MAXILLARY SUTURES AS AN INDICATOR OF ADULT AGE AT DEATH: REDUCING ERROR AND CODIFYING APPROACHES By CARRIE A. BROWN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF T HE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2016
2016 Carrie A. Brown
To Jacob and Isaac, for support and encouragement, but also for lots of laughs, and to Baby Wime, w h o made sure I got this done
4 ACKNOWLEDGMENTS Thanks first go to my committee, Drs. Michael Warren, David Daegling, John Krigbaum, and Lawrence Winner, for pushing me to challenge myself in new realms in this research. An additional thank you and heart felt gratitude go to my Committee Chair, Dr. Warren, for continuously supporting me and fostering my growth as a forensic committee during my time at Chico State, Drs. Eric Ba rtelink, Beth Shook, and John Byrd, thank you for setting me up for success in my doctoral program. The second round of appreciation is for all of my laboratory and academic colleagues from California to Hawaii to Florida and now in Nebraska. I truly woul d not be the anthropologist I am today without your support, encouragement, and, of course, peer reviews! Thanks especially to my frequent co researcher and fellow native Pennsylvanian, Allysha Winburn, for her endless enthusiasm and positivity, and Dr. D my year of data collection. Thank you to the following individuals for providing access to their collections and facilitating my time at them: Ms. Shirley Schermer and Ms. R obin Lillie, Stanford Collection, University of Iowa; Dr. Heather Edgar and her graduate students in the Laboratory of Human Osteology, Maxwell Documented Collection, University of New Mexico; Drs. Dawnie Steadman and Heli Maijanen, William M. Bass Donated Collection, University of Tennessee, Knoxville; Dr. David Hunt, Terry Collection, Smithsonian National Museum of Natural History; Dr. Lyman Jellema, Hamann Todd Collection, Cleveland Museum of Natural History; Dr. Yoshiharu Matsuno and Ms. Chie Koga, Chib a Documented Collection, Chiba University Medical School; and Drs. Yoshinori
5 Kawai and Yoshikatsu Negishi, Jikei Documented Collection, Department of Anatomy, Jikei University Medical School. Thank you also to Nicole and Zach Thomas, Kyle McCormick, Sean Tallman, and Greg and MaryBeth Leifer for providing a home away from home during my research across the U.S., and an additional thanks to Sean Tallman for being a surprise research partner during my time in Japan! Finally, thank you to my family, who has s et me up for success since day one. Sincere thanks to Mom and Dad, who have always encouraged and supported me no matter how far away I have been, and Amanda, who so often lent a sympathetic ear and truly understands what is going on inside my head! Many thanks Bob and Pam for putting up wi th my constant typing and paper shuffling when visiting and checking in on me from afar. These acknowledgments would not be complete without saying thank you to my husband, Jake, and my stepson, Isaac, who both kept me in good spirits throughout this process but also continually remind me what life is all about. The last thank you goes to Baby Wime, who gave me the true deadline for this dissertation. We cannot wait to meet you! This research was made possible by the William R. Maples Dissertation Award from the Department of Anthropology at the University of Florida and a student research fellowship funded by the Oak Ridge Institute for Science and Education and the Defense POW/MIA Accounting Agency (formerly the Joi nt POW/MIA Accounting Command).
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 9 LIST OF FIGURES ................................ ................................ ................................ ........ 15 ABST RACT ................................ ................................ ................................ ................... 18 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 20 Complications in Skeletal Age Estimation ................................ ............................... 20 Sutures and Age Estimation ................................ ................................ ................... 23 Biomechanical Considerations ................................ ................................ ................ 25 Research Goals ................................ ................................ ................................ ...... 28 Research Questions ................................ ................................ ............................... 29 Chapter Outline ................................ ................................ ................................ ....... 33 2 SKELETAL AGE ESTIMATION ................................ ................................ .............. 34 Age and Aging ................................ ................................ ................................ ........ 34 Skeletal Growth and Development ................................ ................................ ......... 38 Age Estimation Using Skeletal Indicators ................................ ............................... 44 Variability ................................ ................................ ................................ .......... 44 Methods ................................ ................................ ................................ ............ 48 The Statistical Basis of Age Estimation ................................ ................................ ... 55 Cranial Suture Age Estimation ................................ ................................ ................ 61 Vault Sutures ................................ ................................ ................................ .... 63 Palatal Sutures ................................ ................................ ................................ 66 Summary ................................ ................................ ................................ ................ 73 3 THE HUMAN PALATE ................................ ................................ ............................ 77 Palatal Form ................................ ................................ ................................ ........... 77 Facial Growth and Development ................................ ................................ ............. 79 Craniofacial Growth Models ................................ ................................ ............. 81 Timing of Craniofacial Gr owth ................................ ................................ .......... 84 Secular Changes in Craniofacial Growth ................................ .......................... 89 Palatal Growth and Development ................................ ................................ ........... 92 Prenatal Palatal Growth and Development ................................ ....................... 92 Postnatal Palatal Growth and Development ................................ ..................... 95 Palatal Developmental Anom alies ................................ ................................ .... 98 Palatal Function ................................ ................................ ................................ ...... 99
7 Mastication ................................ ................................ ................................ ..... 101 Altered Function throu gh Tooth Loss ................................ ............................. 106 The Role of Sutures ................................ ................................ ....................... 108 Palatal Variants ................................ ................................ ................................ ..... 115 Tran sverse Palatine Suture Shape ................................ ................................ 116 Palate Shape ................................ ................................ ................................ .. 116 Tori ................................ ................................ ................................ ................. 117 Palatine Bridging ................................ ................................ ............................ 121 Marginal Crest ................................ ................................ ................................ 122 Lesser Palatine Foramina ................................ ................................ ............... 122 Bone Quality and Porosity ................................ ................................ .............. 123 Summary ................................ ................................ ................................ .............. 123 4 MATERIALS AND METHODS ................................ ................................ .............. 134 Sa mpling Strategy ................................ ................................ ................................ 134 Documented Skeletal Collections ................................ ................................ ......... 139 Data Collection ................................ ................................ ................................ ..... 142 Qualitative Data ................................ ................................ .............................. 143 Quantitative Data ................................ ................................ ............................ 147 Data Analysis ................................ ................................ ................................ ........ 151 Hypothesis Testing ................................ ................................ ............................... 153 Age and Sutural Closure ................................ ................................ ................ 153 Group Affiliation and Sutural Closure ................................ ............................. 154 Biomechanical Variables ................................ ................................ ................ 154 Palatal Variants ................................ ................................ .............................. 155 The Full Picture ................................ ................................ .............................. 156 Error and Limitations ................................ ................................ ............................. 156 Summary ................................ ................................ ................................ .............. 158 5 RESULTS ................................ ................................ ................................ ............. 179 Palatal Suture Closure and Age ................................ ................................ ............ 179 Group Affiliation and Sutural Closure ................................ ................................ .... 184 Biomechanical Variables ................................ ................................ ....................... 186 Descriptive Statistics ................................ ................................ ...................... 186 Relationship to Sutural Fusion ................................ ................................ ........ 188 Relationship to Age ................................ ................................ ........................ 190 Relationship to Group Affiliation ................................ ................................ ..... 191 Relationship to Each Other ................................ ................................ ............. 193 Palatal Variants ................................ ................................ ................................ ..... 194 Relationship to Sutural Fusion ................................ ................................ ........ 194 Relationship to Age ................................ ................................ ........................ 195 Relationship to Group Affiliation ................................ ................................ ..... 195 Relationship to Biomechanical Variables ................................ ....................... 197 Relationship to Other Variants ................................ ................................ ........ 198 Putting it All Together ................................ ................................ ............................ 198
8 6 DISCUSSION ................................ ................................ ................................ ....... 265 Research Question 1 ................................ ................................ ............................ 265 Research Question 2 ................................ ................................ ............................ 268 Research Question 3 ................................ ................................ ............................ 271 Dent al Wear ................................ ................................ ................................ ... 271 Antemortem Tooth Loss ................................ ................................ ................. 272 Sutural Complexity ................................ ................................ ......................... 274 Biomec hanics and Fusion ................................ ................................ .............. 275 Research Question 4 ................................ ................................ ............................ 2 77 The Final Step ................................ ................................ ................................ ...... 283 7 CONCLUSIONS ................................ ................................ ................................ ... 285 APPENDIX A DATA COLLECTION WORKSHEET ................................ ................................ .... 288 B PALATAL TRAIT FREQUENCIES BY GROUP ................................ .................... 290 LIST OF REFERENCES ................................ ................................ ............................. 305 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 322
9 LIST OF TABLES Table page 2 1 A comparison of anthropological studies of age estimation based on all of the palatal sutures ................................ ................................ ................................ .... 75 2 2 Comparison of age estimates for the original and revised maxillary sutu re methods ................................ ................................ ................................ .............. 76 3 1 Timing of palatal growth and development ................................ ....................... 126 3 2 Muscles of the soft palate ................................ ................................ ................. 127 3 3 Muscles of the masticatory system. ................................ ................................ .. 128 4 1 Power analyses for sample size based on previously reported correlation ( r ) values for age and palatal suture closu re ................................ ......................... 159 4 2 Power analyses for sample sized based on desired effect size in ANOVA for group assignment and sutural closure ................................ .............................. 159 4 3 Collection summary information ................................ ................................ ....... 160 4 4 Numerical designators and descriptions for each of the 15 sections of the palatal sutures ................................ ................................ ................................ .. 161 4 5 Palatal variant scoring ................................ ................................ ...................... 163 4 6 Dental inventory categories ................................ ................................ .............. 164 4 7 Dental wear scori ng for the incisors and canines ................................ ............. 164 4 8 Dental wear scoring for the premolars ................................ .............................. 165 4 9 Dental wear scoring for the molars ................................ ................................ ... 165 4 10 Landmark definitions ................................ ................................ ........................ 166 4 11 Measurements and their landmarks ................................ ................................ 169 4 12 Order of measurements in ImageJ ................................ ................................ ... 169 4 13 Total sa mple analyzed for this research ................................ ........................... 174 4 14 Summary of dummy variables developed from nominal variables .................... 178 5 1 Results from ANOVA for suture section scores and age in the15 section/4 phase system ................................ ................................ ................................ ... 220
10 5 2 Results from ANOVA for full suture scores and age in the full suture/4 phase system ................................ ................................ ................................ .............. 220 5 3 Results from ANOVA for full suture scores and age in the full suture/binary system ................................ ................................ ................................ .............. 221 5 4 Differences in mean age by closure score for the 15 section/4 phase system 221 5 5 Differences in mean age by closure score for the full suture/4 phase system .. 222 5 6 Differences in mean age by closure score fo r the full suture/binary system ..... 222 5 7 order correlations for age and individual sutu ral closure in the 15 section/4 ph ase and 3 phase scoring systems ................................ ...... 223 5 8 order correlations for age and individual sutural closure in the full suture/4 ph ase and 3 phase scorin g systems ................................ ....... 223 5 9 order correlations for age and individual sutural closure in the fu ll suture/binary scoring system ................................ ................................ 223 5 10 order correlations for age and individual sutural closure in the control sutures ................................ ................................ ............................ 224 5 11 order correlations for age and summary score in all qualita tive, categorical systems ................................ ................................ ........ 224 5 12 Descriptive statistics for fusion ratios of the measured palatal sutures ............. 224 5 13 nk order correlations for age and sutural fusion in full suture categorical and continuous systems ................................ ................................ 229 5 14 Results from ANCOVA for 15 section/4 phase summary score and age, sex, and ancestry ................................ ................................ ................................ ..... 233 5 15 Inventory scores by tooth ................................ ................................ ................. 234 5 16 Descriptive statistics for dental wear scores by tooth ................................ ....... 234 5 17 Descriptive statistics for mean wear scores ................................ ...................... 235 5 18 Count and percentage of antemo rtem tooth loss by tooth number ................... 236 5 19 Regression results for 15 section summary fusion score and mean ant erior and posterior wear scores ................................ ................................ ................ 241 5 20 Regression results for fusion ratio sum mary score a nd mean wear scores ...... 241 5 21 Regression results for fusion ratio summary score and suture complexity ....... 243
11 5 22 Descriptive statist i cs for sex and mean wear score ................................ .......... 247 5 23 Descriptive statistics for ancestry and mean wear score ................................ .. 248 5 24 Descriptive statistics fo r time period and mean wear score .............................. 249 5 25 Group differences in mean wear score ................................ ............................. 250 5 26 Descriptive statistics for AMTL Index by group ................................ ................. 250 5 27 Descriptive statistics for suture complexity by group ................................ ........ 250 5 28 Relationsh ip of fusion and palatal traits ................................ ............................ 256 5 29 Relationship of age and palatal traits ................................ ................................ 256 5 30 square tests of palatal trait frequency for sex, ancestr y, and time period ................................ ................................ ................. 257 5 31 order correlations of mean wear and palatal variants ........... 258 5 32 order co rrelations of AMTL Index and palatal variants ......... 259 5 33 order correlations of suture complexity and palatal variants 260 5 34 square tests of bilateral palatal traits and order correlations for right and left sides per trait ................. 260 5 35 order correlati ons of palatal traits as compared to each other ................................ ................................ ................................ ................. 261 5 36 Results from multiple regression analyses for 15 section fusion summary score ................................ ................................ ................................ ................. 262 5 37 Results from multiple regression analyses for fusion ratio summary s core ...... 262 5 38 Results from multiple regression analyses for age using the 15 section summary fusion score ................................ ................................ ...................... 263 5 39 Results from multiple regression analyses for age using the fusion ratio summary score ................................ ................................ ................................ 263 5 40 Results from multiple regressi on analyses for age using the 15 section summary fusion score and no interactions ................................ ....................... 264 B 1 Left accessory lesser palatine foramina frequencies by sex ............................. 290 B 2 Right accessory lesser palatine foramina frequenci es by sex .......................... 290
12 B 3 Left marginal crest frequencies by sex ................................ ............................. 290 B 4 Right marginal crest frequencies by sex ................................ ........................... 290 B 5 Left lateral groove bridging frequencies by sex ................................ ................ 291 B 6 Right lateral groove bridging frequencies by sex ................................ .............. 291 B 7 Left medial groove bridging frequencies by sex ................................ ................ 291 B 8 Right medial groove bridging fre quencies by sex ................................ ............. 291 B 9 P alatine torus frequencies by sex ................................ ................................ ..... 292 B 10 Left maxillary torus frequencies by sex ................................ ............................. 292 B 11 Right maxillary torus frequencies by sex ................................ .......................... 292 B 12 Left maxillary exostoses frequencies by sex ................................ .................... 292 B 13 Right maxillary exostoses frequencies by sex ................................ .................. 292 B 14 Maxillary bone quality by sex ................................ ................................ ............ 293 B 15 Palatine bone q uality by sex ................................ ................................ ............. 293 B 16 Palatal porosity by sex ................................ ................................ ...................... 293 B 17 Palate shape frequencies by sex ................................ ................................ ...... 294 B 18 Transverse palatine suture shape frequencies by sex ................................ ...... 294 B 19 Left zygomaticomaxillary shape frequencies by sex ................................ ......... 294 B 20 Right zygomaticomaxillary shape frequen cies by sex ................................ ...... 294 B 21 Left accessory lesser palatine f oramina frequencies by ancestry ..................... 2 95 B 22 Right accessory lesser palatine foramina frequencies by ancestry .................. 295 B 23 Left margina l crest frequencies by ancestry ................................ ..................... 295 B 24 Right marginal crest frequencies by ancestry ................................ ................... 295 B 25 Left lateral groove bridging frequencies by ancestry ................................ ........ 296 B 26 Right lateral groove bridging frequencies by ancestry ................................ ...... 296 B 27 Left medial groove bridging frequencies by ancestry ................................ ........ 296
13 B 28 Right medial groove bridging frequencies by ancestry ................................ ..... 296 B 29 Palatin e torus frequencies by ancestry ................................ ............................. 297 B 30 Left maxillar y torus frequencies by ancestry ................................ ..................... 297 B 31 Right maxillary torus frequencies by ancestry ................................ .................. 297 B 32 Left maxillary exostoses frequencies by a ncestry ................................ ............. 297 B 33 Right maxillary exostoses frequencies by ancestry ................................ .......... 297 B 34 Max illary bone quality by ancestry ................................ ................................ .... 298 B 35 Pa latine bone quality by ancestry ................................ ................................ ..... 298 B 36 Palatal porosity by ancestry ................................ ................................ .............. 298 B 37 Palat e shape frequencies by ancestry ................................ .............................. 299 B 38 Transverse palatine sutur e shape frequencies by ancestry .............................. 299 B 39 Left zygomatic omaxilla ry shape frequencies by ancestry ................................ 299 B 40 Right zygomaticomaxillary shape frequencies by ancestry .............................. 299 B 41 Left accessory lesser palatine foramina frequencies b y time period ................. 300 B 42 Right accessory lesser palatine foramina frequencies by time period .............. 300 B 43 Left marginal crest frequencies by time period ................................ ................. 300 B 44 Right marginal crest frequencies by time period ................................ ............... 300 B 45 Left late ral groove bridging frequencies by time period ................................ .... 301 B 46 Right lateral groove bridging frequencies by time period ................................ .. 301 B 47 Left med ial groove bridging frequencies by time period ................................ ... 301 B 48 Right medial groove bridging frequencies by time period ................................ 301 B 49 Palatine torus frequencies by time period ................................ ......................... 302 B 50 Left maxillary torus frequencies by time period ................................ ................ 302 B 51 Right maxillary torus frequenci es by time period ................................ .............. 302 B 52 Left maxillary exostoses frequencies by time period ................................ ........ 302
14 B 53 Right maxillary exost oses frequencies by t ime period ................................ ...... 302 B 54 Maxillary bone quality by time period ................................ ................................ 303 B 55 Palatine bone quality by time period ................................ ................................ 303 B 56 Palatal porosity by time period ................................ ................................ .......... 303 B 57 Palate s hape frequencies by time period ................................ .......................... 304 B 58 Transverse palatine suture s hape frequencies by time period ......................... 304 B 59 Left zygomaticomaxillary shape frequencies by time period ............................. 304 B 60 Right zygomaticomaxillary shape frequencies by time period .......................... 304
15 LIST OF FIGURES Figure page 3 1 Diagram of the skeletal elements of the ha rd palate and its sutures ................ 124 3 2 Diagrammatic representat ion of the V principle of growth ................................ 125 3 3 Transverse palatine suture shap e ................................ ................................ .... 129 3 4 Palate shape ................................ ................................ ................................ ..... 130 3 5 Tori ................................ ................................ ................................ ................... 131 3 6 Palatine bridging ................................ ................................ ............................... 132 3 7 Marginal crest ................................ ................................ ................................ ... 133 4 1 Palate divided into 15 sections ................................ ................................ ......... 161 4 2 P alate scoring based on the entirety of each palatal suture ............................. 162 4 3 Palatal landmarks used in this study ................................ ................................ 166 4 4 Digital measuremen t set up ................................ ................................ .............. 167 4 5 Setting the scale in ImageJ ................................ ................................ .............. 167 4 6 Example of brightness and contrast adjustment in ImageJ .............................. 168 4 7 Measurements of the AMP suture in ImageJ ................................ .................... 170 4 8 Measureme nts of the PMP suture in ImageJ ................................ .................... 171 4 9 Measurements of the T P suture, right side, in ImageJ ................................ ..... 172 4 10 Measuring sutura l fusion in ImageJ ................................ ................................ .. 173 4 11 Age di stribution of sample ................................ ................................ ................ 175 4 12 Sample distribution by age group and sex ................................ ........................ 176 4 13 Sample distribution by age group and ancestry ................................ ................ 177 5 1 Box and whisker plots of age per closure score for sections of the IN suture, 15 section/4 phase system ................................ ................................ ............... 204 5 2 Box and whiske r plots of age per closure score for sections of the PMP suture, 15 section/4 phase system ................................ ................................ ... 205
16 5 3 Box and whisker plots of age per closure score for sections of the TP suture, 15 section/4 pha se system ................................ ................................ ............... 206 5 4 Box and whisker plots of age per closure score for sections of the AMP suture, 15 section/4 phase system ................................ ................................ ... 207 5 5 Box and whisker plots of age per closure score for the full suture/4 phase system ................................ ................................ ................................ .............. 208 5 6 Comparison of the box and whisker plots of age per closure score for the right and left sides of the T P suture within and outside the GPF, full suture/4 phase system ................................ ................................ ................................ ... 209 5 7 Box and whisker plots of age per closure score for the full suture/binary system ................................ ................................ ................................ .............. 210 5 8 Comparison of the box and whisker plots of age per closure score for the right and left sides of the TP suture within and outside the G PF, full suture/binary system ................................ ................................ ........................ 211 5 9 Box and whisker plot of age per summary score for the 15 section/4 phase system ................................ ................................ ................................ .............. 212 5 10 Box and whisker plot of age per summary score for the full suture/4 phase system not includin g the TP suture within t he GPF ................................ .......... 213 5 11 Box and whisker plot of age per summary score for the full suture/4 phase system includi ng the TP suture within the GPF ................................ ................ 214 5 12 Box and whisker plot of age per summary score for the full suture/binary system not includi ng the TP suture within the GPF ................................ .......... 215 5 13 Box and whisker plot o f age per summary score for the full suture/binary system includi ng the TP suture within the GPF ................................ ................ 216 5 14 Box and whisker plots of age per closure score for the control sutures, full suture/4 p hase system ................................ ................................ ..................... 217 5 15 Distributions of summary scores, all categorical systems ................................ 218 5 16 Scatterplot for age and summary score in th e 15 section/4 phase system ...... 219 5 17 Scatterplots of age and fusion ratio for each of the measured palatal sutures 225 5 18 Scatte rplot of age and fusion ratio summary score ................................ ........... 226 5 19 Distrib utions of fusion ratio scores ................................ ................................ .... 227 5 20 Scatterplot of age and arcsine transformed fusion ratio summary score .......... 228
17 5 21 Summary score by sex ................................ ................................ ..................... 230 5 22 Summ ary score by ancestry ................................ ................................ ............. 231 5 23 Summary score by time p eriod ................................ ................................ ......... 232 5 24 Distri butions of mean wear scores ................................ ................................ .... 235 5 25 Fre quency distribution of AMTL index in the total sample ................................ 237 5 26 Frequency distributions of suture complexity ratios in the total sample ............ 23 8 5 27 Scatterpots of fusion summary and mean wear scores ................................ .... 239 5 28 Scatterpots of fusion summary and mean posterior wear scores ..................... 240 5 29 Scatterpots of fusion summary and mean anterior wear scores ....................... 241 5 30 Scatterplots of summ ary fusion scores and AMTL Index ................................ .. 242 5 31 Box and whisker plots of suture fusion by edentulous statu s ........................... 242 5 32 Scatterplots of individual suture fusion ratios and suture complexity ................ 243 5 33 Scatterplots of age and mean wear scores ................................ ...................... 244 5 34 Scatterplot of AMTL index and age ................................ ................................ .. 245 5 35 Scat terplots of individual suture complexity and age ................................ ........ 246 5 36 Box and whisker plots of mean wear scores by sex ................................ ......... 247 5 37 Box and whiske r plots of mean wear scores by ancestry ................................ 248 5 38 Box and whisker plots of mean wear scores by time period ............................. 249 5 39 Scatterplots of A MTL Index and mean molar wear ................................ ........... 251 5 40 Scatterplots of suture complexity and mean wear score ................................ .. 252 5 41 Scatterplots of suture com plexity and mean posterior wear score .................... 253 5 42 Scatterplots of suture complexity and mean anterior wear score ..................... 254 5 43 Scatterplot s of suture complexity and AMTL Index ................................ ........... 255 A 1 D ata collection worksheet page 1 ................................ ................................ ..... 288 A 2 D ata collection worksheet page 2 ................................ ................................ ..... 289
18 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy MAXILLARY SUTURES AS AN INDICATOR OF ADULT AGE AT DEATH: REDUCING ERROR AND CODIFYING APPROACHES By Carrie A. Brown May 2016 Chair: Michael W. Warren Major: Anthropology This research examines the palate in terms of sutural fusion and age. The study sample is drawn from documented modern and historic U.S. and Japanese skelet al c ollections ( n = 762 individuals ) The research design employed stratified random sampling in order to balance age, sex, and ancestry. Sutural fusion was examined via ordinal scoring systems and quantification through di gital measurement ; sutural complexity for each of the measured palatal sutures was also calculated Summary fusion scores were compared to age, demographic groups, biomechanical proxy variables wear, too th loss, and sutural complexity and palatal vari ants. Results indicate that sutural fusion in terms of age i s best summarized using a 1 5 section/4 phase summary score M easurement of the sutures captures less variation in terms of age than a summary score based on ordinal scoring, like ly due to the ina bility to measure the incisive suture Palatal suture fusion is greater in males and individuals of African ancestry than females and individuals of Asian or European ancestry ; As ian individuals hav e the lowest average fusion No secular trends in palata l suture fusion are noted
19 Fusion has positive association s with wear and tooth loss, and a negative association with sutural complexity. Complexity show s no association with wear, tooth loss, or age ; wear and tooth loss are positively associated with a ge and each other. This indicate s that sutural complexity affects fusion, but that sutural complex ity is not influenced by age, wear, or tooth loss. Simpler sutures show a tendency to fuse before more complex ones, even across the sexes and ancestral gro ups. Very few palatal variants are associated with fusion or age, though almost all show significantly different frequencies by ancestry. Multiple regression analyses for fusion and age indicate that age, sex, sutural complexity, palatine torus expression and their interactions significantly affect fusion, while fusion, sex, ancestry, tooth loss, maxillary tori and exostoses, and their interactions affect the prediction of age from skeletal remains For both models, 50% of the variation in the response v ariable is explained by variation in the predictor variables. A simplified multiple regression model indicate s that age can be predicted with a standard error of 14.6 years
20 CHAPTER 1 INTRODUCTION Age estimation based on skeletal indicators is an impo rtant component of bioarchaeological, paleodemographical, and forensic anthropological analyses It provides key information in describing demographic compositions of human groups, investigating the health of past populations, and contributing to the iden tification of missing persons. In order to be useful for these purposes, age estimates should be close to actual ages at death (accurate) and have small ranges (precise) To estimate age, the biological anthropologist compares the skeletal and dental ele ments of an individual with relevant reference standards. The most commonly used methods rely on growth, development, and macroscopic bony changes in non moveable joints, though microscopic methods are also available and anthropologists often consider the extent of age related pathology and bone quality when estimating age (e.g., osteophytosis, osteopenia, and osteoporosis) Age estimation methods rely on a correlation between chronological and biological age the actual amount of time a person has been and degeneration observed in the skeleton, respectively. Complications in Skeletal Age Estimation Skeletal age estimation is problematic due to an imperfect correlation between chronological a nd biological age, one that decreases further with advanced age (Christensen et al. 2014) This tendency to have increased inaccuracy for age estimates of older individuals is known as the trajectory effect (Nawrocki 2010) With increased age, there is a coincident increase in the dispersion of values from a least squares regression line that relates age at death to a skeletal indicator or method,
21 producing a cone shape that opens to the right. Since g rowth and developmental processes are well understo od, relatively unifo rm across human populations, largely genetically controlled, and changes occur reg ularly and at small intervals, age estimation from the skeletal and dental remains of individuals under the age of 18 years is both accurate and precise. However, o nce growth and development is complete, age estimation must rely on processes of skeletal degeneration, which are far more variable and less well understood resulting in larger age intervals. Larger age intervals can lead to decreased utility of age as a variable in anthropological studies. Variation in the aging process is generally related to inter individual variation. status contribute to the aging process, ances try (geographic origin of his or her ancestors) and sex (Kemkes Grottenthaler 2002) Other factors include individual rates of senescence, mechanical loading patterns, and developmental asymmetry (Kemke s Grottenthaler 2002) It is very difficult to segregate these variables and their effects since there is a complex and poorly understood interaction between genes and environment in terms of aging (Scheuer and Black 2000) The longer a person lives, t he more time he or she is subjected to myriad environmental factors environmental factors may be affected by his or her genetic make up. An additional effect on the estimation of age at death is secular change o ver time. Secular change refers to a non evolutionary short or long term trend; when used in biological anthropology, it commonly refers to a change in skeletal dimensions or shape s or the rate s of maturation over time. In the United States, secular tre nds in
22 height, cranial size and shape, and maturational times are observed between the 19 th and 20 th centuries, with modern Americans generally being taller, having taller and longer cranial vaults but narrower vaults and faces, and maturing earlier than t heir historic counterparts (e.g., Angel 1976; Jantz and Jantz 1999; Jantz and Meadows Jantz 2000; Langley Shirley and Jantz 2010) Thus methods that are developed on reference samples from a certain time period may potentially introduce bias when used for individuals from a different era. This is one of the issues underlying the problem of age mimicry, in which the underl sample artificially imposes this structure on samples aged with the reference met hod (Bocquet Appel and Masset 1982) Variation in the aging process and imperfect methods mean that continued research in skeletal age estimation is necessary in order improve adult age estimation methods. This research entails collecting data on docum ented individuals while also carefully considering the demographic composition of the research sample to make it as balanced as possible (Bocquet Appel and Masset 1982; Hoppa and Vaupel 2002a) It is important to note that skeletal collections may not a ccurately represent the populations from which they are drawn because the sample is most often biase d in terms of sex, ancestry, age composition and socioeconomic status (Hunt and Albanese 2005; Komar and Buikstra 2008; Komar and Grivas 2008) The dem ographic information available for most documented skeletal collections commonly includes sex, ancestry, stature, birth and death dates, from which age at death and time period the individual lived can be calculated. Other variables, such as individual he alth and socioeconomic status nutrition al intake developmental asymmetry, loading
23 patterns, and populational rates of senescence are not variables that are typically recorded and are more challenging to investigate. One solution is to collect additional data that can serve to better elucidate these factors, even in the absence of specific documentation. Examples of this include: linear enamel hypoplasias for health, dental wear for mechanical loading, and observations from both left and right sides to investigate asymmetry. The fundamentals of the aging process and skeletal age estimation are discussed in greater detail in Chapter 2. Sutures and Age Estimation One of the earliest methods developed to estimate age for adult skeletal remains is cranial s uture obliteration. C ranial suture age estimation is based on the age progressive synostosis of cranial sutures with older individuals exhibiting greater amounts of obliteration as compared to younger individuals, who exhibit largely patent sutures (Broo ks 1955; Meindl and Lovejoy 1985) Though cranial suture age estimation is used as early as the 1500s, it is codified as a method employing the vault sutures in a documented skeletal collection in the 1920s (Todd and Lyon 1924; Todd and Lyon 1925a; To dd and Lyon 1925b; Todd and Lyon 1925c ; Ashley Montagu, 1938 ) Since that time, it has been one of the more controversial methods of age estimation due to concerns about accuracy, precision, and variables other than age affecting sutural fusion (Ashley Montagu 1938; Brooks 1955; Hershkovitz et al. 1997; Kroman and Thompson 2009) While the cranial sutures are not typically the first choice for skeletal age estimation, the cranial sutures are still used to estimate age, especially considering the sku ll is the most commonly recognized human element and thus encountered frequently by the biological anthropologist ( Nawrocki, 1998; Garvin and Passalacqua 2012 )
24 Interest in improving cranial suture age estimation has extended examination of the sutures to other areas of the cranium (e.g., facial skeleton and hard palate). In the United States, a ge estimation using the sutures of the hard palate has largely followed the work of Mann and colleagues (1987 ; 19 91 ) but has not been widely employed in skeletal age estimation (Brown 2009) This may be related to general skepticism concerning the validity of suture closure as an indi cator of age or more specifically to problems with applying the palatal method, such as the lack of a standardized protocol and co nfusion with the Mann et al. (1991) system (Brown 2010) The Mann et al. (1991) maxillary suture age method estimates age with rates of inaccuracy and bias that are similar to methods using the vault sutures, sternal ends of the fourth rib, and the pubic symphysis ( Ginter, 2005; Brown 2009) Differing scoring methods and statistical procedures employed in numerous cranial suture closure methods only serve to further obfuscate the u se of suture closure as a means of estimating age at death. Previous stud ies have been temporally and geographically limited (e.g., Mann et al. 1987 ; Mann et al. 1991 ; Ginter 2005; Sakaue and Adachi 2007 ; Beauthier et al. 2010 ; Apostolidou et al. 2011 ) ; there has been no comprehensive, large scale project to examine palat al suture closure. Additional concerns include the unbalanced study samples employed in the reference method ( i.e., unequal representatio n of age/sex/ancestral groups) and a paucity of research investigating how variables other than age affect cl osure of the palatal sutures. Due to the delayed closure of facial sutures as compared to cranial vault sutures, palatal sutures may be useful in estimating age for older adults (Persson and Thilander 1977; Wang et al. 2006) Given the trajectory effect, method s that can provide age
25 estimates in the notoriously difficult period of advanced age are particularly useful Though developed on American individuals of African and European ancestry, the maxillary suture method has also been shown to be useful for estim ating minimum age at death in Japanese individuals (Sakaue and Adachi 2007) Thus age estimation from the maxillary sutures could potentially be useful in biological anthropology, provided that certain limitations are addressed such as: e xamination of sutural closure in a larger, more diverse, and balanced sample; consideration of factors besides age that affect palatal suture closure (e.g., ancestry, sex, secular trends, masticatory forces, or diet); and standardization of the method to enable its use by practitioners. Cranial and maxillary suture age estimation is discussed in greater detail in Chapter 2. Biomechanical Considerations The primary role of the palate i s to support mastication In fact, chewing related activities account for the largest routine loads on the skull as a whole (Rogers 1984) Because of the regular nature of feeding activity, the effect of mastication on palatal structure and sutural fusion should be considered when examining the relationship of sutural closure and age. Ho wever, like the imperfect correlation of skeletal age and chronological age, the form of the palate is not entirely correlated to loading and its biomechanical environment is challenging to characterize (Hotzman 2010) While it is useful to understand h ow function affects form, there are several issues with inferring function from form in the palate. Like many skeletal structures, the facial skeleton is constrained by growth, geometry, and material make up, and its form does not represent optimization o f the structure for solely feeding purposes ( Thomas and Reif, 1993; Hylander and Johnson 1997 ) The skull is a complex structure composed of skeletal elements of varying thicknesses and shape s, joined by fibrous
26 sutures, and it does not exhibit an overal l pattern of deformation ( Herring and Teng, 2000; Herring and Ochareon, 2005 ) This means that unlike long bones and limbs, it is difficult to model the cranial biomechanical environment as a single structure, and isolated, regional approaches are recomme nded ( Herring et al., 2001; Herring and Ochareon 2005 ) Models are particularly beneficial in instances where experimentally obtained values are not possible or very difficult to obtain, such as in living human subjects ( Koolstra, 2002; Herring and Ochar eon 2005; Koc et a l., 2010 ) but they are imperfect representations of a complex and dynamic system. For example, modeling the palate as a shell, plate, and beam did not produce strain values comparable to those that were experimentally obtained (Hotzman 2010). Contributing further to the interpretive challenge of facial function and form are the masticatory muscles. These muscles are heterogeneous, architecturally complex, and mechanically redundant, and attach to cranial bone through aponeuroses with varying orientations rather than connecting directly to the bone ( Koolstra, 2002; Herring 2007 ) Occlusal forces from chewing and incising are transmitted to the alveolar bone not by direct muscle attachments, but via the periodontal ligament, which medi ates force from the teeth to the skeletal structure. The only major joint of the masticatory system is the temporomandibular joint, and this joint is not located on the maxillae or palatines nor is it an efficient lever (Koolstra 2002) Additionally, in tra individual variation in musculature and overall bite force contributes to difficulties in characterizing the mechanical environment of the palate and tooth loss can drastically alter the loads experience by the masticatory complex (Koc et al. 2010)
27 D espite these complications, facial morphology still reflect s jaw movement, and the overall cranial strain pattern during mastication is shear or torsion (Daegling and Hylander 1997; Herring 2007) Generally, individuals who are subjected to greater mast icatory loads have more robust muscles and skeletal structures than those who experience decreased loading during chewing. The skeletal components of the masticatory system are not static and can respond to altered function through bone remodeling. For e xample, in rats fed a soft diet, which reduces loading of the masticatory system, decreased growth rate and change in facial morphology is noted (Kiliaridis et al. 1985) American secular trends of decreased facial width and increased facial height can a lso be attributed at least in part to dietary change following the introduction of processed foods in the 1950s (Jantz and Meadows Jantz 2000; Wescott and Jantz 2005 ; Skorpinski, 2014 ) In evolutionary terms, the introduction of the secondary palate to the mammalian skull increases strength and stiffness during bending and torsion of the maxilla, as demonstrated in experiments on American opossum ( Dibelphis virginiana ) ( Thomason and Russell 1986) Sutures, the fibrous joints of the cranium, are also a n important component of the masticatory system as they allow for both strength and flexibility. Sutures have their own strain environment that differs from the cranium, largely due to differences in their material make up. Like the craniofacial elements sutures have regionally specific loading regimes, and also lend themselves to regional approaches. The local loading environment also contributes to sutural morphology, with highly interdigitated (complex) sutures having greater energy absorption potent ial that straight (simple) sutures due to increased surface area along the sutural margin (Jaslow 1990). Growth, loading
28 environment, and sutural morphology are related, necessitating an understanding of all three to investigate potential effects on mast icatory form but also sutural closure. Palatal growth, development, form, and function is discuss ed in greater detail in Chapter 3. Research Goals This research aims to refine adult age estimation based on the hard palate through the analysis of the fusio n of its sutures and morphological traits that may relate to age or suture closure and by providing a comprehensive, codified, and standardized approach to maxillary suture age estimation. It also attempts to relate the functional role of the palate to v ariation seen in fusion patterns and rates. In order to do so, maxillary suture age estimation methods developed to date are investigated as are the statistical methods used in cranial suture age estimation, the growth and development of the human palate and its sutures, and the effects of function on this area of the skeleton. This research is certainly not the first in any of these areas, but it organizes approximately two decades of work pertaining to age estimation using the maxillary sutures, provide s a comprehensive source for age estimation of the palate, incorporates a greater diversity of skeletal collections than has been previously employed, and accounts for potential covariates with palatal suture closure beyond age Given the small and tempora lly and geographically disjointed samples used in previous research, this research increases the sample size and diversity for the palatal age estimation method with data from multiple documented skeletal collections in the United States and Japan. The sa mple represents individuals from modern and historic time periods from all three major ancestral groups (African, Asian, and European), both sexes, and balanced age ca tegories Data are collected on the fusion of individual
29 palatal and facial sutures and sections of palatal sutures, and three ordinal scoring methods are tested along with the quantification of sutural fusion. Additional data on palatal variants and d ata that serve as proxies for biomechanical forces, such as measures of sutural complexity and shape, antemortem tooth loss, and dental wear are also collected. T he inherent interrelatedness of cranial traits and the inability to examine any trait in isolation requires the consideration of as many factors as possible to better elucidate the rel ationship, if present, of palatal suture closure and age. Research Questions 1. What is the relationship between maxillary suture closure and age? It is expected that there will be some degree of correlation between maxillary sutural fusion and age since th is has been demonstrated in previous studies of the maxillary sutures as an isolated system and when used in conjunction with cranial vault sutures ( Meindl and Lovejoy 1985 ; Mann et al. 1987 ; Mann et al. 1991; Nawrocki 1998) The strength of the relat ionship may vary depending on the number and/or length of locations scored and the way those locations are analyzed ( e.g., as categorical ordinal interval or ratio level variables ). Wheatley (1996) found a lower correlation of sutures with age than Nawrocki (1998) but the former scored only 1 cm sections, while the latter scored the entire suture length. Age estimation using the palatal sutures has largely incorporated ordinal level variables with limited use of interval or ratio level data. In addition to scoring the four palatal sutures, two facial control sutures (nasofrontal and zygomaticomaxillary) are scored in order to examine the relationship between age and closure in oth er areas of the facial skeleton: one that is subjected to similar masticatory loading (zygomaticomaxillary) and one that
30 undergoes less loading during mastication (nasofrontal) ( Rogers 1984; Wroe et al. 2007 ). 2. How d oes group affiliation influence maxillary suture closure? Group affiliation includes sex, ancestry, and t ime period (historic or modern). Previous studies have found differences in suture fusion between males and females, with males undergoing sutural fusion earlier than females, although females exhibit a more regular fusion tempo (Ashley Montagu 1938; Man n et al. 1991; Nawrocki 1998) Ancestral differences are less clear, with some studies finding no difference in cranial suture fusion between European and African Americans ( Meindl and Lovejoy 1985 ; Mann et al., 1991 ) while others find significant dif ferences between these groups (Galera et al. 1998; Nawrocki 1998) Testing the Mann revised method in a Japanese sample, Sakaue and Adachi (2007) did not find that it performed well, though it is useful for providing a minimum age. Likewise, secular tr ends are observed variably in suture closure, with Masset (1989) reporting earlier suture closure in modern individuals and Nawrocki (1998) reporting later closure for modern individuals. Zambrano (2005) does not find any systematic secular trends between historic and modern individuals for vault sutures using the equations provided by Nawrocki (1998). Because of the results of previous studies, it is unclear what differences, if any, might be present among groups in terms of palatal sutural fusion. 3. How do biomechanical factors influence maxillary suture closure? In this research, the main biomechanical variable investigated is bite force, though other palatal variants that may affect the biomechanical environment of the palate are explored (see Research Question 4, below). As less masticatory loading occurs, sutures have the potential to exhibit osseous bridging and/or closure when
31 sutural margins approach. Based on previous experimental studies that have found altered craniofacial form and increased f usion alongside decreased bite force (EngstrÂšm et al. 1986; Hinton 1988; Wheatley, 1996; Skorpinski 2014) it is expected that sutural fusion will be seen more commonly in individuals with decreased bite force as measured by lower rates of dental wear a nd higher rates of antemortem tooth loss ( AMTL ) However, there is the potential that extreme dental wear that results in dentine exposure could lead to dental disease and tooth loss, so the relationship of dental wear and AMTL is also examined. Sutural c omplexity can also be related to bite force and sutural fusion. Simpler sutures are more likely to be seen where loading is predominantly tensile, while more complex sutures are generally indicative of a compressive loading environment ( Rafferty and Herri ng, 1999; Herring and Ochareon, 2005 ) Tensile sutures may resist fusion to some extent because of continuous growth at margins brought about by tensile forces. However, increased sutural complexity may reflect an adaptation to greater loading environmen ts, resulting in similar strain regardless of observed sutural complexity. If this is the case, there should be no relationship of dental wear and sutural complexity, as sutures have adapted to account for greater loads. In terms of fusion, complexity ma y interact with force, producing results that suggest that there is less fusion in individuals with greater bite force, as measured by higher rates of dental wear and little to no AMTL. Fusion may also be more related to sutural morphology than force sinc e diet has also been shown to be unrelated to midpalatal sutural complexity in certain primate species (Hotzman 2004)
32 4. What are the relationships of palatal variants to age, demographic group, and palatal biomechanics? Palatal variants have some degree o ancestry, or age (Hauser and De Stefano 1989) Because of this, they are often used in forensic anthropology, bioarchaeology, and paleodemography to support group membership. The relationship of palatal varian ts to biomechanical factors and palatal sutural fusion, however, is not currently well understood. For example, it has been hypothesized that the presence of a raised area of bone along the midpalatal suture ( torus palatinus ) is related to increased masti catory stresses resulting in the need for buttressing along the center portion of the palate (Hooton 1946) This relationship is not universally agreed upon, with other researchers attributing expression to genes or a combination of genetic pre programmi ng and environmental stressors ( Woo, 1950; Hassett 2006) Additionally, multiple other palatal traits have been studied with only limited investigation of their relationship to masticatory forces or fusion of the palatal sutures. These traits include: transverse palatine suture and overall palate shape, palatine bridging, crista marginalis lesser palatine foramina, and more generally, overall bone quality and porosity. This research investigates the frequencies of palatal variants across three ancestr al groups and in both sexes, and the relationships of variants to age and certain biomechanical variables as discussed in Research Question 3, above. If a variant has demographic classificatory power, it is expected that it will occur with greater frequen cy in a certain group over another. If a variant is related to masticatory function, it is expected that it will be related to dental wear, sutural complexity, and/or AMTL, if those variables are shown to reflect masticatory stresses.
