendix E, column J). H. naledi appeared in a derived position relative to H. erectus in Dembo et al. (2016) and as a sister taxon of H. georgicus in Argue et al. (2017). It appeared as a sister taxon to H. habilis in Irish and Grabowski (2021) and as a sister taxon to Au. sediba in Caparros and Prat (2021). As with H. floresiensis, H. naledi’s position in the supertree confirms the lack of consensus regarding its relationships (figure 8).

98Finally, a new species, H. longi, was proposed in 2021, using a phylogenetic analysis to define its hypodigm (Ji et al., 2021; Ni et al., 2021). Described from the Harbin skull, found in north-eastern China and dated to the late Middle Pleistocene, this taxon includes other contemporary fossils found in China and Tibet (Ni et al., 2021). Since this group includes the Xiahe mandible, it could well be the Denisovans, a group which is very poorly known from the fossil record (Peyrégne et al., 2024). Even more surprisingly, phylogenetic analysis has suggested that H. longi is a sister group to H. sapiens, while in phylogenetic analyses based on molecular data, the Denisovans are a sister group of H. neanderthalensis. Ni et al. (2021) constrained the topology based on molecular conclusions in a separate analysis, which showed only a slight increase in the total length of the tree, an extended hypodigm for H. longi and more general agreement with chronology, geography and previous phylogenetic hypotheses. Recent molecular data retrieved from the holotype of H. longi has confirmed its attribution to the Denisovans (Fu et al., 2025a; 2025b).

99More recently, a new taxon has been proposed based on the Xujiayao fossil sample: H. juluensis sp. nov. Wu and Bae (Bae, 2024). The hypodigm of this species partially covers the one proposed for H. longi (Ni et al., 2021; Bae, 2024; Bae and Wu, 2024; Wu and Bae, 2025). In particular, the Xiahe mandible is part of both hypodigms. As it is to date the most complete fossil securely attributed to the Denisovans, both H. longi and H. juluensis can be considered valid taxonomical names for the Denisovans (although H. daliensis could be more appropriate, but see Bae et al., 2023). No phylogenetic analyses have been conducted to group these fossils together, unlike the work done by Ni et al. (2021) for H. longi, but Wu and Bae (2025) have conducted morphological comparisons based on metric and non-metric traits. Further complicating the issue is the absence of clear proposals as to the phylogenetic relationships of H. juluensis with other species. Wu and Bae (2025) only provide a general scheme in the form of a network suggesting links between H. juluensis, "Middle Pleistocene [Asian] Homo", H. erectus and early H. sapiens from China. We may wonder whether any of these taxa are more closely related to any other.

100While these recent studies all present shortcomings, they have the advantage of bringing the aforementioned Asian fossils back to the centre of paleoanthropologists’ attention, from which they have long been excluded (Liu et al., 2022).

Wanted: a phylogeny to understand human evolution

101Because phylogenies take relationships and the historical dimension of evolution into account, they are an essential tool for tracing the thread of evolutionary history and testing hypotheses about groups of species and their biology (Felsenstein, 1985; Harvey and Pagel, 1991; O’Meara, 2012; Garamszegi, 2014), and hominins are no exception. In this final section, a few selected examples of how phylogenies may be used to understand various aspects of human evolution are presented, reflecting their main uses in evolutionary biology: understanding a speciation pattern, estimating ages of divergence, drawing biogeographical inferences, studying the evolution of a particular trait and carrying out virtual reconstructions of a common ancestor.

102Kimbel et al. (2006) performed a parsimony analysis to test the hypothesis of an ancestor-descendant relationship between Au. anamensis and Au. afarensis. To do so, they studied four populations of Au. anamensis and Au. afarensis and found a pectinate tree topology, with each more derived terminal branch being more recent in the stratigraphy, which led them to conclude in favour of anagenesis linking these two taxa (figure 9). Their conclusion was also supported by the absence of autapomorphies in Au. anamensis, since all the synapomorphies of this taxon are also present in Au. afarensis (Kimbel et al., 2006; Folinsbee et al., 2015). Like Kimbel et al. (2006), Parins-Fukuchi et al. (2019) and Parins-Fukuchi (2021) worked on speciation mechanisms in human evolution, on a broader taxonomic scale. For this purpose, they tested the existence of three different processes relating morphological species during hominin evolution: (1) anagenesis, (2) cladogenesis involving the division of the parent taxon into two daughter taxa and (3) cladogenesis where the parent taxon persists alongside the daughter taxon. This approach aimed to reconstruct hominin evolution more realistically, reflecting the many ancestor-descendant relationships proposed in the literature (e.g., between Au. anamensis and Au. afarensis, between H. heidelbergensis and H. neanderthalensis).

Figure 9

Figure 9

Phylogenetic tree adapted from Kimbel et al. 2006. The analysis focused on the fossils excavated in the four sites that make up the most basal OTUs of the tree. Their position in the tree being correlated to their age (i.e., the OTUs branch successively following a chronological order), the authors conclude as to anagenetic evolution from Au. anamensis to Au. afarensis |
Arbre phylogénétique issue de celui présenté par Kimbel et al. 2006. L’analyse a porté sur les fossiles provenant des quatre sites qui forment les UTO à la base de l’arbre. Leur position dans l’arbre étant corrélée à leur âge (i.e., les UTO se branchent successivement suivant un ordre chronologique), les auteurs en ont conclu à une évolution de type anagénétique liant les espèces Au. anamensis et Au. afarensis

103Dembo et al. (2016) used their phylogenetic analysis to try to assign an age to H. naledi, whose dating was then unknown. The age estimated by this study was far removed from that published the following year from direct dating of fossils and on associated sediments (Dirks et al., 2017), showing once again that, hominin evolution being bushy and non-linear (Groves and Mirazon Lahr, 1994), one cannot assign a precise age to a taxon based on its anatomy alone.

104Furthermore, studies using Bayesian likelihood frequently estimated ages of divergence between OTUs (e.g., Ni et al., 2021), sometimes using molecular data to better calibrate the ages (Püschel et al., 2021 – but see Mongle et al., 2022 for a critique of this study and Püschel et al., 2022 for a response). However, the only phylogenetic analysis based on anatomy that attempted to estimate these ages while focusing on recent Homo species yielded results that are markedly incongruent with those obtained from molecular data (Ni et al., 2021 and see Bergström et al., 2021 for molecular-based estimations). This discrepancy was further discussed in Ni et al. (2021:52).

105Ni et al. (2021) used their results to conduct a paleobiogeographical study of hominin dispersals during the Pleistocene. Strait and Wood (1999), Strait (2013) and Sekhavati and Strait (2024) used the results of their previous phylogenetic analyses to find the geographical regions where common ancestors lived.

106Püschel et al. (2021) used four topologies of hominin phylogeny to reconstruct the evolution of body mass and encephalization quotient at tree nodes. Braga et al. (2017) reconstructed the evolution of two inner ear traits. Organ et al. (2011) attempted to identify dental anatomical changes explained by phylogeny in representatives of the genus Homo, with the aim of identifying the antiquity of food processing (and cooking in particular).

107Gómez-Robles et al. (2013) tried to reconstruct the dental morphology of the common ancestor to H. sapiens and H. neanderthalensis and then compared their results with the fossil record. Their results suggested that this common ancestor had not yet been identified in the fossil record. Mounier and Mirazón Lahr (2016) used the same approach with crania. Their results showed that some Middle Pleistocene fossils often referred to as H. heidelbergensis show resemblances with virtual common ancestors (VCA). Renewed analyses focusing on the common ancestor to all H. sapiens found close resemblances between the VCA and South-African and, to a lesser extent, east African fossils (Mounier and Mirazón Lahr, 2019).

