Taille et conformation crânienne chez les Hominidés de la fin du Pléistocène: Contributions de la morphométrie géométrique au débat sur l'origine de l'Homme moderne 9781841710150, 9781407351261

Citation preview

Taille et conformation cranienne chez les Hominides de la fin du Pleistocene Contributions de la morphometrie geometrique au debat sur l'origine de l'Homme modeme

Martin FrieB

BAR International Series 799 1999

Taille et conformation cranienne chez les Hominides de la fin du Pleistocene Contributions de la morphometrie geometrique au debat sur l' origine de l'Homme modeme

Martin FrieB

BAR International Series 799 1999

Published in 2016 by BAR Publishing, Oxford BAR International Series 799 Taille et conformation cranienne chez !es Hominidr!s de lafin du Pleistocene

© M Frid?, and the Publisher 1999 The author's moral rights under the 1988 UK Copyright, Designs and Patents Act are hereby expressly asserted. All rights reserved. No part of this work may be copied, reproduced, stored, sold, distributed, scanned, saved in any form of digital format or transmitted in any form digitally, without the written permission of the Publisher.

ISBN 9781841710150 paperback ISBN 9781407351261 e-format DOI https://doi.org/10.30861/9781841710150 A catalogue record for this book is available from the British Library BAR Publishing is the trading name of British Archaeological Reports (Oxford) Ltd. British Archaeological Reports was first incorporated in 1974 to publish the BAR Series, International and British. In 1992 Hadrian Books Ltd became part of the BAR group. This volume was originally published by Archaeopress in conjunction with British Archaeological Reports (Oxford) Ltd/ Hadrian Books Ltd, the Series principal publisher, in 1999. This present volume is published by BAR Publishing, 2016.

BAR

PUBLISHING BAR titles are available from:

EMAIL

PHONE FAX

BAR Publishing 122 Banbury Rd, Oxford, OX2 7BP, UK [email protected] +44 (0)1865 310431 +44 (0)1865 316916 www.barpublishing.com

Meinen Eltern

1

2

Quiconque s' est jamais avise de speculer sur ces quatre choses : Qu'y Qu'y Qu'y Qu'y

a-t-il au-dessus? a-t-il au-dessous? avait-il avant le monde? aura-t-il apres ?

Il aurait rnieux valu pour lui qu'il ne filt pas ne.

Talmud de Babylone, Haguigah, llb.

3

4

Avant-propos

Afin de faciliter la lecture de ce travail et d'eviter des explications repetitives, nous tenons a donner a l'interesse les indications suivantes : La methode des plaques minces (« thin-plate splines») decrit le changement entre deux conformations par le moyen d'une grille de transformation de coordonnees cartesiennes. Ces changements n'ont du sens qu'en relation avec !'ensemble des points de reperes utilises. Par consequent, lorsque nous parlons du deplacement de certains points, celui-ci ne doit pas etre pris dans le sens absolu, mais toujours comme deplacement relatif par rapport a une configuration precise de points. Ces aspects seront expliques de maniere plus detaillee dans le chapitre 2 traitant des methodes. Par ailleurs, et suivant une tradition en morphometrie geometrique, nous avons joint un glossaire des principaux termes techniques morphometriques que le lecteur pourra consulter a tout moment (cf Annexe 1).

PN

PS/Meso PSa PSt

RW SQ TPS

u

HE HSA HSM N

Preneandertalien( s) Paleolithique superieur et Epipaleolithique Paleolithique superieur ancien (avant 20.000BP, cf Materiel) Paleolithique supeneur tardif (apres 20.000BP incluant l'Epipaleolithique) Relative warps, les flexions relatives (cf Glossaire) Les hommes fossiles de Skhul et Qafzeh, consideres comme un groupe operationnel. Thin-plate splines, les plaques minces (cf Glossaire) Part uniforme (cf Glossaire)

D' autres abreviations utilisees concement essentiellement les specimens fossiles ; celles-ci sont specifiees dans le tableau 2.1 (cf Materiel). Par ailleurs, nous utiliserons souvent le sigle PS pour designer le regroupement, operationnel, des sujets du Paleolithique superieur ancien et tardif ainsi que ceux de l'Epipaleolithique.

Conformement aux conventions intemationales, les legendes precedent les tableaux et suivent les graphiques. Pour les tableaux et graphiques, nous auront recours aux sigles suivants :

cs

-

Lorsque nous comparons des mesures conventionnelles, nous utilisons les definitions et numeros de Martin (1914, cf KnuBmann 1988). Les references bibliographiques ont ete redigees suivant la convention de ['American Journal of Physical Anthropology.

Centroid size, la taille centro"ide (cf Glossaire) Homo erectus Homo sapiens archa"ique Homo sapiens anatomiquement modeme du Paleolithique moyen Neandertalien(s)

Enfin, pour une orientation plus aisee dans le texte, le lecteur trouvera en haut a droite de la page une petite icone symbolisant les principaux chapitres.