33 Chapter O utline This dissertation is composed of seven chapters. Chapter 1 provides an introduction to the research and outlines research questions. Chapter 2 is an overview of the aging process and skeletal age estimation, with particular focus on the use of sutures to est imate age. Chapter 3 presents the growth, development, and function of the human palate, and ways in which palatal traits are employed for group assignment. Chapter 4 describes the materials and methods used in this research, and Chapter 5 presents resul ts. Chapter 6 discusses these results and their implications, and Chapter 7 summarizes this research.
34 CHAPTER 2 SKELETAL AGE ESTIMATION The estimation of age at death from skeletal remains is an important area of research in biological anthropology beca use it directly contributes to knowledge about human variation; however, it remains particularly challen ging for adult individuals Skeletal age at death estimation is employed in the sub fields of forensic anthropology, paleodemography, and bioarchaeolog y to develop age profiles of individuals or groups. These age profiles contribute to the identification of skeletal remains, the understanding of past diseases through mortality profiles and life expectancies, and the reconstruction of past lifeways and d emographic structures. In order to understand how skeletal biologists conduct age estimation and the inherent challenges in doing so, the aging process is outlined, followed by skeletal growth and development, skeletal age estimation, and the statistics u sed in estimating age at death from skeletal remains. The final section of this chapter deals exclusively with age estimation using the cranial sutures. Age and Aging Age is the length of time an organism has existed, and it can be described chronological ly or biologically. Chronological age is measured by the time a person has lived since his or her day of birth (e.g., years, months days) while biological age is the stage of physiological development of an individual irrespective of how long he or she (Kirkwood and Austad 2000: p 233) but it can broadly be defined as the process of getting ol der.
35 Aging is dictated by internal (genetic) and external (environmental [e.g., health, disease, trauma]) variables, which leads to great interpersonal heterogeneity in the environmental variables throughout his or her life, and the dissimilar genetic make ups among individuals also contribute to variation. Additionally, the interaction between genes and environment is complex, and it is difficult to isolate single causal factors to explain inter individual differences (Scheuer and Black 2000) Aging als o depends on a combination of genes unique to the individual (private) and genes shared among groups of individuals (public) (Kirkwood and Austad 2000) The human life cycle is comprised of multiple progressive stages, beginning with fertilization, and continuing through prenatal life, birth, postnatal life (infant, child, juvenile, adolescent, adult), maturity, senescence, and ending with death (Bogin 1999) Within the human life cycle, the process of aging is generally divided into two distinct stage s: growth/development and degeneration. A third stage, maintenance, may be added on the continuum between growth and development and degeneration, but no age specific processes are seen in this stage (Garvin et al. 2012) Maintenance is the general upk eep of the mature organism on a day to day basis (i.e., maintaining stasis), and this is generally following growth and development when the organism or element undergoes no size or shape changes. During maintenance, the processes of repair and deteriorat ion are balanced. Growth describes a general increase in size and subse quent changes in shape and form (Scheuer and Black 2000) Development involves a process of differentiation at cellular levels that leads to a more specialized and mature state (Bog in 1999) The end result of growth and development is the adult form of a n organism or element of
36 that organism (Enlow and Hans 1996) The processes of growth and development are tightly controlled genetically and generally less affected by extrinsic f actors like environment and disease. Even in cases where growth is temporarily halted, children a normal growth pattern once the insult has been removed (Prader et al. 1963; Tanner 1963) An exc eption to this is a disturbance that occurs during key de velopmental stages, which a ffect s only the sites or processes active at that moment in time, but could cause adverse consequences for the organism as a whole depending on the severity of the insult a nd the sites or processes affected (Brodie 1941) Additionally, children with access to good nutrition will display accelerated growth when compared to similar age cohorts with poor nutrition (Ferembach et al. 1980) Degeneration involves the breakdow n of structures and functions in an organism. It is highly variable, and its mechanisms are poorly understood (Crews 1993) Senescence (the process of getting older) is not simply the end result of the degeneration of biological systems through accumula ted effects on t he organism. This process is complex, and the exact contributions of genes and environment are not yet clearly elucidated (Kenyon 2010; Vaupel 2010) While multiple genes influence senescence, specific genes programmed to increase longe vity likely do not exist and environment certainly plays a key role (Kirkwood and Austad 2000) An interesting paradox emerges with increasing age. Individuals who are able to survive to very old ages may in fact appear biologically younger than indivi duals who did not survive to very old ages (Angel 1984) Schmitt (2002) explains that individuals who live to a very old chronological age are likely people who experience a slower
37 progression of biological age and thus have an increased life expectancy compared to those who undergo a faster progression of biological age. This results in a lag between biological and chronological age. Whether it is genes, environment, or an interaction of the two, something enables a protracted survival, and the key to longevity may in fact be that the organism does not undergo degeneration at the same rate as organisms with decreased relative lifespans. Degenerative processes may display greater variation among individuals not simply because they are more variable than the processes of growth and development, but also because senescence may depend on individual circumstances, such as health and prosperity. An exception to this is for individuals of extreme advanced age (e.g., 100+), who appear to undergo degeneration at the same rate (Vaupel 2010) Evolutionary Theories of Aging : The mutation accumulation, antagonistic pleiotropy, and disposable soma evolutionary theories of aging rely on this basic underlying principle: the force of selection is progressively weaken ed with increasing age (Crews 1993; Kirkwood and Austad 2000) This means that the longer an individual organism has been alive, the less effect natural selection will have. The difference between mutation accumulation and antagonistic pleiotropy is wh ether deleterious genes simply accumulate over time via mutation due to a progressive reduction in the force of selection in older individuals (mutation accumulation; Medawar 1952) or if aging is brought about by genes that are beneficial early in life bu t carry deleterious late life changes (antagonistic pleiotropy; Williams 1957) The disposable soma theory posits that the organism can only allocate so many resources towards certain functions and that some organisms will preferentially allocate to repr oduction at
38 the cost of maintenance and repair, eventually leading to senescence (Kirkwood 1977) The main problem with all three of these concepts is that aging itself can represent an adaptation, and as such is not just a side effect of selection for t raits that maximize reproductive potential (Mitteldorf 2004) Humans present a special case since they not only live longer but also have an extended post reproductive lifespan. In a strict Darwinian sense, life past reproduction is of no utility, yet h umans continue to live well beyond reproductive years. The explanation for longevity following the reproductive years is often found in life history theory. The large brain of humans, which is an adaptation that increases survival in many environments, r epresents a life history trade off for delayed fecundity, a longer developmental timeframe, and increased longevity (Kirkwood and Austad 2000) Kirkwood and Austad (2000) also suggest that humans live longer relative to many other species because they ha ve decreased extrinsic mortality levels, due at least in part to their large brains. This reduced extrinsic mortality level allows for the accumulation of resources to maintain the organism beyond reproduction age (Kirkwood, 1977) yet it complicates age estimation as humans have a longer period of time to be exposed to detrimental environmental conditions that may affect rates of degeneration as well as live long enough to express late life, deleterious genetic effects Skeletal Growth and Development B ones grow through endochondral and intramembranous ossification and in response to genetic and environmental signals Endochondral ossification is the development of osseous tissue from a cartilaginous precursor; this type of ossification characterizes th e majority of the postcranial ske leton and some cranial elements such as the cranial base. Intramembranous ossification is the result of the combined
39 processes of direct mineralization of mesenchymal connective tissue and osseous deposition; this type of ossification occurs in a majority of the cranial elements and a few postcranial elements. Both types of ossification involve the growth phases of initiation, proliferation, histodifferentiation, morphogenesis, and apposition (Brodie 1941) The different iation of precursor cells results in the formation of cells with specific functions for skeletal tissue. Osteoblasts are the cells that build bone through deposition, and osteoclasts destroy bone through resorption. Osteocytes are the third type of cell in skeletal tissue; they are mature osteoblasts. Osteocytes serve important regulatory roles in skeletal tissue, and while they cannot replicate or resorb/deposit bony matrix, they are crucial in signaling these processes to active osteoblasts and osteocl asts. Ossification occurs at primary and secondary centers. The primary centers of ossification are the sites of initial bone development. The majority of the primary centers form prior to birth, and a single element can have multiple primary centers ( Scheuer and Black 2000) At these centers, osteoblasts deposit new bony tissue, which causes size increase have formed, they develop into osteocytes, while osteoblasts continue to deposit new bon e on the external surface. The outside surface of bones, the periosteum, remains osteogenic throughout life (Scheuer and Black 2000) Secondary centers of ossification form after birth and are generally located at the extremities of forming bones. Thes e centers allow for an increase in length during the growth period. During skeletal growth, deposition largely outpaces resorption; in the post growth phase, deposition and resorption are balanced.
40 As new bone matrix is being deposited, skeletal element s increase in length and width and change in shape via modeling and remodeling. In closely related structures, g rowth in one element will cause changes in adjacent elements. The subtraction an d addition of bone in m odeling and remodeling allow skeletal e lements to maintain consistent relationships with other skeletal elements and structures while they increase in size. M odeling affects the development and modification of primary lamellar bone, while r emodeling subtracts already deposited bone and adds n ew secondary cortical bone through complementary processes of deposition and resorption. During remodeling, deposition occurs on the external surface, while resorption is simultaneously occurring on the opposite side of the same surface (Enlow and Hans 1 996) Remodeling enables skeletal tissue to relocate and respond to functional demands or repair trauma (Enlow and Hans 1996) The attainment of adulthood signals an end to large scale skeletal growth and very little modeling takes place following skele tal maturation; remodeling continues to take place throughout adulthood although it occurs at a slower pace than in childhood. When elements change in both size and shape, they change position within the soft tissue matrix. Drift is passive, small scale relocation that results in the change in relative position of a structure within another structure ( Thilander 1995 ; Enlow and Hans 1996) Displacement, or translation, is a large scale process of drift where the entire bone actively moves to a new posi tion, and the change in position is measured in relation to other bony elements. In primary displacement, the relocation is related to growth of the bone, while in secondary displacement, adjacent structures are the
41 impetus for movement (Thilander 1995) All relocation processes involve modeling in the early stages and remodeling throughout life. Bone is a living and dynamic tissue that can alter its size and shape in response to mechanical, physiological, or other environmental factors. A development al trajectory is defined both by intrinsic and extrinsic cues, during which a different combination of internal and external constraints operate to give rise to a certain form (Rasskin Gutman and IzpisÂœa Belmonte 2004) A skeletal element will develop in to a largely recognizable form in the absence of load bearing, but it will not be structurally sound (Lanyon 1984) A certain level of external loading is necessary for bone growth or bones will atrophy, and even in utero there are mechanical forces (Sch euer and Black 2000) The understanding of the process of skeletal modeling/remodeling during life based on external loads and mechanical needs has long been credited to Julius Wolff. l be deposited where it is needed and removed (or resorbed) where it is not needed (White et al. 2012: p 28) Accordingly, a skeletal element can be expected to adapt over time in response to the presence or absence of external load(s): increased activi ty results in an osteogenic response, decreased activity results in bone loss (Lanyon 1984) Using this principle, variation in size and shape among individuals or species can be assessed in terms of the ability carry out a function or action (e.g., the ability to withstand stresses generated in a particular activity or to describe mechanical relationships in elements of a system) (Swartz 1991) Robust skeletal elements are assumed to be able to endure greater mechanical loading without failure, while g racile elements can endure less.
42 The skeletal response to mechanical loading is quite complex, and how much of the past mechanical environment that can be interpreted from skeletal elements is still a source of debate (see Ruff et al. 2006) T he skeleton has other functions besides mechanical competence (Ruff et al. 2006) E ach skeletal structure actually has three biological contexts: functional, developmental, and evolutionary (Wainwright 1988) Therefore, no optimum level of adaptation or ideal fi t of skeletal structure to loading can be assumed (Carter 1984) ; although a context (Ruff et al 2006: p 485) Growth is bounded, meaning that skeletal structures are constrained by geometri c rules and growth and material properties, and they do not generally reach physical limits (Thomas and Reif 1993) Genes, hormones, age, and disease can affect the skeletal response. The remodeling response also depends on the type of bone being subjec ted to the load (cortical versus trabecular), the type of load being applied (static versus dynamic), the frequency of application, the magnitude of application, the sense of application (tension versus compression), the location of the load (e.g., femoral midshaft versus femoral head), direct versus indirect loading, and other disturbance(s) to the skeletal tissue (e.g., trauma). Determining functional adaptation can be difficult due to the properties of bone and theoretical and experimental limitations. Skeletal tissu e is not a homogenous structure. B one is anisotropic, meaning that it has different material properties in different orientations and even within a single element responses to the same applied force can be markedly different (Swartz 1991 ) It can also be difficult to define the load and to what forces the element is subjected W hile the use of models eliminates many
43 confounding variables, models may be too simplistic. Often, engineering models are based only a single function and canno t take into account the complex three dimensional shapes seen in skeletal structures (Swartz 1991) In experimental loading, the choice to conduct in vivo versus in vitro analysis, frequently based on feasibility, will dictate the types of assumptions th at must be made and each is faced with certain limitations (e.g., for in vitro it may be difficult to simulate natural behaviors, while invasive procedures for in vivo may adversely impact muscle function and repair phenomena related to the procedure and n ot loading) (Bertram and Swartz 1991; Swartz 1991) The use of dry bone versus living skeletal tissue in experiments is also a concern. All of these factors indicate that adaptation to mechanical loads is contextual and site specific (Carter 1984) and that mechanical environments may not be easily interpreted from skeletal form, nor can shape be assumed to perfectly correlate with function or load. important component of resear ch in skeletal biology because it demonstrates that growth occurs not only because of genetic signals or progressing age but also due to mechanical constraints and physiological demands (Francillon Vieillot et al. 1990) Ruff et al. (2006) recommend the term bone functional adaptation to describe the response of bone to its loading environment. Experimental research has shown that bone does indeed respond to its mechanical environment, and the consideration of multiple factors in the remodeling response is vital to successful research (e.g., Lanyon et al. 1982 ; O'Connor et al. 1982; Carter 1984; Lanyon 1984 ; Lanyon and Rubin 1984; Meade et al. 1984; Rubin and Lanyon 1985 ; Garman et al. 2007; Ozcivici et al. 2007)
44 Age Estimation Using Skeletal Indicators In biological anthropology, age estimation relies on skeletal indicators that are correlated to biological age, using these biological indicators to predict chronological age. There are three requirements for a good skeletal age indicator: the traits show progressive and unidirectional change with advancing age, features can be reliably classified or measured, and changes occur at approximately the same time in all people (Milner and Boldsen 2012b) If an indicator and its associated traits d o not show the characteristics of the first requirement, age cannot be predicted from that indicator The second requirement speaks to observer error; multiple observers should be able to classify or measure the given features. The third requirement is m ost often affected by po pulational and sex differences. Research in skeletal age estimation aims to find indicators that fulfill all three of these requirements, plus understand variation in the aging process as expressed by the skeleton. Variability Th e dissonance between chronological age and biological age as measured by the skeleton is the result of many factors. Broadly, differences among individuals are due to differences in genes and environment (Hoppa 2000) Specifically, geographic origin; se x; health, nutritional, and socioeconomic status; secular trends (temporality); individual rates of senescence; mechan ical loading patterns; and developmental asymmetry can contribute to these differences (Kemkes Grottenthaler 2002) Additionally, the co rrelation between biological age and chronological ag e decreases with increasing age t he trajectory effect (Nawrocki 2010) This results in older individuals exhibiting greater variability in age related processes (i.e., degeneration)
45 than younger indi viduals and greater difficulty in estimating the age of older adults as compared to younger adults (Nawrocki 1998) Sex differences arise largely from sexual dimorphism, differences in size and shape between males and females, while regional differences are due to both genetic and environmental factors, (e.g., the adaptation of differing body proportions due to climate). Fo r some age estimation methods, there is no variation in population or sex, for others there is variation in both, and for still othe rs variation exists in only one (population or sex). Differences between the sexes and among regions have largely been dealt with by continued research and the development of sex and regional specific age estimation methods However, large, varied sampl es can alleviate the need for many sex and regionally specific methods and will likely aid in reducing age mimicry, in which the estimated age distribution of sample becomes similar to the reference method even though it is actually very demographically d ifferent (Bocquet Appel and Masset 1982; Konigsberg et al. 2008) Secular trends (variation in temporality) and their effects on age estimation have not been as thoroughly investigated as sex and regional variation (Milner and Boldsen 2012a) Individua ls living in different time periods but in the same general geographic area can display differences in body size and proportions, and secular trends in height and other skeletal dimensions have been well documented in biological anthropology (e.g., Jantz a nd Jantz 1999; Jantz and Meadows Jantz 2000) Langley Shirley and Jantz (2010) find secular trends in fusion rates of the medial clavicle, with modern Americans commencing fusion four years earlier than Americans from the early 20 th century If earlier skeletal maturation in more modern populations o ccur s, it is important
46 to recognize this temporal variation as standards developed using historic individuals may not be applicable to modern individuals (and vice versa); for age estimation this could resul t in over estimation of age in skeletal remains (Langley Shirley and Jantz 2010) Additionally, m aturation, like skeletal dimensions and shape, has both environmental and genetic components. If differences are observed between temporal groups, this sugg ests variation of an environmental nature; if differences are observed among ancestral groups, this suggests genetic variation. Developmental and degenerative asymmetry is the differential progression of development/degeneration between the right and lef t sides of the body of one individual. Estimating age can be problematic when different sides of the same skeletal indicator produce dissimilar age estimates or when only a single element is present and asymmetry cannot be assessed. Asymmetry has not tra ditionally played a large role in explaining variability in skeletal age estimation, though methods do often make a recommendation on which side to use if differences are observed (e.g., older or younger side). Development and degeneration may progress as ymmetrically due to individual and populational differences in biomechanical environments, physiological processes, or genetics, and if differences occur in the growth process, these may be further magnified later during degeneration (Overbury et al. 2009 ) Overbury et al. (2009) find asymmetry in pubic symphysis phase assignment using the Suchey Brooks method ( Katz and Suchey, 1986; Brooks and Suchey 1990 ) for over 60% of the individuals in their sample, though accuracy is still maintained if age is est imated with the morphologically older side. McCormick and Kenyhercz (2015) also found side differences of age related traits for the pubic symphyses and auricular surfaces and
47 conclude that component based methods including traits from both sides are pref erable to phase based methods. Conversely, Beresheim (2015) does not find statistically significant differences between left and right sides of the pubic symphysis, attributing observed asymmetry in Overbury et al. (2009) to observer error. The investiga tion of developmental and degenerative asymmetry is as important as understanding how other sources of error might contribute to conclusions about asymmetry. T wo parallel but complimentary forces have driven research aimed at understanding variability in age estimation : forensic science and paleodemography. Following the ruling in the Daubert case (1993) and the recommendations of the National Academy of Sciences (2009) the field of forensic science has been impelled to bette r understand how well methods perform and the error associated with their application This includes methods contributing to the identification of human remains such as age estimation (e.g., Baccino et al. 1999; Martrille et al., 2007; Kimmerle et al., 2 008 ) A powerful critique of a nthropological demography in the 1980s focused on the perceived inability to ever estimate age with any certainty due to age mimicry and a low correlation between skeletal age indicators and chronological age (Bocquet Appel a nd Masset 1982; Hoppa 2002) Reactions to this gloomy prediction served to improve not only single indicator methods, but to rethink the statistical basis for age estimation in anthropology and the ways in which to best combine estimates from multiple i ndicators (e.g., Van Gerven and Armelagos 1983 ; Konigsberg and Frankenberg 1992 ; Konigsberg and Frankenberg 1994 ; Aykroyd et al. 1997; Aykroyd et al. 1999; Hoppa and Vaupel 2002a)
48 Methods Macroscopic methods are the most commonly employed since th ey represent the quickest and cheapest methods and are thus accessible to all osteologists (Falys and Lewis 2011; Garvin and Passalacqua 2012; Milner and Boldsen 2012a) Microscopic methods are used, but they are often time consuming, expen sive, and de structive (e.g., histology). Methods are based on classificatory schemes (categorical data) or measurements of age indicators (continuous data) (Milner and Boldsen 2012a) Clas sificatory methods include phase systems that use overall form to place an in dividual into a distinct phase (e.g., et al., 1984; et al., 1985; Lovejoy et al., 1985a; Brooks and Suchey 1990 ) and component systems that score separate portions of a skeletal element to develop a summary score (e.g., McKern and Stewart, 1957; Gilbert and McKern, 1973; Meindl and Lovejoy, 1985; Boldsen et al. 2002; Buckberry and Chamberlain 2002) Measurement based age estimation methods include microscopic analyses of bone microstructure and measurement of long bone lengths for sub ad ults. Sub adults exhibit active growth and development and are generally less than 20 years of age (Falys and Lewis 2011) Since growth and development are tightly controlled genetically, methods of sub adult estimation have the ability to consistently e stimate age within a few years of known age. Sub adult age estimation methods include: timing of the appearance of primary and secondary ossification centers, dental development and eruption, long bone length, and fusion of ossification centers (skeletal maturation). For young children (fetal age, infant, less than 10 years old), age estimation relies most on the development of the dentition, long bone length, and the appearance of primary growth centers. Still in childhood but prior to puberty
49 (approxi mately 10 14 years), dental development and the appearance of secondary growth centers are employed. For individuals in their mid to late teen years (adolescents), age estimation relies heavily on the fusion of secondary ossification centers and the appea rance of the third molars. While an adult may exhibit some developmental changes such as the late fusing epiphyses of the vertebrae, ilium, and clavicle an adult is an individual who has completed growth and development. Adult a ge estimation is based lar gely on processes of degeneration, making it markedly more difficult, less accurate, and less precise than the age estimation of sub adults. General degenerative processes (e.g., edentulism, osteoarthritis at joint surfaces) can be used to place an indivi dual into broad age categories (e.g., young, middle, older; Listi and Manhein 2012) but these lack specificity Methods employing the pubic symphyses, auricular surfaces, sternal ends of the ribs, and cranial sutures are employed for adult age estimatio n (e.g., McKern and Stewart, 1957; et al., 1984; Lovejoy et al., 1985a; Meindl and Lovejoy, 1985; Brooks and Suchey 1990; Mann et al., 1991; Buckberry and Chamberlain 2002; Osborne et al. 2004) No adult age estimation methods offer particular ly small intervals and most methods produce age interv als that span several decades. T he large age intervals provided by adult age estimation methods likely do not reflect scientific or statistical limitations, but rather are a true indicator of the biol ogical reality of the highly variable aging process ( Nawrocki 1998 ; Kirkwood and Austad l 2000 ) Research also includes modifying existing methods to age very old adults (e.g., Berg, 2008; Beauthier et al. 2010 )
50 Adult single indicator methods The most r eliable and most commonly employed single age indicator for adults is the pubic symphysis phase method developed by Suchey and colleagues (Brooks and Suchey 1990; Buikstra and Ubelaker 1994) This method was developed from a large and diverse modern sam ple and, with the use of standardized casts for each of six component system for the pubic symphysis is also available, though it is generally far less employed by practitione rs ( McKern and Stewart 1957 ; Gilbert and McKern 1973 ) The pubic symphysis may be so useful because it exhibits delayed development, well into the middle aged adult years, though the method also considers degenerative changes. However, the pubic symphy sis, especially in archaeological remains, i s often damaged or not present. Other adult age estimation methods also employ a combination of late development and degenerative changes, with the emphasis mainly on degenerative changes after about 30 years of age. Because of preservational problems with the pubic symphysis, Lovejoy and colleagues develop a similar phase system to estimate age from the auricular surface of the ilium because it is often more well preserved and exhibits age associated changes (Lo vejoy et al. 1985a) This system is modified by Osborne and colleagues (Osborne et al. 2004) in order to provide statistically viable age ranges. Buckberry and Chamberlain (2002) develop ed a component system for the hod), but the method suffers from small sample sizes in the younger age stages, resulting in poor performance for these groups and does not demonstrate broad applicability across many samples ( Mulhern and Jones, 2005; Falys and Lewis 2011 ) A phase syste m is also employed for the sternal end of the fourth rib
51 ( et al. 1984; et al. 1985) but this too suffers from the same preservational issues as the pubic symphysis and the statistical validity of the small age intervals for each of the phases is questionable (Nawrocki N.D.) Cranial sutures a re discussed in greater detail below. These methods do not represent the only ways to estimate adult age at death, but they are the most commonly employed macroscopic indicators in anthropological skeletal analysis (Garvin and Passalacqua 2012) Adult multiple indicator methods T he multifactoriality of the aging process suggests that a single age indicator or bone does not adequately reflect chronological or biological age nor can any one indicator be truly predictive (Kemkes Grottenthaler 2002) More age indicators are certainly better, especially when considering the imperfect correlation between chronological and biological age and the variability of the aging process (Houck et al. 1996) However, there is currently no consensus on how to best com bine multiple age indicators ( Uhl and Nawrocki 2010 ; Garvin and Passalacqua, 2012 ) For example, if the pubic symphysis is considered to be the most reliable indicator and it is present, how much weight is given to other age indicators? What if age indi cators produce different es timates for the same individual ? How are the indicators that are present best combined? The difficulty in constructing a final age estimate is reflected in Buikstra and Ubelaker (1994) where the recommendation is to use these t hree broad age categories: 20 34 years (young adult), 35 49 years (middle adult), or 50+ years (old adult) based on For cases that do not fit into one of the three age categories, the recomme ndation is that more weight be given to postcranial indicators than cranial.
52 In a survey of practicing forensic anthropologists, Garvin and Passalacqua (2012) find that the manners of constructing a final age estimate based on multiple indicators are hig hly varied, experience based and often include statistically invalid assumptions One of the more objective techniques indicated was the use of the lower end of the interval from the method that provided the oldest age and the higher end of the interval from the method that provided the lowest age (colloquially referred to as presents statistical challenges since many of the reference articles do not use the same statistical information (e.g., s tandard deviations versus standard errors) (Garvin and Passalacqua 2012) More statistically rigorous ways of combining multiple indicators include those methods classified as multifactorial (e.g., McKern and Stewart, 1957; 1970; Lovejoy et al. 1985b; Martrille et al. 2007) Three main approaches are currently available for estimating age from multiple indicators: the complex method 1970) ; the multifactorial summary age met hod (Lovejoy et al. 1985b) ; and transition analysis 1 (Boldsen et al. 2002) While analytically different, all three have the ability to collect data on more than one age indicator and then it to develop a summary age. The AcsÂ‡di and NemeskÂŽri (1970) com plex method, employed primarily by European anthropologists, bases age on the average of the ages given by the pubic symphysis, trabecular structure of the humeral and femoral heads, and endocranial 1 Here transition analysis refers to the specific method to combine multiple indicators developed by Boldsen and colleagues. Transition analysis is al so a generalized statistical a pproach in age estimation (see The Statistical B asis of Age Estimation, below).