108Since the publications of the analyses mentioned in this section, new fossil discoveries and other elements have brought new arguments into these debates, which are therefore still open. However, this does not inhibit the role that phylogenetic analyses play in our understanding of human evolution, as shown throughout this section.

Conclusion

109Phylogenetic analyses of recent Homo species have thus helped to fuel debates on and to clarify, in several cases, the taxonomy and relationships of these taxa. More importantly for many paleoanthropologists, they have brought a better understanding of various processes and mechanisms underlying human evolution, such as mechanisms specific to island taxa, migrations in the distant past, anagenetic evolution and cladogenetic events.

110Nevertheless, some taxa are still understudied by means of phylogenetic analyses. For example, too few phylogenetic studies have been made on specimens recently grouped under the species names H. longi and/or H. juluensis (and long referred to as "archaic H. sapiens"). Insofar as the phylogenetic position of these taxa and that of the Denisovans seem contradictory, new analyses are needed.

111Further research on these and other taxa should benefit from the numerous methodological developments described and discussed in this review. Postcranial and endostructural characters, which are mostly lacking in hominin phylogenetics, should be systematically included in such analyses. Considering the wide and increasingly well-documented intra-taxonomic variability, detailed consideration of this in the analyses is also warranted. Finally, the use of geometric morphometrics in phylogenetics to finely document the morphological variations of anatomical elements other than the cranium should shed further light on relationships among fossil hominins.

Acknowledgements: This study was supported by a doctoral thesis grant provided to Pierre Gousset by the Doctoral School 227 DIVONA of the Muséum national d’Histoire naturelle (MNHN). We are grateful to the following colleagues in the MNHN for insightful discussions: Véronique Barriel, Guillaume Billet, Miguel Caparros, Hugo Hautavoine, Aurélien Mounier, Sandrine Prat, Valentin Rineau, Margaux Simon-Maciejewski, Amélie Vialet and Valéry Zeitoun. Finally, we would like to warmly thank two anonymous reviewers for their very useful comments that helped to improve this manuscript and the BMSAP translator for her language revision.

Haut de page

Bibliographie

Ackermann RR, Arnold ML, Baiz MD et al (2019) Hybridization in human evolution: Insights from other organisms. Evolutionary Anthropology: Issues, News, and Reviews 28(4):189-209

Adams DC, Cardini A, Monteiro LR et al (2011) Morphometrics and phylogenetics: principal components of shape from cranial modules are neither appropriate nor effective cladistic characters. Journal of Human Evolution 60(2):240-243

Aranburu A, Arsuaga JL, Sala N (2017) The stratigraphy of the Sima de los Huesos (Atapuerca, Spain) and implications for the origin of the fossil hominin accumulation. Quaternary International 433:5-21

Argue D, Morwood MJ, Sutikna T et al (2009) Homo floresiensis: a cladistic analysis. Journal of Human Evolution 57(5):623-639

Argue D (2015) Variation in the Early and Middle Pleistocene: The phylogenetic relationships of Ceprano, Bodo, Daka, Kabwe and Buia. In: Behie AM, Oxenham M (eds) Taxonomic Tapestries: The Threads of Evolutionary, Behavioural and Conservation Research. ANU Press, Acton, Australia, pp 215-248

Argue D, Groves CP, Lee MSY et al (2017) The affinities of Homo floresiensis based on phylogenetic analyses of cranial, dental, and postcranial characters. Journal of Human Evolution 107:107-133

Asfaw B, Gilbert WH, Beyene Y et al (2002) Remains of Homo erectus from Bouri, Middle Awash, Ethiopia. Nature 416(6878):317-320

Bae CJ (2024) The "muddle in the middle" (~400 ka-~100 ka). In: Bae CJ (ed) The Paleoanthropology of Eastern Asia. University of Hawaii Press, Honolulu, HI, USA, pp 95-131

Bae CJ, Liu W, Wu X et al (2023) "Dragon man" prompts rethinking of Middle Pleistocene hominin systematics in Asia. The Innovation 4(6):100527

Bae CJ, Wu X (2024) Making sense of eastern Asian Late Quaternary hominin variability. Nature Communications 15(1):9479

Bailey SE, Wood BA (2007) Trends in postcanine occlusal morphology within the hominin clade: The case of Paranthropus. In: Bailey SE, Hublin J-J (eds) Dental Perspectives on Human Evolution: State of the Art Research in Dental Paleoanthropology. Springer Netherlands, Dordrecht, Netherlands, pp 33-52

Bailey SE (2002) Neandertal dental morphology: implications for modern human origins. PhD dissertation, Arizona State University, Tempe, AZ, USA, 238 p

Balzeau A, Albessard-Ball L, Kubicka AM et al (2022) Frontal sinuses and human evolution. Science Advances 8(42):eabp9767

Bardin J, Rouget I, Yacobucci MM et al (2014) Increasing the number of discrete character states for continuous characters generates well-resolved trees that do not reflect phylogeny. Integrative Zoology 9(4):531-541

Berger LR, de Ruiter DJ, Churchill SE et al (2010) Australopithecus sediba: A New Species of Homo-Like Australopith from South Africa. Science 328(5975):195-204

Berger LR, Hawks J, de Ruiter DJ et al (2015) Homo naledi, a new species of the genus Homo from the Dinaledi Chamber, South Africa. eLife 4:e09560

Bergström A, Stringer CB, Hajdinjak M et al (2021) Origins of modern human ancestry. Nature 590(7845):229-237

Bermúdez de Castro JM, Arsuaga JL, Carbonell E et al (1997) A Hominid from the Lower Pleistocene of Atapuerca, Spain: Possible Ancestor to Neandertals and Modern Humans. Science 276(5317):1392-1395

Bermúdez de Castro JM, Martinón-Torres M, Arsuaga JL et al (2017) Twentieth anniversary of Homo antecessor (1997-2017): a review. Evolutionary Anthropology: Issues, News, and Reviews 26(4):157-171

Billet G, Bardin J (2019) Serial Homology and Correlated Characters in Morphological Phylogenetics: Modeling the Evolution of Dental Crests in Placentals. Systematic Biology 68(2):267-280

Bocherens H, Jaouen K, Rivals F et al (2022) Chapter 16 - Methodological advances in Neanderthal identification, phylogeny, chronology, mobility, climate, and diet. In: Romagnoli F, Rivals F, Benazzi S (eds) Updating Neanderthals. Academic Press, London, San Diego, Cambridge, Kidlington, pp 303-320

Bouée S, Détroit F (2010) Morphométrie géométrique et classification phylogénétique, application à la question de l’origine d’Homo sapiens en Asie. Bulletins et mémoires de la Société d’anthropologie de Paris 22(3):172-184

Braga J, Bouvier P, Romeyer Dherbey J et al (2017) Echoes from the past: New insights into the early hominin cochlea from a phylo-morphometric approach. Comptes Rendus Palevol 16(5):508-520

Brooks DR, Bilewitch J, Condy C et al (2007) Quantitative Phylogenetic Analysis in the 21st Century. Revista Mexicana de Biodiversidad 78(2):225-252

