Révision de l’espèce Homo erectus (Dubois, 1893)

• Published in: Bulletins et Mémoires de la Société d’Anthropologie de Paris Vol. 12; no. 2; pp. 1 - 200
• Main Author: Valéry Zeitoun
• Format: Journal Article
• Language: English
• Published: Société d'Anthropologie de Paris 01.06.2000
• ISSN: 0037-8984; 1777-5469; 1777-5469
• DOI: 10.4000/bmsap.5963

The hypodigm for Homo erectus is a problem which remains unresolved. Most disagreements are based on chronological rather than morphological data. A methodology based neither on simple global similarity nor on chronological position is required to clarify taxonomical questions. Therefore the identification of apomorphic features for Homo erectus has to be tried. The first step is to seek criteria allowing to define this species as a clade or as a grade; id est as a monophyletic unit or as an archaic evolutionary step of the species Homo sapiens. The reconstruction of this evolutionary pattern should be matched with other data (dating, biogeography) to allow to reconstruct a real evolutionary synopsis.According to several authors (Stringer, 1984; Hublin, 1986; Kennedy, 1991; Brauer and Mbua, 1992) the existence of autapomorphic features for Homo erectus is not proved. Nevertheless, a re-examination of the calvaria of human fossils is made in the light of cladistics to define the taxonomic and the phylogenetic ranks of the species Homo erectus (Dubois, 1893). New morphological features and unpublished metrical data are proposed at the same time that an analysis and a control of published morphological data were made.In the aim of using most of the information in phylogenic study about fossils at individual level, a coding methodology is proposed which allows to use continuous data in cladistics. The purpose of this new coding proposal is to be useful in paleoanthropology where most of the time the studied material is individuals and not populations.There is no reason to decide first that continuous characters are not useful to discriminate among groups of taxa. Morphological features can describe either the position or the size of an element next to another. They can cause a more specific and more complex definition, a precise combination of elements. It is then possible to assert, whether this definition of the feature is respected or not, that the feature is present or absent. When, for instance, the upper part of an anatomical area is larger than its lower part, to evaluate this criterion, it is better to use an index calculated from homologous points which relate to these both parts. This process allows a better definition of the states of the features. It shows where the most revealing dimensions are located (a set can be divided in many subsets and the intensity of a feature’s expression can change among the subset). Before taking measures and calculating indices, reference points must be guaranteed to be anatomically homologous.Statistical methods or descriptive analysis (correspondence analysis, factor analysis) are based on comparisons of individuals and parameters in relation to a barycenter obtained by the calculation of each specimen’s own values. The use of these methods of analysis constitutes an exercise of the pheneticians “total morphological pattern”, since they express a global similarity from the criterions generated by the global weight of the individuals who have been studied. Another approach is more typological.The method proposed in this book is influenced by the gap coding proposed by Thiele (1993), but applied at individual level. The method use a standard which is independent of all specimens taken into account as an average. Ignoring which of the most discriminant parameters allow the reconstruction of phylogeny, all the measurements existing between every reference points are taken into account. Thus each measure can be connected to another (if it cannot, then the “false” precision of some craniometric points should be questioned). The complete data are used to calculate indices. Two sorts of indices are proposed to be kept :— type I (arc AB - chord AB) / chord AB— type II chord AB / chord CD.The indices are continuous metrical data which must be transcribed in a matrix to be used in a cladistic analysis. Before explaining the coding method, mathematical properties linked to the index must be discussed. Indices are relative data. They are relatively independent from the global size (Olivier, 1974). They give indications on relative size. Thus, even if the affine or non-affine values of the measures cannot be distinguished by the use of simple indices (see Bookstein, 1978) the global size effect is weakened. Concerning the rigorous use of index, arguments on the rules which are to be adopted are still published (Atchley et al., 1976; Atchley et Anderson, 1978 versus Albrecht, 1978; Dodson, 1978; Hills, 1978). Those arguments have revealed that the index formulation in logarithmic form usually allows to use them without any problem in the sight of the Laplace-Gauss law. So for a given feature, if the normality is not respected in the reference set then it is not taken into account.An analysis is realized to