53 suture closure. Concerns raised with this method include the averaging of indicators without weighting and the need for the same age indicators as the reference method ( Lovejoy et al., 1985b; Brooks and Suchey, 1990 ) While this method is widely used in Europe following the recommendations of the Workshop of E uropean anthropologists (Ferembach et al. 1980) it is less frequently employed in other regions. The original multifactorial summary age method employs the auricular surface, pubic symphysis, cranial sutures, and radiographic analysis of the proximal fem ur and clavicle, though any combination of methods can be used as long as the entire sample is seriated by method prior to data collection (Lovejoy et al. 1985b) Data on each indicator is collected and then principal components analysis (PCA) is run to weight the indicators employed for the sample. The final age estimate for an individual is the weighted average of the available age indicators (Lovejoy et al. 1985b) The need to seriate biases this method towards paleodemographic usage, though with so me modifications it can be used in forensic identification. Martrille et al. (2007) propose running the PCA on a single case using the individual age indicators collected and then the correlation between the first principal component and the age indicator s as the weights. These authors conclude that as many skeletal and/or dental indicators as possible should be used with particular attention to methods that have higher accuracy for a certain age range when producing the final age estimate (Martrille et al. 2007) Multiple indicator transition analysis employs scoring of separate components of the pubic symphysis (five characteristics), iliac portion of the sacroiliac joint (nine characteristics), and cranial sutures (five segments) (Boldsen et al. 20 02) This method computes the likelihood of death estimates occurring at different ages for each
54 character by looking at the age of transition from a particular stage to the next for each indicator and calculating the probability that an individual died a t a particular age given certain observed skeletal traits (Boldsen et al. 2002) One of the particularly useful facets of transition analysis is that it allows for the estimation of age from incomplete or fragmentary remains and does not require seriatio n of the sample. However, transition analysis, like other Bayesian statistical methods, requires knowledge of an independent but appropriate prior age distribution or the use of uniform priors. Tests of multif actorial methods are equivocal. Saunders et al. (1992) do not find that the Lovejoy et al. (1985b) multifactorial method outperforms single indicators in their unseriated sample but that a simple average of the estimates produced from singl e indicators is more effective. Conversely, Bedford et al. (1993) find th at the multifactorial method is superior to any single indicator in their seriated sample Martrille et al. (2007) in examining the pubic symphysis (Brooks and Suchey 1990) fourth rib end sternal extremity ( et al. 1984; et al. 1985) auricular surface (Lovejoy et al. 1985a) and the anterior teeth (Lamendin et al. 1992) find that the use of multiple indicators (through PCA) has the lowest inaccuracy for all groups, but if the sample is bro ken down by young (25 40), middle (41 60), and old (>60) age groups, single indicators (different for each group) are more accurate than PCA. Bethard (2005) finds that the method of transition analysis proposed by Boldsen et al. (2002) does not perform as well as the authors claim, though a subsequent application of transition analysis by Milner and Boldsen (2012b) produces more favorable results. Uhl and Nawrocki (2010) test various individual indicators (pubic symphysis, auricular surface, sternal end o f the rib, and cranial sutures) and three ways of combining them (average
55 of point estimates, range of spatial overlap of four confidence intervals, and multiple linear regression with forward stepwise selection) and find that combining multiple indicators is always more effective than simply using single indicators. Their results favor the use of linear regression, which has similar inaccuracy val ues to other methods but offers the advantage of producing both a point estimate and an interval. The continu ing development of multifactorial approaches and the best way to combine multiple age indicators necessitates continued research on single age indicators and subsequent research on the most accurate way(s) to combine information from these single indicator s. The continued refinement of single indicator methods is important because it leads to more accurate, precise, and reliable estimates of age at death. As eloquently stated by Milner and Boldsen (2012a: p 99) o better than the individual indicators of age they for both single and multiple skeletal age indicators. To date there has been little consensus among anthropologist s on ways to measure the effectiveness of age estimation methods (Uhl and Nawrocki 2010) The Statistical Basis of Age Estimation The estimation of age at death from skeletal remains necessitates the conversion of observable skeletal age indicators into c hronological ages which is accomplished via statistical inference. The two main schools of statistical inference in skeletal age estimation are frequentist and Bayesian. Anthropologists have traditionally relied on frequency statistics for calculating a ge at death, but the use of Bayesian methods is becoming more common. The preference of one over the other is not clear in biological
56 anthropology, and often the anthropologist will blur the line between the two approaches in order to best interpret the d ata (Klepinger and Giles 1998) Frequency based inference comprises the statis tics and analytical procedures that are most familiar t test, regression). These tests are based on the assumption that the population from which the sa mple is drawn is normally distributed. The sampled data are assumed to be a repeatable random sample from an unchanging underlying populati on that has specified parameters. For a frequentist, probability describes the frequency of a specific outcome ove r many trials (Klepinger and Giles 1998) This type of statistical inference closely mimics the desired scenario in scientific testing ( objective, controlled, replicable experiments), and hypotheses are rejected based on the evidence presented, but they are never proven to be true. Sampling error is a very large compo nent of frequentist inference, meaning that experimental design is an important part of the experiment itself (e.g., adequate sample sizes, appropriate sampling procedures) (Nawrocki 2010) The strength of frequency based approaches is in their use of relatively simple calculations, straightforward, unvarying procedures, and the ability to apply the same formulae in subsequent analyses. Bayesian statistical inference does not rely on pred efined parameters but instead assumes fixed data with parameters that can be probabilistically determined from experiential observation. Bayesian inference incorporates known facts about the universe (or testing population), and this a priori knowledge is then used in hypothesis testing and to update subsequent test iterations. Probability is not based on repeated trials but is the measure of the outcome of a hypothesis (Jefferys and Berger 1992)
57 Hypothesis testing in Bayesian inference produces a prob abilistic statement about the strength of the hypothesis, in the form of a posterior probability or likelihood ratio. Bayesian statistical inference can be quite complex and time consuming; however with the advent of powerful computing capabilities, Bayes ian inference has become more common. Many researchers have recognized that data, especially those of a biological nature, do not always conform to stringent assumptions, and the support for Bayesian inference has grown. The underlying distribution, a pr oblem that can affect the outcome of hypothesis testing in frequentist inference, is not a problem in Bayesian inference because the parameters are always updated to reflect the data. Another distinction that can be made in statistical inference is betwe en parametric and nonparametric options; frequency statistics and Bayesian statistics have both parametric and nonparametric options. A parametric statistical method employs distribution (e.g., a normal distribution for the population; probability distribution as determined by the data) or a fixed model structure. Nonparametric methods make no such assumptions about underlying probability distributions or fixed model structur es. Nonparametric tests are less powerful (meaning they have less ability to find a real effect or an association) than parametric procedures when the population follows an expected distribution or model structure (either normal or defined by priors), but they can be particularly useful for small sample sizes and when the data do not follow an expected distribution because nonparametric tests are more robust (resistant to outliers). The choice of statistical paradigm (frequency or Bayesian inference) and methods (parametric or nonparametric) are generally left to the practitioner, though
58 some choices are more commonly seen in biological anthropology than others (e.g., frequentist inference using parametric methods). The development of age estimation method s based on skeletal indicators starts with data collection from documented samples. A certain indicator or several indicators are observed to be age related, and t hese traits are compared in term s of their relationship with known ages at death in the refe rence sample. Observations can then be lumped into discrete descriptive categories as in phase systems or assigned individual numbers that are summed as in component systems. There are currently no standards for reporting results, though most age estimat ion methods will commonly include at least the age ranges, means (or other measure of central tendency), standard deviation (or other measure of dispersion), and sample sizes per phase, stage, or score (Nawrocki 2010) Age ranges are variably presented, to include prediction or confidence intervals, or are constructed through the use of adding one or two standard deviations to the mean age. Measures of method effectiveness include: calculations of inaccuracy (the absolute difference between estimated a nd actual age per individual, phase, or sample), bias (the positive or negative difference between estimated and actual age per individual, phase, or sample), the correlation coefficient (r; strength and direction of the linear relationship between two var iables), the coefficient of determination (r 2 ; the amount of variation in one variable that can be explained by variance in another variable), and the percentage of correct classification using published data, which can include individuals correctly classi fied by phase or stage, by one or two standard (Murray and Murray 1991; Uhl and
5 9 Nawrocki 2010) Measures of bias, inaccuracy, and covariance ( r and R 2 ) average the total error in a method or phase (or other gr ouping category), but it is also important to include information on the overall pattern of error for a particular study. For example, Lovejoy et al. (1985b) include the largest absolute differences between predicted and actual ages, which inform on h ow m uch a method over or underestimates age. An additional consideration in testing methods following their development is the inclusion of data on inter and intraobserver error to address method reliability. Regression based methods estimate age (the depen dent variable) from the age indicator (s) (the independent variable [s] ). While this is not entirely biologically realistic morphology age is the unknown variable in skel etal analyses and must be predicted from observed traits. The success of regression depends on a strong linear relationship between dependent and independent variables and continuous data that is independent where multiple predictor variables are employed (Adler 2012) If the independent variables are categorical rather than continuous, ANOVA is employed in the same way as regression (Crawley 2013) The use of regression in age estimation is not without problems. Because regression aims to reduce ind ividual deviations from a particular line, age estimates for extrema points tend to regress to the mean, referred to as (Masset 1989) There is often a non linear relationship between indicator and age, and morphological traits are discrete rather than continuous (Kemkes Grottenthaler 2002) When there is a relationship, the minimum acceptable value of the correlation coefficient
60 is not agreed upon (Bocquet Appel and Masset 1982; Lovejoy et al. 1985b) Additionally, indepen dence among different age indicators cannot be assumed. Transition analysis, based in Bayesian inference, does not require data to meet the assumptions of regression, nor does it require a normally distributed population. This analysis models the passag e of individuals from a given developmental stage to the next stage in an ordered sequence using a prior distribution (Konigsberg et al. 2008) Rather than predict age from skeletal traits, transition analysis takes a more biologically realistic approach by condition ing all indicators on age and giving the probability that a set of skeletal remains are from a person who died at a certain age (Hoppa and Vaupel 2002b) While frequency based statistics assign fixed intervals per stage, Bayesian inference a llows for the use of prior knowled ge to adjust the intervals. Theoretically, Bayesian inference appears to be well suited to the statistical needs of biological anthropology, but it has yet to see significant incorporation into the most commonly used age estimation methods Bayesian statistics are believed to solve problems of age mimicry, independence, inaccurate representations of estimation uncertainty, and open ended upper age intervals, but methods like transition analysis still only work as well as their associated reference samples and scoring systems (Garvin et al. 2012) Since a nalysis is based on the estimated age of transition between adjacent phases of a method, it requires discrete stage or phases that are age progressive as well as a known age reference sample that has been previously scored using the same method (i.e., informed priors) (Garvin et al. 2012) Prior distributions have the capability to negatively affect age estimates if they are not appropriately developed and the selectio n of p riors can be highly subjective. F or continuous data,
61 frequentist approaches are better, and with appropriate experimental design are highly effective. Cranial Suture Age Estimation The use of cranial sutures in age estimation is based on a positiv e correlation between suture fusion and age Older individual s tend to exhibit more sutural obliteration than younger individuals, who retain largely patent sutures. Unlike epiphyseal fusion, complete obliteration of the cranial sutures rarely occu rs, th ough complete fusion of the endocranial aspect of sutures is more likely (Ashley Montagu 1938) Because they fuse earlier, endocranial aspects of sutures have the potential to more accurately predict the ages of younger adults, while the ectocranial port ions of the sutures may be preferable for older adults (Perizonius 1984) Craniofacial sutures, most notably the external portions, show an even greater delay in fusion than vault sutures and because of this have great potential for age estimation in old er adults (Wang et al. 2006) According to Kokich (1976) facial sutures often remain open until the eight h decade of life, and facial sutures may be able to distinguish between old and very old adults (Beauthier et al. 2010) Sutural age estimation m ethods most often examine sections of sutures or degrees of fusion along a suture in order to develop an overall picture of sutural fusion and assign an age (e.g., Meindl and Lovejoy, 1985; Mann et al. 1991) Methods that employ the ectocranial portions of the sutures are more common than those that examine endocranial aspects, largely because the external surfaces of sutures are easier to observe than the internal surfaces, especially for the facial sutures ( Wang et al., 2006; Falys and Lewis, 2011 ) Ma croscopic cranial suture age estimation methods
62 are easy to apply and generally have low rates of interobserver error ( Zambrano 2005 ; Milner and Boldsen 2012a) Cranial suture age estimation also has several challenges, several of which are related to th e variability of sutural fusion. The complexity of the skull means that sutural fusion cannot be interpreted as a simple linear relationship between only fusion and age. A lthough sutural growth is related to growth of the brain and facial structures, it is unclear why sutures fuse since patency enables the skull to retain flexibility (Herring, 2008) Variation in fusion rates are observed between the sexes, with m ales fusing earlier than females while the pace of obliteration is more regular in females ( Ashley Montagu 1938) Fusion rates may, or may not differ among ancestral groups Galera and colleagues (1998) and Nawrocki ( 1998) found statistically significant differences between ancestral groups, but Meindl and Lovejoy (1985) d id not. Secular tren ds may also contribute to variation in sutural fusion, and given earlier maturity in more recent cohorts, more modern individuals may also exhibit greater suture closure at a given age than prehistoric or historic individuals ( Masset, 1989; Langley Shirley and Jantz 2010) Conversely, Nawrocki (1998) finds that synostosis occurs at a slightly slower rate in modern individuals, so that for any given level of suture closure the modern sample is actually aged younger than the historic. Zambrano (2005) finds that the equations given in Nawrocki (1998), while based on a historic sample, perform well for modern individuals and that there is no systematic secular trend observed when these equations are applied to indiv idual forensic cases. Somatic dysfunction ( e.g., sacroiliac fusion, ankylosing spondylitis, and scoliosis ) has also been found to have a stronger correlation with cranial suture fusion than documented age (Kroman and Thompson 2009) The
63 effect of fusion on cranial strain pattern and the effects o f mastication on fusion have also been recently considered (Wang et al. 2006) and offer a potential avenue to quantify qualitative criteria, such as thinning or thickening of the vault and bone density changes, which have been used in past to make inferenc es about age estimation based on the cranium ( 1986; Henderson et al. 2005; Wang et al. 2006) Beyond the inherent variability of sutural fusion among individuals and populations, cranial suture age estimation methods themselves als o contribute to challenges in age estimation from the sutures. There is a lack of consensus in scoring cranial suture obliteration, including the number of sites observed, the amount of suture s examined (full suture versus 1 cm sections), and the number o f stages. The distinction between different stages can be difficult to identify and intra and interobserver error increases with greater subdivision of stages (Scheuer and Black 2000) In the history of cranial suture age estimation, analytical techniq ues have been rudimentary, disjointed, and lacking in statistical complexity (Nawrocki 1998) While many studies compare sets of certain sutures (e.g., frontosphenoidal sutures: Dorandeu et al. 2008; maxillary sutures: Mann et al., 1987; Mann et al. 1991 ; vault sutures: Meindl and Lovejoy 1985; squamous and parietomastoid sutures: Saito et al. 2002) studies that incorporate many different sutural sites or sutures into multifactorial methods are less common ( Nawrocki 1998 ; Boldsen et al. 2002) Vault Sutures Age estimation using the sutures of the cranial vault is one of the oldest and most deliberated methods (Ashley Montagu 1938) Methods of age estimation from the vault sutures are based on the w ork of Todd and Lyon from the 19 th century ( Broca, 1875; Todd and Lyon 1924; Todd and Lyon 1925a; Todd and
64 Lyon 1925b; Todd and Lyon 1925c) The Todd and Lyon method scores obliteration of endocranial and ectocranial sutures on a five stage scale: 0 none, 1 one quar ter, 2 one half, and 3 three quarters, 4 complete. Interestingly, Todd and Lyon do not find that their work supports the use of cranial sutures as an accurate age estimation method due to the highly variable nature of sutural fusion. Further researc conclusions about the ability of cranial suture fusion to predict age (e.g., Ashley Montagu 1938; Singer 1953 ; Brooks 1955; McKern and Stewart 1957; Powers 1962 ) These critiques point out that cra nial sutures are unable to estimate age within 10 years of known age at death. However, further research in the 1970s and 1980s revives cranial suture age estimation, not because the methods are able to produce more precise intervals, but because the anth ropological field begins to embrace more age indicator (Meindl and Lovejoy 1985) At this time, vault sutures are also incorporated into multifactorial methods (e.g 1970; Lovejoy et al. 1985b) Of methods currently employed, Meindl and Lovejoy (1985) is the most common for ectocranial methods and AcsÂ‡di and NemeskÂŽri (1970) is the most common for endocranial methods (Falys and Lewis 2011 ) The Meindl and Lovejoy (1985) system stage system to remove the potential ambiguity of three intermediate phases: 0 open, 1 1 50% union, 2 51 99% union, and 3 complete This system also uses predefined landmarks on specific sutures, divides the skull into vault and lateral anterior systems,
65 and does not include any endocranial suture sites. A sum of all landmark scores is used to estimate age, and the lateral anterior system is preferred since the fusion pattern is more regular in these sutures (Meindl and Lovejoy 1985) This differs from the methods used by European anthropologists, who continue to employ th e five phase system of Broca, score endo and ectocranial sutures and calculate a suture coefficient (sum of all obliteration scores divided by total number of observation points) (Ferembach et al. 1980; Masset 1989) Both European and North American anthropologists employ linear regression to predict age at death fro m closure of the vault sutures. Yet the utility of cranial vault suture estimation is still debated. So me practitioners routinel y include cranial fusion as a component of age estimation while (Buikstra and Ubelaker 1994; Nawrocki, 1998; Garvin and Passalacqua 2012 ; Warren personal communication ) In fact, the standards given in Buikstra and Ubelaker (1994) advise that suture closure is only useful when other criteria are not a vailable or when the information is used in conjunction with other skeletal age indicators Practicing forensic anthropologists, regardless of experience, rank cranial vault sutures as one of the least preferred age estimation methods (Garvin and Passalac qua 2012) Conversely, Nawrocki (1998) asserts that the use of sophisticated statistical methods, ones that include models which incorporate multiple suture sites based on their ability to predict age and the proper construction of error intervals makes cranial suture age estimation comparable to other adult age estimation techniques.
66 Palatal Sutures Age and palatal suture closure are first investigated in the dental and orthodontic fields as related to clinical practices of mid palatal expansion (e.g. Latham 1971; Persson and Thilander 1977) A positive correlation between suture closu re and age is noted, and much of this research focuses on sutural microstructure and employs histological methods (e.g., Persson and Thilander 1977) In anthropolog ical research, the fusion of the palatal sutures is summarized by Mann and colleagues, who publish two versions of their method: original (Mann et al. 1987) and revised (Mann et al. 1991) Of these two methods, the revised one is cited more frequently than the original ( Wheatley, 1996; Sakaue and Adachi, 2007; Brown, 2009; Beauthier et al., 2010; Brown, 2010; Apostolidou et al. 2011; Siegel and Passalacqua 2012;) Table 2 1 compares the samples for the published original and revised methods, subseque nt test samples, and newly developed methods. Age estimation from the maxillary sutures has yet to be widely employed for adult age estimation (Brown 2010; Garvin and Passalacqua 2012) The palatal sutures are: incisive (IN), transverse palatine (TP) and median palatine, which is divided into anterior and posterior sections based on location in relation to the TP suture (AMP and PMP, respectively); Chapter 3 discusses the structure of the palate and its sutures in greater detail. Like other vault su tural age estimation methods, the sutures of the hard palate are examined for fusion, and age is estimated based on the varying states of fusion seen throughout the palate. The conversion of observed obliteration to age depends on the particular method em ployed. Sutural obliteration most often proceeds in this order: IN, PMP, TP in the greater pala tine foramen (GPF), TP, and AMP
67 (1987) suture obliteration is measured for each of the four maxillar y sutures and the percent of obliteration present per suture is calculated (Mann et al. 1987) These percentages are added and converted to an obliteration score value between 0 and 4, with the percent of obliteration associated with each value as follow s: 0=0%, 1=1 25%, 2=26 50%, 3=51 75%, 4=76 100% (Mann et al. 1987) A more detailed age prediction model is presented in Mann (1987) where linear inverse prediction formulae are given in order to produce an age estimate from a given obliteration score, regressing age on suture closure. While Mann (1987) and Mann et al. (1987) are largely similar, the published study bases scoring on the half of the maxilla with the least amount of obliteration, while the thesis employs the side with the most obliterati on. Gruspier and Mullen (1991) find the original method to be 27 28% accurate within 1 0 years and 55 71% accurate within 20 years based on two observers. These authors caution that the Mann et al. (1987) method may appear to work well because higher co rrelations of age and suture closure in the younger group mask lower correlations in the older group, causing an overall significant linear regression when in fact one does not exist. Following the testing of the original method by Gruspier and Mullen (19 91) and the publication of the revised method (Mann et al. 1991), the original method does not appear to have gained much traction in the anthropological literature. The revised method is based solely on visual examination for any amount of obliteration a long each of the four sutures as well as the TP suture within the GPF (Mann et al. 1991) For sutures expressed bilaterally, the side with the most obliteration is used. Age is estimated based on the last suture to se e any degree of
68 obliteration, using the ages given in Figure 2 of the reference article ( Table 2 2). For example, if the incisive, posterior median palatine, and transverse palatine sutures show at least one area of obliteration but the anterior median palatine does not, the individual is es timated to be between 35 and 50 years of age. If the same general progression of suture fusion is always observed, this technique is straightforward, albeit lacking in statistical robusticity. However, where the observed pattern dif fers from the expected it is less clear how to estimate age since M ann et al. (1991) only provide the earliest age of fusion per suture seen in their sample. The age estimate can also include the assessment of other subjective palatal traits, such as bone condition, edentulis m, and alveolar resorption. However, these traits are not defined, no codified measurement system is provided, and it is unclear how these traits should be assessed in relation to fusion (i.e., how are traits versus fusion prioritized?). Tests and modifi cations of the revised method include: Wheatley (1996), Ginter (2005), Sakaue and Adachi (2007), Beauthier et al. (2010), Brown (2010), Apostolidou et al. (2011), and Siegel and Passalacqua (2012) Results of these studies are equivocal, with some sugges ting the method performs with high enough accuracy to be employed for age estimation ( Ginter, 2005; Beauthier et al. 2010; Brown 2010 ; Apostolidou et al., 2011 ) while others caution against its use due to low accuracy and precision in age estimates as co mpared to known ages at death ( Wheatley, 1996; Sakaue and Adachi 2007; Siegel and Passalacqua 2012 ) What all of the studies do agree on is the general relationship of palatal suture closure to age, the progression of palatal suture fusion outlined by M ann and colleagues, and that palatal sutures can be used, at a minimum to place individuals into broad age categories (e.g., young, middle
69 aged, old). H owever, the methods employed vary greatly and even where a similar method has been applied, the results seem to be contradictory (e.g., Wheatley, 1996 versus Beauthier et al ., 2010) No standard exists for palatal suture age estimation, and, even in studies that cite the revised method, application varies. Palatal suture fusion is also included in more mul tifactorial approaches: Nawrocki (1998) and Vodanovi et al. (2011) Nawrocki (1998) finds that the correlation of age and all palatal sutures is 0.55, as compared to positive correlations of 0.66 and 0.67 for age and all ectocranial and all endocranial sutures, respectively. In using median palatine suture as one of four methods to estimate age in a Croatian archaeological sample, Vodanovi et al. (2011) find that closure of this suture and dental wear show high degrees of association. Compared to mo re complex methods that require extensive training such as tooth root translucency and pulp/tooth area ratio the use of the palatal sutu re method for age estimation results in less accurate age estimations but is easier to apply 2011) T he use of a single suture and an archaeological sample (where actual ages at death cannot be verified) differs from the Mann method and other tests of this method, but the potential relationship of tooth wear and midpalatal suture closure is interesting W heatley (1996) and Beauthier et al. (2010) both employ systems that score smaller sections of each palatal suture and use multiple regression to relate fusion at each site to age. Though neither study offers vastly improved accuracy as compared to the rev limited utility in estimating age at death, while Beauthier et al. (2010) suggest the method is promising, especially for individuals of advanced age since palatine fusion
70 progres ses more slowly and starts later than vault fusion. They also find that there is agreement between age estimates from palatal and vault suture closure, and there is good agreement between age estimates produced from their method and the revised method. I nterestingly, Wheatley (1996) finds that individuals with partial to complete tooth loss actually display premature fusion of the sutures, resulting in overestimation of their ages, and the presence of a large number of edentulous individuals in her sample may affect method performance. Variation in palatal suture fusion by sex is also not consistent among studies. Mann et al (1991) and Apostolidou et al. (2011) f ind greater obliteration in males as comp ared to females of the same age, while Wheatley (199 6) does not find that sex significantly affects the rate of suture closure. Ginter (2005) finds no statistically significant differences in the accuracy of age estimation between the sexes for either the original or revised methods, though age is more oft en correctly predicted for males than females. This trend of higher correct classification rates for males is also found by Mann et al. (1991) Sakaue and Adachi (2007), and A postolidou et al. (2011) Populational differences in palatal suture closure are harder to assess because test samples often examine individuals from one major group or a small sample from a second group ( Table 2 1). Mann et al. (1991) find minor ancestral differences in palatal suture closure, but this study includes only individual s of African and European ancestry. The study by Ginter (2005) offers a more complete picture of revised method performance in different groups. Ginter (2005) found lower correct classification rates for the ances trally diverse sample group as compared to white individuals, though no statistically significant differences exist
71 among these groups. Using the original method, Ginter (2005) finds that individuals from the ancestrally diverse sample group have slight ly higher correct classifications significant. In a Japanese sample, age is estimated correctly for only 36.9% of males and 25.7% of females, but since ages are not oft en over estimated Sakaue and Adachi (2007) state that palatal suture closure is useful as an indicator of minim um age in Japanese individuals. In a Greek sample, correct classification is much higher, with an overall rate of 87% correct (89% for males, 84 % for females). When combining palatal sutures with vault sutures, palatine sutures are more heavily selected in models for individuals of African ancestry versus European ancestry (Nawrocki 1998) The effects of secular trends on palatal suture closur e are unknown. Examining the samples given in Table 2 1 it can be seen that only three studies include samples from both historic and modern time periods. Even in these samples, only one has sample sizes balanced and large enough for comparison between h istoric and modern individuals (Wheatley 1996) Improving palatal suture closure as an age i ndicator T he large scale applicability of maxillary suture age estimation remains uncertain because of the inability to compare methods due to differences in sam ple composition, the various ways that palatal suture closure has been scored and analyzed, and skepticism concerning the validity of sutural closure to estimate age. In order to improve palatal suture closure as an age indicator and reduce error in estim ating age from palatal suture fusion, these challenges must be addressed. While a poor correlation between age and palatal suture fusion could be contributing to poor method
72 performance, without the investigation of other variables that potentially relate to age and/or fusion it cannot solely be attributed to this. Investigation of the maxillary sutures has been conducted in samples composed largely of indivi duals of European ancestry, some individuals of Af rican ancestry, and only a limited number of indi viduals of Asian ancestry. No one study includes adequate sample sizes from all three major ancestral groups, and most samples to date are not balanced in terms of age, sex, ancestry or time period ( Table 2 1). P oorly distributed samples with small sizes introduce unnecessary error (Nawrocki 1998) Th ere are also methodological inconsistencies in suture scoring and how those scores are translated into age estimates. An interobserver error study conducted by Brown (2010) indicated that while practitioner s show high concordance in observing and recording obliteration of the palatal sutures, there is little agreement on how to develop an age interval from these observations. Being able to apply the method with little e rror for many observers is important, and the inability to do so introduces error Efforts to standardize the method by scoring pre defined suture sites have not always been successful (e.g., Wheatley 1996) and actually result in lower correct classification rates than the solely visual meth od that takes into account the entire length of a single suture. Based on tests of the original and revised methods, there is still some confusion that exists on the different approaches presented in each method. The two methods differ significantly in n ot only the ages assigned per state of fusion ( Table 2 2), but how fusion is scored. The original method requires measurement of the sutures, though the technique is not clearly outlined, and the revised method uses a visual, quantitative method, though s ome continue to use the summary score from the original method.
73 These differences are poorly defined between the two methods, and subsequent tests of the methods reflect this difficulty, contributing to error when applying either method. The improper appl ication of statistics, or the absence of statistical analysis altogether, also contributes to difficulties in estimating age from the palatal sutures. Ver y few of the above methods employ robust statistical analysis or selection procedures, and the revise d method, which subsequent studies use as a guide, does not even provide descriptive statistics for stages of observed suture fusion. There is a clear lack of predictive models for age estimation. Even with the general trend of increasing fusion with inc reasing age confounding variable s, especially in an area as complex as the palate, are important considerations. None of the above studies have truly examined the relationship of sutural fusion, age, masticatory function, and other palatal traits. Equal ly as important is the investig ation of other facial sutures as controls in order to better understand maxillary suture fusion patterns. While multiple studies show a general relationship between maxillary suture fusion and age, in a study of the frontona sal suture, Alesbury et al. (2013) found that fusion of this suture is poorly correlated with age and that no regular pattern of fusion occurs Secular trends may also affect age estimation based on the palatal sutures if individuals in more recent decade s are in fact maturing earlier than their historical counterparts, and, as with other age estimation methods, this should be investigated. Summary This chapter presented descriptions of age and aging, processes of skeletal growth and remodeling, and how an thropologists estimate age from skeletal indicators, including a consideration of the statistics employed and how cranial and palatal sutures are used for age estimation. The next chapter provides a detailed discussion of the
74 human palate and its growth, development, and function. It also includes a consideration of how the palate is used in biological anthropology.
75 Table 2 1. A comparison of anthropological studies of age estimation based on all of the palatal sutures. The order is chronological bas ed on publication date. Information in parentheses further describe the sample. Reference n Ancestry Sex Time period Age range Mann et al. ( 1987 ) 36 Males (14) Females (22) Modern 22 73 (males) 13 79 (females) Mann et al. ( 1991 ) 186 Males (110) Females (76) Historic (171) Modern (15) Not given Gruspier and Mullen ( 1991 ) 83 Males Historic 29 87 Wheatley ( 1996 ) 346 Not given Males (177) Females (169) Historic (146) Modern (200) 13 101 Ginter ( 2005 ) 155 Diverse a Males (96) Females (59) Modern 26 100 (both) Sakaue and Adachi ( 2007 ) 375 Asian (Japanese) 274 males 101 females Not given ~15 80 (both) Beauthier et al. ( 2010 ) 134 European (French, Bel gian) 78 males 56 females Modern (100) Historic (34) 19 96 (males) 19 101 (females) Apostolidou et al. ( 2011 ) 271 European (Greek) 150 males 121 females Modern 20 64+ Siegel and Passalacqua ( 2012 ) 200 European African Not given Historic 10 82 (both) a Classified according to social classification categories in use at time of collection; white is European, diverse includes East Asian and Khoisan, black is African.
76 Table 2 2. Comparison of age estimates for the original (Mann et al. 1987) and rev ised (Mann et al. 1991) maxillary suture methods. Suture fusion observed Age interval Original Age interval Revised IN <25 20 24 PMP 25 42 25 29 TP in GPF N/A 30 34 TP 43 60 35 50 AMP 60+ 50+
77 CHAPTER 3 THE HUMAN PALATE The human craniofacial skeleton, which includes the palate, exhibits a complex developmental history and biomechanical environment. As a part of this structure, the palate is intrinsically related to the growth, development, maturation, form, and function of the skull, facial s keleton, and dentition. Besides it role in alimentation, t he palate is also important in biological anthropology because morphological traits can be used for individuation or placement within a specific group (e.g., sex, ancestry, and age). This section discusses facial and palatal growth, development, and maturation; palatal function; and uses of the palate in biological anthropology. More general processes of skeletal growth and agin g are discussed in Chapter 2. Palatal Form The human palate is compose d of hard and soft tissue, and it serves as a divider between the alimentary and respiratory tracts. The hard palate is made up of the left and right maxillae and palatines, though considerable debate exists on the presence of a separate premaxilla in hum ans (Scheuer and Black 2000) 1 The anterior two thirds of the hard palate is composed of the palatine processes of the right and left maxillae, and the posterior one third is composed of the horizontal plates of the right and left palatines. The soft pa late is made up of the aponeuroses and fibers of the tensor veli palatini levator veli palatini and uvulae muscles (Scheuer and Black 2000) ; the 1 The premaxilla debate can be traced as far back as Vesal ius in 1543 (Ashley Montagu, 1938 ). Since that time, the literature is divided between those that confirm the presence of a separate premaxilla in humans, similar to other primates, and those who deny its presence, setting humans apart from non human primates. This debate is only pertinent here in how it relates to the incisive suture, and whether or not this suture is functionally and developmentally the same as the other sutures of the hard palate or is simply a fissure in the maxilla. For this research, the incisive suture is considered as a sutural junction between the maxillae and prem axillae.
