A MORPHOMETRIC STUDY OF CRANIAL SHAPE IN THE HOMINOIDEA PAUL O'HIGGINS Submitted in accordance with the requirements for the degree of Ph.D. Dcpartmcnt of Anatomy Univcrsity of Leeds January 1989 11 ABSTRACT' This study investigates the applicability of a number of traditional and newer methods of morphological description to the problem of defining hominoid cranial form. The aim has been to produce an objective assessmentof the relative merits of the methods from both a practical and theoretical perspective. The thesis is presented in three parts. In the first, several approacheswhich offer potential in the description of cranial morphology are reviewed and the phenetic relationships of the crania of extant hominoids are examined using data obtained by a variety of new (shape factors, least squares, and Fourier analysis) and more traditional (dimensions, angles and indices) approaches to morphological description. The analysis concentrates on a comparison of the resulting patterns of group dispositions and on an examination of the ability of the various approaches to allow an accurate determination of the affinities of crania of unknown provenance.The results indicate that it there is little difference in the patterns of phcnetic relationships obtained although appears that the analyses based upon linear and angular measurementsand Fourier coefficients provide the widest separation between the groups. The second study employs linear and angular measurements and Fourier coefficients in an examination of within-group cranial variability. In general the results lead to similar conclusions about the patterns of sexual dimorphism in extant hominoid crania and the influence of size on cranial morphology. The analyses of Fourier data differ from those of linear and angular measurements,however, in that purely size related variation is given a smaller weighting relative to morphological variation attributable to other sources. In the third part the fossil record relating to the evolution of Homo is reviewed. The third study employs these same two approaches to morphological description in a study of the patterns of cranial variation between certain fossil hominids. By contrast with the first study the pattern of phenetic relationships between OTUs appears to be considerably influenced by the choice of measurementmethod although there is a common underlying pattern of group dispositions. The reasonsfor these differences are considered in the light of the results of multivariate morphometric studies of cranial form undertaken by other workers. From these studies it is concluded that: landmark a) the results obtained by studies employing landmark dependent and data with reduced differences dependencemay differ to some degreeand that this difference is principally related to in the ways in which the various anatomical regions influence the measurements, for b) the choice of method for craniometric problems should be determined with due regard the task at hand, c) the investigator should be aware of the potential pitfalls and advantagesof each method in furnishing answers to specific questions, d) the investigator should be aware of the fact that the use of different morphological descriptions may give rise to different results. iii ACKNOWLEDGEMENTS I am grateful to the late Professor EH Ashton. It is through his guidance and assistance that this project was begun and it was with his help and encouragement that it continued and developed. His untimely death was a considerable loss of both a friend and a supervisor. I am also grateful to Professor WJ Moore who made this work possible, offered advice, encouragement,and every assistancethroughout this study. Dr RM Flinn of the University of Birmingham has contributed invaluable statistical advice and pointed the way to the application of a number of the more sophisticated methods of shape description. Dr DR Johnson has been a valuable source of advice, criticism, and most of all encouragementthroughout this work. Without the support of these three people, especially following the death of Professor Ashton, this thesis would never have been completed. A great debt of gratitude is owed to Mr TJ McAndrew whose expertise in computing, statistics and good sensewas constantly appreciated. He was tolerant of my continual pestering and without his invaluable assistancethe work would have been considerably delayed, if not abandoned. Several people have been willing to offer constructive advice at various stages during this work. My thanks go to Professor R McNeill Alexander and to the members of his Wednesdaydiscussion group in the Department of Zoology for their constructive comments whenever I have presented aspects of this work to them. Professor KV Mardia and Dr JT Kent and their students, Ian Dryden and Jackie Gough, of the Department of Statistics have always been willing to offer statistical advice and to consider the more esoteric aspects of problems relating to shape measurementwhen I have been lost or confused. Particular thanks go to Professor A Bilsborough