5

Remerciements

Senckenberg, Franfurt am Main) ; M.A. Fugazzola (Soprintendenza di beni culturali, Roma) ; D. GrimaudHerve (IPH, Paris) ; J.L. Heim (IPH, Paris) ; I. Hershkovitz (Sackler school of Medecine, University of Tel Aviv) ; H.E. Joachim (Rheinisches Landesmuseum, Bonn) ; H. Katz (Rockefeller Museum, Jerusalem) ; A. Langaney (Musee de l' Homme, Paris) ; H. de Lumley (IPH, Paris) ; R. Machiarelli (Museo L Pigorini, Roma) ; F. Mallegni (Universita di Pisa) ; P. Menessier (Musee de l'Homme, Paris) ; P. Morel (Service d' Archeologie, Neuchatel) ; M. Oliva (Anthropos Institute, Brno) ; D. Pilbeam (Peabody Museum, Harvard University) ; G. Rossi (Museo Balzi Rossi, Genova) ; G. Spadea (Soprintendenza di beni culturali, Genova) ; C.B. Stringer (Natural History Museum, London) ; A. Vigliardi (Soprintendenza di beni culturali, Firenze) ; J. Zias (Rockefeller Museum, Jerusalem).

Au terme de ce travail, nous tenons a remercier toutes les personnes qui nous ont apporte leur soutien et sans lequel ce travail n'aurait pas vu le jour. Le Professeur B. Vandermeersch pour nous avoir accueilli dans son laboratoire, pur avoir dirige ce travail et pour nous avoir donne tout le soutien scientifique, financier et logistique necessaire a nos recherches. Le professeur W. Henke, notre mentor et codirecteur, qui nous a forme en paleoanthropologie et essaye d'ouvrir notre esprit au raisonnement evolutionniste. Si nous en avons fait preuve, c'est grace a lui, si nous en avons manque, c'est en depit de son enseignement. Le professeur G. Giacobini qui nous a fait l'honneur de juger ce travail et de participer au jury. Nous en sommes tres honore.

Nos remerciements vont plus particulierement vers les personnes qui, en dehors de l' acces au materiel, nous ont fait un accueil chaleureux et donne des conditions de travail ideales :

M. Baylac nous a guide en matiere de morphometrie geometrique. Sa pedagogie et sa disponibilite ont ete essentielles pour ce travail.

B. Arensburg (Sackler School of Medecine, University of Tel Aviv) ; M. Chech (Musee de l'Homme, Paris) ; V. Formicola (Universita de Pisa) ; J-J. Hublin (Musee de l' Homme, Paris) ; R. Kruszynski (Natural History Museum, London) ; S. Lee-Bruce (Peabody Museum, Harvard University) ; G. Manzi (Universita di Roma « La Sapienza») ; A. Mann (University Museum, Philadelphia) ; T. Molleson (Natural History Museum, London) ; J. Monge (University Museum, Philadelphia) ; C. Simon (Departement d' Anthropologie et d'Ecologie, Universite de Geneve) ; M. Teschler-Nicola (Naturhistorisches Museum, Wien); B. Vandermeersch (Universite Bordeaux I). Le professeur R. Fenart a enrichi notre travail avec de precieuses discussions methodologiques et des indications bibliographiques. Le professeur 0. Bar-Y osef nous a donne son soutien et nous avons beneficie de son esprit critique vis-a-vis de l' anthropologie.

H. Duday nous a fait l'honneur de participer a ce jury. Durant notre doctorat, nous avons eu le bonheur de pouvoir profiter de ses qualites d' anatomiste et anthropologue. Notre travail a egalement beneficie de sa bienveillance en tant que directeur de l'equipe. Nous lui sommes tres reconnaissant. Nous souhaitons egalement exprimer notre plus grande reconnaissance envers le professeur 0. Dutour, qui a bien voulu accepter de presider ce jury. Sa participation nous a ete un grand honneur. Tous les membres du Laboratoire d' Anthropologie de l'Universite de Bordeaux 1 qui ont accompagne, chacun a sa maniere, notre travail pendant plus de trois ans : M. Bessou, J. Bruzek, D. Castex, J. Cathalaa, E. Cleuvenot, C. Couture, H. David, H. Duday, M. Elyaqtine, D. Gallardo, D. Gambier, F. Houet, B. Maureille, V. Peyrier, S. Prat, P. Sellier, M. Seurin, V. Sladek, A-M. Tillier, C. Verna.

E. Crubezy et H. Duday ont bien voulu nous faire part de leurs connaissances en matiere de morphologie cranienne. Nous leurs devons beaucoup d'indices de travail et de commentaires. Le groupe de travail de morphometrie geometrique du MNHN (Paris), particulierement M. Baylac et X. Penin, a enrichi nos connaissances en morphometrie geometrique et statistiques multivariees.

Nous remercions particulierement D. Castex et D. Gambier pour l' aide, les commentaires et les discussions enrichissantes dont nous avons beneficie. Tous les membres de l'Institut fur Anthropologie der Johannes Gutenberg-Universitat, Mainz, en particulier E. Frauendorf et U. Krenzer. Maintes personnes nous ont permis d'etudier et de photographier le materiel fossile. Nous les remercions pour leur confiance, leur comprehension pour notre gourmandise en matiere d'espace et de temps de travail et, enfin, pour leur patience face a nos tentatives de realiser un agenda a la convenance de tout le monde :

D. Castex a bien voulu nous aider dans la redaction du texte en franliauee 38.1 38.1

Groupe

N

1

Valeurs IJ>ropres 1.05696

Japonais

54

2

0.76611

19.9

58.0

Melanesiens

44

3

0.62909

13.5

71.5

Tasmaniens

33

4

0.5378

9.9

81.4

Roumains

27

5

0.48888

8.1

89.5

Africains

50

6

0.38021

4.9

94.4

Total

208

7

0.35751

4.4

98.8

8

0.18822

1.2

100.0

Ence qui conceme l'arriere-crane, l' ACP nous montre que la plupart de la variance conceme d'un cote la hauteur de l'ecaille occipitale (lambda et inion, axe 1) et de l'autre cote la position relative de l'opisthocranion (axe 2, cf Figure 4.59). Si la hauteur de !'occipital ne parai't pas privee de sens biologique, en revanche la repartition des scores individuels sur cet axe n'indique pas de variation specifique a un des groupes actuels. Sur ce premier axe on

peut constater que, au fur et a mesure que l' ecaille gagne en hauteur, !'orientation de la suture parieto-masto"idienne devient plus horizontale. Le deuxieme axe, outre la position relative de l'opisthocranion, decrit egalement la position du porion relativement a l'incisure parietale.