78 muscles of the soft palate are listed and described in Table 3 2. The only muscles that connect to the hard palate do so via the soft palate at the posterior palatines; no muscles have direct origins or insertions on the hard palate. The maxillae are paired bones that form the floors of the eye orbits, the floor and lateral walls of the nasal cavity, and the ro of of the mouth. The maxillae also house sinuses (one on each side) and the upper dentition 10 deciduous teeth and 16 permanent teeth. The left and right maxillae form the majority of the hard palate, joining with each other at the anterior median pala tine (AMP) suture and with the palatines at the transverse palatine (TP) suture (Figure 3 1 ). The maxillae also articulate with the frontal, nasals, lacrimals, ethmoid, inferior nasal conchae, vomer, zygomatici and sphenoid. The anterior portions of the maxillae articulate with the premaxillae at the incisive (IN) suture, though this suture is often fused and not externally visible in adults. The palatines are paired bones that form the posterior portion of the roof of the mouth and walls and floors of t he nasal cavity. The left and right palatines are joined with each other at the posterior median palatine (PMP) suture and with the maxillae at the TP suture ( see Figure 3 1 ). The TP suture descends into the greater palatine foramen (GPF), which is forme d at the lateral junction of the alveolar process of the maxilla and the horizontal plate of the palatine on each side. The palatines also articulate with the vomer, inferior nasal conchae, ethmoid, and sphenoid. Sutures are the fibrous joints that inte rlock to form tight connections between adjacent bones in the hard palate, permitting flexibility in growth maximum durability in adulthood and simultaneous movement and cohesion (Rogers 1984 ; Thilander 1995) Palatal sutures change in morphology duri ng growth, generally progressing from wide
79 and straight to narrow and more sinuous. In adulthood, the two main sutures of the palate are the median palatine and transverse palatine (see Figure 3 1); the incisive suture is less prominent as it has a tende ncy to fuse in late adolescence or early adulthood (Mann et al. 1991) Facial Growth and Development The facial skeleton supports the orbits, nose, jaw, and soft tissues of the face, as well as vision, olfaction, respiration, and alimentation. The face is composed of fourteen bones that are formed via intramembranous ossification and joined by sutures. Six of the bones are paired maxillae, lacrimals, nasals, inferior nasal conchae, zygomatics, palatines, and two are unpaired vomer and mandible The growth of the face is complex and highly integrated ( Hinrichsen and Storey 1968 ; Enlow and Hans 1996 ) None of the elements that make up the facial skeleton grow or exist in isolation, and growth is related to complex biological and mechanical demands i n the entire cranial system. Growth is a composite change of all components in this system, although some areas of the cranium and face contribute greater percentages to total growth than others (Enlow and Hans 1996) An increase in size is connected to complex remodeling changes that ensure that craniofacial shape, proportions, and relationships among elements are maintained (Enlow 1966) The timing of facial growth is covered below as well as the growth of the elements of the palate. The coordinate d growth of multiple elements of the skull is important in maintaining the functional integrity of the entire system. The face itself is built on the cranial base, thus the growth of the face is intrinsically linked with the growth of the neurocranium. T he human brain is a powerful driving factor in facial growth because it
80 is the most rapid growing organ in infancy (Bogin 1999: p 72) Additionally, the coordinated growth of teeth and the craniofacial complex is important in maintaining the functional i ntegrity of the masticatory system. Without this harmony between teeth and bon e growth, an individual would not survive (Bogin 1999) There are important size and shape differences between prenatal, infant, child, and adult facial forms, and the relati ve positions of facial structures change with age (Feik and Glover 1998) ; the adult face is not simply a larger infant face. An example of these size and shape differences can be seen in the changing relationship of the inferior borders of the nasal cavi ty and the eye orbit. At birth, these two landmarks are at approximately the same level. Displacement and drift during childhood and adolescence steadily increases the distance between these landmarks. This results in the transition of a short, small, a nd round neonate face to an elongated, larger, and more rectangular adult face. Growth of the facial skeleton is dominated by increases in height, followed by depth, and then width. Overall, the face appears to grow downward and outward as unit, but this only describes the direction that the face moves in relation to the rest of the cranium, not the direction of growth (Brodie 1941) Facial bones have surfaces that are resorptive in nature, which are complemented by depository surfaces, meaning that the face actually grows from behind and above, displacing elements down and to the front (Enlow 1966) The location of these paired surfaces allows for integrated remodeling and growth of the face. Because of the varied regional orientations of the face and the complex relationships of shape and dimensions among the skeletal elements, the pairing of resorptive and depository surfaces is not easily summarized directionally. Each bone
81 grows in specific ways in order to compensate for size and shape changes in the growing face. The V principle describes the general progression of skeletal growth through both remodeling and displacement (Enlow and Hans 1996) (Figure 3 2 ). For facial elements, deposition occurs on the internal surface of the V (+ symbols), an d resorption occurs on the outer surface ( symbols). Together, these deposition and resorption events represent the process of remodeling. The gray V shapes depict the former locations of the element and indicate displacement during growth. The directi on of growth is from the narrow end of the V to the wider end (indicated by the solid black arrow). In the facial elements, it is not simply the anterior surface that is resorptive and the posterior surface that is depository ; even a single element can ex hibit multiple orientations for resorptive and depository surfaces. However, the V principle provides a descriptive framework for interpreting skeletal growth; in the face, the surface that points the same direction as growth is depositional, while the su rface that points away from the direction of growth is resorptive. Craniofacial Growth Models Disagreement s on the number of facial ossification centers and how craniofacial growth progresses highlight some of the difficulties in modeling craniofacial gro wth ( Woo 1949 ; Scheuer and Black 2000) There are two main, competing models of craniofacial growth: nasal septum (Scott 1953; Scott 1954) and functional matrix (Moss 1968; Moss and Salentijn 1969) These models describe what happens during growth and offer explanations of some growth stimuli, but neither addresses the influence of genes or the impact of the interaction of genes and environment on the growth process. While these are integral to the growth and development of any
82 organism and cannot be separated from what happens during these processes, much like age is the result of complex gene and environmental interactions (see Chapter 2), an investigation of the cellular and genetic mechanisms of facial growth is beyond the scope of this research Prior to the development of the nasal septum and functional matrix models, the predominant idea was that sutures were intrinsic growth sites, pushing apart cranial elements as they grew, similar to the growth plates seen in long bones (Moss 1969) It is now understood that the stimulus for bone growth/remodeling along a suture is actually related to tension that is produced by displacement of the bone where the suture is located and not by the suture itself. In fact, compression at a sutural junction actually leads to resorption of the bone, not bony deposition (Enlow and Hans 1996) While sutures are sites of osteogenesis and major growth centers of the cranium, they are not the driving force of displacement in the cranium nor are they independent and self initiating growth sites (Lenton et al. 2005) The nasal septum model is based on the displacement of the maxilla, which leads to bony growth/remodeling (Scott 1953; Scott 1954) The observation of abnormal growth in the midface following the removal of or damage to the nasal septal cartilage forms the basis for this model (Scott 1953) Growth of cartilage in the nasal region causes tension in adjacent sutures, resulting in the anterior and inferior movement of facial bones from the cranial base and one another, with the exception of the mandible. This is possible because cartilage is adapted to pressure related growth sites, and it can allow growth in compression (Enlow and Hans 1996) In this model, cartilage has intrinsic growth potenti al (i.e., it is a growth center).
83 Opponents of this theory argue that this is a singular explanation for a multifactorial process, and it does not account for the remainder of craniofacial growth (Enlow and Hans 1996) For example, the cartilage of the mandibular condyle is not a growth center even though it participates in growth early in life and has the ability to absorb pressure forces later in life (Thilander 1995) Unlike the calvaria, the facial skeleton does not have a major organ like the bra in to drive growth, and although the nasal septum theory attempts to adapt this concept to the facial skeleton, it really only applies to the nasal region. In the absence of the nasal septum, while the nose does not grow, maxillary development is still fa irly normal (Feik and Glover 1998) The functional matrix model states that bones react to changes in the functional units they support, and it places the initiation of growth on the enclosing soft tissues of the cranium, with the skeletal components foll owing suit (Moss 1968; Moss 1969; Moss and Salentijn 1969) In the cranium, multiple functions take place, and each function is carried out by a particular functional matrix and protected or supported by a skeletal unit (Moss and Salentijn 1969) For example, midfacial growth is driven by respiratory function, and growth of the masticatory apparatus drives growth of the mandible and parts of the maxilla. In this model, the nasal septal cartilage is a locus of secondary, compensatory, and mechanical g rowth rather than an initiator of growth (Thilander 1995) The concept of the functional matrix is an important component of functional morphology and the interpretation of function from form. One of the points of opposition to the functional matrix th eory is that functional forces do not operate in utero and therefore have no effect on growth during this time (Mooney et al. 1989) The nasal septal cartilage theory purportedly accounts for this
84 prenatal growth. However, studies have found that there are fetal facial and mastication functions that are carried out, albeit small (Humphrey 1969) Furthermore, nasal septal cartilage, while important in prenatal growth, shows less clear contributions to postnatal growth, though it does play a significant biomechanical role in maintaining normal midfacial form (Thilander 1995) Both the nasal septum and functional matrix models posit that passive skeletal changes of the face are related to active growth of adjacent soft tissues (Mooney et al. 1989) Wh ere they differ is the role played by the nasal septum versus the functioning spaces of the face (e.g., respiration, alimentation) and what is causing the force that leads to tissue separation (soft tissue matrix or cartilage) (Mooney et al. 1989) There may also be a difference in growth during the prenatal and postnatal time periods, which would result in a better model fit based on what material is being studied. However, both models demonstrate that facial skeletal growth is the result of the growth of other structures and is not mediated by an intrinsic force. Timing of Craniofacial Growth Normal craniofacial growth patterns are studied in order to better understand the patterns themselves, their timing, and deviations from the norm. Studying human craniofacial growth presents several challenges, one being access to study material because the initiation of growth centers commences in utero Histological, cephalometric, radiographic, and other types of studies contribute to the knowledge of growth ti ming and patterns pre and postnatally and form the basis for current knowledge in this area. The timing of growth and growth patterns for the face are outlined temporally below, starting with the prenatal period.
85 Prenatal craniofacial growth Prenatal g rowth of the face involves the coordination of multiple specialized tissues. Facial growth begins with neural crest cells from the brain, which migrate to form the facial growth centers and later to form connective tissue (cartilage, bone, and ligaments). These growth centers are located first in the pharyngeal (branchial) arches, which are composed of mesenchyme covered externally by ectoderm and internally by endoderm. The pharyngeal arches provide the framework for future development, and each of the five bilateral pairs of pharyngeal arches gives rise to specific facial bones with all of their associated veins arteries, nerves and muscles. The first pharyngeal arch is the origin for the maxilla and mandible and all muscles for mastication. The fac ial growth centers from the first pharyngeal arch form around the stomodeum (the primitive mouth). These growth centers give rise to the frontonasal, maxillary, and mandibular prominences, which grow and later fuse to form the face. At four weeks in uter o the stomodeum marks the future location of the mouth, the entire head is composed mainly of brain with thin layers of ectoderm and mesoderm, and the eyes are located on the lateral surfaces of the head (Enlow and Hans 1996) By the end of the fourth we ek and into the fifth week of intrauterine life, the basic organization of the face commences as the frontonasal prominence and the paired maxillary and mandibular prominences from the pharyngeal arches come together (Barnes 2012) Skeletal growth is i nitiated in the facial skeleton following the appearance and fusion of facial growth centers that lay the framework for the basic form of the face. The appearance of ossification centers in utero taken from Enlow and Hans (1996) is : maxilla end of we ek six; premaxilla seven weeks; mandible six to eight weeks;
86 zygomatic eight weeks; nasal eight weeks; lacrimal eight and a half weeks. The facial bones then enlarge from their ossification centers. The greatest period of growth of the facial b ones is from 24 mm to 36 mm crown rump length stage (approximately 9 to 10 weeks gestational age) (Avery and Devine 1959) During this time, growth is greatest in the antero posterior plane, with limited vertical growth (Avery and Devine 1959) The for m of the face is visible between four and ten weeks in utero (Scheuer and Black 2000) Prior to 14 weeks in utero facial bones grow from their ossification centers, but no significant remodeling takes place (Enlow and Hans 1996) Prenatal remodeling do es occur, but the majority will not take place until morphologically definitive skeletal elements appear, which is at or after 14 weeks (Enlow and Hans 1996) Some remodeling may occur as early as week 10, but it is limited to two locations: bone around tooth buds and the endocranial surface of the frontal bone (Enlow and Hans 1996) From week 14 onward, growth entails enlargement and remodeling, although large scale growth and development of the facial skeleton, with the exception of the eyes, is rela ted to dental and masticatory muscle development. These functions are not given priority in fetal development, resulting in the large head and eyes to small face proportions seen in human newborns (Scheuer and Black 2000) Postnatal craniofacial growth W hile prenatal growth is important in establishing a baseline form, work by Richtsmeier and colleagues (1993) demonstrates that postnatal growth patterns contribute significantly to adult form, and the majority of facial growth occurs postnatally. At birth the skull as a unit is closer to adult size and proportions than any other skeletal element, but the calvarium is six to eight times the size of the face due to a difference in
87 growth rates between the neuro cranium and viscerocranium (Brodie 1941) Elem ents that make up the neurocranium follow a neural growth curve, which preferentially accommodates rapid growth required by the large brain of human infants (Briggs and Martakis 1998; Feik and Glover 1998) The elements of the viscerocranium follow an S shaped somatic growth curve, displaying slow prenatal growth followed by a period of rapid growth postnatally, similar to growth of the postcranial elements (Feik and Glover 1998) By birth, the craniofacial skeleton has completed 30 60% of its total gro wth; following birth the size of the neurocranium increases by about 50%, as compared to a 200% increase in height and a 75% increase in width of the facial skeleton (Thilander 1995) Growth of the face is most rapid during the first three years of life (Feik and Glover 1998) The development of the dentition and masticatory muscles greatly influences postnatal facial growth and development. The emphasis on brain development at the expense of facial development in the infant skull precludes the early development of adult sized dentition, thus two sets of teeth are found in humans: deciduous and permanent (Rogers 1984) There are four distinct time periods of dental eruption that are related to facial growth and development: deciduous teeth during t he second year of life; permanent incisors and first permanent molars between six and eight years; permanent canines, premolars, and second molars between ten to twelve years; and third molars around eighteen years (White et al. 2012) At each of these s tages, the facial skeleton must accommodate these new additions by increasing in size and altering in shape.
88 At or around the time of puberty, the facial skeleton also undergoes changes that are related to the acquisition of secondary sex characteristics. Adult female faces are smaller and have a more rounded contour than males (Rogers 1984) Changes in brow and chin shape related to greater robusticity in males as compared to females also occur. The face is considered to have reached skeletal maturity between 12 to 15 years in males and 10 to 13 years in females (Feik and Glover 1998) Adult craniofacial growth Facial growth and development does not cease in adults, though the pace is greatly decreased, and changes in the adult face may also be related to degeneration, disease, loss of the teeth, and the accumulated effects of masticatory stresses, rather than growth. Much of the aging of the face has been discussed in terms of soft tissue change (e.g., wrinkling) but adult facial growth still occurs f or skeletal components and dentition. While growth as a process is generally attributed to subadults, growth of the head and face do occur in adulthood, with changes in facial size and shape occurring largely between 16 30 years (Albert et al. 2007; Behr ents 1985) Changes in facial form can continue into the fifth and sixth decades of life, though the rate of change is much slower than that of children and adolescents, as is the order of magnitude (1 2 mm over several years to several decades) (Behrent s 1985) Displacement no longer occurs, but the craniofacial skeleton changes in horizontal, vertical, and sagittal dimensions via remodeling (Albert et al. 2007) While changes are small, they can be clinically significant, especially for the implanta tion of prosthetic devices. A sexually dimorphic trend exists, with female faces exhibiting increased vertical growth over time as compared to males, who exhibit increased horizontal growth (Albert et al. 2007)
89 Females also show greater shape changes, perhaps due to hormonal changes associated with menopause ( Doual et al. 1997 ; Albert et al., 2007 ) Secular Changes in Craniofacial Growth In the United States, secular changes in cranial and craniofacial morphology include an increase in cranial capacity vault height, base length, and total length, and a decrease in vault and facial width and cranial vault thickness (Angel 1976; Nawrocki 1995; Jantz and Meadows Jantz 2000; Jantz 2001; Wescott and Jantz 2005) Shape changes of the vault are more pro nounced than those of the face, but facial dimensions do show a trend of narrowing and becoming higher over time (Jantz and Meadows Jantz 2000) Additionally, cranial shape changes are more pronounced than size changes. Change is often attributed to impr ovements in health and nutrition, such as access to better healthcare, decreased prevalence of diseases, and increased caloric intake. However, while these environmental factors can certainly contribute to taller stature and cranial height in U.S. populat ions, they do not explain the narrowing seen in the cranium and face (Skorpinski 2014) It is possible that changes of the face are related to changes in the cranial base, since an increase in cranial capacity does result in inferior movement of the cran ial base (Jantz 2001; Wescott and Jantz 2005) but th is is not the only explanation. There are intrinsic difficulties in interpreting the effects of better nutrition and health on secular change in the United St ates. While ample information is available for documented skeletal collections, informati on on nutrition was not routinely collected and is thus not available Many skeletal collections, and specifically those considered to be historic (individuals born in the 19 th century), are composed of indiv iduals of low
90 socioeconomic status (Hunt and Albanese 2005) These individuals may exhibit less positive growth changes as they likely had decreased access to improvements in health care and alimentation when compared to individuals of higher socioeconom ic status. The composition of the reference samples therefore may bias the interpretation of potential secular trends if these trends are rooted in improved health and nutrition. Another alternative explanation for secular change in cranial shape and si ze is a change in the consistency of food consumed and not in its nutritional value (Jantz and Meadows Jantz 2000; Wescott and Jantz 2005 ; Skorpinski 2014 ) With the introduction of processed foods to the American diet in the mid 20 th century, the over all quality of food has decreased, as has its toughness. More nutrition, including overnutrition, does not necessarily mean better nutrition, and changes in size and shape could be related to eating less tough or gritty foods, which represents a biomechan ical, dietary explanation ( Wescott and Jantz 2005 ; Skorpinski, 2014 ) In fact, Kiliaridis et al. (1985) find that rats fed food with a softer consistency did show changes in craniofacial morphology as compared to the control group, which was fed a standa rd laboratory diet of pellets The effect of diet on palatal form is discussed in further detail below. The above explanations for secular change are purely environmental. However, there is also a genetic component to growth and development, and Jantz (2 001) attributes observable secular trends to a combination of phenotypic plasticity and genetic change over time. With the improved health hypothesis, individuals growing up in the same environment should exhibit the same change even among different ances tral groups. Yet, secular changes in facial dimensions are not the same between European and African Americans. The increase in facial height over time is only
91 significant in Europeans, and facial depth, the anterior to posterior dimension of the palate, shows an increase in European Americans and a decrease in African Americans (Angel 1976; Jantz and Meadows Jantz 2000 ; Jantz, 2001 ) similar morp hology even though they are largely living in the same environments with similar access to better food and healthcare (Jantz and Meadows Jantz 2000) Sparks and Jantz (2002) performing a re analysis of cranial and facial measurements work on cranial form (Boas, 1910) find that environmental factors on cranial form are minimal when compared to differences seen among ancestral groups (i.e., there is less difference between a European born immigrant and his/her American born child in te rms of cranial measurements than there is between individuals from different ancestral backgrounds). Their study suggests a high heritability of cranial form traits and not a large amount of phenotypic plasticity, as Boas claimed. Conversely, Gravlee et al. (2003 a; 2003b ) in their own reanalysis of plasticity and not the high heritability of cranial traits though they do not argue against the heritability of cranial form Thus differences in cranial form, of the lack thereof, may be interpreted differently, though what these studies do suggest is that multiple variables are likely contributing to cranial form, rather than solely heredity or environment. The invest igation of secular trends in craniofacial growth needs to adequately address multiple variables, which is challenging. However, by collecting data that reflects genetic contributions to size and shape, such as the sex and ancestry of
92 individuals, and envi ronmental factors, such as dental wear as a proxy for changes in food consistency, it is possible to better understand observed trends in terms of secular change. These changes do not occur in isolation and, much like growth and development, are the resul t of complex interactions of genes and environment. Palatal Growth and Development This section discusses the growth and development of the hard tissue components of the palate as well as the aspects of the dentition that directly contribute to growth and development of the palate. The timing of palatal growth and development, including some key events in dental development that affect the hard palate, is given in Table 3 1, compiled from Scheuer and Black (2000) Like the stud y of craniofacial growth pa latal growth is often studied to inform abnormal growth patterns, such as cleft palate. Prenatal Palatal Growth and Development The first maxillary and palatine ossification centers appear early in intrauterine life and are located at the anterior and me dial aspects of the nasal capsule, respectively. The primary palate, also described as the premaxilla, is visible by the end of the fourth week in utero Another ossification center in the premaxilla appears between nine and ten weeks. Woo (1949) identi fies three ossification centers for each maxilla (six total), with one of these for the maxilla itself and the other two for the premaxilla, and research by Avery and Devine (1959) confirms the presence of at least two premaxillary growth centers based on histological observation of normal and cleft palate embryos. The palatal shelves form from the maxillary processes at four to six weeks in utero
93 the oral cavity, but ru n alongside it), and the tongue is located in between them because of the small size of the oral cavity. The transition from vertical to horizontal shelves occurs between seven and eight weeks in utero and is related, in part, to the expansion of the inf erior portion of the lower face. Mouth opening reflexes commence prior to palatal shelf elevation and have also been shown to significantly affect tongue withdrawal from the vertical shelf via traction from mandibular depression (Humphrey 1969) though t the movement of the tongue inferiorly, a vacuum is created that then draws the shelves towards midline; this also results in a larger oral cavity. Once the tongue has descended, th e palatal shelves expand horizontally and join at midline. Bone then forms at this location, which becomes the secondary palate. The secondary palate displays an antero posterior gradient of palatal closure 2 beginning at the primary palate (Burdi and Fai st 1967) During the time of palatal shelf elevation and closure, there is also a significant increase in overall depth and height of the facial region, and the palate moves into a position that is approximately 90 degrees relative to the cranial base, s imilar to its postnatal position (Diewert 1983) At the time of midline closure, the maxillary processes fuse with the nasal septum and posterior portion of the primary palate. Cleft palate anomalies often occur during this time frame (see Palatal Devel opmental Anomalies, below). The palatine processes meet in the midline later than the maxillary processes at about 18 weeks in utero The maxilla does not completely ossify until late fetal life nor does it attain adult form or proportions by birth, and the premaxilla remains separate from the maxilla until 2 Note that closure here refers to the joining of the adjacent sides of the secondary palate, not the obliteration of the sutural junctions at this location.
94 about month four or five in utero By the sixth month in utero the facial aspect of the incisive suture is closed. During fetal life, deposition of bone occurs mainly on the anterior maxilla (labial surface), and the fetal maxillary arch lengthens horizontally in posterior and anterior directions (Enlow and Hans 1996) The palatine reaches adult form by mid fetal life, although it does not reach adult proportions until later (i.e., the perpendicula r plate ends up being much greater in height due to changes in the size of the face and nasal cavity). Fetal palatal growth is tied very closely with fetal dental d evelopment ( Table 3 1). While in utero tooth germs develop. As the teeth begin to form, bone also forms around them. This bone extends on the buccal and lingual surfaces and between the teeth in the form of thin walls. This bone growth is the foundation for the alveoli. Sutural growth is also an important component of prenatal palatal growt h and development. In a histological examination of the mid palatal suture, Latham (1971) finds that the interpremaxillary suture develops almost coincident with the premaxillary ossification centers and is definitively established no later than seven wee ks in utero Sutural formation along the maxillary mid palate is present at 10.5 weeks, with definitive formation of an intermaxillary suture by 12 weeks (Latham 1971) The difference in timing of sutural appearance is likely related to union of the sep arate right and left halves of the premaxillae and maxillae, with the former occurring several weeks prior to the latter. Following sutural formation, sutural growth occurs until 16 weeks in utero with growth and remodeling occurring together after this time and continuing postnatally. Based on macroscopic examination of the premaxillary area, Sejrsen et al. (1993)
95 conclude that the development of the incisive suture is most likely related to the development of the anterior teeth. Postnatal Palatal Growt h and Development At birth, there is no longer any evidence of a separate premaxilla on the facial aspect of the cranium, though the internal surface of the palate often shows evidence of this separation (the incisive suture or fissure). Also at birth, o nly one primary center of ossification is still present on each side of the maxilla, the maxilla has a small body with tooth germs close to the orbital floor, and the maxillary sinuses are small (Scheuer and Black 2000) The infant palate is composed mai nly of fine cancellous bone that quickly remodels in conjunction with changes in the dentition, nasal cavity, and eye orbits, contributing to rapid growth. In the immediate postnatal period, the fine cancellous bone of the palate is replaced by cortical bo ne with medullary spaces, and the medial ends of the palatal processes gradually thicken (Latham 1971) During the first two years of life the inferior cortical layer of the palate remains cancellous due to deposition on this surface, the intermaxillary suture increases in height while also narrowing, and the sutural margins become parallel and exhibit continuous cortical bone (Latham 1971) At around three years of age compact cortical bone and clear medullary spaces are seen in the thickened medial ar ea with sutural tissue composed of fiber bundles running parallel to the sutural bone margins (Latham 1971) Between years two and four midpalatal sutural growth slows and then s tops. Also within the first two and a half years following birth, the decid uous dentition erupts (Hillson 1996) The maxillary dentition emerges slightly later than the mandibular.
96 The maxilla has the appearance of growing out and down, but it is actually the superior and posterior appositional growth of this element that cause s both the lowering of the nasal floor and the continued anterior projection of the alveoli and nasal spine in childhood, resulting in a downward and forward pattern of displacement (Brodie 1941) The postnatal palate, largely dominated by the maxilla, g rows in three dimensions: length (anterior to posterior), width (transverse), and height (superior to inferior). Growth of the palate takes place via paired processes of deposition and resorption that involve different facial components for each dimensio n. Transverse growth occurs via expansion of the midpalatal suture through deposition with some resorption, remodeling and resorption along the labial and buccal surfaces of the alveolar border, and deposition along the lingual surface of the maxillary al veolar bone. Bony deposition on the buccal surface of the maxillary tuberosity (posterior alveolar border) also affects width. Growth in length occurs via expansion and bony deposition at the transverse palatine suture and maxillary tuberosities, as well as remodeling of the maxillary tuberosity, while the anterior (labial) surface of the maxilla is resorptive. Growth in height occurs via remodeling and resorption of the nasal side of the hard palate with coincident bony deposition on the oral side of th e palate; dental eruption also plays a large role in height growth. While growth of the face is largely affected by facial structures, the expansion of the middle cranial fossa of the neurocranium also affects growth in all three dimensions. Growth in wi dth via deposition at the alveolar margins generally ceases around age 7, growth in height generally stops after age 9, and growth in length occurs largely in adolescence until adult life, mainly driven by dental development (i.e., t he emergence of the thi rd molars).
97 Dental development and eruption affects palatal growth and development because of the close relationship between teeth and bone. Eruption is the process by which teeth move towards the occlusal plane, and it concludes when the tooth reaches th at plane. All maxillary permanent teeth develop superior to the deciduous dentition, with the exception of the second and third molars. The eruption of the permanent teeth therefore serves to displace the deciduous dentition and alter the alveolar bone. Alveolar bone is also altered by root growth since the roots continue to grow after the tooth has erupted. Continued eruption of the teeth results in mesial drift, where the ng up space in the back of the mouth for more teeth. The combined processes of eruption and drift result in adjustment of the dentition in relation to the face, which is permitted by the periodontal ligament. The relationship between the periodontal liga ment and the alveolar bone is also important because it allows the bone to remodel in order to adapt to changing forces in the mouth (see also Mastication, below) In the period from childhood to adolescence, palatal sutures changes from simple straight li nes to lines of increased complexity. The transverse palatine suture develops into a squamous suture, while the median palatine becomes sinuous and interdigitated (Melsen 1975) The palate also changes from flat to increasingly concave on its interior s urface. The palatal surfaces of younger individuals tend to have rough, bumpy surfaces that smooth with increasing age (Bass 2005) and Mann et al. (1987) find that in advanced age the maxilla has a flat and smooth lingual surface. Growth generally ceas es once the palate has attained adult size and shape, which is most often around the age of 18 (Scheuer and Black 2000)
98 Palatal Developmental Anomalies Cleft palate is one of the most commonly researched developmental anomalies of the palate, and it is the failure of one or both maxillary prominences to merge with the fused nasal prominences. This anomaly is associated with problems prior to cell differentiation and the disruption of growth processes following this stage (Scheuer and Black 2000) Gro wth disruption can result in clefts of the soft palate only, clefts of the soft and hard palate, complete unilateral clefts of the lip and palate, and complete bilateral clefts of the lip and palate and can affect the primary and secondary palate (Kirschne r and LaRossa 2000) Palatal clefts are troublesome because they can lead to problems with alimentation and respiration, as well as issues with hearing, den tition speech, and midfacial growth. A cleft palate indicates a disruption to normal palatal de velopment and can be attributed to a number of different factors. Most often the problem occurs at the time of elevation of the palatal shelves (see Prenatal Palatal Growth and Development, above). During this time period, the following things can occur to result in cleft palate: inhibition of cell division/migration, which means the palatal shelves are too small to meet at midline; failure of shelf elevation at correct time; excessive head width, which is also related to sex differences since females pa late elevation occurs approximately one week later than males; failure of shelf fusion; or post fusion rupture (Ferguson 1987) A failure to displace the tongue can also cause a cleft. The etiology of cleft palate is not entirely understood. There can b e environmental factors that affect palatal growth, and certain types of clefts can be associated with genetic disorders. However, most clefts are multifactorial in origin or result from a mutation or change at a major single gene locus (Kirschner and LaR ossa
99 2000) There are also demographic differences in the occurrence and frequency of cleft palate. According to Kirschner and LaRossa (2000) Asians are most likely to exhibit cleft lip and palate, followed by Whites, then African Americans, and cleft palate occurs more commonly in females (Barnes 2012) The hard palate can also exhibit developmental cysts. There are three types of cysts seen in the hard palate: median anterior maxillary in or near the incisive foramen; median palatal between the palatal processes (in the vicinity of staurion ); and globulomaxillary at the lateral junction of the premaxilla and maxilla between the lateral incisor and canine (Barnes 1994) These cysts form when there is a delay in the retraction of overlying ect odermal tissue during development or, in the case of the globulomaxillary cyst, in relation to odontogenesis ( Little and Jakobsen 1973 ; Barnes 1994) A retraction delay results in the retention of epithelial tissue between skeletal elements as they unit e, forming a cyst that is lined with epithelium and contains either a fluid or semisolid substance (Little and Jakobsen 1973) Cysts are generally rounded or oval, although some can be irregular in shape, and may or may not have sclerotic margins. They are not associated with any significant functional problems and appear to be largely asymptomatic (Stafne 1969) Palatal Function When describing and measuring loads there are two basic concepts employed: stress and strain. Stress is the normalized int ensity of a force, and it is measured as the load per unit area (Swartz 1991) Strain is the physical change in dimension of a loaded body, and it is quantified by measuring the deformation of a given body when a load is applied (Swartz 1991) Stress i s often much more difficult to measure, especially in biology, so strain is more commonly used. Strain can be predicted
100 theoretically via engineering models or measured in extant and extinct organisms through strain gage analysis. In strain gage analysis small gages measure the amount of deformation of a material, such as a skeletal element, when loaded. The cranium protects the brain and other organs from impact loads, and the craniofacial region as a whole is subjected to a wide variety of loading fact ors. These factors are difficult to measure or describe due to the irregular shapes and thicknesses of cranial elements, presence of fibrous joints (sutures), varied muscular and occlusal loads, and cross bracing in the skull (Herring and Ochareon 2005) There is no overall pattern of cranial deformation, making it difficult to study the skull as a single mechanical entity (Herring and Ochareo n, 2005) a uniform and optimal strain environment cannot be assumed, and regional or single element approaches ar e very important when investigating form and function in the skull since strain regimes can be region and element specific ( Hylander and Johnson 1997 ; Herring et al. 2001) Functional loading refers to routine activities rather than infrequent traumati c loads, though infrequent traumatic loads are certainly a consideration when it comes to resisting damage to tissue (Lanyon and Rubin 1985) Based on in vivo studies, most craniofacial bones experience a level of strain comparable to limb elements, but the strain pattern differs from long bones because the most common pattern of strain is shear or torsion, rather than bending, and strain magnitudes are very different depending on the location of the skeletal element and applied force (Herring and Ochareo n 2005) Additionally, the orientation of maximum stiffness in cranial cortical bone and the anatomical axis of the structure are not coincident (Herring and Ochareon 2005; Wang and Dechow 2006) and the assumption that craniofacial bones do not
101 receiv e heavy loading is a false one ( Hylander and Johnson 199 7; Herring and Ochareon, 2005 ) In examining loads in the cranial region, it is important to consider not only the type of force and load, but also the frequency and magnitude of that load. Mechani cal loading is also imperative for bone growth and maintenance, and loss of muscle function w ill also lead to altered loading (Herring 2007) (See Chapter 2, Mechanics in Bone Growth). Mastication While the skeletal elements of the maxillae and palatines s upport alimentation, respiration, and sight, the primary function of the palate is to provide str uctural support for mastication The main loading on the palate and the cranium as a whole is from chewing related activities mastication and incision, whic h are largely cyclic in nature (as opposed to static). During mastication, the elevator muscles of the mandible pull it superiorly, while depressors perform the opposite function. T hese loads also twist the mandible and maxillae, and cause shear in the m axillae, tension in the palate, and compression in the nasals (Herring 2008; Herring and Ochareon 2005). Loading on the maxilla is indirect since it occurs primarily on the teeth Outside of the palate, the masseter muscle pulls downward on the zygomat ic arch, creating vertical tensile strains at its origin, and the downward pull of the masseter plus the upward force on the teeth creates sheering stress under the eye orbits (Rogers 1984) Strain is minimal in the brow ridge (forehead) and higher in th e zygomatic arches, mandible, and around the eye orbit (Rogers 1984; Wroe et al. 2007) The facial skeleton also undergoes torsion during mastication, and strain depends on which side is loaded (i.e., which side the individual chews on). The muscles in volved in mastication are listed in Table 3 3.
102 The maxilla is structurally and morphologically very different from the mandible, even though both house the dentition. The palate is braced by the zygomatic and temporal bones in order to match in strength t he impact applied by the mandible via the masticatory muscles (Rogers 1984) and no masticatory muscles attach d irectly to the hard palate ( Table 3 3). Compared to the mandible, the bone of the m axilla is actually less stiff and the maxilla is subjected to less stress during mastication and incision (Hotzman 2010). Values for mandibular bone mineral density are twice as high as compared to the maxilla, and the maxilla has thinner and less cortical bone than the mandible (Devlin et al. 1998) Even thou gh there are no direct masticatory muscle attachment sites on the bone of the hard palate, the periodontal ligament plays a crucial role in mastication. This ligament is the connection between the teeth, where direct bite force is applied, and the alveola r bone. Bite force is a result of the jaw elevator muscles and is modified by jaw biomechanics and reflex mechanisms (Koc et al. 2010) While the masseter muscle is a large source of maxillary strain, tooth contact is also important (Herring et al. 200 1) When biting occurs, force is transmitted from the occlusal surface of the teeth to the alveolar bone via the periodontal ligament. This occlusal contact is an important part of mastication, and loads transmitted via the periodontal ligament dictate b one remodeling in the maxillary structure providing a type of feedback loop between the teeth and alveolar bone The application of force to the tooth and root is important in maintaining alveolar bone growth and development, and signaling bite reflexes that stop the chewing motion when a particularly hard substance is encountered.