Brown P, Sutikna T, Morwood MJ et al (2004) A new small-bodied hominin from the Late Pleistocene of Flores, Indonesia. Nature 431(7012):1055-1061

Bull JJ, Cunningham CW, Molineux IJ et al (1993) Experimental molecular evolution of bacteriophage T7. Evolution 47(4):993-1007

Cameron DW, Patnaik R, Sahni A (2004) The phylogenetic significance of the Middle Pleistocene Narmada hominin cranium from central India. International Journal of Osteoarchaeology 14(6):419-447

Cameron DW (2003) Early hominin speciation at the Plio/Pleistocene transition. HOMO 54(1):1-28

Cann RL, Stoneking M, Wilson AC (1987) Mitochondrial DNA and human evolution. Nature 325:31-36

Caparros M (1997) Homo sapiens archaïques : un seul ou plusieurs taxons (espèces) ? : analyse cladistique et analyse morphométrique. PhD dissertation, Muséum national d’Histoire naturelle, Paris, France, 384 p

Caparros M, Prat S (2021) A Phylogenetic Networks perspective on reticulate human evolution. iScience 24(4):102359

Carter K, Worthington S, Smith TM (2014) News and views: non-metric dental traits and hominin phylogeny. Journal of Human Evolution 69:123-128

Chamberlain AT, Wood BA (1985) A reappraisal of variation in hominid mandibular corpus dimensions. American Journal of Physical Anthropology 66(4):399-405

Chamberlain AT, Wood BA (1987) Early hominid phylogeny. Journal of Human Evolution 16(1):119-133

Chen F, Welker F, Shen C-C et al (2019) A late Middle Pleistocene Denisovan mandible from the Tibetan Plateau. Nature 569 (7756):409-412

Churchill SE, Vansickle C (2017) Pelvic Morphology in Homo erectus and Early Homo. The Anatomical Record 300(5):964-977

Collard M, Wood BA (2000) How reliable are human phylogenetic hypotheses? Proceedings of the National Academy of Sciences 97(9):5003-5006

Collard M, Wood BA (2007) Hominin homoiology: An assessment of the impact of phenotypic plasticity on phylogenetic analyses of humans and their fossil relatives. Journal of Human Evolution 52(5):573-584

Conaway MA, von Cramon-Taubadel N (2022) Morphological integration of the hominoid postcranium. Journal of Human Evolution 171:103239

Congreve CR, Lamsdell JC (2016) Implied weighting and its utility in palaeontological datasets: a study using modelled phylogenetic matrices. Palaeontology 59(3):447-462

Coon CS (1962) The origin of races. Knopf, New York, NY, USA, 844 p

Curnoe D (2001) Early Homo from southern Africa: a cladistic perspective: research letter. South African Journal of Science 97(5):186-190

Curnoe D (2003) Problems with the use of cladistic analysis in palaeoanthropology. HOMO 53(3):225-234

Darlu P, Tassy P, d’Haese C et al (2019) La reconstruction phylogénétique. Editions Matériologiques, Paris, France, 430 p

de Koning AP, Gu W, Castoe TA et al (2012) Phylogenetics, likelihood, evolution and complexity. Bioinformatics 28(22):2989-2990

de Pinna MCC (1991) Concepts and Tests of Homology in the Cladistic Paradigm. Cladistics 7(4):367-394

Demarchi B, Hall S, Roncal-Herrero T et al (2016) Protein sequences bound to mineral surfaces persist into deep time. eLife 5:e17092

Dembo M (2016) Exploring morphological phylogenetics of fossil hominins. PhD dissertation, Simon Fraser University, Vancouver, Canada, 244 p

Dembo M, Matzke NJ, Mooers AØ et al (2015) Bayesian analysis of a morphological supermatrix sheds light on controversial fossil hominin relationships. Proceedings of the Royal Society B: Biological Sciences 282(1812):20150943

Dembo M, Radovčić D, Garvin HM et al (2016) The evolutionary relationships and age of Homo naledi: An assessment using dated Bayesian phylogenetic methods. Journal of Human Evolution 97:17-26

Dennell RW, Martinón-Torres M, Bermúdez de Castro JM (2011) Hominin variability, climatic instability and population demography in Middle Pleistocene Europe. Quaternary Science Reviews 30(11):1511-1524

DeSilva JM, McNutt E, Benoit J et al (2019) One small step: A review of Plio-Pleistocene hominin foot evolution. American Journal of Physical Anthropology 168(S67):63-140

DeSilva JM (2009) Functional morphology of the ankle and the likelihood of climbing in early hominins. Proceedings of the National Academy of Sciences 106(16):6567-6572

Détroit F, Mijares AS, Corny J et al (2019) A new species of Homo from the Late Pleistocene of the Philippines. Nature 568(7751):181-186

Dirks PH, Roberts EM, Hilbert-Wolf H et al (2017) The age of Homo naledi and associated sediments in the Rising Star Cave, South Africa. eLife 6:e24231

Dubois E (1894) Pithecanthropus erectus: eine menschenaehnliche Uebergangsform aus Java. Landesdruckerei, Batavia, Indonesia, 58 p

Edwards AW, Cavalli-Sforza LL (1963) Reconstruction of evolution. Annals of Human Genetics 27(4):105-106

Eldredge N, Tattersal I (1975) Evolutionary models, phylogenetic reconstruction, and another look at hominid phylogeny. Contributions to Primatology 5:218-242

Fabre V, Condemi S, Degioanni A (2009) Genetic Evidence of Geographical Groups among Neanderthals. PLOS ONE 4(4):e5151

Farris JS (1969) A Successive Approximations Approach to Character Weighting. Systematic Biology 18(4):374-385

Felsenstein J (1981) Evolutionary trees from DNA sequences: A maximum likelihood approach. Journal of Molecular Evolution 17(6):368-376

Felsenstein J (1985) Phylogenies and the Comparative Method. The American Naturalist 125(1):1-15

Folinsbee KE, Evans DC, Fröbisch J et al (2015) Quantitative Approaches to Phylogenetics. In: Henke W, Tattersall I (eds) Handbook of Paleoanthropology. Springer Reference, Berlin, Germany, pp 167-215

Fong-Zazueta R, Krueger J, Alba DM et al (2025) Phylogenetic Signal in Primate Tooth Enamel Proteins and its Relevance for Paleoproteomics. Genome Biology and Evolution 17(2):evaf007

Fu Q, Bai F, Rao H et al. (2025) The proteome of the late Middle Pleistocene Harbin individual. Science 389(6761):704-707

Fu Q, Cao P, Dai Q et al. (2025) Denisovan mitochondrial DNA from dental calculus of the >146,000-year-old Harbin cranium. Cell 188(15):3919-3926.e9

Garamszegi LZ (ed) (2014) Modern Phylogenetic Comparative Methods and Their Application in Evolutionary Biology: Concepts and Practice. Springer Berlin Heidelberg, Berlin, Heidelberg, Germany

Gautney JR (2023) A new approach to exploratory data analysis in hominin phylogenetic reconstruction. Journal of Human Evolution 182:103412

Gilbert CC, Frost SR, Strait DS (2009) Allometry, sexual dimorphism, and phylogeny: A cladistic analysis of extant African papionins using craniodental data. Journal of Human Evolution 57(3):298-320

Gilbert WH (2008) Hominid systematics. In: Gilbert WH, Asfaw B (eds) Homo erectus: Pleistocene Evidence from the Middle Awash, Ethiopia. University of California Press, Berkeley, Los Angeles, London, pp 349-371