121

Resultats

Calotte (avec U) Flexions relatives 1 et 2 0.06

• +•

+

0.04

+

□

□□

0.02

•~

~ 0.00 C\I r--:

,p

LO

C\I

~

+

-0.02

• .p

a: -0.04

•

• IJ

□

)~

,_

0

@.

8d

o+

□

/ii:) ~

q.. □ □

;2,-.

0

-

c9 0

0

•

0

□

0

c:P" □

0

0

«> 0

+

•• • c::P i

0

•

□

+ + + 0□

• •

+

•

0

•

jo

e-o

0

0

□

0

-0.06

□

• -0.08 -0.08

-0.06

-0.04

+ -0.02

0.00

0.02

0.04

0.06

0.08

0.10

•

Jap. Mela. Tas. Roum.

Afr .

RW1 (27.32%)

Figure 4.46 : Calotte des sujets actuels, flexions relatives. Scores individuels des sujets sur les 2 premiers axes.

122

Resultats

Consensus

,/

"

/

/

/

~

~

'\I\. '\

I

'\

I

i

•2

\

1

7 IU

ei

I

Ii -u

a) / /

RWUl pos

------

~

V

/

-

r.(

b)

-- -

~

'\.

"" "\.

\

I

\

12

-

'1

9l

'7

L

'

I

~7: Tl

-u-

RWU2pos

~v

'

//

I

~\

'

I 11---'\. I

,

I

'

rrf'

1 'fL.

"i

1 9l:

'7

IU

L u

Figure 4.47 : Calotte des sujets actuels, flexions relatives. Changements des conformations le long des 2 premiers axes. Visualisation par a) grille b) deplacement vectoriel.

123

Resultats

Crane et mandibule (avec U) Flexions relatives 1 et 2 0.08

0.06

0.04

o• ?fl co

•

• • Oo

0.02

O

I

□

e

O

•

LO

O

I

B

e

•

0

C\i ,....

(\J

0.00

•

q.

□ □

0

~

a:

0

-0.02 0

0

-0.04

□

□

0

0

cs'4- ~ 1r~~

vey1

□

0

□

0

O

oO

0 0

0

0

0

o+ ,m .. v//5

~tf'2~c

m □ c!Jo + 0 0 shn ~'ep oo°bl[_~o + Oo □ m1 sw§ 0

okf c9 □ oo pobg1 skl5 q:6biaipa oooo ~c al q,o l?:ltJo o □ o .a. ~iioDl 0 ~ Oo i, ct:riicf 0 0

.

O O

0

84b o0 mld1 □ 'l-lay4 □ o o"o q, sk14' o lbi"o 'fl o o.t. o sac1 ~ av g mllfl 5 h2 Drdb

0

00

1

0 0

0.10

>-

sing ■

n

0

□

Do 0

q9

0

...

C

0

0

0 0

0

-0.10

'

0

0 0

0 0

-0.14

-0.18 -0.18

□

0

+ Neg

.t.

□

ii ■

-0.14

-0.10

-0.06

-0.02

0.02

0.06

0.10

0.14

U_X

Figure A 20 : Arriere-crane, repartition des scores individuels de la part uniforme.

Tableau A 12 : Arriere-crane : statistiques descriptives de la part uniforme. U_X M PS N PN HSA HSM HE Tous Groupes

X -0.00240697 0.0287059 -0.01543198 -0.00668321 -0.01732091 0.00527326 -0.09259419 0.000002

U_Y O'

0.04813998 0.04414654 0.02506158 0.02099651 0.05492112 0.03498712 0.00500812 0.04827469

X -0.00238853 0.01240204 -0.0107307 0.00911717 0.02830319 0.00686269 -0.03649146 -3.0669E-17

231

O'

0.05169391 0.04748775 0.01965778 0.03056377 0.07517236 0.0375609 0.03861868 0.05081904

N 208 31 6 5 6 5 2 263

•

Mod. PS N HSM PN HSA HE

Annexe Part uniforme (U) face 1 0.12 0

0

0

0.08

0

0

eea

0

I>

0 0

0

0

'b 0

8

0

0.04

0

0

0

00

Lal + no

n

-0.04

0

0

0

mlh37 □

0 0

0

0

11h23

"

oolhl DoI> 0 O

0

o0

□

0

~

g

0

0

...

ac

0 0

0

abg.p

Oo 0 0

0 u4b ac5D

0

r~

oo

k2 □

0

0 00

'b0 flh16

0

Hay4 ch['j □

0

u4a 0

□

[tJ

0

00

"'

g0

00

s~4

o

0 0

0

0

"'•q, ;r,, olbi

0

0

0

0

08

...