103 The palate is part of a complex mechanical environment that Hotzman (2010) finds difficult to characterize. Attempting to model the palate as a shell, beam, and plate, Hotz man (2010) does not find that any of these models accurately predict experimentally obtained strain values. One of the reasons the mechanical environment of the maxilla is so difficult to characterize is because no direct forces are applied to the maxilla via muscle attachments, and it exhibits no direct articulation with the temporomandibular joint (Hotzman 2010) An additional complication is that bite force can vary depending on several factors, such as individual variation in sex, age, craniofacial m orphology, dental disease, dental restorations, and occlusion, and bite force values can vary depending on the recording method and device (Koc et al. 2010) For example, aging might lead to loss of muscle force though the effect of age on bite force is presumed to be small (Koc et al. 2010) Strain gage analysis is also particularly challenging in the palate, and measuring compressive and tensile forces on the internal and external surfaces of skeletal material for in vivo studies is less than perfect (Herring and Ochareon 2005; Wang et al. 2010) An additional concern in the relationship of form and function in the palate is that the facial skeleton may also be over adapted for routine food processing and therefore form may not fully reflect these behaviors (Daegling and Hylander 1997; Hylander and Johnson 1997) Humans exhibit low strains even during powerful biting, which when considered with the thick dental enamel that protects the teeth, may indicate that normal mechanical loads sustained du ring mastication may not have a large effect on facial form ( Swartz 1991 ; Hylander and Johnson, 1997) Based on their work with macaques, Hylander and Johnson (1997) find that the high strain areas of the
104 zygomatic arch are not exclusively correlated wit h thick layers of dense cortical bone, though the geometry and bone mass density in this area are related to countering loading from mastication. An optimized facial structure should display maximum strength with minimal material (e.g., robust bone struct ure in areas of high strain and decreased robusticity in areas of low strain ; Hylander and Johnson 1997) However, as with many skeletal structures that are limited by other factors such as growth restraints, the facial skeleton does not display an optim ized structure for the sole purpose of feeding. This can make the inference of function from form challenging. Though facial form and mastication are not perfectly correlated, facial morphology still reflects jaw movement (Herring 2007) Changes in jaw movement are transferred to muscles via ligaments and/or aponeuroses, which are transferred to the osseous structure, influencing, at least in part, its morphology. In fact, feeding is an important part of skeletal growth and development. I n experiments on American opossum crania, Thomason and Russell (1986) find that the secondary palate contributes significantly to torsional strength and stiffness of the rostrum and maxillae in latero medial bending. Once the palatal shelves meet in the midline, they d etect a notable increase in torsional strength and stiffness and conclude that the secondary palate is important to resist ing forces exerted on the upper dent ition during mastication. Therefore while the facial skeleton may not be optimally adapted to res ist masticatory forces, its form is related at least in part to function. Given the difficulty in interpreting the strain environment of the palate from experimental studies, theoretical models, and palatal morphology, it is useful to examine other variabl es that have the potential to serve as proxies and provide information on
105 forces in this skeletal structure. For example, since bite force and mandibular bone strain are highly correlated, mandibular strain measurements are informative for indicating bite force in the maxilla (Swartz 1991) However this still does not entirely describe the strain environment of the palate, which is structurally ver y different from the mandible. The contribution of diet Since t he human masticatory system is responsive to changes in food texture (Lucas et al. 2002) dietary changes have the ability to alter the cranial skeleton For example, d ecreased bite force, such as that seen when softer foods are consumed, may result in a less robust craniofacial structure, a narro wer palate, sutural fusion, and decreased dental wear. This is because softer foods require less force to process. Sutural fusion is disc ussed in further detail below. In an experiment comparing rats f ed a soft diet and those fed a standard l aboratory di et of hard pellets, Kiliaridis et al. (1985) find that the softer diet leads to a changes in cranial shape that include more anteriorly direct facial growth and increased facial height, though no overall difference in skull size are found. The soft diet g roup also displays a decreased growth rate in the gonial angle. Kiliaridis et al. (1985) conclude that masticatory muscles influence both local and overall cranial growth/remodeling and softer fo ods produce a narrower palate. Since the teeth come into dir ect contact with food and transmit bite force to alveolar bone, dental wear is potentially informative for interpreting the biomechanical environment of the masticatory system as well as looking at dietary change over time (Skorpinski 2014) Greater dent al wear could potentially be related to greater bite forces. For historic and prehistoric populations extensive wear may be related to large
106 bite force, required to process tough or gritty foods. In modern populations eating less tough or gritty foods, d ental wear could be reduced. Altered Function through Tooth Loss Tooth loss is generally interpreted as a sign of advanced age, although it can occur at any age due to infection or extreme wear, leading to disease and eventual tooth loss. Because of the r elationship of the teeth, periodontal ligament, and alveolar bone in transmitting forces during mastication, t he partial to complete loss of the dentition alters the biomechanical environment of the masticatory system. Most notably, the loss of teeth mean s a loss of direct loading on the occlusal surface of the teeth, which leads to a loss of loading on the periodontal ligament and the alveolar bone. Since osteoblasts reside in the lining of the tooth roots, the loss of teeth means the eventual loss of al veolar bone due to the absence of loading via the periodontal ligament. However, if only the crown of the tooth is lost, it is possible to maintain alveoli as long as the root is present. Tooth loss also can entail dietary changes, including the decrease d ability to process tough foods and a transition to a softer diet, which brings about decreased b ite force. For those individuals who live past tooth loss, remodeling and eventual resorption of the alveolar bone occurs, with the alveolar process reducing in height and becoming rounded or even sharp in advanced stages (Scheuer and Black 2000) Overall, the maxilla reduces in size. Morphological changes also result in a relaxation of the tongue to fill in areas where bone once filled and the approximation of alveolar bone margins in occlusion, changing the way the jaw moves. The loss of teeth can also weaken bone by decreasing bone density especially in individuals with osteoporosis or osteopenia Devlin et al. (1998) find a signi ficant
107 negative correlat ion between bone mineral density and age in the anterior maxilla and mandible. Bone mineral density changes whether due to increasing age, edentulism, or a possible interaction of the two will in turn impact the sustainable loads on the jaw, with decrease d bone density leading to earlier failure as bone is less able to resist loads of larger magnitudes. T here may be coincident structural changes in the palate to account for this loss of strength. These structural changes could include increased instances of fusion to increase strength or a lack of fusion to ensure flexibility in brittle bone. Stiffness of the bone could also be affected, with a decrease in bone density leading to less stiff bone and again a decreased ability to sustain the cyclic, high m agnitude forces resulting from mastication (Herring 2008) The maxilla is more regionally variable than the mandible in elastic properties when alveolar cortical bone, maxillary body cortical bone, and palatal cortical bone are compared (Peterson et al. 2006) Therefore, it is unclear to what degree changes in bone mineral density will a ctually affect the structure. Another consideration is the implantation or use of devices to aid edentulous individuals. In modern populations where dental care is prac ticed, the installation of dentures, implants, or other prosthetic devices is common. Dentures or prosthetic devices affect maxillary form because they alter the mechanical environment and bone, while implants are insert ed into the alveolar bone For dentures, loading is still a cyclic force on bone during mastication, but one that differs from forces experience d during occlusion of the dental arcade due to differen ces in material properties between the structure of teeth and that of the dentures. In the case of implants, the desired effect is one of osteointegration with bone that will closely mimic the missing teeth, but it
108 still does not perfectly imitate these s tructures since implants lack periodontal ligaments (Yacoub et al. 2002) Yacoub et al. (2002) investigate bone strain in craniofacial bone adjacent to and distant from dental implants in order to determine the effect of implants on the maxilla and other craniofacial bones. Their results indicate that implants affect an area much greater than just that adjacent to them, including the zygomaticotemporal suture and supraincisor cortical bone (Yacoub et al. 2002) Strain is transmitted through multiple ro utes from the implant site, and while no comparisons were made with non edentulous individual s in this study, it demonstrates t hat complex strain patterns are produced via dental implant loading (Yacoub et al. 2002) The Role of Sutures Sutures allow for a combination of strength and flexibility in the palate, and they play a key role in growth and mastication. Cranial sutures largely withstand cyclic forces of mastication and to a lesser extent soft tissue and organ growth (Jasinoski et al. 2010) The design of the cranium as a whole in terms of sutures, bone, and remodeling capability allows for rigidity while maintaining a structure that is not overly cumbersome (Yu et al. 2004) While sutures do not have intrinsic growth potential, they do allow fo r expansion of the palate, with growth occurring along sutural margins. During mastication, patent palatal sutures also enable movement of the hard palate in response to loading. Loading of the sutures is not possible until the sutural margins approach o ne another, thus linking growth and function (Herring 2008; Zollikofer and Weissmann 2011) The presence of sutures is one of the main reasons the skull cannot be considered as a single functional entity because loading in one area of the cranium is no t necessarily transmitted efficiently to other areas ( Herring et al., 2001; Herring and
109 Ochareon 2005) than bone, meaning they are more flexible and deform at lower stress/strain than bo ne ( Linge, 1970; Jasinoski et al. 2010) Sutures are also anisotropic, meaning that their mechanical response is dependent on orientation (Yu et al. 2004) Energy absorption is complex and may be related to sutural morphology, fiber orientation, and lo ading rate ( J aslow 1990; Rafferty and Herring 1999 ; Jasinoski et al., 2010 ) It has been suggested that sutures damp force transmission or absorb energy ( Jaslow, 1990; Herring and Teng, 2000; Herring and Ochareon 2005) Like force transmission in bone, force transmission in sutures is also regionally specific and affected by local muscle actions and signals rather than overall cranial loading ( Herring and Teng, 2000; Herring et al., 2001; Herring and Ochareon 2005; Zollikofer and Weissmann 2011) Spe cific sutures display specific strain regimes, and sutures rarely alternate between regimes (e.g., the midpalatal suture has a tensile strain regime; Herring and Ochareon 2005). The presence of the midpalatal suture bisecting the palate also may mean tha t force is not effectively transmitted across the entire palate during mastication. However, Herring et al. (2001) find that in pigs strains are not significantly different by side for left versus right side chewing pigs. The local strain regime determin es the structure of the suture, with compressed sutures being interdigitated and tensed sutures displaying a flat (beveled or straight) form (Herring and Ochareon 2005). These shapes related to the ways in which the collagen fibers of the suture attach t o bone: in tensed sutures fibers are straight or cruciate and in compressed sutures the arrangement is oblique (Herring and Ochareon 2005). Even if the suture as a whole has a particular strain regime, it should be noted
110 that because of the arrangement of sutural fibers, the load on the sutural margins is always tensile; without a tensile load on the margin, the bone would actually resorb (Enlow and Hans 1996; Herring and Ochareon 2005) Therefore even in a behavior like mastication that would intuit ively seem to produce compressive forces, the sutural margins do not sustain this compression. If sutures removed from particular strain environment, they have shown the ability to adapt to new environment, and the fibers will arrange so they are subjecte d to tensile stresses (Herring 2008; Jasinoski et al. 2010) Sutural fusion Sutural fusion is not well understood from a functional perspective (Wang et al. 2006) In humans, palatal sutures show a tendency to fuse, but usually do not completely oblit erate even in old age. Most often, fusion consists of small areas of bony bridging rather than entire sections of a suture. This means that some degree of patency is generally maintained throughout life. Patent sutures serve to segregate certain regions and result in a structure that is less rigid and not mechanically integrated (Wang et al. 2010) Experimental data from pigs indicate loaded patent sutures exhibit strains an order of magnitude greater than adjacent bone ( Herring and Teng, 2000; Herring et al., 2001; Herring and Ochareon 2005) though fused sutures do not appear to have a significant effect on bite force (Wang et al. 2010) There is an expectation that the strain environment where sutures are fused should be more unified since synosto sed sutures are still less stiff and less mineralized than surrounding bone and thus differ in mechanical properties ( Herring et al., 2001; Grau et al. 2006 ) Thus there is a basic structural difference between elements that are fused together and act as a single element versus elements that maintain separation but are connected via a fibrous
111 joint (Herring 1972) While fusion may strengthen skull structure, often the tradeoff is reduced flexibility Herring (1972). One functional hypothesis for sutural fusion posits that it is directly affected by loading. Osteogenesis along sutural margins via applied tensile force is the basis for orthodontic practices like midpalatal expansion, and Linge (1970) finds that expansion increases bony deposition at the mi dpalatal suture in rhesus macaques. Because tension at the sutural margins is required to maintain bone and stimulate bone growth, and cyclic loading has been shown to increase osteogenesis along sutural margins (Kopher and Mao 2003) a lack of loading c ould result in growth cessation and narrowing of the sutural margin. It is important to note that while tensile forces are osteogenic along sutural margins, there is a difference between growth at the margins, resulting in expansion, and fusion of the sut ure itself, effectively halting expansion. Narrowing can be the result of reduced masticatory function; for example, the consumption of softer foods, injury, or the loss of the dentition (edentulism). Comparing rats fed a soft diet with rats fed a stand ard laboratory diet of hard pellets, EngstrÂšm et al. (1986) find that in the soft diet group the internasal, nasopremaxillary suture, and interpremaxillary sutures exhibit narrowing, a decrease in bony spicules, and for the internasal suture, significant o bliteration. From these results, EngstrÂšm et al. (1986) conclude that decreased masticatory function, which alters the tension placed on bone from sutural and periosteal fibers, leads to sutural fusion. Hinton (1988) also finds that decreased masticatory loading via consumption of a soft diet or incisor clipping leads to decreased cartilage growth in the intermaxillary suture in rats, with the suture becoming largely fibrous rather than maintaining secondary cartilage.
112 Experimental studies do not point u nequivocally to a decrease in loading as an explanation for sutural fusion. Actual strain levels sustained in sutures may actually be far too small to induce osteogenesis in the first place (Henderson et al. 2004) and there are likely myriad factors con tributing to fusion beyond functional necessity. Additionally, and counter to reduced loading inducing fusion, Heller et al. (2007) find that applying mechanical stress to fusing posterior frontal and patent sagittal rat sutures results in significantly m ore fusion at locations subjected to oscillating compressive and tensile stress as compared to a static control. Finally, movement of the skeletal elements may be what retains patency. Latham (1971) suggests that the lack of synostosis in the mid palatal suture could be due to the slight movement of bones caused by the range of motion in mastication However, in humans there is not a wholesale lack of fusion in palatal sutures (e.g., Mann, 1987; Mann et al., 1991; Gruspier and Mullen, 1991; Wheatley, 199 6; Ginter 2005) so movement does not entirely account for sutural patency. If sutural fusion is related to function p artial to complete loss of the dentition or low rates of dental wear could alter the process of sutural fusion, possibly resulting in p remature fusion, absence of fusion, or abnormal sequences of fusion. Wheatley (1996) finds that individuals with partial to complete edentulism also displayed ages. Thus, edentulism may affect function but also the ability to accurately predict age from palatal suture closure. Sutural complexity Sutures are sinuous structures with varying levels of complexity. A complex suture is one that displays numerous interdig itations, resulting in a total length longer
113 than its straight line (chord) length. A simple suture is one that has nearly equivalent total and chord lengths. Sutural interdigitations can play a role in transmitting force from one bone to another (Herrin g 1972) Increased interdigitations of sutures creates increased surface area of the sutural margin and increased collagen along those margins, giving greater energy absorption potential to more complex sutures and strength during bending (Jaslow 1990) Sutural morphology is affected by stress during ontogeny (Herring 1972) but growth is not enough to cause interdigitations. A suture must be loaded in order for these to develop, and greater loading results in greater bone growth (Herring 2008) Th us, complexity is related to growth processes and individual variations in strain environment (Zollikofer and Weissmann 2011) Interdigitations are seen both parallel and perpendicular to the main direction of applied force, but a suture can adapt for mu ltiple loading scenarios by modifying the number of interdigitations and the directions of its fibers (Herring 1972) This means that sutural morphology is not perfectly correlated with a specific loading scenario or strain environment. Since growth and loading are linked in terms of sutural morphology, with increasing age, sutures may also increase in complexity (Zollikofer and Weissmann 2011) Bony shape is related in part to the collagen fibers that make up the sutures themselves. In sutures where t he dominant loading regime is tensile, fibers are straight or cruciate. For compressed sutures, oblique fibers are present. Sutures with numerous interdigitations are associated with compressive loads, while butt ended, shallow, beveled, and straight (i. e., simple) sutures are associated with tensile loads (Rafferty and Herring 1999; Herring and Ochareon 2005). In examining butt ended,
114 moderately interdigitated, and complexly interdigitated sutures, Jasinoski and Reddy (2012) find that strain energy is highest in the butt ended sutures and lowest in the complexly interdigitated sutures, indicating that a more complex sutural morphology may be better at dissipating strain. However, regardless of the strain regime of the suture or portion thereof, the su tural margins must always be in tension so that the bone at this area does not resorb (Enlow and Hans 1996). This means that describing overall strain regimes of sutures are likely overly simplistic summaries of loading as it relates to sutural morpholog y. Because of the relationship of force and sutural morphology, complexity has also been examined in terms of diet. In individuals or species who routinely chew harder foods, these loads may lead to increased sutural complexity as sutures become more in terdigitated to increase the surface area of the suture and prevent the disarticulation of cranial elements with greater loading (Herring 1993) However, Hotzman (2004) does not find that species who consume harder foods have a significantly higher mid p alatal suture complexity than those s pecies who consume softer foods, and complexity is likely related to other factors than just diet, such as age. Increased loads associated with harder foods or greater masticatory loading may also impact sutural fusion Herring (1972) finds that fusion begins in the more complex portion of the intermaxillary suture in pigs, but that this trend is not the same in other species, where fusion begins along the less complex portion. In the palate, sutures are not generally very complex, with more complex regions usually located in the posterior median palatine suture. Human palatal sutures rarely completely fuse, lending support to the concept that simple sutures may resist fusion, perhaps due to their tendency to
115 be subje cted to mainly tensile loads. However, because complex sutures can withstand greater loading, stress and strain may not play a major role in sutural fusion, with fusion related more to sutural morphology than how it is loaded. Palatal Variants The underst anding of growth, development, form, and function of the palate translates into the ability to use certain morphological traits to draw conclusions about the age, sex, and population origin (ancestry) of individuals and groups. Skeletal non metric traits generally fall into one of the following categories: (1) bone shape, (2) bony feature morphology, (3) suture shape, (4) presence/absence of trait, and (5) feature prominence/protrusion (Hefner 2009) These traits do not generally appear in isolation, an d certain groups may share similar suites of traits, enabling the anthropologist to even when suites of traits are found together, indicating a certain level of inter t rait association, just how much association exists is not well understood (Hauser and De Stefano 1989) As with processes of aging (see Chapter 2) skeletal traits are also the result of complex interactions between genes and environment, and it is still largely unknown why certain traits arise in some but not others and how function or group affiliation may or may not affect their expression. The palatal variants discussed below have been used with varying degrees of success to differentiate individual s and groups by age, sex, and/or ancestry A brief description is provided along with the utility of the trait for differentiation Not all palatal traits are included only those that are most commonly referenced or employed for differentiation based o n group membership. While certain generalizations are provided for group assignment based on trait expression, it is important to note that single traits
116 are rarely informative for group assignment due to their variable expression across multiple groups. Additionally, traditional scoring techniques are largely based on visual assessment, rather than metric methods. Transverse Palatine Suture Shape The medial portion of the TP suture displays varying forms, as can be seen in Hauser and De Stefano (1989). Anthropologists commonly condense these categories and score the appearance of the suture as straight, bulging/arched (anteriorly or posteriorly), and jagged/M shaped ( Rhine 1990 ; Gill, 1998; Hefner, 2009 ) These shapes are depicted in Figure 3 3. The straight shape is generally associated with individuals of Asian ancestry, the anterior bulging/arching with individuals of African ancestry, and the jagged/M shaped with individuals of European ancestry. There are no known sex or age differences in TP su ture shape. Shape of the TP suture may be affected by the presence of prognathism, in which the mid or lower face projects explanation for an M shaped suture is less c lear, as individuals with this shape suture do not usually display pronounced prognathism (i.e., individuals of European ancestry). Palate Shape Palate shape refers to the shape of the alveoli, to include the dental arcade. This shape can generally be d ivided into three categories: parabolic, hyperbolic, and elliptic ( Rhine, 1990; Gill, 1998 ) More specifically, Rhine (1990: p 20) defines the categories as follows: parabolic hyperboli c in Figure 3 4. As with TP suture shape, palatal shape is scored based on visual assessment and associated with ancestral groups as follows: parabolic European, hyperbolic
117 African, elliptic A sian. Palatal shape is largely tied to ancestral differences, though females and males may also differ in palatal shape, which may be tied to sizes due to sexual dimorphism (Rogers 1984). Rogers (1984) offers general differences between the sexes, with males displaying larger/U shaped palates as compared to smaller and more parabolic female palates. However, no frequencies are given. It is unclear whether or not age affects palatal shape. Subjective descriptions of palatal shape are potentially probl ematic when considering replicability and interobserver error (Maier 2013) Examination of palatal shapes via three dimensional digitization and machine learning methods shows that the previously used shape categories of parabolic, hyperbolic, and ellipt ic are discrete from one another, and that shape alone is accurate in classifying ancestry in only 58% of cases (Maier 2013; Maier et al. 2015) Maier (2013) did not note a secular change in palatal shape with digitization of shape. Tori In the mastic atory complex, a torus is a rounded prominence or ridge along the median palatine suture ( torus palatinus ), the lingual side of the maxillary alveolar bone usually in the molar region ( torus maxillaris ), or the lingual side of the mandible below the alveol ar margin and generally in the vicinity of the second premolar ( torus mandibularis ) (Figure 3 5; Hauser and De Stefano 1989; Rogers 1984) Bony growths along the buccal aspect of the dentition, generally in the vicinity of the molars, are maxillary exos toses. The majority of research on tori in the maxilla has focused on palatine tori, with relatively little investigation of maxillary tori and exostoses, and the summary that follows largely pertains to the former.
118 Tori are normal, non pathological var iants of the human skull composed of spongy bone (Miller and Roth 1940; Woo 1950) In the palate, this spongy bone is a result of an inferior enlargement of the diploe of the maxillary and palatine bones, with e underlying bones (Miller and Roth 1940; Woo 1950) Tori may be present in isolation or a single individual may display multiple tori. Suzuki and Sakai (1960) find a statistically significant correlation of palatine and mandibular tori in 309 living J apanese patients, and Woo (1950) reports that skulls with maxillary and mandibular tori are present in higher percentages in skulls with palatine tori versus those where palatine tori are absent. The etiology of tori is largely unknown and continues to be debated (Hassett 2006) The growth of tori is hypothesized to be genetically determined, the result of greater masticatory forces requiring buttressing of the masticatory system, or a combination of genes and forces genetic signals triggered in people wi th greater masticatory stresses what Hassett (2006) refers to as a threshold trait. Hooton (1946) masticatory apparatus, which produces pressure in the median palatine ar ea. Because of this pressure, bone on either side of the suture adapts by thickening, to serve as a type of buttress for resisting pressure. Woo (1950) disagrees with this, citing greater masticatory stress on the molar teeth and the anterior to posterio r arrangement of lamellae in palatine tori that do not configure to the expected lateral medial arrangement for a structure adapted to countering mid palatal masticatory stress. Thus, Woo (1950) attributes tori presence to heredity as does more recent, cl inical literature (e.g.,
119 Eversole 2011) Suzuki and Sakai (1960) find a higher percentage of tori occurrences in childre n whose parents also have tori. Tori are present in various forms, and the form of the torus palatinus can vary from a small ridge a long the median palatine suture to a massive structure that occupies the majority of the hard palate and may even inhibit speech or mastication. Woo (1950) states that the palatine torus generally tapers gradually anteriorly, with a more abrupt end poster iorly, and it can be restricted to only one area of the palate The palatine the median palatine suture, with a deep groove down the center. Hooton (1946) describes the various forms of the palatine torus as mound, ridge, or lump and Woo (1950) further categorizes them as ridge narrow and nearly uniform in width, mound wide with anterior and posterior tapering, lump irregular shape. Smaller palatine tori are mor e common than large (Woo 1950). V arying frequencies of t ori have been reported across populations Clinical dental literature reports that palatal tori are most common in individuals of Asian ancestry (Eversole 2011) though Chohayeb and Volpe (2001) fi nd the highest frequencies of palatal tori in African American women as compared to Caucasian, Hispanic, Asian, and Native American women in the Washington, DC area. Miller and Roth (1940) report the overall occurrence of any degree of expression of torus palatinus in their sample of 10 40 New York dental patients to be 24.2%. Studies on skeletal materials range from a single instance of a palatine torus out of 600 skulls of varying geographic origins (skull from British Columbia; Berry and Berry 1967) 3 to 10% 3 Nor did they report any instances of maxillary tori in the same study.
120 expression of some amount of palatal torus in British skulls (Brothwell 1981) to a high Woo 1950). In this same study, A merican limited, though Brothwell (1981) reports frequencies from 2.5 to 17%, with no ancestral group specified. The frequency of tori by sex is les ambivalent than ancestry, with most research agreeing that females are more likely to display palatine tori than males ( Miller and Roth, 1940; Woo, 1950; Eversole 2011) In their patients, Mi ller and Roth (1940) (1950) research agrees with greater expression in females but at a lower frequency than reported by Miller and Roth (1940). Clinical literature states that t ori develop following puberty (Eversole 2011) but a more specific relationship of tori presence and age is not clear. Miller and Roth (1940) find that a palatine torus was rarely seen before the age of five years, and the average age of occurrence for a ny degree of expression was 35.9 years. In their sample, Miller and Roth (1940) find that increased expression is seen with increased age, from which they conclude that development of a palatine torus is gradual, progressive, and associated with increasin g age (i.e., slight tori are seen more often in younger individuals while moderate to marked tori are seen in older indiv iduals). However, Woo (1950) finds that the torus does not increase with age, and attributes this difference with Miller and Roth (194 0) to a statisti cal error. Woo (1950) concludes that growth of the
121 palati ne torus stops at age 20 years, along with skeletal growth, and does not increase with age. Chohayeb and Volpe (2001) also find no relationship between age and the presence of a pal atine torus. These results bring into question the ways in whi ch this trait has been measured, and the irregular form of tori may lead to difficulty in standardizing observations. Scoring ranges from simple absence/presence (Berry and Berry 1967) to mo re complex systems including elevation, width, length, size, and shape ( Miller and Roth, 1940; Hooton, 1946; Woo, 1950; Hauser and De Stefano 1989) Suzuki and Sakai (1960) measure the palatine torus in living patients as trace palpation only, not visi ble by sight; slight visible by sight; and marked; if asymmetrical, the side with a greater degree of expression is scored. Since shape is so variable, systems that take into account relative size as compared to the overall masticatory complex or certai n skeletal elements may be more effective at summarizing tori expression. Palatine Bridging Grooves that serve as passage for vessels and nerves originate from the greater palatine foramen. The lateral palatine groove ( sulcus palatinus lateralis ) runs al ong the alveolar border to the canine and is larger than the medial palatine groove ( sulcus palatinus medialis ), which can have the appearance of bifurcating from the lateral groove (Hauser and De Stefano 1989) If the lateral groove bifurcates anteriorl y, an alveolar groove may also be present along the alveolar border and lateral to the lateral groove. Palatine grooves can exhibit tubercles and/or spines along their borders, and then when these tubercles or spines connect, they form bridges. Scoring o f palatine bridging can follow degree of completeness, position, and number (Hauser and De Stefano 1989) Degrees of expression of palatine bridging are depicted in Figure 3 6.
122 There is limited research concerning the demographic distribution of palatine bridging. There is no conclusive sex incidence or relationship of age and bridging frequency (summarized in Hauser and De Stefano 1989) reports the presence of bridging in fetal and newborn skulls, but other studies show no relationship with age, an increase in frequency with age, and a decrease with age (summarized in Hauser and De Stefano 1989). Marginal Crest Stieda (1 891) describes an osseous crest along the posterior end of the horizontal lamina of the palatine bone marginal crest or crista marginalis At this location the palatine can exhibit spicules or variably expressed ridges of bone. The location of this cre st is depicted in Figure 3 7. No demographic frequencies have been reported for this trait (Hauser and De Stefano 1989) Lesser Palatine Foramina Lesser palatine foramina can lie on both sides of the posterior border of the hard palate posterior to the greater palatine foramen (Berry and Berry 1967) Absence of lesser palatine foramina is infrequent (Hauser and De Stefano 1989) Scoring lesser palatine foramina can be done by presence/absence of any number of foramina greater than 1 (Berry and Berry 1967) or by more detailed means that include number, shape, size, and position in relation to the marginal crest (Hauser and De Stefano 1989) Expression is usually symmetrical though in cases of asymmetry there is no pronounced side difference (Hauser and De Stefano 1989) Lesser palatine foramina are visible in Figures 3 6 and 3 7. Hauser and De Stefano (1989) report low heritability estimates and the following frequency of the trait in eight human populations: Egypt (all eras)=48.6%, Nigeria
123 (Asha nti)=41.0%, Palestine (Lachish)=13.2%, Palestine (modern)=23.3%, India (Punjab)=48.0%, Burma = 32.0%, North America (British Columbia)=71.0%, South America (Peru)=59.4%. Males may exhibit greater than one lesser palatine more frequently than females, and the trait appears m ore frequently in older adults as opposed to younger ones though this trait is visible at birth (Hauser and De Stefano 1989) Bone Quality and Porosity Bone quality generally decreases with advanced age, meaning that bone density decre ases and bone may appear more porous and lightweight. Devlin et al. (1998) find that bone mineral densities in the anterior maxilla and the mandibular body are significantly correlated with age, though the posterior maxilla and hard palate are not found t o hav e a significant relationship. Cortical bone has greater porosity with increased age, but trabecular bone does not show the same effects (von Wowern and Stoltze 1978) Therefore, decreased bone quality and increased porosity in the maxilla may be in dicative of old age, even if only generally. Summary This chapter presented growth and development of the facial skeleton and palate, the form and function of the palate, and how palatal traits are employed by biological anthropologists to classify individ uals. It is important to understand the complex developmental and biomechanical environment of the palate in order to investigate the relationship of age on this region of the skeleton. The next chapter outlines the research methods and materials used in this study.
124 Figure 3 1 Diagram of the skeletal elements of the hard palate and its sutures. Note: this diagram depicts an adult with all permanent dentition; the incisive suture and premaxillae are depicted to show location only since an adult is u nlikely to display a completely open incisive suture and separation of the maxilla at the region of the central incisors. IN = incisive, AMP = anterior median palatine, TP = transverse palatine, PMP = posterior median palatine, GPF = greater palatine fora men. This diagram is not to scale.
125 Figure 3 2 Diagrammatic representation of the V principle of growth, following Enlow and Hans (1996). The solid black V represents the current location of the element; the gray Vs represent the former location a nd the process of displacement; the plus and minus symbols indicate deposition and resorption, respectively, which, when combined result in remodeling; and the direction of growth is represented by the solid black arrow.
126 Table 3 1. Timing of palatal gr owth and development, compiled from Scheuer and Black (2000: p 134, 138, 151). This table includes some key dental development and eruption events. Stage Timing/Age Event Fetal ( in utero ) Week 6 Maxillary ossification centers appear Weeks 7 8 Palatine perpendicular plate ossification centers appear By Week 8 Maxillary body and 4 processes identifiable Weeks 9 10 Appearance of premaxillary ossification centers Week 10 Palatine orbital and sphenoidal processes begin development Weeks 10 12 Maxill ary sinuses begin development Week 11 Maxillary deciduous dentition crypts begin formation Weeks 14 16 Maxillary deciduous tooth germs begin formation Weeks 17 18 Maxillary deciduous crypts complete Week 18 Palatine palatal processes fuse Mid fet al life Palatine adult form attained but not proportions At birth (neonate) 0 years Maxilla: main p a rts of bone and rudimentary sinuses present Palatine: adult form except horizontal and perpendicular plates are of equal width and height, orbital proc ess has no air cells Dental: crowns of deciduous maxillary dentition in crypts, roots of deciduous teeth start to form Infancy/ Childhood 0 12 years Maxillary body and sinus size increase gradually, eruption/replacement of deciduous dentition 0 1 yea r Permanent first molar and anterior teeth begin formation/mineralization 2 4 years Mineralization of premolars and second molars By 3 years Deciduous dentition emerged, completed root formation Childhood 3+ years Palatine perpendicular plane increa ses in height 6 8 years First permanent molar erupts posterior to second deciduous molar, deciduous incisors lost, permanent incisors erupt; formation of third molar begins Juvenile 10 12 years Deciduous canines and molars lost; permanent canines, prem olars, and second permanent molars erupt; formation of third molar continues (crown mineralization complete 4 years following formation commencement) Puberty ~10 years for girls/~12 years for boys Palatine attains adult proportions Adolescence No later than 12 14 years All permanent teeth erupted (except M3s) ~18 years Eruption of third permanent molars
127 Table 3 2 Muscles of the soft palate. Muscle name Function(s) Origin Insertion Elevators Levator veli palatini Elevates soft palate posterior ly and superiorly towards pharyngeal plate Contracts during swallowing to avoid food entering nasopharynx Provides velopharyngeal closure during speech Inferior surface of the petrous portion of the temporal bone Palatal aponeurosis Musculus uvulae Shorte ns/broadens uvula Helps close nasopharynx during swallowing Role in speech Nasal spine on palatine Near uvula Depressors Palatoglossus Elevates posterior tongue Aids in swallowing Maintains palatoglossal arch Palatal aponeurosis Lateral portions of the posterior tongue Palatopharyngeus Tenses to pull pharynx over food bolus Aids in swallowing and breathing Lowers palate Lower surface of palatal aponeurosis Lateral wall of pharynx, posterior border of thyroid cartilage Other Tensor veli palati ni Assists in opening or swallowing to equalize air pressure in Eustachian (auditory) tube Medial pterygoid plate of sphenoid, cartilage of Eustachian tube Palatal aponeurosis, horizontal portion of the palatine
128 Table 3 3 Muscles of the masticatory sy stem, compiled from White et al. (2012: p 99). Muscle name Function Origin Insertion Temporalis Elevates the mandible Closes the mouth Lateral cranial vault inferior to the superior temporal line Lateral sides, apex, and anterior surface of the coronoid process of the mandible Masseter Elevates the mandible Closes the mouth Inferior surface of the zygomatic arch Lateral surface of the mandibular ramus and the gonial angle of mandible Medial pterygoideus Elevates the mandible Closes the mouth Superior h ead: postero superior maxilla Inferior head: Medial surface of the lateral pterygoid plate of the sphenoid Medial surface of the mandibular gonial angle Lateral pterygoideus Protracts the mandible Assists in depressing the mandible Pushes jaw forward Allo ws for lateral medial movement of mandible Superior head: sphenoid greater wing Inferior head: lateral pterygoid plate of the sphenoid Neck of the condyloid process of the mandible and the articular disc and fibrous capsule of the temporomandibular join t
129 A B C D Figure 3 3. Transverse palatine suture shape. A) Straight, B) anterior deviation/bulging, C) posterior deviation/bulging, D) jagged/M shaped. Photo graphs courtesy of Carrie A. Brown.
130 A B C Figure 3 4. Palate shape. A) Parabolic, B) hyperbolic, C) elliptic. Photographs courtesy of Carrie A. Brown.
131 A B Figure 3 5. Tori. A) Small palatine torus, B) large palatine and bilateral maxillary tori an d small to medium maxillary exostoses. Photographs courtesy of Carrie A. Brown.
132 Figure 3 6. Palatine bridging; spines, left; bridging, right. Photograph courtesy of Carrie A. Brown.
133 A B Figure 3 7. Marginal crest. A) Anteriorly positio ned ridge, B) posteriorly positioned ridge. Photographs courtesy of Carrie A. Brown.