Goloboff PA (1993) Estimating Character Weights During Tree Search. Cladistics 9(1):83-91

Goloboff PA (1997) Self-Weighted Optimization: Tree Searches and Character State Reconstructions under Implied Transformation Costs. Cladistics 13(3):225-245

Goloboff PA (2003) Parsimony, likelihood, and simplicity. Cladistics 19(2):91-103

Goloboff PA, Catalano SA (2016) TNT version 1.5, including a full implementation of phylogenetic morphometrics. Cladistics 32(3):221-238

Goloboff PA, Mattoni CI, Quinteros AS (2006) Continuous characters analyzed as such. Cladistics 22(6):589-601

Gómez-Olivencia A, Been E (2019) The Spine of Late Homo. In: Been E, Gómez-Olivencia A, Ann Kramer P (eds) Spinal Evolution: Morphology, Function, and Pathology of the Spine in Hominoid Evolution. Springer International Publishing, Cham, Switzerland, pp 185-211

Gómez-Robles A, Bermúdez de Castro JM, Arsuaga JL et al (2013) No known hominin species matches the expected dental morphology of the last common ancestor of Neanderthals and modern humans. Proceedings of the National Academy of Sciences 110(45):18196-18201

González-José R, Escapa I, Neves WA et al (2008) Cladistic analysis of continuous modularized traits provides phylogenetic signals in Homo evolution. Nature 453(7196):775-778

Gousset P, Bardin J, Rouget I et al (2026) Homo luzonensis and the role of homoplasy in the evolution of hominin insular species. Cladistics (early view)

Green DR, Uno KT, Miller ER et al (2025) Eighteen million years of diverse enamel proteomes from the East African Rift. Nature 643(8072):712-718

Green RE, Krause J, Briggs AW et al (2010) A Draft Sequence of the Neandertal Genome. Science 328(5979):710-722

Grine FE, Mongle CS, Fleagle JG et al (2022) The taxonomic attribution of African hominin postcrania from the Miocene through the Pleistocene: Associations and assumptions. Journal of Human Evolution 173:103255

Groves CP, Mirazon Lahr M (1994) A bush not a ladder: Speciation and replacement in human evolution. Perspectives in Human Biology 4:1-11

Groves CP (1986) Systematics of the great apes. In: Swindler D, Erwin J (eds) Comparative Primate Biology, Volume 1: Systematics, Evolution and Adaptations. Alan R Liss, New York, NY, USA, pp 187-217

Harvati K, Ackermann RR (2022) Merging morphological and genetic evidence to assess hybridization in Western Eurasian late Pleistocene hominins. Nature Ecology & Evolution 6(10): 1573-1585

Harvati K, Weaver TD (2006) Human cranial anatomy and the differential preservation of population history and climate signatures. The Anatomical Record Part A: Discoveries in Molecular, Cellular, and Evolutionary Biology 288A(12):1225-1233

Harvey PH, Pagel MD (1991) The Comparative Method in Evolutionary Biology. Oxford University Press, Oxford, England

Hawks J (2004) How much can cladistics tell us about early hominid relationships? American Journal of Physical Anthropology 125(3):207-219

Hennig W (1965) Phylogenetic Systematics. Annual Review of Entomology 10(1):97-116

Hooijer DA (1954) Pleistocene vertebrates from Celebes: XI. Molars and a tusked mandible of Archidiskodon celebensis Hooijer. Zoologische Mededelingen 33(15):103-120

Howell FC (1999) Paleo-Demes, Species Clades, and Extinctions in the Pleistocene Hominin Record. Journal of Anthropological Research 55(2):191-243

Hublin J-J (2009) The origin of Neandertals. Proceedings of the National Academy of Sciences 106(38):16022-16027

Irish JD, Bailey SE, Guatelli-Steinberg D et al (2018) Ancient teeth, phenetic affinities, and African hominins: Another look at where Homo naledi fits in. Journal of Human Evolution 122:108-123

Irish JD, Grabowski M (2021) Relative tooth size, Bayesian inference, and Homo naledi. American Journal of Physical Anthropology 176(2):262-282

Irish JD, Guatelli-Steinberg D, Legge SS et al (2013) Dental Morphology and the Phylogenetic "Place" of Australopithecus sediba. Science 340(6129):1233062

Ji Q, Wu W, Ji Y et al (2021) Late Middle Pleistocene Harbin cranium represents a new Homo species. The Innovation 2(3)

Jung H, von Cramon-Taubadel N (2022) Morphological modularity in the anthropoid axial skeleton. Journal of Human Evolution 172:103256

Keating JN, Garwood RJ, Sansom RS (2023) Phylogenetic congruence, conflict and consilience between molecular and morphological data. BMC Ecology and Evolution 23(1):30

Kimbel WH (2013) Hesitation on hominin history. Nature 497(7451):573-574

Kimbel WH, Lockwood CA, Ward CV et al (2006) Was Australopithecus anamensis ancestral to A. afarensis? A case of anagenesis in the hominin fossil record. Journal of Human Evolution 51(2):134-152

Kimbel WH, Rak Y, Johanson DC (2004) The Skull of Australopithecus afarensis. Oxford University Press, New York, NY, USA, 273 p

Kitching IJ, Forey P, Humphries C et al (1998) Cladistics: The Theory and Practice of Parsimony Analysis. Oxford University Press, New York, NY, USA, 252 p

Kjær KH, Winther Pedersen M, De Sanctis B et al (2022) A 2-million-year-old ecosystem in Greenland uncovered by environmental DNA. Nature 612(7939):283-291

Kluge AG (1989) A Concern for Evidence and a Phylogenetic Hypothesis of Relationships among Epicrates (Boidae, Serpentes). Systematic Biology 38(1):7-25

Kluge AG (1997) Testability and the Refutation and Corroboration of Cladistic Hypotheses. Cladistics 13(1-2):81-96

Komza K, Viola B, Netten T et al (2022) Morphological integration in the hominid midfoot. Journal of Human Evolution 170:103231

Krings M, Stone A, Schmitz RW et al (1997) Neandertal DNA Sequences and the Origin of Modern Humans. Cell 90(1):19-30

Larson SG (2007) Evolutionary transformation of the hominin shoulder. Evolutionary Anthropology: Issues, News, and Reviews 16(5):172-187

Lee MSY, Palci A (2015) Morphological Phylogenetics in the Genomic Age. Current Biology 25(19):R922-R929

Lewis PO (2001) A Likelihood Approach to Estimating Phylogeny from Discrete Morphological Character Data. Systematic Biology 50(6):913-925

Lieberman DE, McBratney BM, Krovitz G (2002) The evolution and development of cranial form in Homo sapiens. Proceedings of the National Academy of Sciences 99(3):1134-1139

Lieberman DE, Wood BA, Pilbeam DR (1996) Homoplasy and early Homo: an analysis of the evolutionary relationships of H. habilis sensu stricto and H. rudolfensis. Journal of Human Evolution 30(2):97-120

Liu W, Athreya S, Xing S et al (2022) Hominin evolution and diversity: a comparison of earlier-Middle and later-Middle Pleistocene hominin fossil variation in China. Philosophical Transactions of the Royal Society B: Biological Sciences 377(1847):20210040

Lovejoy CO, Suwa G, Spurlock L et al (2009) The Pelvis and Femur of Ardipithecus ramidus: The Emergence of Upright Walking. Science 326(5949):71-71e6