0

0

0

OO 0

Laq

Amd + 0

0 "

om1°

o oo o

0

0

0

~

0

■

□

0

0

O

0 o O:0

m~b~

sacs13.c2 '"" skis+ ,f" _..o

□

0

□

ooo o o

a1>a

o

l

0 0

acpr

OO

m~2

0

o■

ces +

0

0 0

0 00 0

kab

0

0 0 0

0

□

0

0 0

0 0

0

□

bg2 □

0

cm! ~1

0

-0.08

□ ~

.t.

Mod. PS N HSM

"

PN

0

0

□

+ -0.12 -0.14

-0.10

-0.06

-0.02

0.02

0.06

0.10

0.14

U_X

Figure A 21 : Face, repartition des scores individuels de la part uniforme.

Tableau A 13 : Face : statistiques descriptives de la part uniforme. U_X

M PS N HSA HSM PN Tous Groupes

X 0.00283192 0.00602812 -0.06077819 -0.11567141 -0.01371138 -0.05744961 0.0006

U_Y O"

0.02777531 0.03489447 0.02608063 0.04732505 0.0277715 0.03202834

X 0.00509392 -0.03127056 -0.00475664 -0.03027043 -0.02611577 -0.01696924 -0.00014875

232

O"

0.03834108 0.03882679 0.02411945 0.04175562 0.01530385 0.03961245

N 188 24 5 1 5 3 226

■

HSA

--------------------~A=n=n=e=x=e

D

_______________________

Annexe 6 : Analyses discriminantes : parametres statistiques des fonctions. V aleurs propres, coefficients de correlation canonique (R canonique), Lambda de Wilks, valeurs du Chi2, degres de liberte (Dl), niveau de signification (p) et probabilites a priori I posteriori des sujets. Les specimens mal classes sont signales par un * precedant leur nom. Tableau A 14 : Crane et mandibule de l'echantillon complet, analyse discriminante. Test du Chi 2 sur les racines. Racine

Valeur Propre

R canonique

Lambda de Wilks

Chi2

DI

p

0

2.5035

0.8453

0.1006

404.1691

102

0.00"+00

1

1.0650

0.7182

0.3525

183.5089

66

5.84"-13

2

0.3737

0.5216

0.7280

55.8831

32

5.62"-03

Tableau A 15 : Crane et mandibule de l'echantillon complet, analyse discriminante. Probabilites de classification. A) Probabilites a priori des sujets non ou mal classees. Classif. Observee

1 p=0.88265

2 p=0.07653

3 p=0.02041

4 p=0.02041

PS --PS

M

PS

HSM

N

M

M

PS

HSM PS HSM

*Abri Pataud Tabun 1 *Uzzo 4b

N N

B) Probabilites a posteriori des sujets non ou mal classees. Classif. Observee

M p=0.88265

PS p=0.07653

N p=0.02041

HSM p=0.02041

PS --PS

0.977

0.023

0.000

0.000

0.000

0.000

0.000

1.000

0.861

0.139

0.000

0.000

*Abri Pataud Tabun 1 Uzzo 4b

Tableau A 16 : Calotte (N=52), analyse discriminante. Test du Chi 2 sur les racines. Racine

Valeur propre

R canonique

0

12.0503

1

1.3015

2 3

Chi'

DI

0.9609

170.5973

64

1.35"-11

0.7520

0.1933

66.5603

45

2.01"-02

0.8306

0.6736

0.4449

32.8005

28

2.43"-01

0.2279

0.4308

0.8144

8.3135

13

8.23"-01

Tableau A 17 : Calotte, analyse A) Probabilites a priori des sujets non classes.

ES Singa

p

Lambda de Wilks 0.0148

discriminante.

Probabilites

de

Classif. Observee

1 p=0.61538

2 p=0.13462

3 p=0.07692

4 p=0.05769

5 p=0.11538

-----

HSM HSM

PN PS

N N

PS PN

HSA HSA

B) Probabilites a posteriori des sujets non classes.

ES Singa

Classif. Observee

PS p=0.61538

N p=0.13462

HSA p=0.07692

PN p=0.05769

HSM p=0.11538

-----

0.01545

0.03411

0.00322

0.15090

0.79632

0.01453

0.00003

0.00000

0.00000

0.98544

233

classification.

Annexe

Tableau A 18 : Neuro- viscerocrane (N=32), analyse discriminante. Test du Chi2 sur les racines. Racine

Valeur propre

R canonique

Chi'

DI

p

0.9971

Lambda de Wilks 0.0011

0

169.1707

1

4.3963

105.7490

56

6.80"-05

0.9026

0.1853

26.1285

27

5.12"-01

Tableau A 19 : Neuroviscerocrane, A) Probabilites a priori des sujets non classes.

analyse

Classif. Observee

-------

Kabwe Saccopastore 1 Tabun 1

discriminante.

Probabilites

1 P=0.75000 N

2 P=0.12500 HSM

3 P=0.12500 PS

HSM

PS

N

HSM

PS

N

de

classification.