134 CHAPTER 4 MATERIALS AND METHODS Sampling Strategy Power analyses were first conducted to develop a sampling strategy prior to data collection. This type of analysis ex amines the relationship among sample size ( n ), significance or confidence level ( ), power ( ), and population effect size and is used to predict one of these variables given the other three (Cohen 1992). One of the most common uses of power analysis in research is finding the sample size required in order to obtain a significant effect if one is present, given a specified degree of confidence and power ( Cohen 1992 ; Adler 2012; Crawley ) speaks to the risk of committing a Type I error, which occurs when the null hypothesis is mistakenly rejected; generally, = 0.05 is used unless a more or less stringent test is desired (Cohen hypothesis (Type II error), with representing the probability of accepting a false null hypothesis (Crawley 2013). Effect size is a statistic that quantifies the degree that the null hypothesis is believed to be false (Cohen 1992; Vacha Haase and Thompson 2004). For example, i n a two sided t test an effect size of zero indicates that there is no difference between group means (Vacha Haase and Thompson 2004). Effect size is more challenging to determine in experimental design and is dependent on what statistical test is chosen for data analysis (Cohen 1992). For this research the significance level and power we re set at conventional standards ( = 0.05 and 1 = 0.8 [ = 0.2], respectively) (Cohen 1992; Crawley 2013). Effect size wa s based on previously published values for sutural closure or recommended values for small, medium, and larg e effect sizes taken from
135 Cohen (1992) when no such val ues we re available. All predicted sample sizes we re rounded up to next highest full integer since it is not possible to collect a portion of a sample (i.e., 322.157 is rounded to 323 since 0.157 of an individual is not a valid amount). Power analyses we r e conducted in R version 3.0.0 (RCoreTeam 2014). In the investigation of age and suture closure, correlation can be used to talk about the strength of that relationship. Correlation values for suture closure and age we re drawn from previous research by Nawrocki (1998) and Wheatley (1996). Using the value of r = 0.55 given by Nawrocki (1998) for the correlation between age and all palatine sutural fusion, power analysis indicate d that a minimum of 23 samples should be collected. As a comparison, values drawn from Wheatley (1996), which are correlations between age and specific palatal sutures, range from r = 0.271 for the left lateral incisive suture to r = 0.494 for the posterior median palatine suture. For the lower value, power analysis indicate d th at a minimum of 105 samples should be collected, while for the higher value a minimum of 30 samples. The correlation value provided by Nawrocki (1998) is preferred as it takes into account all palatal sutures, not just the relationship between single sutu res and age. These power analyses are summarized in Table 4 1. Power analyses we re also conducted to find the sample sizes needed to find significant differences if they exist in sutural closure due to sex, ancestry, and temporality (i.e., secular trend s). These differences are investigated via t tests for two groups or the analysis of variance (ANOVA) for more than two groups. Both tests compare differences in mean values among groups. For ease of comparison, ANOVA power analyses we re conducted for a ll groups (sex, ancestry, and secular trends) using
136 k = 2 for sex (male or female) and secular trends (historic or modern) and k = 3 for ancestry (African, Asian, European); k indicates the number of groups being compared. In these analyses effect sizes o f 0.10 (small), 0.25 (medium), and 0.40 (large) we re employed (Cohen 1992); the effect sizes are denoted by f The results of these analyses are given in Table 4 2. Power analyses for biomechanical variables and palatal variants and their relationship to both age and sutural closure are complicated by myriad variables to be investigated and the lack of published r values for many of these statistical tests. For example, while increased loading in the palate may lead to more complex sutural morphology, an d more complex sutures could potentially delay fusion (see Chapter 3), there are currently no published r values for the relationship of sutural complexity and fusion. The same limitation is true for r values for the relationship of palatal variants, ante mortem tooth loss, and dental wear to age and palatal suture closure. These relationships can be tested through correlation and regression analysis, but without known r values, it is not possible to estimate effect size. Additionally, several of these va riables necessitate the use of nonparametric statistical tests as they are categorical or ordinal level variables. Since a certain distribution is not assumed when using a non parametric test, the more straightforward power analyses described above for pa rametric tests cannot be applied. Therefore, no power analyses we re conducted for biomechanical variables and palatal variants, and the assumption wa s made that the previously described power analyses and sample sizes are more than adequate for these addi tional variables.
137 Beyond determining the sample size needed to find a significant effect if one is present, the knowledge of required sample size also ensures that the sample can be evenly distributed in terms of demographic variables. In age estimation studies this is particularly germane since the underlying reference sample structure can affect method performance (see Chapter 2). The desired goal of the sampling strategy is to have groups of relatively equal size so that the entire sample is balanced Power analysis identifies the number of samples to be collected but does not specify possible interactions among variables or sample group compositions (e.g., Asian females between the ages of 30 and 39 years). In order to maintain a conservative appro ach, a group wa s designated as being age sex and ancestry specific for data collection purposes. In this manner, larger samples a re available for factor specific analyses (e.g., ancestry and sutural closure), and by collecting the recommended sample s ize for the most specific group age, sex, and ancestry together sample sizes for all factors are more than adequate when examined on an individual basis. Ages we re broken down into 10 year intervals, starting at age 20 and ending with a terminal categ ory of 70+ years, for a total of 6 age groups 20 29 years, 30 39 years, 40 49 years, 50 59 years, 60 69 years, and 70+ years. No individuals under the age of 20 years we re included as other methods of age estimation are more accurate for these ages (see Chapter 2), and skeletal collections often do not include large numbers of sub adult individuals. Sex is a binary category male or female, and ancestry is divided into the three major ancestral groups African, Asian, and European. While time of birt h and/or death is also considered, it is not used in group assignment due to the composition of the skeletal collections. Collections with
138 individuals born during the 19th or early part of the 20th centuries (generally pre 1940) we re designated as histori c, while collections with individuals born during the second half of the 20th century (generally post 1940) are designated as modern. These designations serve to aid in the investigation of potential secular trends. The highest r and f values produce d ne eded sample sizes between 22 and 26 individuals ( Tables 4 2 and 4 3). Therefore, the ideal sample size to be collected per age sex and ancestry specific group wa s set at 22 individuals, with the understanding that sutural closure or other variables m ay not differ significantly for some groups. This ideal sample size is limited by the availability of samples in and the demographic composition of the skeletal collections, but almost all groups have sample sizes at or very close to this number. The onl y group not meeting this threshold is European females between the ages of 20 and 29 years. Even examining four large U.S. collections, only 12 samples from this group were obtained, and 10 were viable following data scrubbing (see below). Data were also collected on individuals from other groups or mixed ancestral groups (i.e., Hispanic, European/Native American), but because of the limited number of these individuals overall, they we re not included in further data analyses. If demographic information a bout the documented collections could be collected ahead of time, a sampling strategy was devised prior to arrival (Hamann Todd, Chiba, Bass); if the sample composition was not available to the researcher prior to data collection, a sampling strategy was d evised upon arrival to the collection (Maxwell, Terry, Jikei). In general, a stratifi ed random sampling strategy wa s prefe rred; this entailed choosing individuals at random from within the pre defined age, ancestry, and
139 sex groups. Ancestry largely led t he decision on what collections to visit (e.g., the Terry Collection has some of the only documented individuals of African ancestry in the U.S., and the Bass Collection has very few non balanced sample overall was also important. Documented Skeletal Collections Table 4 3 summarizes information about the collections visited for this research and the total number of samples collected. These numbers reflect all samples examined for this research, but not the to tal sample number analyzed due to subsequent data scrubbing (see below). All skeletal collections contain both male and female individuals, although the general trend is that they contain more males than females. This is why the sampling strategy detaile d above is used. Data was collected from both historic and modern collections, as it was impossible to collect a large enough sample size using only one time period. For example, in the U.S., neither of the largest modern collections have a substantial n umber of individuals of African ancestry while over 50% of the historic Terry Collection is comprised of individuals of African ancestry res idents of New Mexico and is housed at the Maxwell Museum of Anthropology at the University of New Mexico in Albuquerque. Collection of documented human remains began in 1975 and continues to date (Komar and Grivas 2008). Documentation for the individual s in this collection is self/next of kin reported or obtained from the medical examiner/Department of Anatomy, but not all of the individuals in the collection are documented (Komar and Grivas 2008). Individuals with unknown demographic information are e xcluded from this study. The majority of individuals in this collection
140 are of European ancestry, and both sexes are represented. This is a modern collection; all individuals have a documented year of death within the last 30 years. The William M. Bass D onated Skeletal Collection, housed at the University of Tennessee, Knoxville, represents a collection of individuals from the donated body program, established by Dr. Bass in 1981. The majority of individuals in this collection are from Tennessee and the southeastern U.S. ( University of Tennessee Knoxville, N.D.). Documentation is self next of kin or medical examiner reported. Collection and curation continues to this day, and most individuals in this collection have birth years post 1940 ( Universit y of Tennessee Knoxville, N.D.). The majority of individuals in this collection are of European ancestry, though there are a larger number of non European individuals than the Maxwell Collection. The Robert J. Terry Anatomical Skeletal Collection, hous ed at the Smithsonian National Museum of Natural History in Washington, DC and Suitland, MD, contains skeletons from medical school cadavers collected by Dr. Robert Terry from 1898 to 1941 and continued by Dr. Mildred Trotter until 1967 (Hunt and Albanese 2005). Documentation for the skeletons in this collection is from morgue records. Individuals in this collection have birth years between 1828 and 1943, and the majority of individuals died between ages 20 and 80 years (Hunt and Albanese 2005). The col lection is composed of a relatively similar number of male and female individuals of African and European ancestry. This is considered a historic collection based on the birth years for the majority of the individuals in this collection. The Hamann Todd Osteological Collection contains skeletons retained from cadavers used by medical students, and collection began in 1912 by Dr. T. Wingate
141 Todd. It is housed at the Cleveland Museum of Natural History in Cleveland, OH. The collection represents individua ls who were born in the 19th century (Komar and Buikstra 2008). Like the Terry Collection, the Hamann Todd collection is historic and is composed of males and females of African and European ancestry. Documentation is available for the majority of the s keletons; however some ages are given as intervals rather than specific numbers, and in some cases it is unclear exactly how documentation was obtained. White et al. (2012) state that only about 16% of the individuals in this collection have reliable enou gh age data for skeletal aging studies. Due to this and the presence of mid sagittal sectioning in some of the cranial remains, the sample obtained from this collection is small and only represents individuals for which a single, known age is recorded. The Chiba Documented Collection consists of skeletons obtained from individuals who died while incarcerated and is composed solely of individuals of Asian ancestry. This collection is housed at the Chiba University School of Medicine in Chiba, Japan. The majority of the individuals in this collection are Japanese, although a few individuals with a birthplace of Korea are included. This collection represents individuals born during the 19th and earlier part of the 20th centuries (birth years of 1851 to 19 23) and is therefore considered to be historic. Individuals whose documentation do es not include age at death are not included in this research. The Jikei Documented Collection, housed at the Jikei University School of Medicine in Tokyo, Japan, consists o f full skeletal remains and isolated skulls obtained from individuals who donated their bodies to the medical school. This collection is more recent that the one housed at Chiba University, with dates of death in the second half of
142 the 20th century, and r epresents modern individuals. Individuals with unknown or approximate ages at death are not included in this research. Data Collection A data sheet, designed prior to data collection and tested at the Stanford Collection, University of Iowa, prior to visi ting other collections, was used to manually record data for each individual ( Appendix A). Once data was collected, it was input into a Microsoft Excel spreadsheet. Both sides of the sheet contain a header that includes the specimen identification number who collected the data, whether the data are collected by in person examination or via photograph, and the date of data collection. No one other than the researcher (Carrie A. Brown) collected data for this project, and all skulls were examined in perso n. The Excel worksheet also includes a log page where dates of data collection, photography, collection of biological profile information, digitization, and additional notes are recorded. Age, ancestry, sex, and any other biological information were unkn own at the time of data collection. Additionally, no other areas of the skeleton were examined in order to avoid introducing bias from other age indicators; only the maxillae, select facial sutures, and dentition were inspected. Each skull was examined fi rst for the presence of items that obscure observation of the sutures, dentition, or palatal variants. This included: postmortem breakage of skeletal material, sampling, midpalatal or other sectioning, and incomplete processing that results in the retent ion of soft tissue or adipocere. Individuals without intact palates or craniofacial skeletons were excluded, though individuals were scored that did not have all observable traits. No individuals were excluded based on damage or trauma to other parts of the skull or palatal/facial abnormalities (e.g., disease, unusual growth).
143 Qualitative Data Data were collected for each individual following the order on the worksheet (Appendix A). Closure of the palatal sutures and two facial control sutures was scor ed using a 4 phase ordinal variable system following Meindl and Lovejoy (1985): 0 open, 1 1 50% union, 2 51 99% union, 3 complete. The palate was initially scored in 15 sections ( Wheatley 1996 ; Beauthier et al.2010) (Figure 4 1 and Table 4 4), a nd then each of the four palatal sutures was scored in its entirety (Nawrocki 1998) (Figure 4 2). Additionally, the right and left halves of the TP suture and the right and left TP suture within the greater palatine foramen were also scored to examine po tential side differences in sutural fusion. The closure per suture and for sides of the TP suture was further converted to a binary score: 0 no fusion or 1 any fusion. This binary system represents the Mann et al. (1991) method, which requires only the recognition of any amount of closure along a suture. The TP suture within the GPF was scored separately from the TP suture as a whole due to age differences noted in timing of fusion at this location versus the entirety of the TP suture (see Chapter 2 ). The facial control sutures were scored for degree of closure following Meindl and Lovejoy (1985) and in their entirety, not by section, due to their smaller overall length. For each of the three systems (15 section/4 phase, full suture/4 phase, and f ull suture/binary) a summary score was also calculated. The summary score for the 15 section system ranges from 0 to 45, for the 4 phase full suture system from 0 to 12, and for the binary full sutures system from 0 to 4; neither of these full suture syst em summary scores include the TP suture within the GPF. Therefore, for the 4 phase full suture and binary full suture systems, a second summary score that includes the TP suture within the GPF was produced. To determine which side to use for these second
144 summary scores, the right and left fusion scores for the TP suture within the GPF were statistically compared for both the 4 phase and binary scoring systems using Mann Whitney U tests, the nonparametric equivalent of the t test. Nonparametric tests were used because the suture fusion scores are ordinal level variables. There were no significant differences between right and left fusion scores of the TP suture within the GPF for either the four phase ( p = 0.437) or binary ( p = 0.259) scoring systems. Gi ven rho a nonparametric method for measuring the relationship between two variables, was calculated for age and fusion score for both the right and left TP sutures in the GPF in the 4 phase and binary scoring systems. The right TP suture in the GPF scores have slightly higher positive correlations with age than the left for both 4 phase and binary scoring systems (four phase: right = 0.316, left = 0.291; binary: right = 0.321, left = 0.303). Therefore, the right side is used i n the second summary score that includes the TP suture in the GPF. The summary scores that include the TP suture in the GPF range from 0 to 15 for the 4 phase system and from 0 to 5 for the binary system. Palatal variants w ere scored based on Hauser and D e Stefano (1989, and references therein), Gill (1998), and Hefner (2009); descriptions of these traits and their use in biological anthropology are given in Chapter 3. Table 4 5 gives the specific scoring system used in this research. Some of the referen ced scoring systems were modified for this research, while the scoring of other variables was developed during this data collection due to absence of scoring systems for these variables in the relevant literature. Shapes of the transverse palatine and zyg omaticomaxillary sutures were scored on initial examination, while palatal shape, lateral and medial groove bridging,
145 marginal crest accessory lesser palatine foramina, maxillary and palatine bone quality, porosity, tori, and exostoses were scored using n otes made during initial observation and a re examination of the photographs. The zygomaticomaxillary suture shape was scored because its shape may influence closure of this suture; it is included in the palatal variants table for ease of reference. The dentition was inventoried following a modified version of the inventory categories presented by Buikstra and Ubelaker (1994: p 49) ( Table 4 6, see also Appendix A). The first five categories were maintained, while category 6 was combined with category 8 since no radiographic examination was conducted as part of this research. The new category 6 indicates a tooth with intact and non resorbed alveolar bone that was unobservable, either due to congenital absence or because the tooth was unerupted. Addition al observations, including dental restorations and the presence of Wear for each tooth present was recorded following Smith (1984) for the incisors, canines, and premolars, Scott (1979) for the molars, and general recommendations by Buikstra and Ubelaker (1994). Tables 4 7 to 4 9 outline the dental wear scoring systems. In order to summarize the total wear per individual, regardless of the number of teeth present, wear was summed and d ivided by the total number of teeth present for the entire dentition, the posterior dentition (molars and premolars), and the anterior dentition (canines and incisors). The overall mean wear score gives a picture of wear per individual and the mean poster ior and anterior wear scores serve to analyze teeth that have contact surfaces of different sizes.
146 Because of the difficulty of determining if a tooth was congenitally absent or lost prior to death for the third molars and, in more limited circumstances, o ther teeth, an AMTL index was developed to compare antemortem tooth loss among all individuals without relying on simple counts of teeth lost. This index gives a standardized amount of AMTL per person, and it is calculated as follows: ([Total teeth poss ible Teeth no AMTL]/Total teeth possible)*100 (4 1) The total number of teeth possible per individual is a count of all inventory scores except 3 (missing, no associated alveolar bone) and 6 (unobservable, either due to congenital absence or non eruptio n), for a total number of teeth no greater than 16 (the number of teeth in the maxillary arch if all teeth are present). For the AMTL Index, values of 0 indicate an individual with no tooth loss, while values of 100 indicate complete tooth loss (totally e dentulous). Values may fall anywhere on the continuum from 0 to 100. Photographs of the cranium were taken using a Nikon D700 digital camera, AF S DX Nikkor 18 55mm lens (overview), AF S micro Nikkor 105mm lens (close up), and a Sigma EM 140 DG ring fla sh. In general, 10 photographs were taken per cranium: documentary containing the specimen identification number, overview in anterior view, overview in inferior view, detail of the entire palate, detail of each of the greater palatine foramina, detail o f each of the zygomaticomaxillary sutures, and two details of the nasofrontal suture. One detail of the frontonasal suture was taken without a scale to record presence or absence of the supranasal suture, to be used in future research. All of these photo graphs were taken with a scale and with the lens parallel to the surface being documented so that there was no distortion of the image to enable accurate
147 digital measuring. Additional photographs were taken as needed, to include documenting trauma or vari ants and the use of different camera settings to improve detail/lighting for certain samples. Photographs were documented as taken by marking the appropriate box on page 2 of the data collection worksheet and then burned directly to a disk for transfer to a computer. Quantitative Data Suture closure and character state variables, as described above, are qualitative summaries of the available data and either categorical or ordinal level variables. These types of variables can limit the ability to conduc t parametric statistical tests and potentially obscure smaller scale variation. In order to compare these variables quantitatively and capture more precise information, interval and ratio level variables were developed based on detailed photographs of th e palate. To prepare for digitization, all images were sorted into digital folders by specimen number and then labeled sequentially with specimen number. Each detail photograph of the palate was copied into a separate folder containing only palate photog raphs in order to enable ease of reference when working in ImageJ (Rasband 1997 2014). Digitally measuring sutures was chosen as a technique due to time constraints at the skeletal collections and the ability to more accurately measure total sutural leng th and fusion in the digital realm because of the sinuous nature of sutures. Digitally measuring the sutures of the hard palate in human adult skulls based on images does not significantly differ from the traditional caliper method (Moreira et al. 2006). Landmarks used for measurement in this study are based on features of the palate or extrema points Table 4 10 lists and defines the landmarks used in this study, and Figure 4 3 depicts th ese landmarks on the palate and in relation to other palatal
148 featu res. These landmarks were chosen from reference materials or developed for this study based on their ability to best quantify sutural chords and total lengths. No landmarks incorporating the TP suture within the GPF were employed due to the difficulty of photographing this region without distortion. The antero lateral transverse palatine (altp) landmark was used in order to best quantify deviations along the length of the TP suture prior to its descent into the greater palatine foramen. While Szrkat et a l. (2003) measure the length of the transverse palatine suture using summi palati on both sides, similar to the way that altp is defined for this study, the definition for this landmark is not entirely clear. It is also possible to take the length of the TP suture from a defined feature the most anterior point on the greater palatine foramen, where the suture descends into the foramen, but this potentially artificially increases measures of complexity since in most palates the TP suture exits the GPF ant eriorly in the sagittal plane for several millimeters before making an abrupt 90 degree turn medially. Therefore, the altp marks the most antero lateral point on the TP suture on both right and left sides and does not include any portion of the TP suture that enters into the GPF or the abrupt deviation of this suture. Incisulare (inc), described by Skrzat et al. (2003), was chosen over orale (Bass 2005) for the measurement of the AMP suture because inc is located on the posterior border of the incisive fo ramen. Orale defined as the midline of the hard palate where a line drawn tangentially to the posterior margins of the central incisor alveoli crosses the midline (Bass 2005), can also be difficult to locate where antemortem and/or postmortem tooth loss is present. Additionally, in the majority of palates examined, no
149 suture was visible either within or anterior to the incisive foramen, making the choice of incisulare over orale the logical one for this research. The landmark employed for the most post erior portion of the PMP suture was the posterior nasal spine (pns) (Bass 2005). Neither staphylion nor alveolon w as used as the posterior landmark for the PMP suture because they both are defined in relation to the posterior or alveolar border of the pa late rather than the most posterior point on this suture (Steele and Bramblett 1988). The choice of the posterior nasal spine enabled the complete measurement of the PMP suture since the use of the aforementioned points could result in artificial truncat ion of the suture in cases where these points are found anterior to the most posterior point of the intersection of the palatine bones. It is also impossible to precisely locate staphylion and/or alveolon in individuals with alveolar bone loss. In cases where the posterior nasal spine and the posterior edge of the PMP suture did not correspond, the suture was measured to the posterior edge of the PMP suture rather than the posterior nasal spine. Finally, the standard landmark of staurion (sr), the inter section of the median and transverse sutures of the hard palate, was employed with no modifications. In cases where the right and left TP sutures asymmetrically intersected the median palatine suture, the more posterior point of intersection was used for the AMP suture measurement. This results in slightly truncated PMP suture measurements, but the truncation is standardized across the entire sample. Using these landmarks, the total length (including all oscillations), total amount of suture closure, and sutural chord (shortest distance from point A to point B) per suture were digitally measured in ImageJ (Rasband 1997 2014) using photographs of the
150 palate and a Wacom Tablet touchscreen monitor and stylus (Figure 4 4). Measurements were taken in centimet ers, to the one thousandth decimal place. Prior to any measurement, the scale was set in ImageJ (Rasband 1997 2014) by using the 1 known distance (Figure 4 5). The scale was set for each sample prior to digitization. If necessary, image brightness and/or contrast were adjusted to optimal levels for viewing or distinguishing sutures (Figure 4 6). Measurements and their corresponding landmarks are given in Table 4 11, and the order the measurements are taken for each individual is given in Table 4 12. Landmarks can be referenced in Figure 4 3, and Figures 4 7 to 4 10 show examples of measu rements of sutures and fusion. Ordinal observations (see above, Qualitative Data) we re not referenced at any time during the digital measuring process. The incisive suture was excluded from the digitization process since most individuals in this sample displayed nearly or completely obliterated incisive sutures, which made it impossible to measure total sutural and chord lengths. In individuals where fusion was not continuous (i.e., multiple sites of closure present along the length of suture with breaks in between fusion sites), each separate instance of fusion was measured and then sum med to produce a total fusion score for that suture (see Figure 4 10). The length, chord, and fusion of the TP suture were measured separately for right and left sides, and these scores were then summed to create total length, chord, and fusion variables for this suture. Specifically for the chord measurement, this strategy was used to avoid incorporating large scale anterior or posterior deviations of the entire course of the suture since this is a different variable,
151 which is described qualitatively usi ng the categorical scale for transverse palatine suture shape ( Table 4 5). From these measurements, ratio level variables were produced to standardize measurements for comparison among individuals. These variables are: sutural complexity and the degree o f suture closure for the AMP, PMP, and TP sutures. Sutural complexity was calculated using the suture length ratio (Rafferty and Herring 1999) : Total Suture Length/Suture Chord (4 2) Lower ratio values (close to 1) indicate sutures that are less complex while larger ratio values (greater than 1) indicate sutures that are more complex. Suture length ratio was used since it does not require sutures to qualify as fractals, is more straightforward than fractal analysis, and is found to better quantify sutu ral complexity when compared to fractal dimension analysis (Hotzman 2010). The fusion ratio was calculated by dividing the amount of fusion by the total length of the suture. A higher fusion ratio (closer to 1) indicates more fusion; while a smaller val ue (closer to 0) indicates less fusion. Scores of 1 indicate full fusion of the suture; scores of 0 indicate the absence of any degree of fusion. A summary score for the fusion ratio was calculated by adding fusion for the PMP, TP, and AMP sutures, and d ividing this number by the sum of the total lengths of these sutures: (PMP.fus+TP.fus+AMP.fus) / (PMP.len+TP.len+AMP.len) (4 3) Data Analysis This study uses frequentist statistics due to the structure of the data and certain limitations that limit the application of Bayesian based inference. Because of the previous scoring systems used in cranial suture age estimation, there are no discrete,
152 age progressive phases or stages in the methods developed to date for palatal sutures, nor is there a single ref erence sample that has been scored using that method (Garvin et al. 2012) Additionally, p rior distributions, which are required for Bayesian methods are not easily obtainable for samples derived from skeletal collections (Konigsberg et al. 2008) Bay esian based methods may be a possibility in the future, if discrete phases of palatal suture closure can be defined. The research design and sampling strategy outlined above provide a balanced sample, which supports the use of frequency statistics at this stage (Nawrocki 2010) Following data collection, the total sample wa s scrubbed to eliminate individuals not classified into one of the three major ancestral groups and individuals who ha d missing data points for one or more suture fusion scores, inclu ding the facial control sutures. This resulted in the total sample used for analysis, reflected in Table 4 13. Individuals with missing data for palatal variants were not eliminated from the analyzed sample. More than 25 samples for European males and f emales over the age of 70 years were present in the un scrubbed sample, and these groups were culled so that their total sample sizes were not over 25 per group to avoid biasing the sample towards very old individuals. The age distribution is given in Fig ure 4 11; Figures 4 12 and 4 13 display the sample distribution by age and sex and age and ancestry. For the sample as a whole, the mean age is 50.94 years, median age is 50 years, and minimum and maximum ages are 20 and 102 years, respectively. Due to t he power analyses that informed data collection in terms of sample size, the balanced demographic composition of the sample, and the overall large size of the sample, parametric statistics were employed as long as the data reasonably meet
153 assumptions for t he particular method used. When data did not meet or come reasonably close to meeting the assumptions inherent to parametric methods, nonparametric methods we re employed (e.g., scores were ordinal or non normally distributed in a correlation analysis). C hapter 5 (Results) details the particular test used a simpler test is always preferred over a more complex one). All statistical analyses were conducted in R version 3.1.2 (RCoreTeam 2014) In tests of statistical significance, a result is considered to be significant if it falls below the p = 0.05 threshold. For correlation analyses, p values are not reported since they can be affected by large sample sizes and onl y inform on the difference from that value from 0. Hypothesis Testing Data were analyzed in the order of the research questions outlined in Chapter 1: age and sutural closure, group affiliation and sutural closure, biomechanical proxies, and palatal varia nts. This order serves as a type of model building in that it first determines the relationships of each type of variable to the two variables being investigated age at death and fusion and then uses those explanatory variables with the strongest rela tionship with age and/or fusion to move to the next level. The final step is to incorporate all data. Specific details inherent to the testing of each of the four main research questions are given below and in Chapter 5. Age and Sutural Closure Degree of palatal suture closure is expressed both categorically and continuously. Categorical data include nominal and ordinal scoring of sutural closure for full sutures and sections of sutures as well as summary scores developed from those scores. Continuou s data include ratios of fusion for individual sutures and a fusion ratio
154 summary score. The relationship between known age and degree of palatal suture closure based on categorical scoring was first visually examined to determine the appropriate statisti cal method to use in analyzing these data, and the ability to treat summary scores as continuous data was also investigated. Due to the largely categorical nature of the closure scores, nonparametric tests were used to compare age and closure even for the continuous scoring system as it enabled easier comparison of this system with ordinal systems. Scoring systems were also investigated via ANOVA to determine if certain scoring categories could be collapsed and if there was any need to eliminate certain v ariables from the 15 section/4 phase system. The main goal of testing this hypothesis was to determine which system had the strongest relationship to known age at death. Group Affiliation and Sutural Closure The relationship of group affiliation and sutur al closure was first visually examined, and then data were checked to see if they met the assumptions for parametric comparisons of means, namely constant variance and normally distributed error. For this stage of analysis, the scoring system/combination of sutures with the highest relationship to age was employed. In cases where the data violated parametric assumptions, nonparametric tests were employed. A full model, excluding groups with non significant differences in means/medians was also run to exa mine potential interactions among variables. Biomechanical Variables Each of the three biomechanical variables wear, AMTL, and complexity was compared to fusion, known age at death, and group, as well as to one other. Because of the continuous natur e of the biomechanical variables, all three were compared to
155 both the 15 section/4 phase summary score and the fusion ratio scores, as appropriate. Due to the way that complexity scores were measured and calculated, it was not possible or valid to combine complexity scores across palatal sutures. Therefore, complexity scores were analyzed in terms of fusion of separate sutures order correlations due to the non normal frequency distributions of t he majority of biomechanical variables. With continuous data, regression was employed to look at the effects of variables on one another. For comparison across groups, data were checked first for constant variance. With relatively constant variance and large sample size, parametric comparison of groups means ANOVA was employed; in cases where the data did not reasonably meet these assumptions, nonparametric alternatives were used. Palatal Variants Palatal trait expression was analyzed in terms of age, group affiliation, biomechanical variables, and the relationship of palatal traits to one another. order correlations, appropriate for categorical data, were used to summarize relationships of palatal traits with other variables and e ach other. Frequencies of traits by sex, ancestry, and time period were compiled. While frequencies are generally informative for looking at between and among group chi sq uare tests to determine if differences between observed and expected frequencies were statistically significant. When at least one expected cell count was < 5 in a table
156 an alyzed, no further tests of significance were conducted at this phase of analysis, and rho values were employed to inform variable selection in subsequent analyses. The Full Picture Based on testing of the four hypotheses, models were developed using thos e variables shown to have a significant effect on or otherwise moderate to strong association with age or fusion generally set at a threshold of correlation value > 0.100 or < 0.100. Details of specific testing and variables used can be found in Chapte r 5. The goal of this final step of analysis was to best describe fusion in terms of all variables investigated and understand age variation across the palate. Error and Limitations Error in this study could potentially come from several sources, both ran dom and systematic. Random error is most often attributed to human or observer error, and while it is always present, it is also statistically quantifiable and limited by repeated measurement and controls ( Youden 1998 ; Brach and Dunn 2004). A good way to control for random error is to include tests of intra and interobserver error. Systematic error is more difficult to control for as it relates to variation of the sample mean value in relation to the true popula tion mean value (Brach and Dunn, 2004). This type of error can be minimized by repeatedly refining measurement techniques in order to better calibrate them with what they are measuring, but it lacks true statistical quantification (Brach and Dunn, 2004). In this research, the potential for rand om error was mainly during data collection. It includes: inexact measurements in ImageJ (Rasband 1997 2014), transposition of numbers during data collection and entry, and the incorrect association of individuals with photographs or scores. Several pro tocols are put in place to minimize the
157 introduction of random error, to include standardized data collection worksheets with headers containing specimen identification numbers, double checking of manual and transcribed data entry, a log sheet to record da tes when data were collected, the photography of specimen identification number before each set of photographs as well as the retention of this identifying photograph with associated images throughout image transfer, and the relabeling of all photos to inc lude the specimen identification number. Error from measurement in ImageJ (Rasband 1997 2014), especially of total suture length including all oscillations, is believed to be negligible. An additional human introduced source of error that cannot be cont rolled for in this study is problems with documentation in the skeletal collections, to include transposition of numbers or the association of remains with the wrong demographic information. While the current research cannot resolve these particular issue s, by collecting a relatively large sample size, it is hoped that these types of error will be minimized by the amount of data points analyzed. Systematic error in this research is largely related to human variability and biology. The variability of aging is discussed in detail in Chapter 2. The biological process of sutural fusion means that sutures that are largely obliterated are difficult to measure total length (including all oscillations). Therefore, for individuals with a large amount of obliterat ion, total length may approach or be equal to chord length, meaning that sutural complexity will be estimated as very low. Another issue with the measure of sutural complexity is that it only measures surface complexity. Because of their structure as thr ee dimensional objects sutures can vary in complexity and morphology beyond what is seen on the surface (Hotzman 2010). Thus the suture length ratio may
1 58 not give the complete or accurate picture of sutural complexity as it relates to suture structure and function. Accounting for this variability could be done with methods that include computed tomography imaging, but that is beyond the scope of the current investigation. Summary This chapter outlined the sampling strategy, sample, data collection methods and data analyses employed in this research. It also addressed possible error and limitations in this specific research and research in skeletal biology. The following chapter discusses the results of the data analyses.
159 Table 4 1. Power analyses for sample size based on previously reported correlation ( r ) values for age and palatal suture closure. n r Significance level ( ) Power (1 ) Reference 105 0.27 0.05 0.8 Wheatley (1996) 30 0.49 0.05 0.8 Wheatley (1996) 23 0.55 0.05 0.8 Nawrocki (1998) Table 4 2. Power analyses for sample sized based on desired effect size in ANOVA for group assignment and sutural closure. n k a F b Significance level ( ) Power (1 ) 394 2 0.10 0.05 0.8 323 3 0.10 0.05 0.8 64 2 0.25 0.05 0.8 53 3 0.25 0.05 0.8 26 2 0.40 0.05 0.8 22 3 0.40 0.05 0.8 a Number of groups. b Desired effect size, larger numbers indicate larger effects.
160 Table 4 3 Collection summary information. This table represents all individuals examined prior to data scrubbing. Collection Location Total Size Ages (in years) Period Ancestry Number of individuals examined for this research Maxwell Documented University of New Mexico 278 Fetal 100 Modern European 173 William M. Bass Donated University of Tennessee, Knoxville ~1000 Fetal 101 Modern European, African, Asian 177 Robert J. Terry Smithsonian National Museum of Natural History 1728 14 102 Historic European, African 225 Hamann Todd Osteological Cleveland Museum of Natural History ~3100 Fetal 105 Historic European, African 15 Chiba Documented Chiba University School of Medicine, Japan 199 17 83 Historic Asian 153 Jikei Documented Jikei University Schoo l of Medicine, Japan 283 skeletons, 757 skulls Fetal 95 Modern Asian 114
161 Figure 4 1. Palate divided into 15 sections. The drawing is not to scale. Table 4 4. Numerical designators and descriptions for each of the 15 sections of the palatal sutur es. Numerical designator Suture Section Side 1 Incisive Lateral Right 2 Incisive Lateral Left 3 Incisive Medial Right 4 Incisive Medial Left 5 Posterior median palatine Anterior Midline 6 Posterior median palatine Posterior Midline 7 Transverse pala tine Greater palatine foramen Right 8 Transverse palatine Greater palatine foramen Left 9 Transverse palatine Lateral Right 10 Transverse palatine Lateral Left 11 Transverse palatine Medial Right 12 Transverse palatine Medial Left 13 Anterior median palatine Anterior Midline 14 Anterior median palatine Mid section Midline 15 Anterior median palatine Posterior Midline
162 Figure 4 2. Palate scoring based on the entirety of each palatal suture. The drawing is not to scale.
163 Table 4 5. Palatal v ariant scoring. More detailed trait descriptions are available in Chapter 3. Score Variant 0 1 2 3 Reference Accessory lesser palatine foramina a None present 1 present 2 present 3 present b Berr y and Berry (1967), Hauser and D e Stefano (1989) Cris ti marginalis a Absent Present N/A N/A Stieda (1891) Lateral/medial groove bridging a Absent Spurs/ridges present Spurs/ridges present, tendency towards bridging Complete Hauser and D e Stefano (1989); modified Palatine torus Absent Small, including thin m ounding in vicinity of suture Moderate Pronounced Hauser and D e Stefano (1989), modified Maxillary tori a Absent Small Moderate Pronounced Hauser and D e Stefano (1989), modified Maxillary exostoses a Absent Small Moderate Pronounced Developed Maxillary/pa latine bone quality Good Moderate Thin/Poor N/A Developed Porosity Absent Present N/A N/A Developed Palate shape Elliptic Hyperbolic Parabolic Trapezoidal Gill (1998), modified Transverse palatine shape Straight Anterior deviation Anterior and posterio r deviation (M shaped) Posterior deviation Hefner (2009) Zygomaticomaxillary suture shape a No angles; greatest lateral projection at inferior zygomatic 1 angle; greatest lateral projection near midline 2+ angles; variable position for greatest lateral pro jection; S shaped/jagged N/A Hefner (2009) a Sides scored separately. b LPF scored as number present. Majority less than 3, but some scores of 4 and 5 observed.