Macchiarelli R, Bayle P, Bondioli L et al (2013) From outer to inner structural morphology in dental anthropology. Integration of the third dimension in the visualization and quantitative analysis of fossil remains. In: Scott GR, Irish JD (eds) Anthropological Perspectives on Tooth Morphology - Genetics, Evolution, Variation. Cambridge University Press, New York, NY, USA, pp 250-277

Madupe PP, Koenig C, Patramanis I et al (2025) Enamel proteins reveal biological sex and genetic variability in southern African Paranthropus. Science 388(6750):969-973

Margush T, McMorris FR (1981) Consensus n-trees. Bulletin of Mathematical Biology 43(2):239-244

Martin JM, Leece AB, Neubauer S et al (2021) Drimolen cranium DNH 155 documents microevolution in an early hominin species. Nature Ecology & Evolution 5(1):38-45

Martin RMG, Hublin J-J, Gunz P et al (2017) The morphology of the enamel-dentine junction in Neanderthal molars: Gross morphology, non-metric traits, and temporal trends. Journal of Human Evolution 103:20-44

Martinón-Torres M, Bermúdez de Castro JM, Gómez-Robles A et al (2007) Dental evidence on the hominin dispersals during the Pleistocene. Proceedings of the National Academy of Sciences 104(33):13279-13282

Martinón-Torres M, Bermúdez de Castro JM, Gómez-Robles A et al (2012) Morphological description and comparison of the dental remains from Atapuerca-Sima de los Huesos site (Spain). Journal of Human Evolution 62(1):7-58

Martinón-Torres M, Bermúdez de Castro JM, Martínez de Pinillos M et al (2019) New permanent teeth from Gran Dolina-TD6 (Sierra de Atapuerca). The bearing of Homo antecessor on the evolutionary scenario of Early and Middle Pleistocene Europe. Journal of Human Evolution 127:93-117

Martı́nez I, Arsuaga JL (1997) The temporal bones from Sima de los Huesos Middle Pleistocene site (Sierra de Atapuerca, Spain). A phylogenetic approach. Journal of Human Evolution 33(2): 283-318

Mayr E (1950) Taxonomic Categories in Fossil Hominids. Cold Spring Harbor Symposia on Quantitative Biology 15:109-118

Meneganzin A, Bernardi M (2023) Were Neanderthals and Homo sapiens ‘good species’? Quaternary Science Reviews 303:107975

Meyer M, Arsuaga JL, de Filippo C et al (2016) Nuclear DNA sequences from the Middle Pleistocene Sima de los Huesos hominins. Nature 531(7595):504-507

Meyer MR, Williams SA (2019) The Spine of Early Pleistocene Homo. In: Been E, Gómez-Olivencia A, Ann Kramer P (eds) Spinal Evolution: Morphology, Function, and Pathology of the Spine in Hominoid Evolution. Springer International Publishing, Cham, Switzerland, pp 153-183

Mongle CS, Pugh KD, Strait DS et al (2022) Modelling hominin evolution requires accurate hominin data. Nature Ecology & Evolution 6(8):1090-1091

Mongle CS, Strait DS, Grine FE (2019) Expanded character sampling underscores phylogenetic stability of Ardipithecus ramidus as a basal hominin. Journal of Human Evolution 131:28-39

Mongle CS, Strait DS, Grine FE (2023) An updated analysis of hominin phylogeny with an emphasis on re-evaluating the phylogenetic relationships of Australopithecus sediba. Journal of Human Evolution 175:103311

Mounier A, Caparros M (2015) Le statut phylogénétique d’Homo heidelbergensis - étude cladistique des homininés du Pléistocène moyen. Bulletins et Mémoires de la Société d’Anthropologie de Paris 27(3):110-134

Mounier A, Balzeau A, Caparros M et al (2016) Brain, calvarium, cladistics: A new approach to an old question, who are modern humans and Neandertals? Journal of Human Evolution 92:22-36

Mounier A, Marchal F, Condemi S (2009) Is Homo heidelbergensis a distinct species? New insight on the Mauer mandible. Journal of Human Evolution 56(3):219-246

Neubauer S (2015) Human Brain Evolution: Ontogeny and Phylogeny. In: Bruner E (ed) Human Paleoneurology. Springer International Publishing, Cham, Switzerland, pp 95-120

Ni X, Ji Q, Wu W et al (2021) Massive cranium from Harbin in northeastern China establishes a new Middle Pleistocene human lineage. The Innovation 2(3):100130

O’Meara BC (2012) Evolutionary Inferences from Phylogenies: A Review of Methods. Annual Review of Ecology, Evolution, and Systematics 43:267-285

Organ C, Nunn CL, Machanda Z et al (2011) Phylogenetic rate shifts in feeding time during the evolution of Homo. Proceedings of the National Academy of Sciences 108(35):14555-14559

Palci A, Lee MSY (2019) Geometric morphometrics, homology and cladistics: review and recommendations. Cladistics 35(2):230-242

Paradis E, Schliep K (2019) ape 5.0: an environment for modern phylogenetics and evolutionary analyses in R. Bioinformatics 35(3):526-528

Parins-Fukuchi C (2021) Morphological and phylogeographic evidence for budding speciation: an example in hominins. Biology Letters 17(1):20200754

Parins-Fukuchi C, Greiner E, MacLatchy LM et al (2019) Phylogeny, ancestors, and anagenesis in the hominin fossil record. Paleobiology 45(2):378-393

Paterson RS, Mackie M, Capobianco A et al (2025) Phylogenetically informative proteins from an Early Miocene rhinocerotid. Nature 643(8072):719-724

Peyrégne S, Slon V, Kelso J (2024) More than a decade of genetic research on the Denisovans. Nature Reviews Genetics 25(2):83-103

Philippe H, Brinkmann H, Lavrov DV et al (2011) Resolving Difficult Phylogenetic Questions: Why More Sequences Are Not Enough. PLOS Biology 9(3):e1000602

Picin A, Hajdinjak M, Nowaczewska W et al (2020) New perspectives on Neanderthal dispersal and turnover from Stajnia Cave (Poland). Scientific Reports 10(1):14778

Pilbeam D (1996) Genetic and Morphological Records of the Hominoidea and Hominid Origins: A Synthesis. Molecular Phylogenetics and Evolution 5(1):155-168

Prat S (2000) Origine et taxinomie des premiers représentants du genre Homo. PhD dissertation, Bordeaux 1, Talence, 687 p

Prat S (2004) Les premiers représentants du genre Homo, en quête d’une identité. Apports de l’étude morphologique et de l’analyse cladistique. Bulletins et Mémoires de la Société d’Anthropologie de Paris 16(1-2):17-35

Profico A, Buzi C, Di Vincenzo F et al (2023) Virtual excavation and analysis of the early Neanderthal cranium from Altamura (Italy). Communications Biology 6(1):1-8

Prüfer K, Racimo F, Patterson N et al (2014) The complete genome sequence of a Neanderthal from the Altai Mountains. Nature 505(7481):43-49

Pugh KD (2022) Phylogenetic analysis of Middle-Late Miocene apes. Journal of Human Evolution 165:103140

Püschel HP, Bertrand OC, O’Reilly JE et al (2021) Divergence-time estimates for hominins provide insight into encephalization and body mass trends in human evolution. Nature Ecology & Evolution 5(6):808-819