B) Probabilites a posteriori des sujets non classes. Classif. Observee

PS

-------

Kabwe Saccopastore 1 Tabun 1

N

HSM P=0.12500 0.000

P=0.75000 0

P=0.12500 1

0

0

1

0

0

1

Tableau A 20 : Frontal (N=46), analyse discriminante. Test du Chi2 sur les racines. Racine

Valeur propre

R canonique

Chi2

DI

p

0.9341

Lambda de Wilks 0.0317

0

6.8437

1

0.9421

125.9928

48

6.70"-09

0.6965

0.2485

50.8133

33

2.46"-02

2 3

0.7577

0.6566

0.4827

26.5850

20

1.47"-01

0.1786

0.3893

0.8485

5.9979

9

7.40"-01

Tableau A 21 : Frontal, analyse discriminante. Probabilites de classification. A) Probabilites a priori des sujets non classes. Classif. Observee ES *Skl4

---

P=0.58696 HSM

2 P=0.15217 PS

3 P=0.06522 PN

4 P=0.06522 N

5 P=0.13043 HSA

HSM

PN

HSA

HSM

N

PS

---

PS P=0.58696 0.250

N P=0.15217 0.000

HSA P=0.06522 0.000

PN P=0.06522 0.001

HSM P=0.13043 0.749

HSM

0.000

0.047

0.195

0.687

0.071

1

B) Probabilites a posteriori des sujets non classes. Classif. Observee ES *Skl4

Tableau A 22 : Arriere-crane (N=46), analyse discriminante. Test du Chi2 sur les racines. Racine

Valeur propre

R canonique

Chi'

DI

p

0.7529

Lambda de Wilks 0.1338

0

1.3086

1

1.0814

90.5188

40

8.97"-06

0.7208

0.3089

52.8692

28

2

3.06"-03

0.4124

0.5404

0.6429

19.8815

18

3.40"-01

3

0.0960

0.2960

0.9080

4.3428

10

9.31 "-01

4

0.0049

0.0695

0.9952

0.2180

4

9.94"-01

234

-------------------~A=n=n~e=x=e

Tableau A 23 : Arriere-crane, analyse A) Probabilites a priori des sujets non ou mal classes. Classif. Observee

*Fq

PN --PN

*LaF

N

*Biache 1 ES

*Ndutu *LH18 *Q6 *Q9 *Sac1 Singa *Skl4 *Skl9 *Swans *Tabun 1 *U4a

D

______________________

discriminante.

Probabilites

classification.

de

1

2

3

4

5

6

P=0.58491

P=0.11321

P=0.09434

P=0.07547

P=0.09434

P=0.03774

N

PN PN PS

PS PS PN PN HSA HSA PN PN HSM HSM HSM PN

HSA HSM

HSM HSA HSM HSA

HE HE HE HE HE HE HE HSA HE HE HE HE HE HE HE

N

HSA PS PS PS PS PS PS PS PS PS PS PS PN

HSA HSA HSM HSM PN --HSM HSM PN PN PS

N

HSM HSM HSM HSM PN PN HSA HSM PN HSM PS

N

PN HSM

N

HSM PN PN N N N N

PN HSA HSM HSA N

N N

HSA HE HSA HSA N N

HSA N

HSA

B) Probabilites a posteriori des sujets non classes. Classif. Observee

PS

N

PN

HSA

HSM

HE

P=0.58491 0.073

P=0.11321 0.796

P=0.09434 0.120

P=0.07547 0.003

P=0.09434 0.008

P=0.03774 0.000

0.230

*Fq

--PN

0.138

0.185

0.412

0.035

0.000

*LaF

N

0.613

0.164

0.158

0.016

0.050

0.000

HSA HSA HSM HSM PN --HSM HSM PN PN PS

0.533

0.001

0.077

0.135

0.254

0.000

0.793

0.009

0.029

0.069

0.099

0.000

0.556

0.008

0.142

0.003

0.291

0.000

0.821

0.001

0.012

0.000

0.166

0.000

0.683

0.041

0.175

0.002

0.099

0.000

0.815

0.018

0.105

0.011

0.051

0.000

0.883

0.013

0.024

0.050

0.029

0.000

0.771

0.000

0.018

0.016

0.194

0.000

0.428

0.147

0.322

0.007

0.096

0.000

0.622

0.006

0.089

0.017

0.266

0.000

0.206

0.044

0.520

0.041

0.189

0.000

ES

*Ndutu *LH18 *Q6 *Q9 *Sac1 Singa *Skl4 *Skl9 *Swans *Tabun 1 *U4a

Tableau A 24 : Face (N=23), analyse discriminante. Test du Chi2 sur les racines. Valeur propre

R canonique

Lambda de Wilks

Chi 2

DI

p

14.9039

0.9681

0.0166

106.5682

48

2.55"-06

1

1.2060

0.7394

0.2639

34.6376

30

2.56"-01

2

0.7178

0.6464

0.5821

14.0673

14

4.45"-01

Racine 0

235

Annexe Tableau A 25 : Face (N=37), analyse A) Probabilites a priori des sujets non ou mal classes. Classif. Observee

1 P=0.64865

--HSM

HSM PN

Kabwe *Skl4

discriminante.

Probabilites

de

2 P=0.13514 N

3 P=0.13514

4 P=0.08108

PN

HSM

N

PS PS

---

PS P=0.64865 0.000

N P=0.13514 0.421

HSM P=0.13514 0.574

PN P=0.08108 0.005

hsm

0.000

0.004

0.060

0.936

classification.

B) Probabilites a posteriori des sujets non ou mal classes. Classif. Observee Kabwe *skl4

Tableau A 26: Calvarium (N=23), analyse discriminante. Test du Chi 2 • Racine 0

Valeur propre

R canonique

56.0866

0.9912

Tableau A 27 : Calvarium (LMl), A) Probabilites a priori des sujets non classes.