164 Table 4 6. Dental inventory categories, modified from Buikstra and Ubelaker (1994) Score De scription 1 Present, not in occlusion 2 Present, development complete, in occlusion 3 Missing, no associated alveolar bone 4 Missing, antemortem loss, alveolar bone resorbing or fully resorbed 5 Missing, postmortem loss, no alveolar resorption 6 Mis sing, unobservable, congenital absence, unerupted; alveolar bone intact and no resorption 7 Present, damaged Table 4 7. Dental wear scoring for the incisors and canines, following Smith (1984) and described in Buikstra and Ubelaker (1994) Score Desc ription 1 Unworn, polished, or only small wear facets; no dentin exposure 2 Point or hairline of dentin exposure 3 Line of dentin with distinct thickness 4 Moderate dentin exposure, no longer resembles a line 5 Large dentin area, enamel rim still comp lete 6 Large dentin area, enamel rim lost on one side or very thin enamel 7 Large dentin area, enamel rim lost on two sides or small remnants of enamel remain 8 No enamel remains, complete loss of crown, crown surface same shape as roots
165 Table 4 8. Dental wear scoring for the premolars, following Smith (1984) and described in Buikstra and Ubelaker (1994) Score Description 1 Unworn, polished, or small facets; no dentin exposure 2 Moderate cusp removal, blunting 3 Full cusp removal and/or moderat e dentin patches 4 Minimum of one large dentin exposure on one cusp 5 Two large dentin areas, possible coalescence of these areas 6 Dentin areas coalesced, enamel rim still complete 7 Full dentin exposure, loss of rim on at least one side 8 Severe los s of crown height, crown surface takes on shape of roots Table 4 9. Dental wear scoring for the molars, following Scott (1979) and described in Buikstra and Ubelaker (1994) Note: the molar is divided into four quadrants, and each quadrant is scored a nd then summed to produce a wear score between 4 and 40. Score Description 0 a No data available; tooth not in occlusion, unerupted, or absent antemortem or postmortem. 1 Wear facets invisible or very small 2 Wear facets large, large cusps present, surf ace features evident; pinprick sized dentin exposure or dots possible. 3 Cusp(s) rounded and not clearly defined; cusp coming obliterated but not flat 4 Area and cusp(s) flat; no dentin exposure or only very small pinprick sized dot 5 Area and cusp(s) f lat; dentin exposure one fourth of quadrant or less 6 Greater dentin exposure than 5, dentin exposure more than one fourth of quadrant, but enamel still present; quadrant surrounded by enamel on all sides 7 Enamel on only two sides of quadrant 8 Enamel on only one side of quadrant, usually outer rim; enamel thick to medium on remaining side 9 Enamel on only one side of quadrant; enamel is very thin; sides may be worn through 10 No enamel on any part of quadrant; complete dentin exposure; wear below cer vicoenamel junction and into root a Not employed in this research.
166 Table 4 10. Landmark definitions, listed alphabetically. Name Description Reference Antero lateral transverse palatine (altp) The most antero lateral point on the transverse palatine s uture. This point does not include any descent into the greater palatine foramen but may vary in relative location among individuals. Defined for this study Incisulare (inc) The most anterior point on the median palatine suture directly posterior to the posterior margin of the incisive foramen. Skrzat et al. (2003: p 124) Posterior nasal spine (pns) The midpoint of the posterior edge of the hard palate. In cases where the posterior termination of the PMP and the pns did not correspond, the measurement w as taken to the most posterior point of the suture. Steele and Bramblett (1988) Staurion (sr) The intersection of the median and TP sutures. Martin (1928) Figure 4 3. Palatal landmarks used in this study.
167 Figure 4 4. Digital measurement set u p: Wacom Tablet touchscreen monitor, stylus, and Image J. Photograph courtesy of Carrie A. Brown. Figure 4 5. Setting the scale in ImageJ. Photograph courtesy of Carrie A. Brown.
168 A B Figure 4 6. Example of brightness a nd contrast adjustment in ImageJ. A) Palatal photograph before adjustment, B) palatal photograph after adjustment. Photographs courtesy of Carrie A. Brown.
169 Table 4 11. Measurements and their landmarks. All measurements taken in cm. Suture From To AMP inc sr PMP sr pns TP R altp L altp TP R sr R altp TP L sr L altp Table 4 12. Order of measurements in ImageJ. # Measurement 1 AMP chord 2 PMP chord 3 R TP chord 4 L TP chord 5 AMP length 6 PMP length 7 R TP length 8 L TP length 9 AMP fus ion 10 PMP fusion 11 R TP fusion 12 L TP fusion
170 A B Figure 4 7. Measurements of the AMP suture in ImageJ. A) AMP chord, B) AMP total length. Photographs courtesy of Carrie A. Brown.
171 A B Fi gure 4 8. Measurements of the PMP suture in ImageJ. A) PMP chord, B) PMP total length. Photographs courtesy of Carrie A. Brown.
172 A B Figure 4 9. Measurements of the TP suture, right side, in ImageJ. A) Right TP chord, B) ri ght TP total length. Photographs courtesy of Carrie A. Brown.
173 Figure 4 10. Mea suring sutural fusion in ImageJ ; each location of fusion is measured individually (yellow line), and total fusion per suture is calculated as the sum of all individual m easurements. Yellow arrows indicate other areas of fusion to be measured. Photograph courtesy of Carrie A. Brown.
174 Table 4 13. Total sample analyzed for this research, broken down by demographic groups with mean and median ages per group. Ancestral gro up Sex Age group n Mean age Median age African Female 20 29 22 25.09 24.5 30 39 20 34.75 35 40 49 20 44.85 45 50 59 19 54.79 55 60 69 21 64.38 65 70+ 22 83.32 81.5 African Male 20 29 23 24.74 25 30 39 20 34.80 35 40 49 22 44.91 45.5 50 59 20 54.55 55 60 69 22 64.36 64 70+ 19 78.32 77 Asian Female 20 29 22 24.23 24 30 39 22 33.91 33 40 49 21 43.24 43 50 59 20 53.65 53 60 69 22 64.36 64 70+ 23 77.09 76 Asian Male 20 29 22 24.36 24 30 39 22 34.14 34 40 49 24 43.58 43 50 59 22 53.59 53 60 69 22 64.09 65.5 70+ 22 77.82 77 European Female 20 29 10 26.00 26 30 39 22 35.00 35.5 40 49 22 44.95 44.5 50 59 20 53.55 53 60 69 22 64.82 64.5 70+ 25 84.00 82 European Male 20 29 19 24.47 24 3 0 39 25 35.52 36 40 49 22 45.18 46 50 59 17 54.47 54 60 69 19 65.21 66 70+ 25 79.44 78 TOTAL 762 50.94 50
175 Figure 4 11. Age distribution of sample ( n = 762).
176 Figure 4 12. Sample distribution by age group and sex.
177 Figure 4 13. Sample distribution by age group and ancestry ( AF = African, AS = Asian, EU = European )
178 Table 4 14. Summary of dummy variables developed from nominal variables. Original score is listed first, followed by dummy variable underneath. Variable Dummy variables Sex Female Male 0 1 Ancestry A frican A sian E uropean 1,0 0,1 0,0 Time period Historic Modern 0 1 Zygomaticomaxillary suture shape Straight (0) 1,0 1 angle (1) 0,1 2+ angles (2) 0,0 TP suture shape Straight (0) Anterior dev iation (1) M shaped (2) Posterior deviation (3) 1,0,0 0,1,0 0,0,1 0,0,0 Palatal shape Elliptic (0) 1 ,0,0 Hyper bolic (1) 0 ,1,0 Para bolic (2) 0,0,1 Trap ezoidal (3) 0,0,0 Marginal crest Absent 0 Present 1 Dental restorations Absent 0 Present 1 Full edentulism Absent 0 Present 1 Porosity Absent 0 Present 1
179 CHAPTER 5 RESULTS Palatal Suture Closure and Age Three qualitative, categorical systems were employed in the first phase of data collection: 15 section/4 phase, full suture/4 phase, and full suture/binary. The relationships of age and closure for the qualitative, categorical scoring systems are summarized with box and whisker plots (Figures 5 1 through 5 14). The boxes display the median age and first and third quartiles for age per clo sure score for sections of sutures, individual sutures, and summary scores. In all plots a general trend of increasing age with increasing closure score is noted, though the dispersion of values per closure score is quite large. It also appears that redu cing the number of categories by collapsing certain scores may be warranted (e.g., scores of 1 and 2 for the TP suture in the 15 section/4 phase system [see Figure 5 3 and below]). Distributions of summary scores show the most frequent scores are those to wards the lower end of the fusion spectrum (Figure 5 15). The box and whisker plot for age and summary score of the 15 section/4 phase system is quite complex and therefore also graphed as a scatterplot for comparison (Figure 5 16). When graphed as a scat terplot it can be seen that summary score approximates continuous data due to the number of possible scores available ( n = 45). Figure 5 16 includes a least squares regression line (black), but the scatterplot does not show a strong or clear relationship between summary score and age. Since the data do not appear to conform to any known models that could potentially strengthen or linearize the relationship between age and suture closure summary score, the data were not transformed. A nonparametric smooth er was not helpful in summarizing the
180 data due to the approximation of continuous data (i.e., the data still behave as discrete data even with the large number of possible scores). To compare the relationship of suture closure to age for each of the suture s or sections of sutures within the categorical systems, ANOVA tests were run. In order to do this, palatal fusion scores were treated as ordinal factors and the mean age per fusion score was compared for sections of sutures and full sutures. Homogeneity of variance for scores produced by the three categorical systems was tested using Fligner Killeen tests (Crawley 2013). The majority of sutural fusion scores showed constant variance, making the choice of a parametric test appropriate. Tables 5 1 throu gh 5 3 display the results of these tests and indicate that the IN and PMP sutures differ significantly in closure score by age for the 15 section/4 phase and full suture/4 phase systems, while the PMP and right TP sutures differ significantly for the bina ry system. The AMP suture shows differences in score by age for only one section in the 15 section/4 phase system. In terms of summary scores for each of the systems, one way ANOVA tests for age and each summary score, treating each possible combination of scores as a factor, indicate that there are significant differences in mean age by each level of summary score ( p < 0.001, all three systems). Because the boxplots appear to show some degree of overlap in fusion scores for the 4 phase systems, the p otential of collapsing certain fusion scores was also investigated. Tables 5 4 through 5 (HSD) post hoc comparisons of differences in mean age per fusion score; the binary system is presented for comparison. Most of the significant differences are found in unfused (0) versus partial (1 or 2) or complete (3) fusion categories for both iterations of
181 the 4 phase system (Tables 5 4 and 5 5). In the 15 section/4 phase system, differences in mean age for fusion s cores 1 and 2 are significant only for right and left portions of the medial IN suture. The full suture/binary system shows significant differences in mean age for scores of no fusion (0) and any amount of fusion (1) for all sutures except the IN suture ( Table 5 6). Based on these results, fusion scores 1 and 2 were experimentally collapsed for the 15 section/4 phase and full suture/4 phase systems to determine if a 3 phase system was preferable to a 4 phase or binary system. This collapsing resulted in a 3 phase system with scores of: 0 no fusion, 1 partial fusion, and 2 complete fusion, and new summary scores were calculated. Comparison of the new 3 phase system with known age was done alongside this same comparison for the 4 phase and binary sys tems. order correlation was used to examine the relationship between age and closure since suture closure scores are ordinal level data. Correlation values ( rho ) were computed for sections of sutures, full sutures, and summary scores for t he qualitative, categorical scoring systems (Tables 5 7 through 5 11). All rho values have positive correlations (i.e., a larger value in age is associated with a higher sutural fusion score), although values for sections of sutures and full sutures are a lways less than 0.400. The lowest value is for the IN suture in the full suture/binary scoring system ( rho = 0.052; Table 5 9), and the highest value is for the left medial portion of the IN suture in the 15 section/4 phase scoring system ( rho = 0.368; Ta ble 5 7). The control sutures have rho values lower than most values for individual sutures or sections of sutures in all systems (Table 5 10). Correlation values for the summary suture scores are generally greater than the individual sutural values, wit h the highest correlation
182 between age and score in the 15 section/4 phase system (Table 5 11). However, this value is not markedly different from that of the full suture/4 phase system, nor does the inclusion of the TP suture within the GPF drastically af fect the correlation between age and summary score (e.g., full suture/4 phase rho = 0.423 versus full suture/4 phase with GPF rho = 0.419; Table 5 11). When comparing the 4 phase scoring system to the collapsed 3 phase scoring system there is no remarkabl e difference between the two systems in terms of rho (Tables 5 7, 5 8, and 5 11). Of the qualitative, categorical scoring systems, the summary score from the 15 section/4 phase system has the strongest relationship between known age and sutural closure (T able 5 11). Descriptive statistics for fusion ratios in the quantitative system are given in Table 5 12, and the relationship of fusion ratio and age is summarized with scatterplots (Figures 5 17 and 5 18). Only the AMP, PMP, and TP sutures (but not withi n the GPF) are available for the quantitative system, since the IN suture was largely fused in adults and could not be scored or measured and the TP suture within the GPF was not measured from photographs (see Chapter 4). While the fusion ratios are conti nuous, their distributions are not normal; all are skewed to the right with the majority of ratio s around 0 (Figure 5 19 and Table 5 12). As can be seen in Figures 5 17 and 5 18 the data do not conform well to a linear model (indicated by the black lines) and values tend to cluster close to 0. Arcsine transformation of the fusion ratio summary score, appropriate for values between 0 and 1 and following Herring (1972) and McDonald (2014) was conducted. This transformation produced the same relationship of age and summary score ( rho = 0.349) and did not greatly reduce residual error as compared to the untransformed fusion score data, but it did produce a better distribution of points
183 around the least squares regression line (Figure 5 20). Because no larg e scale improvement was made by arcsine transformation, these values were not used further. Log transformation, commonly used in biological data, is not possible for the fusion ratio summary scores because of multiple instances of non fusion of AMP, PMP, and TP sutures, resulting in scores of 0. order correlation values for fusion ratios for the summary score and individual sutures are given in Table 5 13 along with a comparison of rho values for the same individual sutures and summary scores scored categorically. As compared to categorical systems, the suture fusion ratio results are similar, although suture ratio rho values are lower that full suture/4 phase values for all except the AMP suture. The fusion scores for the full suture/ 4 phase system were also compared to rho values indicate moderately high levels of agreement among the two systems (PMP suture rho = 0.790, TP suture rho = 0.671, AMP suture rho = 0.679), though these values indicate the relationship is less than perfect. Plotting the r esiduals for all summary score systems indicates that the qualitative, categorical systems fare slightly better in terms of residual errors as compared to the quan titative system. Of the various qualitative systems tested, the summary score of the 15 section/4 phase system has the highest correlation with known age and the most normally distributed error. When treating summary scores as continuous variables and co mparing them to age via regression analyses, all have significant relationships with age ( p < 0.001). However, the R 2 values, which explain the amount of variation in the response variable that is accounted for by variation in the explanatory variable, in dicate
184 that the effect of age on fusion, while significant, is not fully explained by variation in age. For the 15 section/4 phase summary score, R 2 = 0.161, indicating that only 16% of the variation in fusion summary score is being accounted for by varia tion in age. The R 2 value for age and the fusion ratio summary score is 0.076, and the other two categorical systems fall between these two values (full suture/4 section system R 2 = 0.156, full suture/binary system R 2 = 0.111). Group Affiliation and Sutur al Closure Sex, ancestry, and time period were visually compared to the sutural closure summary score with the highest rho (15 section/4 phase system; Tables 5 11 and 5 13) using box and whisker plots (Figures 5 21 through 5 23). These plots indicated tha t there appear to be differences in closure summary score for sex and ancestry, but not for time period. For sex, there are several outliers in the higher summary scores for females (see Figure 5 21). Homogeneity of variance for sex, ancestry, and time p eriod in terms of summary score were checked with Fligner Killeen tests (Crawley 2013). Variance is not constant for sex ( p < 0.001), is constant but approaches significance for ancestry ( p = 0.086), and is constant for time period ( p = 0.918). Error di stributions are examined by plotting the residuals for each of the three groups. Based on Normal Q Q plots for all three groups, error approaches a normal distribution. However, due to non constant variance in two of the groups and several outliers for s uture fusion in females, a parametric model that included all three groups and their interactions falsely identified a significant difference in sutural fusion between historic and modern individuals when no such difference exists ( see Figure 5 23). There fore, initial testing was done by group and results from parametric and non parametric tests were compared prior to performing multiple group comparisons.
185 Wilcoxon rank sum tests, the non t test (Crawley 2013), ind icate that there is a statistically significant difference in summary fusion score between males and females ( p < 0.001) but not between historic and modern individuals ( p = 0.876). A Kruskal Wallis test of medians, the non parametric equivalent of ANOVA, indicates that there is a significant difference in summary fusion score among ancestral groups ( p < 0.001). These tests confirm the visual differences seen in Figures 5 21 through 5 23. Parametric tests had similar results tests for sex a nd time period produced p values of < 0.001 and 0.844, respectively, and ANOVA for ancestry showed differences among groups that were significant at p < 0.001. Based on these results, time period was removed from subsequent group analyses. Given the simil ar values in nonparametric and parametric tests and the large sample size, ANOVA was used to further investigate potential interactions between sex and ancestry in terms of suture fusion using the 15 section summary score, which approximates a continuous v ariable. While both sex and ancestral groups were still significantly different for suture fusion ( p values < 0.001), no significant interaction effect was present for sex and ancestry ( p = 0.861). To test for the effects of sex and ancestry while contro lling for age, an ANCOVA was run, and interactions among all terms were explored. These results are displayed in Table 5 14. Notably sex, ancestry, and age display significant differences in terms of suture closure, and there is an interaction between s ex and age. No other interactions are significant at the p < 0.05 level, though the interaction of sex, age, and ancestry is just above the p value cutoff.
186 Biomechanical Variables Descriptive Statistics Inventory scores by tooth are presented in Table 5 15 to display the overall picture of the dentition in the study sample. The most frequent scores, regardless of tooth number, are 2 (present and in occlusion) and 4 (absent antemortem). Unobservable teeth (score of 6) are most common for the third molar s, while postmortem tooth loss (score of 5) is most common for the incisors. Teeth that are present but not in occlusion (score of 1) along with teeth that are missing with no associated alveolar bone (score of 3) are the least common in the sample. Descr iptive statistics for dental wear scores for the maxillary teeth are presented in Table 5 16. Because of the categorical scale of dental wear scores, the median is presented along with the mean to give a picture of the distribution of wear scores per toot h. For the molars, the distributions are slightly skewed right (mean > median), indicating that higher scores are less common, while for the premolars, canines, and incisors, the distributions approach normal (mean = median). Of all tooth positions, the third molar had the lowest number of observable teeth, which corresponds with the results presented in Table 5 15. Comparison of wear scores by side per tooth number (e.g., first right molar and left first molar) was undertaken by testing if the absolute value of the differences in wear per side by tooth were statistically significantly different from t tests indicated that the mean of the absolute values for wear were significantly different than 0 for all teeth ( p <0.001). Wh en teeth were compared within class (e.g., third mola t test to examine if the difference in wear was significantly different from 0, significant differences in wear by tooth position were noted for all pairs ( p <0.001)
187 Because wear differs significantly by side and tooth class, m ean wear scores for all maxillary dentition, posterior dentition, and anterior dentition were calculated for each individual; descriptive statistics and wear score distributions are displayed in Table 5 17 and distributions in Figure 5 24. All mean wear scores are skewed to the right, indicating that a low level of wear is more frequent in the sample than a high level of wear. Posterior wear scores have higher means and medians than anterior wear scores, and the overall mean wear falls between wear scores for the posterior and anterior dentition. These results correspond to those displayed in Table 5 16, where the molars display higher mean and median wear scores than the anterior dentition, although the premolars are nearly identical to the canines and incisors in terms of wear. The frequency of antemortem tooth loss (AMTL) by tooth number and in the total sample, as summarized by the AMTL Index, is presented in Table 5 18 and Figure 5 25 The first and second molars represent the highest percentages of teeth lost antemortem, and the lowest percentage of teeth lost antemortem are the canines and third molars. The mean AMTL Index is 39.55 and the median is 25.00, and its distribution in thi s sample is bimodal, with the most frequent scores representing individuals with no to very little tooth loss and individuals with near to complete tooth loss. Individuals with moderate tooth loss, at the center of the histogram, are less common. In the total sample, 167 individuals are fully edentulous, compared to 595 individuals with at least one tooth present (21.92% and 78.08% of the sample, respectively). Frequency distributions for sutural complexity for each of the three measured palatal suture s (PMP, TP, and AMP) are given in Figures 5 26. Mean sutural
188 complexity is greatest in the TP suture (1.53) and less but nearly equal in the PMP and AMP sutures (1.24 and 1.25, respectively). All three histograms show distributions that are skewed to the right. Along with the mean complexity scores, this indicates that very complex sutures in the palate are rare. Relationship to Sutural Fusion The relationship of mean dental wear sco res and fusion, as summarized by the 15 section and fusion ratio summary scores is given in Figures 5 27 through 5 29 order correlation values for wear and fusion; non parametric correlation was employed due to non normal distributions of wear scores. For all mean wear scor es, values are positive and similar between qualitative and quantitative systems, indicating that with increased wear, increased fusion is observed. The relationship of mean anterior wear and fusion is the highest of the three. Separate regression analys es were conducted for summary fusion scores and overall wear and then scores and anterior/posterior wear, since combining the wear variables violates the assumption of independence. These analyses indicate that overall mean wear has a significant effect o n fusion ( p < 0.001, R 2 = 0.026), while when considered separately, only anterior wear has a significant effect (Table 5 19; R 2 = 0.104). Regressing wear on the fusion ratio summary score, overall wear also has a significant effect on fusion ( p < 0.001, R 2 = 0.052), as do both anterior and posterior wear (Table 5 20; R 2 = 0. 104 ). These R 2 values indicate that when posterior and mean wear score are considered, around 10% of the variation in fusion summary score is explained by variation in dental wear. B oth summary fusion scores were then compared with the AMTL Index (Figure 5 30). The least squares regression lines and rho values indicate that there is a
189 positive correlation between fusion summary score and AMTL, though the relationship for AMTL and fus ion ratio summary score is less strong than AMTL and 15 section summary fusion score. Linear regression of summary scores and the AMTL Index indicates that tooth loss is a significant effect on sutural fusion regardless of fusion system ( p < 0.001, both s ystems), but the R 2 values indicate that the amount of variation in fusion is being explained by only a small amount of the variation in AMTL Index (15 section summary score R 2 = 0.136, fusion ration summary score R 2 = 0.041). Comparing fusion summary sco res with edentulous status individuals with at least one tooth present versus those who were fully edentulous shows that individuals without any dentition have higher fusion summary scores than those with at least one tooth present (Figure 5 31). Wilc oxon rank sum tests, performed due to the non normal distributions of both fusion and edentulism in the sample, indicated that the difference in edentulous and non edentulous individuals in terms of fusion is significant, regardless of summary system used ( p < 0.001). Sutural complexity and fusion have a negative relationship for all three measured sutures (Figure 5 rank order correlation values indicate that the relationship between complexity and fusion, while negative, is moderate in str ength, with PMP suture complexity having the strongest relationship to fusion ratio. Negative rho values indicate that with an increase in complexity there is a decrease in fusion. Linear regression of fusion ratio and suture complexity per suture shows that PMP and TP complexity have significant effects on fusion, while AMP suture complexity does not (Table 5 21). The adjusted R 2 value for regression of fusion on sutural complexity is
190 0.187, indicating that variation in complexity accounts for about 19% of variation in suture fusion. Fusion of the control sutures was also compared to wear and tooth loss. It was not compared to complexity since this variable was not measured for these sutures. Fusion of the nasofrontal and zygomaticomaxillary sutures h as low associations with overall mean dental wear. For the nasofrontal, the relationship is negative ( rho = 0.040) and for the zygomaticomaxillary, the relationship is positive (left rho = 0.048, right rho = 0.068). Comparing posterior wear and fusion, rho values for the nasofrontal and left zygomaticomaxillary sutures are negative ( rho = 0.067 and 0.011, respectively), and the right zygomaticomaxillary suture is positive ( rho = 0.012). For anterior wear, all associations are positive, with a low asso ciation for the nasofrontal suture ( rho = 0.071) and moderate associations for the left and right zygomaticomaxillary sutures (left rho = 0.211, right rho = 0.232). All associations for AMTL Index and control suture fusion are positive, with moderately st rong relationships (nasofrontal rho = 0.184, left zygomaticomaxillary rho = 0.242, right zygomaticomaxillary rho = 0.238). Relationship to Age Age and mean dental wear scores display a positive correlation, based on order correlations (Figu re 5 33). Older individuals show larger mean wear scores, while younger individuals have smaller wear scores. The strongest relationship of wear and age is for the mean anterior wear score. Regression of mean wear on age indicates that age is a signific ant effect for wear (all mean wear scores, p < 0.001). Coefficients of determination for mean wear are as follows: overall mean wear R 2 = 0.056, mean posterior wear R 2 = 0.061, and mean anterior wear R 2 = 0.230.
191 The regression of age on wear was not per formed as this does not reflect biological reality (i.e., wear does not explain variation in age) and because the overall mean wear score includes posterior and anterior dentition. AMTL Index and age also has a positive correlation, and the rho value is moderately strong ( Figure 5 34). In this sample, an increase in tooth loss is associated with increased age, and vice versa. Linear regression of the AMTL Index on age showed that the age does have a significant effect on AMTL Index ( p < 0.001, R 2 = 0.3 46). Sutural complexity of the PMP, TP, and AMP sutures shows little to no relationship with age (Figure 5 35). The rho values, while extremely small, are also negative. Linear regression for complexity by suture and age indicates that age does have a si gnificant effect on PMP and AMP suture complexity at the p < 0.05 level (PMP p = 0.027, AMP p = 0.033), but the R 2 values are quite low (PMP R 2 = 0.006, AMP R 2 = 0.006). The effect of age on TP suture complexity is not significant ( p = 0.516, R 2 = 0.001). Relationship to Group Affiliation Mean wear scores by sex, ancestry, and time period are displayed in Figures 5 36 to 5 38, and descriptive statistics are in Tables 5 22 to 5 24. For all mean wear scores, females have lower wear scores than males, though there are high wear outliers for both sexes across all three scores. Europeans have lower mean wear score than Africans and Asians for overall and posterior wear, but Asians are lower than both Europeans and Africans for anterior wear. As with sex, all groups display outliers for high wear. Comparing modern and historic, historic individuals always have higher mean wear scores than modern individuals, regardless of mean wear score system.
192 There are a few high wear outliers in the historic group, but fa r less than for sex or ancestry. Due to non constant variance, differences in mean wear scores by sex, ancestry, and time period were compared non parametrically (Table 5 25). Significant differences are present across all groups and mean wear scores exc ept for mean anterior wear and time period. The AMTL Index across sex, ancestry and time period was analyzed via ANOVA and summary statistics by group are provided in Table 5 26. While variance does exhibit significant differences for population and era, results from non parametric tests were almost identical to ANOVA, so the parametric option with post hoc tests was preferred. There are no significant differences in mean AMTL Index by sex ( p = 0.591), but there are significant differences in mean AMTL I ndex by ancestry and time period ( p = 0.012 and p < post hoc tests indicate that ancestral differences in mean tooth loss are significant between Europeans and Asians ( p = 0.013) and approach significance between European s and Africans ( p = 0.066), with Europeans having a higher mean AMTL Index than Africans and Asians. Modern individuals have a higher mean AMTL Index than historic individuals. Suture complexity scores for the PMP, TP, and AMP sutures were also analyzed via ANOVA for each of the three groups. Variance was found to be significantly different prior to running ANOVA, but parametric tests were still run because of the large sample size. Descriptive statistics for suture complexity by group are given in Tab le 5 27. Sutural complexity displays significantly different mean values by sex and ancestry for all three sutures ( p < 0.001) and by time period for the AMP suture ( p = 0.033). post hoc tests indicate that significant differences for PMP sut ure
193 complexity exist between Asian African ( p < 0.001) and Asian European ( p < 0.001), for TP suture complexity between Asian African ( p < 0.001) and African European ( p < 0.001), and for AMP suture complexity between Asian African ( p < 0.001) and European Asian ( p = 0.020), with the African European comparison approaching significance ( p = 0.078). Relationship to Each Other The relationship of mean tooth wear scores and antemortem loss is depicted in Figure 5 39, which order c orrelation coefficients. There is a positive correlation between AMTL Index and wear, though the relationship for overall mean wear and posterior mean wear is weak. Mean anterior wear score shows a moderate, positive relationship. Regressing AMTL on ove rall mean wear shows that overall mean wear score has a significant effect on AMTL ( p < 0.001, R 2 = 0.020). Regressing AMTL on posterior and anterior mean wear scores shows that anterior wear has a significant effect on AMTL ( p < 0.001) while the effect o f posterior wear score approaches significance ( p = 0.071). In this model, adjusted R 2 = 0.183. As with antemortem tooth loss, dental wear was compared to sutural complexity by mean wear score for each of the three measured sutures (Figures 5 40 to 5 42). The scatterplots indicate that the relationship of wear and complexity is weak, with little order correlation coefficients, given in each figure confirm that the relati onships are weak and largely negative (i.e., decreased complexity is associated with increased wear). No rho values exceed the absolute value of 0.200 and most are very close to 0. Regressing complexity on wear, mean wear scores have no significant effec ts on PMP, TP, or AMP complexity ( p values from 0.370 to 0.725).
194 Sutural complexity and AMTL Index scatterplots are given in Figure 5 43. Complexity and AMTL do not show a strong relationship, regardless of suture or wear score employed. Correlation valu es are close to 0. Regressing complexity on tooth loss with simple linear regression shows that while AMTL has a significant effect on PMP and AMP complexity ( p = 0.001 and p < 0.001, respectively), R 2 values of the amount of variation of PMP and AMP comp lexity explained by variation in AMTL Index are quite low ( R 2 = 0.014 and R 2 = 0.031, respectively). The effect of AMTL on TP complexity was not found to be significant ( p = 0.787). The reverse regression regressing AMTL on complexity was not performe d, as complexity is unlikely to explain tooth loss and the correlation values were so low. Palatal Variants Relationship to Sutural Fusion The relationship of palatal traits and sutural fusion, as measured by the 15 section/4 phase summary score, was ex order correlations (Table 5 28). The majority of traits show little to no relationship with sutural fusion, though there are slight positive correlations between maxillary and palatine bone quality ( rho = 0.175 and 0.231, re spectively). Because a higher bone quality score indicates poorer bone quality, this relationship is indicative of increased fusion being associated with a loss in bone quality. The palatine torus is also positively correlated with fusion ( rho = 0.105), though the relationship is not strong. All other traits do not have rho values greater than 0.100 or less than 0.100 in regards to fusion. The relationship of sutural shape, scored ordinally, and fusion was also investigated for the TP and zygomaticomaxi llary sutures. The correlation values are low for all comparisons. For the TP suture, when compared to the fusion score from the full
195 suture/4 phase system, rho = 0.038, and when compared to measured fusion, rho = 0.025. For the left and right zygomati comaxillary sutures and fusion score from the full suture/4 phase systems, values are rho = 0.015 and 0.005, respectively. Relationship to Age The order correlations, are given in Table 5 29. Most traits exhibit very low values, indicating little to no relationship to age. Traits with absolute rho values between 0.200 and 0.300 include: left and right maxillary tori and exostoses and maxillary and palatine bone quality. Tori an d exostoses show a negative correlation, indicating that as age increases, frequency and expression of tori and exostoses decreases. For bone quality, a higher score indicates bone that is of poor quality, so the positive correlation found with age indica tes that as age increases, poor quality bone is more common. Relationship to Group Affiliation Palatal trait frequencies by group are given in Appendix B: sex Tables B 1 to B 20; ancestry Tables B 21 to B 40; and time period Tables B 41 to B 60. T o test to see if the palatal traits different significantly in frequency by sex, ancestry, or time chi square tests were employed (Table 5 30). These values show that significant differences by sex occurred for the lesser palatine forami na, palatine and maxillary tori, maxillary exostoses, palatal porosity, and zygomaticomaxillary suture shape. Males appear to display higher counts of additional lesser palatine foramina than females (Tables B 1 and B 2). Females show a tendency to exhib it more pronounced palatal tori, while for males this is the case for maxillary tori and exostoses (Tables B 9 to B 13). An absence of palatal porosity is more common in females as opposed to males (Table B 16), and for traits that could not be observed d ue to
196 obliteration, there was always a higher number of males than females that were scored as unobservable (Tables B 18 to B 20). The significant differences in zygomaticomaxillary shape may be due to these higher frequencies of obliteration; however sco res of 1 also appear more commonly in males than females (Tables B 19 and B 20). In terms of ancestry, all traits display significant differences except the lesser palatine foramina. While the frequency tables are harder to interpret for this three way comparison, they are informative in showing the presence and significance ancestral variation across palatal traits. Notably, palatal shape and TP shape show marked differences in frequencies among groups (Tables B 37 and B 38). Results for time period ar e also significant, except for the lesser palatine foramina, palatine and right maxillary tori, and TP and right zygomaticomaxillary suture shapes. Modern individuals tend to have higher numbers of accessory palatine foramina and slightly increased presen ce of marginal crests as compared to historic individuals (Tables B 41 to B 44). Full bridging of medial and lateral grooves is markedly higher in historic individuals, and a proclivity towards bridging is also higher for historic (Tables B 45 to B 48). Presence of pronounced left maxillary tori is more common in historic individuals, but there is no such difference for the right side (Tables B 50 and B 51). Moderate expressions of maxillary exostoses are observed with higher frequencies in historic indi viduals, while extreme expression is more common for modern individuals (Tables B 52 and B 53). Moderately poor to poor bone quality is more frequent for historic individuals, while the presence of porosity is slightly higher in modern individuals (Tables B 54 to B 56). Hyperbolic and trapezoidal palate
197 shapes are more common for historic individuals, while parabolic shapes are observed with higher frequency in modern (Table B 57). Finally, zygomaticomaxillary sutures with one angle are more common in mo dern individuals, although the difference is not statistically significant for the right side (Tables B 59 to B 60). Relationship to Biomechanical Variables order correlations were calculated for the three mean wear scores and each of the p alatal traits (Table 5 31). These results indicate little to no relationship between wear and trait, regardless of mean wear score used or trait. Values are less than 0.100 or greater than 0.100, with the exception of palate and TP suture shapes and pal atal porosity for the overall and posterior mean wear scores. order correlation coefficients for AMTL Index and palatal traits are presented in Table 5 32. The AMTL Index shows moderate relationships with left and right maxillary tori and exostoses and maxillary and palatine bone quality. The tori and exostoses have negative rho values, indicating that with an increase in AMTL Index there is an associated decrease in expression of maxillary tori and exostoses. For bone quality, a positive correlation indicates that as bone quality declines there is an associated rise in tooth loss, due to the scoring of bone quality in which a higher number indicates poorer bone quality. Sutural complexity and palatal traits have only a small association with the exception of the palatine torus (Table 5 33). The remainder of the rho values are generally near 0.100 or 0.100. The negative correlation between palatine torus and sutural complexity indicates that a larger torus is associated with less compl ex sutures, regardless of suture measured.