Püschel HP, Bertrand OC, Reilly JE et al (2022) Reply to: Modelling hominin evolution requires accurate hominin data. Nature Ecology & Evolution 6(8):1092-1094

Pyron RA (2015) Post-molecular systematics and the future of phylogenetics. Trends in Ecology & Evolution 30(7):384-389

R Core Team (2023) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria

Ragan MA (1992) Phylogenetic inference based on matrix representation of trees. Molecular Phylogenetics and Evolution 1(1):53-58

Rathmann H, Reyes-Centeno H (2020) Testing the utility of dental morphological trait combinations for inferring human neutral genetic variation. Proceedings of the National Academy of Sciences 117(20):10769-10777

Reich D, Green RE, Kircher M et al (2010) Genetic history of an archaic hominin group from Denisova Cave in Siberia. Nature 468(7327):1053-1060

Revell LJ (2024) phytools 2.0: an updated R ecosystem for phylogenetic comparative methods (and other things). PeerJ 12:e16505

Rodríguez-Ezpeleta N, Brinkmann H, Roure B et al (2007) Detecting and Overcoming Systematic Errors in Genome-Scale Phylogenies. Systematic Biology 56(3):389-399

Ronquist F, Klopfstein S, Vilhelmsen L et al (2012) A Total-Evidence Approach to Dating with Fossils, Applied to the Early Radiation of the Hymenoptera. Systematic Biology 61(6):973-999

Ruff CB (2018) Biomechanical Analyses of Archaeological Human Skeletons. In: Katzenberg MA, Grauer AL (eds) Biological Anthropology of the Human Skeleton. John Wiley & Sons, Ltd, Hoboken, NJ, USA, pp 189-224

Scerri EML, Chikhi L, Thomas MG (2019) Beyond multiregional and simple out-of-Africa models of human evolution. Nature Ecology & Evolution 3(10):1370-1372

Schliep KP (2011) phangorn: phylogenetic analysis in R. Bioinformatics 27(4):592-593

Schwartz JH (2015) Defining Hominidae. In: Handbook of Paleoanthropology. Springer, Berlin, Heidelberg, Germany, pp 1791-1835

Scotland RW, Olmstead RG, Bennett JR (2003) Phylogeny Reconstruction: The Role of Morphology. Systematic Biology 52(4):539-548

Scott GR, Irish JD (2013) Anthropological Perspectives on Tooth Morphology: Genetics, Evolution, Variation. Cambridge University Press, Cambridge, England, 576 p

Sekhavati Y, Prang TC, Strait DS (2025) A phylogenetic perspective on the evolution of early hominin foot morphology. Journal of Human Evolution 203:103682

Sekhavati Y, Strait DS (2024) Estimating ancestral ranges and biogeographical processes in early hominins. Journal of Human Evolution 191:103547

Sereno PC (2007) Logical basis for morphological characters in phylogenetics. Cladistics 23(6):565-587

Signorell A (2024) DescTools: Tools for descriptive statistics. R Package Version 0.99.54

Simões TR, Caldwell MW, Palci A et al (2017) Giant taxon-character matrices: quality of character constructions remains critical regardless of size. Cladistics 33(2):198-219

Simon-Maciejewski M, Manzi G, Zeitoun V et al (2024) Cladistics with geometric morphometric data: The variability of the calvarium in the genus Homo. Bulletins et Mémoires de la Société d’Anthropologie de Paris 36(1)

Skelton RR, McHenry HM (1992) Evolutionary relationships among early hominids. Journal of Human Evolution 23(4):309-349

Skelton RR, McHenry HM, Drawhorn GM et al (1986) Phylogenetic Analysis of Early Hominids [and Comments and Reply]. Current Anthropology

Slon V, Mafessoni F, Vernot B et al (2018) The genome of the offspring of a Neanderthal mother and a Denisovan father. Nature 561(7721):113-116

Smith HF, Grine FE (2008) Cladistic analysis of early Homo crania from Swartkrans and Sterkfontein, South Africa. Journal of Human Evolution 54(5):684-704

Sneath PHA (1995) Thirty Years of Numerical Taxonomy. Systematic Biology 44(3):281-298

Sondaar PY (1994) Paleoecology and evolutionary patterns in horses and Island mammals. Historical Biology 8(1-4):1-13

Strait DS (1998) Evolutionary integration in the hominid cranial base. PhD dissertation, State University of New York, Stony Brook, NY, USA

Strait DS (2001) Integration, phylogeny, and the hominid cranial base. American Journal of Physical Anthropology 114(4):273-297

Strait DS (2013) The Biogeographic Implications of Early Hominin Phylogeny. In: Reed KE, Fleagle JG, Leakey RE (eds) The Paleobiology of Australopithecus. Springer Netherlands, Dordrecht, Netherlands, pp 183-191

Strait DS, Grine FE (2004) Inferring hominoid and early hominid phylogeny using craniodental characters: the role of fossil taxa. Journal of Human Evolution 47(6):399-452

Strait DS, Grine FE, Fleagle JG (2015) Analyzing Hominin Phylogeny: Cladistic Approach. In: Henke W, Tattersall I (eds) Handbook of Paleoanthropology. Springer Reference, Berlin, Germany, pp 1989-2014

Strait DS, Grine FE, Moniz MA (1997) A reappraisal of early hominid phylogeny. Journal of Human Evolution 32(1):17-82

Strait DS, Wood BA (1999) Early hominid biogeography. Proceedings of the National Academy of Sciences 96(16):9196-9200

Stringer CB (2022) The development of ideas about a recent African origin for Homo sapiens. Journal of Anthropological Sciences - Rivista Di Antropologia 100:5-18

Stringer CB (1987) A numerical cladistic analysis for the genus Homo. Journal of Human Evolution 16(1):135-146

Stringer CB (1993) Understanding the fossil human record: past, present and future. Rivista di Antropologia 71:91-100

Stringer CB, Humphrey LT, Compton T (1997) Cladistic analysis of dental traits in recent humans using a fossil outgroup. Journal of Human Evolution 32(4):389-402

Tocheri MW, Orr CM, Jacofsky MC et al (2008) The evolutionary history of the hominin hand since the last common ancestor of Pan and Homo. Journal of Anatomy 212(4):544-562

Trinkaus E (1990) Cladistics and the hominid fossil record. American Journal of Physical Anthropology 83(1):1-11

Trueman JWH (2010) A new cladistic analysis of Homo floresiensis. Journal of Human Evolution 59(2):223-226

Tsutaya T, Sawafuji R, Taurozzi AJ et al (2025) A male Denisovan mandible from Pleistocene Taiwan. Science 388(6743):176-180

Van den Hoek Ostende LW (2018) Cladistics and insular evolution, an unfortunate marriage? Another tangle in the Deinogalerix analysis of Borrani et al (2017). Cladistics 34(6):708-713

Van der Geer AAE (2005) Island ruminants and parallel evolution of functional structures. In: Crégut-Bonnoure E (ed) Les Ongulés holarctiques du Pliocène et du Pléistocène. Association Française pour l’étude du Quaternaire, Paris, France, pp 231-240

Vialet A, Prat S, Wils P et al (2018) The Kocabaş hominin (Denizli Basin, Turkey) at the crossroads of Eurasia: New insights from morphometric and cladistic analyses. Comptes Rendus Palevol 17(1):17-32

von Cramon-Taubadel N, Smith HF (2012) The relative congruence of cranial and genetic estimates of hominoid taxon relationships: Implications for the reconstruction of hominin phylogeny. Journal of Human Evolution 62(5):640-653