of the University of Durham, Dr CB Stringer of the British Museum (NH), Dr AT Chamberlain of the University of Liverpool, Dr M Edley and Dr F Edwards for their considerable assistance in proof reading and in providing constructive criticism. Professor PHA Sneath of the Department of Microbiology, University of Leicester was most helpful and encouraging. He was willing to have spend time with me in order to clarify a great number of confusing issues. Several people taken time to discuss problems and explain things to me, these include Dr N Creel, Dr WL Jungers, Dr S Hartman, several members of the Department of Anatomy at SUNY, Stonybrook, Dr PE Lestrel and Professor CE Oxnard. The opportunity to speak at various departments has provided me with helpful comments and advice. I am grateful in this respect to the heads and members of the departments of Statistics, Zoology, Genetics, at this University and to the heads and membersof the departmentsof Anatomy at the Universities of Aberdeen, Liverpool and Sheffield. Miss Linda Cockroft typed the tables and some sections of this thesis, she also spent hours reformatting the documents,setting the types and generally improving the presentation,I am grateful for her constant willingness in the face of pressure. I should like to thank Mr Steve Paxton who devoted hours of his own time to helping with photocopying and the typesetting and Mr Tim Lee who drew several of the diagrams which relate to methods of shape description. I should like to thank my colleagues and friends in the Anatomy Department for constructive discussion and encouragement. Thanks must go to the trustees and staff of the British Museum (NH) and of the Powell- Cotton Museum, Birchington, Kent for allowing accessto the material in their care. To all these people I am most grateful, their help has prevented me from making a number of serious errors, any that remain are entirely my own. iv CONTENTS TABLE OF CONTENTS CHAPTER 1 CRANIAL FORM IN THE HOMINOIDEA -A GENERAL INTRODUCTION 1 CHAPTER 2 A CONSIDERATION OF THE AVAILABLE METHODS FOR THE DESCRIPTION OF CRANIAL FORM 7 CHAPTER 3 PATTERNS OF VARIATION OF CRANIAL FORM WITHIN CERTAIN GROUPS OF EXTANT HOMINOIDS 131 CHAPTER 4 PATTERNS OF CRANIAL VARIATION WITHIN AND BETWEEN CERTAIN GROUPS OF FOSSIL HOMINOIDS 209 CHAPTER 5 CRANIAL FORM IN THE HOMINOIDEA - CONCLUDING REMARKS 352 REFERENCES APPENDIX A LINEAR AND ANGULAR MEASUREMENTS FROM THE CASTS OF FOSSIL CRANIA APPENDIX B THE IDENTIFICATION OF FOSSIL GROUPS FOR CANONICAL ANALYSIS APPENDIX C UNIVARIATE BAR CHARTS LIST OF TABLES AND ILLUSTRATIVE MATERIAL V TABLE OF CONTENTS CHAPTER 1 CRANIAL FORM IN THE HOMINOIDEA A GENERAL INTRODUCTION 1 General Introduction 2 Objectives 2 Context 2 Structure of this thesis 5 CHAPTER 2 TION OF THE AVAILABLE METHODS 7 INTRODUC'T'ION 8 A REVIEW OF METHODS WHICH OFFER POTENTIAL IN THE DESCRIPTION OF CRANIAL SHAPE 11 1. 'Homology' dependent methods of shape description 11 II. The description of biological form without landmarks 19 'Shape factors' 21 Bending energy 24 Moments 26 Medial axis transforms 28 The curvature of outlines 30 34 Polar co -ordinates Fourier Analysis 37 45 Shape description - concluding remarks A STUDY TO EMPIRICALLY EXAMINE THE RELATIVE PERFORMANCE OF A NUMBER OF NEW BIOMETRIC TECHNIQUES IN THE STUDY OF CRANIAL FORM 46 INTRODUCTION 46 MATERIALS AND METHODS 48 A. Materials 48 B. Methods 51 I. Dcfinition of size 51 vi II. Definition of shape 56 a) Cranial angles and linear dimensions 57 Method of measurement 60 Precision of measurement 64 The effectsof size 64 b) Cranial indices 66 Method of calculation 66 Precision of measurement 70 Theeffects of size 70 c) The method of least squares 71 Method of measurement 72 Precision of measurement 74 Effectsof size 74 d) Normative shape factors 76 Method of measurement 77 Precision of measurement 78 The effect of size 78 e) Fourier analysis 79 Methodof measurement 81 Precision of measurementand selection of components 81 Effectsof size 84 III. Statistical methods 85 The Development and application of a suite of programs for the statistical analysis of the data 85 Univariate statistical analysis 90 Multivariate statistical analyses 90 Should the data be logged? 90 How different are the results obtained using midline data from those obtained using three dimensional data? 92 VII How different are the results obtained by different methods of cranial font': description? 93 Linear and angular dimensions 93 Angles and indices 94 Method least of squares 0 95 Shapefactors 95 Fourier analysis 95 The comparison of the results obtained from each set of data 97 RESULTS 98 A. Univariate analyses 98 B. Multivariate studies 103 1. Differences between the results obtained using midline data and those from three dimensional data 103 Linear and angular dimensions 103 Angles and indices 105 2. Differences in the results obtained by different methods of cranial form description 107 The extent of group separation and the ability to identify individuals of unknown provenance 107 Pattern of group dispositions 116 DISCUSSION 125 1.