Lambda de Wilks 0.0175

analyse Classif. Observee

Skl5 Tabun 1

DI

p

52.5794

16

8.93E-06

discriminante. 1 p=0.86957 N

---------

Ndutu Sac1

Chi'

Probabilites 2 p=0.13043

PS PS N PS

N

PS N

B) Probabilites a posteriori des sujets non classes. Classif. Observee Ndutu Sac1 Skl5 Tabun 1

PS P=0.86957 0

---------

236

N P=0.13043 1

0

1

1

0

0

1

de

classification.

----------------~A~nn=e=x=e

D

____________________

Annexe 7 : Analyse des allometries.

Tableau A 28 : Correlations multiples intra- et intergroupes entre taille centro1de et parametres de conformations. Separement pour flexions partielles (part non uniforme) et part uniforme. Analyse

Calotte

Tous 2roupes Neand. PS PS HSM

R molt. 0.63

Non uniforme R2 R2 p ajuste 0.19 0.4 0.05

0.3

uniforme R2 R2 ajuste 0.09 0.06

0.08 55

0.43 0.3 0.55 0.91

0.19 0.09 0.3 0.8

-0.21 0.03 0.23 0.72

0.66 0.25 0.02 0.07

-

-

-

-

-

-

-

-

27

0.41

0.17

0.1

-

-

-

20

0.5 0.55

0.25 0.3

0.16 -0.046

-

-

-

0.0001 38

0.63

0.4

0.36

0.001 38

n

Rmult

55

-

-

-

-

0.77 0.92

0.6 0.85

0.27 0.59

0.12 0.05

-

-

-

-

7 32 23 6

Tous groupes Neand. PS PS HSM

0.97

0.94

0.65

0.07

35

-

-

-

-

Tous 2roupes Calvarium Neand. PS PS HSM

0.83

0.69

0.32

0.13

Neuroviscerocrane

Face

Frontal

Arrierecrane

p

n

7 32 23 6

0.1 27

-

-

-

-

0.75

0.56

-0.68

0.89

-

-

-

-

Tous 2roupes PS PS N HSM

0.94

0.88

0.81

0.8

0.66

0.14

0.37

-

-

-

-

24 14 5 5

0.32 0.62 0.98 0.24

0.1 0.38 0.96 0.05

0.01 0.27 0.92 -0.88

0.33 0.07 0.04 0.9

Tous 2roupes PS PS N HSM

0.77

0.59

0.45

0.001

47

-

-

-

-

0.83 0.9

0.69 0.81

0.43 0.47

0.045 0.1

27 20

-

-

-

-

-

-

-

-

Tous 2roupes PS PS N HSM

0.45

0.21

0.1

0.07

55

0.03

0.001

-0.03

0.96 55

0.7 0.47

0.5 0.22

0.38 -0.05

0.005 0.57

-

-

-

-

31 24 6 5

0.44 0.41 0.88 0.89

0.19 0.17 0.79 0.8

0.14 0.09 0.64 0.6

0.048 0.14 0.09 0.19

237

-

0.08 20 0.48 7 -

24 14 5 5

31 24 6 5

Annexe

Tableau A 29 : Correlations multiples intra- et intergroupes entre taille centro'ide et parametres de conformations. Pour flexions partielles (part non uniforme) et part uniforme combinees. Analyse

Calotte

Tous 12roupes Neand. PS PS HSM

R molt 0.64

uniforme + non uniforme R2 R2 p ajuste 0.41 0.16 0.09

N 55

0.54 0.24 0.94

0.288 0.056 0.89

1.45 0.02 0.59

0.21 0.19 0.09

-

-

-

-

7 32 23 6

Tous 12roupes Neand. PS PS HSM

0.97

0.95

0.83

0.23

35

-

-

-

-

Tous 12roupes Calvarium Neand. PS PS HSM

0.86

0.73

0.3

Neuroviscerocrane

Face

Frontal

Arrierecrane

0.197 27

-

-

-

-

0.89

0.79

-0.29

0.71

-

-

-

-

Tous 12roupes PS PS N HSM

0.95

0.9

0.83

0.82

0.68

-0.067

0.59

24

-

-

-

-

5 5

Tous 12roupes PS PS N HSM

0.85

0.73

0.61

0.86 0.98

0.74 0.95

0.44 0.82

0.06 0.02

-

-

-

-

Tous 12roupes PS PS N HSM

0.49

0.24

0.11

0.1

55

0.8 0.7

0.64 0.49

0.51 0.21

0.01 0.16

-

-

-

-

31 24 6 5

238

20

0.001 38

0.00001 47 27 20

--------------------~A=n=n=e=x=e

_______________________

D

the frontal and occipital portions. In this paper, we will present analyses of the vault and the face. As a global size variable, centroid size was caculated for each specimen using GRF-ND (Slice 1994). Shape variation of the samples was assessed by Bookstein's "relative warps analysis" using Rohlfs tpsrelw software (Rohlf 1997). Relative warps are a principal components analysis of shape variation based on the procrustes distance between each specimen and the mean or consensus configuration of landmarks. The principal components of shape variation in a given population can be shown as deviations from the consensus using cartesian transformation grids by calculating an interpolation function, the thin-plate spline. Hence, as a main difference with the basic thin-plate spline analysis (Bookstein 1991, Yaroch 1994), the emphasis will be laid on within-group variation and not on the comparison of individuals. Furthermore, based on the relationship between shape components and centroid size, allometry is analysed a posteriori. Here, we used regression and correlation to examine wether centroid size has any influence on shape components. The splines for visualizing allometries by means of regression were calculated using Tpsreg (Rohlf 1997). Statistica (version 5.1) was used for statistical computations.