select the discriminant index, those allowing the grouping or the splitting of different individuals. The common parameters to all the specimens are not required to distinguish the specimens. Individuals instead of a priori constituted sets of individuals are taken into account since the subject of the study is to determine the taxonomic belonging and the phylogenetic position of the specimens. Only indices outside the variation of the reference population are kept as discriminant. Thus, for each index when there is a superposition between the reference population and the several specimens taken into account, the index is considered as free from any phylogenetic significance at least within the defined group. When there is different values between specimens and the standard, the differences are used to determine classes. From the extreme value observed (maximum or minimum) for the standard, gaps with an amplitude of one standard deviations are used to define the class. The use of a standard connects this method to the combination coding.A group of individuals will be defined by a space equal to only one standard deviation of the standard population. So the number of real groups should be emphasized but in the end of the analysis such “real” - subgroups- can be linked. To the contrary, use of a double standard deviation space to define individuals group tend to lump “real” subgroups. The process uses to define class for grouping is not founded on any theoretical a priori. It is only a mathematic way to lump specimens in groups.There isn’t numerous specimens with preserved face which are affiliated to Homo erectus. Further more mandibles are affiliated to Homo erectus, but there are isolated. Classical definition of Homo erectus is mainly founded on feature from the calvaria. Most of the other anatomical fragments are allocated to Homo erectus according to their dating not to their morphology. Sixty six specimens within Paranthropus boisei, Australopithecus africanus, Pan troglodytes, Gorilla gorilla species and modern humans allowed to establish classes corresponding to characters states for metrical data.Six hundred and thirteen measurements are taken for a complete skull. Two thousand two hundred and sixty four indices are calculated from these data. By comparison of the standard and the whole specimens taken into account, it is not possible to distinguish modern humans from them (Apes, Australopithecine and Homo) for 1919 parameters. Distinctions are able for only 345 parameters.Morphological characters sum 123, collected according to anthropological literature and personal observations. The anatomical analysis of the specimens leads to consider unequal number of criteria for each bone of the skull (182 criteria for the frontal bone, 25 morphological data and 157 metrical data; 67 for the parietal bone, 16 and 51; 159 for the temporal bone, 62 and 97; 60 for the occipital bone, 20 and 40). This fact is due to the unequal number of homologous reference points present on each bone, to the difference of the complexity of each anatomical part and to the selection of the most discriminant parameters compared to modern humans taken as reference. The main illustration of the exposed process is proposed for a sample restricted to 35 specimens dealing with the case of Homo erectus.The choice of the specimens is guided by three main considerations:— Homo erectus and specimens affiliated to Homo erectus offering the largest amount of available information;— taxa affiliated to Homo habilis on the one hand and archaïc Homo sapiens on the other hand have been taken into account since the limits between the three taxa are not sharply cut.— a few fragmentary specimens, situated in the heart of numerous discussions taken from literature, have been added to the previously chosen specimens.A mixed coding of the states of characters is applied to the 35 specimens so defined. That is to say that some characters with multiple states are treated as ordinate and some others are treated non ordinate. The choice is based on ontogenetic considerations.The choice of the option for the treatment of character of multiple state (ordinate or non ordinate) is based on morphological observations and on an analysis on metrical indices. On the subject of morphological data it consists in observing Pan troglodytes, Gorilla gorilla and Homo sapiens sapiens immature individuals. There is no major problems in recognizing immature individuals in those taxa. On the subject of metrical data, the study of modern human immature specimens was realized to determine the putative polarity of indices. When it was possible to follow an ontogenetic development of the states of character, it had been verified if it was gradual and consequently if it was marked by a polarization. In that case, it implied that states of characters had to be ordinate. In the opposite case, there was no reason to suppose that a given state of character was an obligatory intermediate passage to go from one state to another. Indeed if one considers that an index can follow a linear evolution only, one passes over the heterochrony of the development of a