Evidence for new Neanderthal teeth in Tabun Cave (Israel) by the application of self-organizing maps (SOMs)

https://doi.org/10.1016/j.jhevol.2006.11.009Get rights and content

Abstract

Morphological and metrical study suggested that seven human teeth from Tabun Cave, Israel were part of the upper dentition of a single, probably Neanderthal, individual renumbered as Tabun BC7. An enamel fragment gave ESR age estimates of 82 ± 14 ka (early U-uptake) and 92 ± 18 ka (linear uptake) and an age estimate of 90+30−16 ka using U-series disequilibrium. Although metrical analyses suggested Neanderthal affinities, definitive assessment was difficult as the values often fell into the ranges of both Neanderthal and Levantine early modern human samples. Therefore, two further classification analyses were conducted (neural networks using self-organizing maps and homogeneity analysis). Both identify Tabun BC7 as a Neanderthal. Neural networks are a promising tool for paleoanthropological studies as they can provide reliable classifications even with incomplete data.

Introduction

Seven human teeth from Tabun Cave, Israel, (right I1, I2, M1, M2, M3, and left M2, M3) were tentatively identified in the Catalogue of Fossil Hominids (Oakley et al., 1975) as deriving from Tabun Layer A (Bronze Age to Recent). However, further morphological and metrical study has suggested that they represent part of the upper dentition of a single individual, probably a Neanderthal, renumbered as Tabun BC7 (Coppa et al., 2005). A fragment of dental enamel of Tabun BC7 was removed for ESR and U-series analysis, and was assigned to Layer B based on its dose values. This enamel yields preliminary ESR age estimates of 82 ± 14 ka (early U-uptake) and 92 ± 18 ka (linear uptake), and an age estimate of 90+30−16 ka using U-series disequilibrium. Metric analysis of the crowns, and in particular of the buccolingual diameters of the two incisors, placed them in the range of variability characteristic of the Neanderthals rather than anatomically modern Homo sapiens (AMH). On the other hand, analysis of the molars was insufficient for a definitive taxonomic assessment since the values obtained fell into the ranges of both the Neanderthals and Levantine AMH (Coppa et al., 2005).

The absence of other skeletal remains relating to this sample makes assignment to a specific taxon difficult. The uses of dental metric and morphological traits in the study of taxonomic and phylogenetic relationships were highlighted by Smith (1989). Earlier studies (Gorjanovic-Kramberger, 1906, Gorjanovic-Kramberger, 1907, Keith, 1925, McCown and Keith, 1939) underlined the particular importance of morphological traits in phylogenetic studies. After Weidenreich's study (1937), which pointed out how morphological traits could represent a possible expression of regional variation as well as evolutionary grade, dental nonmetric characters lost importance for classification purposes, and tooth size has been emphasized instead (Smith, 1977a, Smith, 1977b, Frayer, 1978, Frayer, 1984, Brace, 1979, Wolpoff, 1980). However, the morphological traits considered in many previous studies were limited to characters such as taurodontism, the lingual tuberculum of the anterior teeth, Carabelli's cusp, and wrinkled enamel. It is only in the last few years that Neanderthals have been examined for numerous dental morphological traits (e.g., Stringer et al., 1997, Lucci, 2001, Bailey, 2002).

To assist further in the assessment of the BC7 specimen, we analyze its dental morphological traits following the ASUDAS system (Scott and Turner, 1997) and compared them with the large database of Eurasiatic and North African Late Pleistocene samples scored by two of the authors (A.C. and R.V.). We then apply two different classification methods (neural network analysis and homogeneity analysis) to these data to evaluate the strength of the classification and usefulness of these methods.

Section snippets

Materials and methods

Twenty-three dental morphological traits were scored on the Tabun BC7 dentition (Table 1). All comparative specimens are referable to Neanderthals and AMH from the Middle Paleolithic in the Levant, and the Upper Paleolithic-Mesolithic in Europe (Italy, France, Czech Republic, UK), the Middle East (Israel), and North Africa (Algeria, Morocco) (Coppa et al., 2001, Lucci, 2001).

The scoring values of the traits were dichotomized and a first matrix with only presence/absence was thus obtained. The

Neural network analysis by the application of self-organizing maps (SOMs)

The purpose of the analysis was the classification of a new dental sample (TBN-BC7) by comparing it with available data. For this reason, the traits used in the analysis are those which were scored on the specimen. This method can be applied to a very large number of samples and traits (Manni et al., in press). As a reference sample, we added 125 known samples that belong to populations of Neanderthals and AMH from the Middle and Upper Paleolithic. These samples range from Europe to the Middle

Conclusions

The consistency of the two different classification analyses we conducted (neural networks/self-organizing maps and homogeneity analysis) allows us to conclude that Tabun BC7 represents a Neanderthal. The use of dental morphological traits has been an excellent instrument for the taxonomic phylogenetic classification of teeth alone.

This classification study prompted us to introduce neural networks in paleoanthropological studies since these networks can provide reliable classifications and

Acknowledgements

This research was supported in part by MURST COFIN05 and “Progetti di Ateneo” University of Rome “La Sapienza”. We are grateful to Pascal Croiseau for his help in the statistical analysis. We also thank the editor of the journal and the four anonymous reviewers whose precious comments allowed us to improve the manuscript.

References (39)

  • D. Frayer

    Evolution of Dentition in Upper Paleolithic and Mesolithic Europe

    (1978)
  • D. Frayer

    Late Pleistocene and Early Holocene in Europe

  • K.R. Gabriel

    The biplot graphical display of matrices with application to principal component analysis

    Biometrika

    (1968)
  • K. Gorjanovic-Kramberger

    Der diluviale Mensch von Krapina in Kroatien: Ein Beitrag zur Palaoantropologie

    (1906)
  • K. Gorjanovic-Kramberger

    Die kronen und wurzeln der mahlzahne des Homo primigenius und ihre genetische bedeutung

    Anat. Anz.

    (1907)
  • R.W. Heiser et al.

    Homogeneity analysis: exploring the distributions of variables and their nonlinear relationships

  • W. Heiser et al.

    Nonlinear methods for the analysis of homogeneity and heterogeneity

  • S. Kaski

    Data exploration using Self-Organizing-Maps

    Acta Polytech. Scand. Math. Manage. Comput. Eng. Ser.

    (1997)
  • A. Keith

    The Antiquity of Man

    (1925)
  • Cited by (0)

    View full text