English summary Cranial morphology still remains of major interest for the assessment of the evolutionary status of fossils, and numerous authors have stressed the importance of separating shape differences from those related to size (Brauer 1984; Relethford 1984; Kidder et al. 1992; Aiello 1993). Recent advances in the field of morphometrics have given rise to a new approach that is summarized as "geometric morphometrics" (Bookstein 1991; Rohlf & Marcus 1993). It is claimed to be more powerfull since it takes into account the geometry of the object under study (Rohlf & Marcus 1993) and produces a pictorial result of shape differences, making the morphological interpretation of statistics easier. Within the background of the modem human origins debate, this paper tempts to improve knowledge of variation in cranial size and shape among later Pleistocene hominids from Europe, the Near East and Africa. Since opposing models on the origin of modem Homo sapiens have been for long time assessed by traditional multivariate morphometrics (e.g. Brauer 1984; Stringer 1994; Habgood & Walker 1986; Corruccini 1992), we assume that size differences between fossil specimens dominate these models, whereas shape variation is insufficiently known. Therefore, it can be expected that the application of geometric morphometrics results in a re-evaluation of the origin and dispersal of the modem cranial form.

Results Variation in size Analysis of variance (Anova) in modem human populations apparently indicates that size varies both with regard to sex and geographic origin, the latter being statistically dominant over sexual dimorphism. However, due to the relatively weak female sample size, sexual dimorphism cannot be definitely proven (tables 4.9 to 4.41).

Material and methods In this study, we concentrate on cranial affinities between European, Near Eastern and African hominids during the later Pleistocene. The main fossil sample (n=63, table 3.1) includes crania assigned to archaic Homo sapiens, "classic" Neandertals and Preneandertals, as well as to anatomically modem Homo sapiens from the Middle and Upper Paleolithic. For exploratory purpose, two Asian Homo erectus have also been included. All data were collected from original specimens, except for 6. A large representative sample of modem humans (n=210) has been used to compare fossil variation with that seen today (table 3.2).

Between-group differences of phylogenetic units and chronological subdivisions reveal the existence of 3 major assemblages in terms of vault size, with Homo erectus showing the least size, followed by Preneandertals and finally by Neandertals, archaic and anatomically modem Homo sapiens which all show statistically identical values for size (table 4.12). In terms of chronological or evolutionary trends, a global increase of size cannot be clearly inferred. Despite this absence of a generalized trend in the evolution of size, a strongly linear increase is existent among the African transitional group as well as between Pre-Wurmian and "classic" Neandertals, as is shown in figure 4.7. Therefore, size increase should not be interpreted as a specifically modem trend. Facial size unsurprisingly reveals that Neandertals show the highest values, followed by both Pre-Neandertals and the Skhul/Qafzeh group, whereas the Upper Paleolithic and Epipaleolithic show the lowest values, identical to modem variation. In terms of diachronic variation, facial size seems to undergo a global reduction (table 4.36).

A basic photogrammetric setup has been used to ascertain raw data acquisition as far as fossil specimens are concerned, and outline drawings for the modem samples (cf. FrieB 1997, 1998 for technical details). Cartesian coordinates have been recorded from the computer screen using tpsdig (Rohlf 1996). The photographic record has been tested by comparing a series of direct measurements with the same made on the screen. The mean difference between the two measurement series was less than 1 mm (FrieB 1997, 1998). This difference was statistically not significant, hence the raw data can be considered as being reliable. Cranial shape was assessed by a series of 6 to 17 coordinate pairs (depending on the state of preservation) and recorded for each specimen in the lateral view. These landmarks mainly correspond to standard craniometric points, although few type II and III landmarks (Bookstein 1991) had to be used in order to describe the cranial vault (figures 2.8 and 2.9). The morphometric analyses were performed on several fossil datasets, representing various landmark configurations and covering shape features of the vault, the face, as well as 239

Annexe Due to insufficient sample size, allometric relationships in fossil humans were estimated by calculating correlation coeffcients between centroid size and the individual factor scores of the relative warps analysis. For the same reason, detailed within-group analysis was not performed. As is summarized in tables 4.93 and 4.102, the only significant correlations between size and shape, i.e. the principal components (RWU), were found on each first axis, which reveals itself to be a moderate function of size in the case of the vault and highly dependent of size in the case of the facial skeleton. In both analyses, these correlations can be mainly attributed to the specimens dated prior to the Upper Paleolithic, that is the Neandertal lineage, and the archaic to modem Homo sapiens transition.