Wainscoat JS, Hill AVS, Boyce AL et al (1986) Evolutionary relationships of human populations from an analysis of nuclear DNA polymorphisms. Nature 319(6053):491-493

Wake DB (1991) Homoplasy: The Result of Natural Selection, or Evidence of Design Limitations? The American Naturalist 138(3):543-567

Welker F, Ásmundsdóttir RD, Mylopotamitaki D et al. (2025) Palaeoproteomic Contributions, and Current Limitations, to Understanding Middle and Late Pleistocene Human Evolution: Special Issue: What’s in a Name? Late Middle and Early Late Pleistocene Hominin Systematics. PaleoAnthropology 2025(2):325-339

Welker F, Ramos-Madrigal J, Gutenbrunner P et al (2020) The dental proteome of Homo antecessor. Nature 580(7802):235-238

Widianto H, Zeitoun V (2003) Morphological description, biometry and phylogenetic position of the skull of Ngawi 1 (east Java, Indonesia). International Journal of Osteoarchaeology 13(6):339-351

Wiens JJ (1995) Polymorphic Characters in Phylogenetic Systematics. Systematic Biology 44(4):482-500

Wiens JJ (2001) Character Analysis in Morphological Phylogenetics: Problems and Solutions. Systematic Biology 50(5):689-699

Willerslev E, Hansen AJ, Poinar HN (2004) Isolation of nucleic acids and cultures from fossil ice and permafrost. Trends in Ecology & Evolution 19(3):141-147

Wolpoff MH, Hawks J, Caspari R (2000) Multiregional, not multiple origins. American Journal of Physical Anthropology 112(1):129-136

Wood BA, Collard M (1999) The Human Genus. Science 284(5411):65-71

Wu X, Bae CJ (2025) Xujiayao Homo: A New Form of Large Brained Hominin in Eastern Asia: Special Issue: What’s in a Name? Late Middle and Early Late Pleistocene Hominin Systematics. PaleoAnthropology 2025(2):356-369

Xing S, Martinón-Torres M, Bermúdez de Castro JM et al (2015) Hominin teeth from the early Late Pleistocene site of Xujiayao, Northern China. American Journal of Physical Anthropology 156(2):224-240

Xing S, Martinón-Torres M, Bermúdez de Castro JM (2018) The fossil teeth of the Peking Man. Scientific Reports 8(1):2066

Xing S, Martinón-Torres M, Bermúdez de Castro JM de et al (2014) Middle Pleistocene Hominin Teeth from Longtan Cave, Hexian, China. PLOS ONE 9(12):e114265

Xing S, Martinón-Torres M, Deng C et al (2021) Early Pleistocene hominin teeth from Meipu, southern China. Journal of Human Evolution 151:102924

Xing S, Sun C, Martinón-Torres M et al (2016) Hominin teeth from the Middle Pleistocene site of Yiyuan, Eastern China. Journal of Human Evolution 95:33-54

Young NM, MacLatchy L (2004) The phylogenetic position of Morotopithecus. Journal of Human Evolution 46(2):163-184

Zanolli C, Kaifu Y, Pan L et al (2022) Further analyses of the structural organization of Homo luzonensis teeth: Evolutionary implications. Journal of Human Evolution 163:103124

Zeitoun V (1996) Cladistique et paléoanthropologie : le cas de l’espèce Homo erectus (Dubois, 1894). PhD dissertation, Bordeaux 1, Talence, France, 344 p

Zeitoun V (2001) The taxinomical position of the skull of Zuttiyeh. Comptes Rendus de l’Académie Des Sciences - Series IIA - Earth and Planetary Science 332(8):521-525

Zeitoun V, Barriel V, Widianto H (2016) Phylogenetic analysis of the calvaria of Homo floresiensis. Comptes Rendus Palevol 15(5):555-568

Zeitoun V, Détroit F, Grimaud-Hervé D et al (2010) Solo man in question: Convergent views to split Indonesian Homo erectus in two categories. Quaternary International 223-224:281-292

Zou Z, Zhang J (2016) Morphological and molecular convergences in mammalian phylogenetics. Nature Communications 7(1):12758

Haut de page

Documents annexes

  • Appendix A (application/vnd.openxmlformats-officedocument.spreadsheetml.sheet – 17k)
  • Appendix B (application/vnd.openxmlformats-officedocument.spreadsheetml.sheet – 13k)
  • Appendix C (application/vnd.openxmlformats-officedocument.spreadsheetml.sheet – 11k)
  • Appendix D (application/vnd.openxmlformats-officedocument.spreadsheetml.sheet – 11k)
  • Appendix E (application/vnd.openxmlformats-officedocument.spreadsheetml.sheet – 14k)
Haut de page