Variation in shape The first principal component of shape variation (49.1 %, figure 4.60) allows for a relatively clear-cut distinction between modem and what can be called archaic vault shape. Near Eastern Middle Paleolithic fossils from Skhul and Qafzeh fall into the modem range, whereas the archaic pool includes not only Neandertals but also most of the archaic modem humans and, at the extreme, the Sangiran 2 Homo erectus. The Singa and Eliye Springs archaic Homo sapiens fall roughly into the modem group. Figure 4.61 shows the thin-plate splines associated with this axis. The main shape characteristics of the archaic group lie in a relatively low and elongated vault with strong supraorbital and occipital projections. It should be recalled that in the case of the Eli ye Springs cranium, whose morphology in this regard appears to be as modem as Skhul 5 and Mladec 5, the supraorbital region is not preserved. In this study, we used the intermediate reconstruction proposed by Brauer & Leakey (1986b) who describe its morphology as definitely not modem. If one does not take into account the supraorbital portion, it can still be stated that Eliye Springs, as does Singa, shows a rather high and rounded vault according to our results. The second axis (10.9 %) basically seperates European Neandertals from their Near Eastern counterparts as well as from archaic Homo sapiens, but Omo 2 lies close to the range of European Neandertals. From what is seen in the cartesian transformation associated with this component (figure 4.61), it occurs that the main difference between these two groups is the absence of an occipital bun in the African archaic group. The subsequent axes do not reveal any significant seperation or grouping relative to commonly defined taxonomic or chronological units and thus account only for individual variation. In terms of facial shape variation, the relative warps analysis does not reveal the same affinities. In fact, as can be seen from figure 4.73, the first two axes (50.13 %) lead to a relatively distant group of archaic faces having negative factor scores on both axes, but supposedly modem human fossils, such as Qafzeh 9 and Mallaha H37 (Natufian), do show an archaic pattern of facial shape. The combined cartesian transformations of the first two axes (figures 4.74 and 4.75) reveal a strongly uniform shape change resulting in a well marked alveolar prognathism, relatively reduced alveor arch length, and a strong supraorbital torus development. The midfacial portion, as represented by the processus frontalis of the zygomatic bone, appears to retreat relatively to the supraorbital and alveolar parts.

However, as far as the cranial vault is concerned, the apparent allometric relationship is widely due to the extreme position, in size and shape of the Sangiran 2 specimen. Once this fossil is excluded from the analysis, no statistically significant allometry can be detected. In any case, it can be concluded from these results that the differences between modem and more archaic vault morphology can be explained, above all, by shape rather than by size. Conversely, facial shape shows a strong allometric component (figure 4.89), leading to pronounced archaic features when face size increases. Discussion and conclusion According to Kidder et al. (1992) it is necessary to identify the limits of variation of cranial size and shape encountered in modem populations in order to establish at what time modem cranial shape appeared during evolution. Our results indicate that the evolution of the modem cranial vault was primarily due to a shape change, whereas size increase seemed to be an unspecific trend occuring in the African transition but also in the Neandertal lineage. When vault shape, as revealed by relative warps, is related to time, there seems to be no gradual evolution from specimens like Ndutu to late archaic and anatomically modem Homo sapiens (Omo 1, Skhul and Qafzeh). The latter, however, fall inside the range seen in Upper Paleolithic cranial shape, a result that supports models of a Near Eastern origin of modem humans (Vandermeersch 1981; Mann 1995). Most of the African archaic Homo sapiens used in this study share archaic shape features with other groups such as Neandertals (sensu lato), but do not expose clearly modem cranial shape. However, they also share the same size with Neandertals, as well as with the Skhul/Qafzeh group, and it can be concluded from this that numerous preceding studies using standard multivariate statistics, i.e. without size control (Mosimann 1970), resulted in shape description that was affected by residual size. Two specimens considered as belonging to archaic Homo sapiens, Singa and Eliye Springs, revealed a position in shape space close to Upper Paleolithic fossils. We prefer to be cautious concerning this result, because these specimens might not be suitable for comparison given the pathology of Singa (Spoor et al. 1998) and the incomplete state of preservation of the supraorbital region in Eliye Springs (Brauer & Leakey 1986). As far as the Omo 1 specimen is concerned, his modem cranial shape claimed by others (Day & Stringer 1991) is supported by our results, but as long as the fossil is lacking reliable absolute dating, it cannot be

Size effects According to the statistical results from the regression of shape coefficients on the centroid size (4.91), the latter has a rather weak, although highly significant influence on shape of the face. Transformations associated with increasing size mainly consist of a horizontal extension accompanied by a more pronounced supraorbital profile as well as a weaker occipital convexity (figure 4.81). It is noteworthy that facial allometry leads, besides the pronounced supraorbital arch, to an increased alveolar prognathism and a shortened alveolar arch (figure 4.82).

240

--------------------~A=n=n=e=x=e

_______________________

considered as the best proof of any African origin model (cf. Smith 1992). Considering size only, it is interesting to note the important variation inside the Neandertal lineage and the African transitional group. "Classic" Neandertals can be easily differentiated from Preneandertals solely by their increased vault size, whereas their shape is merely the same. From the morphometric point of view, any evolutionary trend towards "neandertalization" (Condemi 1992) cannot be confirmed. Moreover, given the sexual dimorphism of size among modem humans, the same can be expected in fossil groups, so that the observed size differences inside the Neandertal lineage could be attributed to the fact that most Preneandertals used here are claimed to be females. Similar patterns of sexually related size variation cannot be ascertained for the African transitional group, where size range is biggest compared to the other groups used here. Explanations might be given by taxonomic considerations,

D

or simply by the greater time range, but cannot be definitely affirmed at this state of our research. The evolution of the face tends to show strong allometric changes of shape, in the sense that size heavily influences the expression of archaic morphological traits, i.e. strong alveolar prognathism, shortened alveolar arch length, zygomatic retreat and a heavily pronounced supraorbital torus. These features, although often considered as being a Neandertal apomorphy (Rak 1986; Trinkaus 1987), can be detected in some of the specimens that show clearly modem cranial vault shape, for example Qafzeh 9, Skhul 4 and 5 and Mallaha H37. But given the centroid size of these fossils, this morphology should be interpreted as a size dependent general plesiomorphic trait, rather than an apomorphy. This is also supported by the comparison with modem populations who reveal similar allometric shape changes in the face.

241