Table des illustrations

Titre Figure 1
Légende Historical timeline of major methodological steps in research on hominin phylogenetic analyses (references: © Gerd Hennig for Willi Hennig picture; Zeitoun 2000 for the matrix; unsplash.com for the lower limbs and DNA and Simon-Maciejewski et al. 2024 for the cranium with landmarks) | Frise chronologique retraçant quelques-unes des grandes étapes des recherches portant sur l’analyse phylogénétique des hominines (références : © Gerd Hennig pour la photo de Willi Hennig ; Zeitoun 2000 pour la matrice ; unsplash.com pour les membres inférieurs et pour l’ADN et Simon-Maciejewski et al. 2024 pour le crâne avec les landmarks)
URL http://journals.openedition.org/bmsap/docannexe/image/17895/img-1.png
Fichier image/png, 351k
Titre Figure 2
Légende Pie chart representing the taxonomic levels of OTUs in phylogenetic analyses on hominins. On the right are analyses that focus on the phylogenetic relationships of recent Homo species (n=24); on the left are analyses that focus on earlier hominins (n=25). Recent Homo species are Homo species less than one million years old. Gilbert (2008) features in each of the three main categories as he performed three analyses, each one at a different taxonomic level. Pie charts generated from the data presented in appendix B |Diagramme circulaire présentant le rang taxonomique des UTO par étude phylogénétique sur les hominines. À droite, les analyses phylogénétiques se concentrant sur les relations de parenté entre espèces récentes du genre Homo (n=24) et à gauche les analyses se concentrant sur des hominines plus anciens (n=25). Les espèces récentes du genre Homo sont celles ayant vécu au cours du dernier million d’années. Gilbert (2008) a été enregistré dans chacune des trois principales catégories puisqu’il a réalisé trois analyses à chaque fois à un niveau taxonomique différent. Graphique généré à partir des données présentées dans l’appendice B
URL http://journals.openedition.org/bmsap/docannexe/image/17895/img-2.png
Fichier image/png, 212k
Titre Figure 3
Légende Diagram summing up the advantages (in green) and drawbacks (in red) of the different taxonomic levels that can be used for OTUs. The coloured points over each taxonomic level represent individuals. The colour represents the classification for each individual depending on the taxonomic level (i.e., one colour per OTU). The "n" corresponds to the number of OTUs. Polymorphism is coloured in grey since increased polymorphism can be considered as advantageous or disadvantageous depending on the authors and/or situations (see main text) |Schéma résumant les avantages (en vert) et inconvénients (en rouge) des différents niveaux taxonomiques pouvant être utilisés comme UTO. Les points colorés au-dessus de chaque niveau taxonomique représentent des individus. La couleur représente la façon dont les individus sont classés selon le niveau taxonomique de l’UTO (i.e., une couleur par UTO). Les "n" correspondent aux nombres d’UTO. Le polymorphisme est colorié en gris car l’augmentation du polymorphisme peut être considéré comme avantageux ou désavantageux selon les auteurs et/ou les situations (voir le texte principal)
URL http://journals.openedition.org/bmsap/docannexe/image/17895/img-3.png
Fichier image/png, 262k
Titre Figure 4
Légende Bar graph representing the various anatomical regions studied by phylogenetic analyses on hominins (n=47). The graph can be read as follows: 83% of the publications considered show more than one character describing the calvarium. Graph generated from the data presented in appendix C |Diagramme en bâton présentant les parties anatomiques étudiées par analyse phylogénétique portant sur les hominines (n=47). Le graphique peut être lu comme suit : 83 % des publications considérées présentent plus d’un caractère portant sur le calvarium. Graphique généré à partir des données présentées dans l’appendice C
URL http://journals.openedition.org/bmsap/docannexe/image/17895/img-4.png
Fichier image/png, 67k
Titre Figure 5
Légende Bar graph showing the association of anatomical regions studied by phylogenetic analysis on hominins (n=47). Graph generated from data presented in Appendix C. ">1 region" means that two or more of the categories defined in figure 4 are present in the analysis. Abbreviation: C=Calvarium; F=Face; M=Mandible; T=Teeth; P=Postcranium |Diagramme en bâton présentant les associations de parties anatomiques étudiées par analyse phylogénétique portant sur les hominines (n=47). Graphique généré à partir des données présentées dans l’appendice C. ">1 region" signifie que deux ou plus des catégories définies dans la figure 4 sont présentes dans l’analyse. Abréviations : C=Calvarium ; F=Face ; M=Mandibule ; T=Dents et P=Postcrâne
URL http://journals.openedition.org/bmsap/docannexe/image/17895/img-5.png
Fichier image/png, 46k
Titre Figure 6
Légende Diagram showing the various coding procedures used to integrate ASUDAS characters in a matrix. The upper part of the figure (over the upper horizontal black line) shows the primary data. This data may be transformed in two different ways: by using a mean value rounded to the nearest integer (method of Bailey 2002) or by counting the individuals over a breakpoint character state (method of Stringer et al. 1997). The black crosses represent individuals; the continuous and vertical black line represents the breakpoint; values in a green square are those implemented in the matrix |Schéma présentant les différents codages utilisés pour intégrer les caractères ASUDAS dans une matrice. La partie supérieure de la figure (au-dessus de la ligne noire horizontale la plus haute) représente la donnée primaire. Elle peut être transformée de deux façons : en réalisant une moyenne puis en arrondissant cette moyenne à l’unité près (méthode de Bailey 2002) ou en codant le pourcentage d’individus au-dessus d’un état de caractère servant de seuil (méthode de Stringer et al. 1997). Les croix noires représentent des individus, la ligne continue verticale représente la valeur seuil et les valeurs dans un encadré vert sont celles implémentées dans la matrice
URL http://journals.openedition.org/bmsap/docannexe/image/17895/img-6.png
Fichier image/png, 202k
Titre Figure 7
Légende Pie chart representing the type of characters used for each phylogenetic analysis on hominins (n=47). Were considered as continuous characters discretized characters for which an explicit discretization logic and method do exist, excluding for example Cameron et al. 2004 and Mounier and Caparros 2015. *GM stands for Geometric Morphometrics. Graph generated from the data presented in appendix D |Diagramme circulaire présentant le type de caractères utilisés pour chaque analyse phylogénétique des hominines (n=47). Ont été considérés comme caractères continus discrétisés des caractères pour lesquels une argumentation et une méthode existent pour la discrétisation, excluant par exemple Cameron et al. 2004 et Mounier et Caparros 2015. Graphique généré à partir des données présentées dans l’appendice D. *GM signifie morphométrie géométrique
URL http://journals.openedition.org/bmsap/docannexe/image/17895/img-7.png
Fichier image/png, 72k
Titre Figure 8
Légende Supertree generated from 37 phylogenetic analyses on hominins. Green points represent clades found in each tree considered, orange points represent clades found in more than 75% of the trees but less than 100% and red points represent clades found between 50% and 75% of the trees. Taxa of which the position is discussed in the text are signalled by a blue star. Purple stars point to the fossils frequently included in H. heidelbergensis sensu lato |Superarbre généré à partir de 37 analyses phylogénétiques portant sur les hominines. Les points verts représentent des clades trouvés dans tous les arbres considérés, les points orange des clades dont la fréquence est supérieure à 75 % et strictement inférieure à 100% et les points rouges des clades dont la fréquence est comprise entre 50 % et 75 %. Les taxons dont la position est discutée dans le texte sont signalés par une étoile bleue. Les étoiles violettes désignent les fossiles communément inclus dans H. heidelbergensis sensu lato
URL http://journals.openedition.org/bmsap/docannexe/image/17895/img-8.png
Fichier image/png, 415k
Titre Figure 9
Légende Phylogenetic tree adapted from Kimbel et al. 2006. The analysis focused on the fossils excavated in the four sites that make up the most basal OTUs of the tree. Their position in the tree being correlated to their age (i.e., the OTUs branch successively following a chronological order), the authors conclude as to anagenetic evolution from Au. anamensis to Au. afarensis |Arbre phylogénétique issue de celui présenté par Kimbel et al. 2006. L’analyse a porté sur les fossiles provenant des quatre sites qui forment les UTO à la base de l’arbre. Leur position dans l’arbre étant corrélée à leur âge (i.e., les UTO se branchent successivement suivant un ordre chronologique), les auteurs en ont conclu à une évolution de type anagénétique liant les espèces Au. anamensis et Au. afarensis
URL http://journals.openedition.org/bmsap/docannexe/image/17895/img-9.png
Fichier image/png, 77k
Haut de page

Pour citer cet article

Référence électronique

Pierre Gousset, Florent Détroit et Jérémie Bardin, « Hominin phylogenetics: methods and insights »Bulletins et mémoires de la Société d’Anthropologie de Paris [En ligne], 38 (1) | 2026, mis en ligne le 28 mars 2026, consulté le 21 avril 2026. URL : http://journals.openedition.org/bmsap/17895 ; DOI : https://doi.org/10.4000/15z70

Haut de page

Auteurs

Pierre Gousset

UMR 7194 HNHP, CNRS, UPVD, MNHN, Paris, France ; pierre.gousset[at]edu.mnhn.fr ; https://orcid.org/0009-0003-2607-5123

Articles du même auteur

Florent Détroit

UMR 7194 HNHP, CNRS, UPVD, MNHN, Paris, France; https://orcid.org/0000-0001-5208-6203

Articles du même auteur

Jérémie Bardin

UMR 7207 CR2P, Sorbonne Université, MNHN, CNRS, Paris, France ; https://orcid.org/0000-0003-2382-4259

Articles du même auteur

Haut de page

Droits d’auteur

CC-BY-NC-SA-4.0

Le texte seul est utilisable sous licence CC BY-NC-SA 4.0. Les autres éléments (illustrations, fichiers annexes importés) sont susceptibles d’être soumis à des autorisations d’usage spécifiques.

Haut de page
Search OpenEdition Search

You will be redirected to OpenEdition Search