198 Relationship to Other Variants In terms of bilateral expression, bilateral traits show no significant differences in frequency of expression and show moderate to high levels of association between left and right sides (Table 5 34). Because the variables show no significant differences in left and right expression, the left side was used for comparisons among variables (Table 5 35). The highest rho value is between maxillary bone and palatine bone qualities, foll owed by the association of maxillary tori and exostoses; both correlations are positive. For tori and exostoses this signifies that large, more pronounced exostoses are associated with large, more pronounced tori. For bone quality, the positive associati on between maxillae and palatines indicates that similar bone quality is found between these two areas of the hard palate (i.e., poor bone quality in one is associated with poor bone quality in the other). Other associations with rho > 0.100 include: lat eral bridging with medial bridging, maxillary tori, and maxillary exostoses; medial bridging with maxillary tori and exostoses; and palatine tori with maxillary tori. Associations with rho < 0.100 include: marginal crest with palatine bone quality; pala tine tori with maxillary bone quality and porosity; maxillary exostoses with maxillary and palatine bone quality; and palatine bone quality with porosity. The lesser palatine foramina show no marked association with other palatal variables. Putting it All Together The above results were all considered when looking at the significant factors that contribute to palatal suture fusion and age. In these considerations, summary fusion scores were used, as they provide the most complete picture of the state of p alatal fusion per individual and serve to simplify analyses that already include many variables. The control sutures were not included. Dummy variables (DV) were employed for
199 nominal variables found to have a significant effect on or relationship with fu sion or age (see Chapter 4). For ancestry, DV0 is European, DV1 is African, and DV2 is Asian. Additionally, geometric means of the AMP, PMP, and TP suture chords were calculated for each individual and included in larger models for fusion and age predict ion to aid in determining if size is a significant factor in fusion or age prediction or if it interacted with any significant terms. Variables determined to have a significant effect on suture fusion include: age, sex, ancestry, mean wear, AMTL Index, e dentulism, PMP and TP suture complexities, palatine torus, and maxillary and palatine bone qualities. The highest correlation for wear is between summary score and anterior mean wear, but overall wear was used in these analyses for its ability to summariz e all dental wear in one variable. Additionally, because of the asymmetry of wear detected, it is most appropriate to use a summary score that encompa sses all teeth present and scor able. The AMTL Index and edentulism were not included in the same equatio n as these two variables are not independent, and the AMTL Index was preferred as it provides a greater level of detail per individual. The interactions of variables were also considered based on relationships and significant effects as outlined above. For fusion, variables with potential interactions include: sex and age; wear and sex/ancestry/time period; AMTL Index and ancestry/time period; PMP/TP/AMP suture complexity and sex/ancestry; AMP suture complexity and time period; AMTL Index and wear; AMTL Index and PMP/AMP suture complexity; palatine torus and sex/ancestry; bone quality and ancestry/time period; bone quality and AMTL Index; and palatine torus and complexity. Variables
200 determined to have non significant effects on fusion include time peri od and the interaction of mean wear and complexity. Multiple regression analyses were conducted in a manual stepwise procedure by first running an analysis with all variables determined to have a significant effect on or relationship to fusion as well as p otentially significant interaction effects. For the 15 section summary fusion score, this initial model included 1 4 variables and 34 two way interactions. Of these, age, sex, PMP and TP suture complexity, palatine torus, and the interactions between sex and suture complexity (all three measured sutures) were significant at the p < 0.10 threshold. A higher threshold was used for this initial analysis because of the large number of variables compared and potential compounding effects. The adjusted R 2 for this initial model was 0.559. The same large scale initial analysis was also conducted for fusion ratio summary score and found these significant effects ( p < 0.10): age, sex, PMP and TP suture complexity, palatine torus, and interactions of age and sex, sex and PMP and TP suture complexities, and TP suture complexity and palatine torus. The adjusted R 2 value for the fusion ratio summary score was 0.489. The geometric mean of the chords was not found to be significant for either fusion summary score. Additional models were then run for each of the fusion summary scores, eliminating all non significant terms during each iteration. At this stage, effects were considered to be significant if the p value was < 0.05. Final models for fusion are presented in Tables 5 36 and 5 37. The adjusted R 2 values for the 15 section summary score and fusion ratio sum mary score are 0.506 and 0.470, respectively. Collectively,
201 these results indicate that about 50% of variation in fusion is accounted for by variation in age, sex, sutural complexity, and expression of the palatine torus. Variables determined to be sign ificantly related to age include: fusion, dental wear, AMTL Index, PMP and AMP suture complexity, maxillary tori and exostoses, and maxillary and palatine bone quality. For fusion, summary scores are employed, and overall mean wear score is used for ease of comparison. Variables with potential interactions include fusion and sex/ancestry; wear and sex/ancestry/time period; AMTL Index and ancestry/time period; PMP, TP, and AMP suture complexity and sex/ancestry; AMP suture complexity and time period; wear and AMTL Index; AMTL Index and PMP and AMP suture complexity; maxillary tori/exostoses and sex/ancestry/time period; bone quality and ancestry/time period; and AMTL Index and maxillary tori and exostoses/bone quality. TP suture complexity does not have a significant relationship with age. As with fusion, multiple regression analyses were conducted in a manual stepwise procedure. The first model was run using the 15 secti on summary score and included 1 7 variables and 5 7 two way interactions. Of these, si gnificant terms ( p < 0.010) include: summary fusion score, le ft and right maxi llary tori, and palatine bone quality, and interactions between fusion score and sex; fusion score and ancestry DV1; wear and ancestry; AMTL Index and ancestry DV 1; AMTL Index and time period; PMP suture complexity and ancestry; left and right maxillary tori and ancestry; palatine bone quality and ancestry; maxillary bone quality and time period; and AMTL Index and right maxillary torus and left maxillary exostoses. The adjuste d R 2 value for this model was 0.4 81 The same large scale analysis was also run using the fusion ratio summary
202 score. Significant effects included: fusion ratio summary score, left and right maxillary tori, and palatine bone quality, and interactions be tween fusion ratio summary score and sex; fusion ratio summary score and ancestry DV1; wear and ancestry; wear and time period; AMTL Index and ancestry DV1; AMTL Index and time period; PMP suture complexity and ancestry DV1; AMTL Index and wear; left and r ight maxillary tori and ancestry; palatine bone quality and ancestry; maxillary bone quality and time period; and AMTL Index and right maxillary torus and left maxillary exostoses. These results are nearly identical to the 15 section fusion summary score, with the exception of the addition of the wear and time period interaction. The adjusted R 2 model for the fusion ratio summary score wa s 0.4 56 Models were re run, removing non significant terms ( p < 0.05) until the simplest model was found. For 15 secti on summary fusion score, the model remains quite complex with many interactions (Table 5 38). All interaction coefficients are significant, but the main effects of ancestry and left maxillary exostoses on their own are not significant. The adjusted R 2 fo r the final model is 0.445. Using the fusion ratio summary score, the model remains complex with many interactions (Table 5 39), and the adjusted R 2 value for this model is 0.422. These results indicate that just less than 50% of the variation in age is accounted for by variation in fusion, sex, ancestry, AMTL Index, and expression of maxillary tori and exostoses. Predicting age using the models in Tables 5 3 8 and 5 39 is complex due to the number of variables shown to have a significant effect. In orde r to facilitate prediction, the model was further simplified to include eliminating bilateral traits and interaction terms. At this stage, only the left side was employed since no difference was previously
203 found between sides in bilateral traits. The 15 section summary score was employed instead of the fusion ratio summary score due to its higher individual correlation with age and the ease in scoring the sutures visually rather than metrically. Non significant results are included in this model as they are known to affect other variables in the model. Results of the simpler, no interaction model are in Table 5 40; the adjusted R 2 value for this model is 0.412. Comparing the R 2 values for the more complex model with interactions and the simpler model wi thout interactions, it can be seen that they are not drastically different. A point estimate for age can thus be derived by entering information for summary fusion score, sex, ancestry, AMTL Index, and left maxillary torus expression and multiplying by th e coefficients given in Table 5 40. The standard error of the estimate is 14.6; doubling this number and adding to/subtracting from the age point estimate gives an approximately 95% prediction interval for age (Nawrocki 1998). Because palatal size cou ld be affecting certain palatal variables, such as complexity or fusion, the geometric means of the chords were taken and compared to fusion and age as with the other variables.
204 Figure 5 1. Box and whisker plots of age per closure score for sections of the IN suture, 15 section/4 phase system. A) Right lateral IN suture, B) left lateral IN suture, C) right medial IN suture, D) left medial IN suture.
205 Figure 5 2. Box and whisker plots of age per closure score for sections of the PMP suture, 15 se ction/4 phase system. A) Anterior PMP suture, B) posterior PMP suture.
206 Figure 5 3. Box and whisker plots of age per closure score for sections of the TP suture, 15 section/4 phase system. A) Right TP suture in GPF, B) left TP suture in GPF, C) rig ht lateral TP suture, D) left lateral TP suture, E) right medial TP suture, F) left medial TP suture
207 Figure 5 4. Box and whisker plots of age per closure score for sections of the AMP suture, 15 section/4 phase system. A) Anterior AMP suture, B) mid AMP suture, C) posterior AMP suture.
208 Figure 5 5. Box and whisker plots of age per closure score for the full suture/4 phase system. A) IN suture, B) PMP suture, C) TP suture, D) AMP suture.
209 Figure 5 6. Comparison of the box and whisker plots of age per closure score for the right and left sides of the TP suture within and outside the GPF, full suture/4 phase system. A) Right TP suture in GPF, B) left TP suture in GPF, C) right TP suture, D) left TP suture.
210 Figure 5 7. Box and whisker pl ots of age per closure score for the full suture/binary system. A) IN suture, B) PMP suture, C) TP suture, D) AMP suture.
211 Figure 5 8. Comparison of the box and whisker plots of age per closure score for the right and left sides of the TP suture with in and outside the GPF, full suture/binary system. A) Right TP suture in GP F, B) left TP suture in GPF, C) right TP suture, D) left TP suture.
212 Figure 5 9. Box and whisker plot of age per summary score for the 15 section/4 phase system. Compare to Fi gure 5 16, which displays the same data as a scatterplot.
213 Figure 5 10. Box and whisker plot of age per summary score for the full suture/4 phase system not including the TP suture within the GPF.
214 Figure 5 11. Box and whisker plot of age per sum mary score for the full suture/4 phase system including the TP suture within the GPF.
215 Figure 5 12. Box and whisker plot of age per summary score for the full suture/binary system not including the TP suture within the GPF.
216 Figure 5 13. Box and whisker plot of age per summary score for the full suture/binary system including the TP suture within the GPF.
217 Figure 5 14. Box and whisker plots of age per closure score for the control sutures, full suture/4 phase system. A) Right zygomaticomaxi llary suture, B) left zygomaticomaxillary suture, C) nasofrontal suture.
218 Figure 5 15. Distributions of summary scores, all categorical systems. A) 15 section/4 phase system, B) full suture/4 phase system, no TP suture in GPF, C) full sutu re/4 phase system, including TP suture in GPF, D) binary/4 phase system, no TP suture in GPF, E) binary/4 phase system, including TP suture in GPF. 15-section/4-phase System Summary Score Frequency 0 10 20 30 40 0 50 100 150 Full Suture/4-phase System No GPF Summary Score Frequency 0 2 4 6 8 10 12 0 50 100 150 Full Suture/4-phase System With GPF Summary Score Frequency 0 5 10 15 0 50 100 150 Full Suture/Binary System No GPF Summary Score Frequency 0 1 2 3 4 0 50 100 150 200 Full Suture/Binary System With GPF Summary Score Frequency 0 1 2 3 4 5 0 50 100 150 200 250 300 350 A B C D E
219 Figure 5 16. Scatterplot for age and summary score in the 15 section/4 phase system. The black line in dicates th e least squares regression line
220 Table 5 1. Results from ANOVA for suture section scores and age in the15 section/4 phase system. Variable Df Sum of Squares Mean Square F value Probability IN.lat.R.1 3 21451 7150 24.085 <0.001 a IN.lat.L.1 3 1433 478 1.609 0.186 IN.med.R.1 3 15499 5166 17.403 <0.001 a IN.med.L.1 3 2981 994 3.347 0.019 a PMP.ant.1 3 11385 3795 12.783 <0.001 a PMP.post.1 3 1048 349 1.177 0.318 TP.GPF.R.1 3 1 825 608 2.049 0.106 TP.GPF.L.1 3 465 155 0.522 0.667 TP.lat.R.1 3 1292 431 1.450 0.227 TP.lat.L.1 3 291 97 0.326 0.806 TP.med.R.1 3 1062 354 1.192 0.312 TP.med.L.1 3 368 123 0.413 0.743 AMP.ant.1 3 2564 855 2.879 0.035 a AMP.mid.1 3 1148 383 1.289 0.277 AMP.post.1 3 253 84 0.284 0.837 Residuals 719 212561 297 a Difference is significant at the p < 0.05 level. Table 5 2. Results from ANOVA for full suture scores and age in the full suture/4 phase system. Variable Df Sum of Squares Mean Square F value Probability IN.C 3 34768 11589 38.607 <0.001 a PMP.C 3 13402 4467 14.881 0.001 a TP.GPF.R.C 3 1686 562 1.872 0.133 TP.GPF.L.C 3 636 212 0.706 0.548 TP.R.C 3 1221 407 1.356 0.255 TP.L.C 3 234 78 0.260 0.854 TP.C 3 322 161 0.537 0.585 AMP.C 3 1817 606 2.018 0.110 Residuals 738 221538 300 a Difference is significant at the p < 0.05 level.
221 Table 5 3. Results from ANOVA for full suture scores and age in the full suture/binary system. Variable Df Sum of Squares Mean Square F value Probability IN.B 1 674 674 2.106 0.147 PMP.B 1 23216 2321 6 72.578 <0.001 a TP.GPF.R.B 1 6890 6890 21.539 <0.001 a TP.GPF.L.B 1 38 38 0.119 0.731 TP.R.B 1 2563 2563 8.012 0.005 a TP.L.B 1 258 258 0.807 0.369 TP.B 1 281 281 0.877 0.349 AMP.B 1 840 84 0 2.626 0.105 Residuals 753 240865 320 a Difference is significant at the p < 0.05 level. Table 5 4. Differences in mean age by closure score for the 15 section/4 phase system. Comparison Suture Section Side 0 1 0 2 0 3 1 2 1 3 2 3 IN Lateral Right 0.916 0.902 0.421 0.999 0.000 a 0.000 a IN Lateral Left 0.989 1.000 0.463 0.934 0.000 a 0.000 a IN Medial Right 1.000 0.030 a 0.000 a 0.001 a 0.000 a 0.000 a IN Medial Left 0.688 0.305 0.000 a 0.000 a 0.000 a 0.000 a PMP Anterior Midline 0.000 a 0. 000 a 0.000 a 0.480 0.035 a 0.447 PMP Posterior Midline 0.102 0.000 a 0.000 a 0.687 0.002 a 0.031 a TP GPF Right 0.349 0.000 a 0.000 a 0.713 0.482 0.981 TP GPF Left 0.096 0.001 a 0.000 a 0.914 0.905 0.999 TP Lateral Right 0.000 a 0.000 a 0.000 a 0.698 0.686 0.983 TP Lateral Left 0.000 a 0.000 a 0.000 a 0.901 0.505 0.792 TP Medial Right 0.001 a 0.001 a 0.016 a 0.917 0.459 0.699 TP Medial Left 0.000 a 0.001 a 0.025 a 0.999 0.686 0.746 AMP Anterior Midline 0.000 a 0.000 a 0.007 a 0.111 0.336 0.970 AMP Mid Midline 0.000 a 0.000 a 0.062 0.204 0.627 0.995 AMP Posterior Midline 0.000 a 0.000 a 0.218 0.240 0.935 0.993 a Difference is significant at the p < 0.05 level.
222 Table 5 5. Differences in mean age by closure score for the full suture/4 phase system. Comparison Suture 0 1 0 2 0 3 1 2 1 3 2 3 IN 0.931 0.661 0.286 0.003 a 0.000 a 0.000 a PMP 0.081 0.000 a 0.000 a 0.001 a 0.000 a 0.185 TP in GPF, right 0.222 0.000 a 0.000 a 0.828 0.585 0.968 TP in GPF, left 0.098 0.001 a 0.000 a 0.915 0.909 0.999 TP, right 0.000 a 0.000 a 0.003 a 1 .000 0.353 0.391 TP, left 0.000 a 0.000 a 0.001 a 1.000 0.306 0.358 TP 0.000 a 0.000 a 0.003 a 0.990 0.301 0.399 AMP 0.000 a 0.000 a 0.375 0.040 a 0.845 1.000 a Difference is significant at the p < 0.05 level. Table 5 6. Differences in mean age by closure scor e for the full suture/binary system. Comparison Suture 0 1 IN 0.173 PMP 0.000 a TP in GPF, right 0.000 a TP in GPF, left 0.000 a TP, right 0.000 a TP, left 0.000 a TP 0.000 a AMP 0.000 a a Difference is significant at the p < 0.05 level.
223 Table 5 7. order correlations for age and individual sutural closure in the 1 5 section/4 phase and 3 phase scoring systems. Suture Section Side rho (4 phase) rho (3 phase) IN Lateral Right 0.285 0.286 IN Lateral Left 0.272 0.271 IN Medial Righ t 0.358 0.344 IN Medial Left 0.368 0.345 PMP Anterior Midline 0.335 0.332 PMP Posterior Midline 0.340 0.337 TP GPF Right 0.318 0.317 TP GPF Left 0.292 0.293 TP Lateral Right 0.294 0.290 TP Lateral Left 0.296 0.294 TP Medial Right 0.220 0.2 18 TP Medial Left 0.228 0.228 AMP Anterior Midline 0.260 0.252 AMP Mid section Midline 0.249 0.246 AMP Posterior Midline 0.244 0.239 Table 5 order correlations for age and individual sutural closure in the full suture/4 phase and 3 phase scoring systems. Suture rho (4 phase) rho (3 phase) IN 0.362 0.345 PMP 0.352 0.325 TP in GPF, right 0.316 0.316 TP in GPF, left 0.291 0.291 TP, right 0.292 0.295 TP, left 0.295 0.298 TP 0.290 0.298 AMP 0.269 0.259 Table 5 9. Spear order correlations for age and individual sutural closure in the full suture/binary scoring system. Suture rho IN 0.052 PMP 0.302 TP in GPF, right 0.321 TP in GPF, left 0.303 TP, right 0.291 TP, left 0.294 TP 0.287 AMP 0.259
224 Table 5 10 order correlations for age and individual sutural closure in the control sutures. Suture rho Nasofrontal 0.141 Zygomaticomaxillary, right 0.227 Zygomaticomaxillary, left 0.225 Table 5 11 s rank order correla tions for age and summary score in all qualitative, categorical systems. System rho 15 section/4 phase 0.438 15 section/3 phase 0.436 Full suture/4 phase 0.423 Full suture/3 phase 0.404 Full suture/4 phase with GPF 0.419 Full suture/3 phase with GPF 0.412 Full su ture/binary 0.352 Full suture/binary with GPF 0.369 Table 5 12 Descriptive statistics for fusion ratios of the measured palatal sutures. Suture Mean St Dev Median Minimum Maximum PMP 0.296 0.304 0.173 0.000 1.132 a TP 0.067 0.124 0.000 0.00 0.799 AM P 0.070 0.129 0.000 0.000 0.771 Summary 0.105 0.127 0.05 3 0.00 0.70 0 a This maximum represents slight error introduced from estimating staurion due to obliteration.
225 Figure 5 1 7 Scatterplots of age and fusion ratio for each of the measured palata l sutures (PMP, TP, and AMP). The black line in each plot indicates the least squares regression line. A) PMP suture, B) TP suture, C) AMP suture. 0.0 0.2 0.4 0.6 0.8 1.0 20 40 60 80 100 A PMP Fusion Ratio Age 0.0 0.2 0.4 0.6 0.8 20 40 60 80 100 B TP Fusion Ratio Age 0.0 0.2 0.4 0.6 0.8 20 40 60 80 100 C AMP Fusion Ratio Age
226 Figure 5 1 8 Scatterplot of age and fusion ratio summary score (AMP + PMP + TP). The black line indi cates the least squares regression line. 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 20 40 60 80 100 Fusion Ratio Summary Score Age
227 Figure 5 19. Distributions of fusion ratio scores. A) PMP suture fusion, B) TP suture fusion, C) AMP suture fusion, D) fusion ratio summary score. PMP Fusion Ratio Score Fusion Score Frequency 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 50 150 250 TP Fusion Ratio Score Fusion Score Frequency 0.0 0.2 0.4 0.6 0.8 0 100 300 500 AMP Fusion Ratio Score Fusion Score Frequency 0.0 0.2 0.4 0.6 0.8 0 100 300 500 Summary Fusion Ratio Score Fusion Score Frequency 0.0 0.2 0.4 0.6 0 100 200 300 A B C D
228 Figure 5 20 Scatterplot of age and arcsine tr ansformed fusion ratio summary score. The black line indicates the least squares regression line. Compare to Figure 5 18 with untransformed data. 0.0 0.2 0.4 0.6 0.8 1.0 20 40 60 80 100 Arcsine Tranformed Fusion Ratio Summary Score Age
229 Table 5 order correlations for age and sutural fusion in full suture categorical an d continuous systems. rho Designator Full suture/4 phase Full suture/binary Suture ratio PMP suture 0.352 0.302 0.315 TP suture 0.290 0.287 0.224 AMP suture 0.269 0.259 0.280 Summed 0.423 0.352 0.349
230 Figure 5 21. Summary score by sex (F = f emale, M = male). Note female outliers at the top of the plot but overall lower female summary scores as compared to males.
231 Figure 5 22. Summary score by ancestry (AF = African, AS = Asian, EU = European).
232 Figure 5 23. Summary score by time pe riod (H = historic, M = modern).
233 Table 5 1 4 Results from ANCOVA for 15 section/4 phase summary score and age, sex, and ancestry. Variable Df Sum of Squares Mean Square F value Probability Sex 1 12793 12793 280.28 <0.001 a Ancestry 2 4646 2323 50 .89 <0.001 a Age 1 11271 11271 246.94 <0.001 a Sex:Ancestry 2 47 23 0.51 0.601 Sex:Age 1 691 691 15.14 <0.001 a Ancestry:Age 2 163 82 1.79 0.168 Sex:Ancestry:Age 2 269 135 2.95 0.053 Residuals 750 34231 46 a Differ ence is significant at the p < 0.05 level.
234 Table 5 15. Inventory scores by tooth. Score Tooth 1 2 3 4 5 6 7 All RM3 40 158 22 208 30 288 16 762 RM2 0 340 10 345 25 15 27 762 RM1 0 324 4 380 20 2 32 762 RP4 0 365 2 302 64 2 27 762 RP3 0 356 1 288 80 2 35 762 RC 3 369 1 213 103 1 72 762 RI2 0 277 1 259 135 5 85 762 RI1 0 261 3 280 128 2 88 762 LI1 0 264 1 284 122 1 90 762 LI2 0 276 1 266 128 6 84 762 LC 3 351 3 228 94 0 83 762 LP3 1 356 2 298 65 0 40 762 LP 4 1 327 2 327 70 4 31 762 LM1 0 327 1 381 9 2 42 762 LM2 1 326 14 357 25 12 27 762 LM3 33 165 26 213 31 275 19 762 Table 5 16. Descriptive statistics for dental wear scores by tooth. Tooth n (observable) a n (unobservable) b Median Mean St Dev Min Max RM3 156 606 5 6.615 4.167 4 28 RM2 321 441 7 7.960 3.727 4 24 RM1 290 472 9 9.586 4.498 4 27 RP4 347 415 2 1.859 1.051 1 7 RP3 344 418 2 2.038 1.208 1 8 RC 360 402 2 2.381 1.200 1 6 RI2 266 496 2 2.030 1.159 1 7 RI1 246 516 3 2.614 1.136 1 7 LI1 250 512 3 2.624 1.173 1 7 LI2 263 499 2 1.989 1.096 1 6 LC 339 423 2 2.425 1.258 1 8 LP3 343 419 2 2.009 1.186 1 8 LP4 314 448 2 1.834 1.041 1 7 LM1 305 457 9 9.875 4.646 4 30 LM2 302 460 7 8.106 4.054 4 34 LM3 163 599 5 7.190 5.835 4 40 a Th e presence of minor crown damage and restorations resulted in sample sizes for wear score per tooth that are less than the sample sizes for teeth present (recorded as a score of 2 in Table 5 15). b Wear not recorded du e absence or damage of tooth.
235 Table 5 17. Descriptive statistics for mean wear scores. Teeth n (observable) n (unobservable) Median Mean St Dev Min Max All 554 208 3.79 4.60 2.69 1.00 18.33 Posterior 534 228 4.78 5.65 3.44 1.00 23.50 Anterior 468 294 2.08 2.42 1.12 1.00 8.00 Figure 5 24. Distributions of mean wear scores. A) Mean wear score, all maxillary teeth, B) mean posterior wear score, C) mean anterior wear score. Mean Wear Score Count 0 5 10 15 20 0 50 100 150 200 250 300 Mean Posterior Wear Score Count 0 5 10 15 20 25 0 50 100 150 200 Mean Anterior Wear Score Count 1 2 3 4 5 6 7 8 0 20 40 60 80 100 120 A B C
236 Table 5 18. Count and percentage of antemortem tooth loss by tooth number. Tooth Count All % AMTL RM3 208 762 27.30 RM2 345 762 45.28 RM1 380 762 49.87 RP4 302 762 39.63 RP3 288 762 37.80 RC 213 762 27.95 RI2 259 762 33.99 RI1 280 762 36.75 LI1 284 762 37.27 LI2 266 762 34.91 LC 228 762 29.92 LP3 298 762 39.11 LP4 327 762 42.9 1 LM1 381 762 50.00 LM2 357 762 46.85 LM3 213 762 27.95
237 Figure 5 25. Frequency distribution of AMTL index in the total sample. Note that the distribution is bimodal. AMTL Index Count 0 20 40 60 80 100 0 50 100 150 200 250 300
238 Figure 5 26. Frequency distributions of suture complexity ra tios in the total sample. A) PMP suture complexity, B) TP suture complexity, C) AMP suture complexity. PMP Complexity Ratio Count 0 1 2 3 4 0 100 200 300 400 500 600 TP Suture Complexity Ratio Count 0 1 2 3 4 0 50 100 150 200 AMP Complexity Ratio Count 0 1 2 3 4 0 100 200 300 400 500 600 A B C
239 Figure 5 27 Scatterpots of fusion summary and mean wear scores. The black line in each plot indicates the least squares regression line. A) 15 section summary and mean wear scores, rho = 0.192; B) fusion ratio summary and mean wear scores, rho = 0.227. 5 10 15 10 20 30 40 A Mean Wear Score 15-Section Summary Fusion Score 5 10 15 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 B Mean Wear Score Fusion Ratio Summary Score
240 Figure 5 28 Scatterpots of fusion summary and mean posterior wear scores. The black line in each plot indicates the least squares regressi on line. A) 15 section summary and mean posterior wear scores, rho = 0.141; B) fusion ratio summary and mean posterior wear scores, rho = 0.201. 5 10 15 20 10 20 30 40 A Mean Posterior Wear Score 15-Section Summary Fusion Score 5 10 15 20 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 B Mean Posterior Wear Score Fusion Ratio Summary Score
241 Figure 5 29 Scatterpots of fusion summary and mean anterior wear scores. The black line in each plot i ndicates the least squares regression line. A) 15 section summary and mean anterior wear scores, rho = 0.389; B) fusion ratio summary and mean anterior wear scores, rho = 0.318. Table 5 19 Regression results for 15 section summary fusion score and mean anterior and posterior wear scores. Variable Estimate Std Error t value Probability Intercept 9.5890 0.9851 9.734 <0.001 a Mean posterior wear 0.1029 0.1295 0.794 0.427 Mean anterior wear 2.6871 0.3835 7.006 <0.001 a a Difference is signif i cant at the p < 0.05 level. Table 5 2 0 Regression results for fusion ratio summary score and mean wear scores. Variable Estimate Std Error t value Probability Intercept 2.719e 05 1.329e 02 0.002 0.998 Mean posteri or wear 5.953e 03 1.747e 03 3.407 0.001 a Mean anterior wear 2.358e 02 5.174e 03 4.557 <0.001 a a Difference is signif i cant at the p < 0.05 level. 1 2 3 4 5 6 7 8 10 20 30 40 A Mean Anterior Wear Score 15-Section Summary Fusion Score 5 10 15 20 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 C Mean Anterior Wear Score Fusion Ratio Summary Score
242 Figure 5 30. Scatterplots of summary fusion scores and AMTL Index. The black lines indicate the least squares regression lines. A) 15 section/4 phase summary fusion score and AMTL index, rho = 0.397; B) fusion ratio summary score and AMTL Index, rho = 0.260. Figure 5 31. Box and whisker plots of suture fusion by edentulous st atus (N=at least one tooth present, Y=completely edentulous). A) Summary fusion score and edentulism, B) fusion ratio summary score and edentulism. 0 20 40 60 80 100 10 20 30 40 A AMTL Index 15-Section Summary Fusion Score 0 20 40 60 80 100 0.0 0.2 0.4 0.6 B AMTL Index Fusion Ratio Summary Score
24 3 Figure 5 32. Scatterplots of individual suture fusion ratios and suture complexity. The black line in each plot indicates the fitted exponential decay curve A) PMP fusion ratio and suture complexity, rho = 0.591; B) TP fusion ratio and suture complexity, rho = 0.391; C) AMP fusion ratio and suture complexity, rho = 0.206. Table 5 21 Regression resu lts for fusion ratio summary score and suture complexity Variable Estimate Std Error t value Probability Intercept 0.421 0.025 16.613 <0.001 PMP complexity 0.059 0.012 4.802 <0.001 TP complexity 0.144 0.017 8.560 <0.001 AMP complexity 0.018 0.017 1.054 0.292
244 Figure 5 3 3 Scatterplot s of age and mean wear scores The black line in each plot indicates the least squares regression line. A) Age and mean wear score, rho =0.240; B) age and mean posterior wear score, rho = 0.209; C) age and mean anterior wear score, rho = 0.501. 20 40 60 80 100 5 10 15 A Age Mean Wear Score 20 40 60 80 100 5 10 15 20 B Age Mean Posterior Wear Score 20 40 60 80 100 1 2 3 4 5 6 7 8 C Age Mean Anterior Wear Score
245 Figure 5 34. Scatterplot of AMTL index and age. The black line indicates the least squares regression line; rho = 0.631. 20 40 60 80 100 0 20 40 60 80 100 Age AMTL Index
246 Figure 5 35. Scatterplots of individual suture complexity and age. The black lines indi cate the least squares regression lines. A) PMP suture complexity and age, rho = 0.075; B) TP suture complexity and age, rho = 0.044; C) AMP suture complexity and age, rho = 0.017. 20 40 60 80 100 1 2 3 4 5 6 A Age PMP Suture Complexity 20 40 60 80 100 1.0 1.5 2.0 2.5 3.0 B Age TP Suture Complexity 20 40 60 80 100 1.0 1.5 2.0 2.5 3.0 3.5 C Age AMP Suture Complexity
247 Figure 5 36. Box and whisker plots of mean wear scores by sex (F = female, M = male) A) Overall mean wear score, B) mean posterior wear score, C) mean anterior wear score. Table 5 2 2 Descriptive statistics for sex and mean wear score Females Males Teeth Mean St Dev Min Max Mean St Dev Min Max All 4.11 2.36 1.00 18.33 5.08 2.89 1.00 18.00 Posterior 5.13 3.12 1.00 23.50 6.14 3.65 1.00 22.33 Anterior 2.19 1.02 1.00 8.00 2.65 1.17 1.00 7.00
248 Figure 5 37. Box and whisker plots of mean wear scores by ancestry (A F = African, AS = Asian, EU = European ) A) Overall mean wear score, B) mean posterior wear score, C) mean anterior wear score. Table 5 2 3 Descriptive statistics for ancestry and mean wear score. African Asian European Teeth Mean St Dev Min Max Mean St Dev Min Max Mean St Dev Min Max All 4.38 2.36 1.00 18.33 5.42 3.34 1.00 18.00 3.84 1.70 1.00 10.50 Posterior 5.35 3.37 1.00 23.50 6.71 3.85 1.00 22.33 4.65 2.46 1.00 16.00 Anterior 2.56 1.25 1.00 8.00 2.22 1.05 1.00 5.50 2.50 1.01 1.00 6.00
249 Figure 5 38. Box and whisker plots of mea n wear scores by time period (H = historic, M = modern) A) Overall mean wear score, B) mean posterior wear score, C) mean anterior wear score. Table 5 2 4 Descriptive statistics for time period and mean wear score. Historic Modern Teeth Mean St De v Min Max Mean St Dev Min Max All 5.13 2.99 1.00 18.33 3.92 2.06 1.00 16.00 Posterior 6.27 3.76 1.00 23.50 4.82 2.73 1.00 17.00 Anterior 2.47 1.20 1.00 8.00 2.36 1.02 1.00 6.00
250 Table 5 2 5 Group differences in mean wear score. p value Teeth Se x a Ancestry b Time Period a All <0.001 c <0.001 c <0.001 c Posterior <0.001 c <0.001 c <0.001 c Anterior <0.001 c 0.004 c 0.490 a Wilcoxon signed rank test b Kruskal Wallis test c Difference is significant at the p < 0.05 level. Table 5 26. Descriptive statistics for AMTL Index by group. Group Median Mean St Dev Minimum Maximum Female 26.67 40.32 39.49 0 100 Male 21.43 38.81 39.61 0 100 African 25.84 37.69 36.26 0 100 Asian 18.75 35.76 39.48 0 100 European 34.52 45.45 42.14 0 100 Historic 18.75 31.71 35.31 0 100 Modern 36.61 47.27 41.92 0 100 Table 5 27. Descriptive statistics for suture complexity by group. Suture Group Median Mean St Dev Minimum Maximum PMP Female 1.210 1.377 0.506 0.784 6.282 Male 1.065 1 .104 0.140 0.795 2.198 African 1.092 1.181 0.284 0.795 3.821 Asian 1.132 1.333 0.548 0.784 6.282 European 1.120 1.194 0.245 0.950 3.088 Historic 1.106 1.217 0.338 0.784 3.821 Modern 1.115 1.258 0.440 0.795 6.282 TP Female 1.543 1.589 0.293 1.128 2.994 Male 1.419 1.479 0.248 1.044 2.814 African 1.393 1.447 0.250 1.044 2.994 Asian 1.562 1.393 0.291 1.073 2.807 European 1.517 1.517 0.266 1.095 2.814 Historic 1.460 1.508 0.266 1.044 2.994 Modern 1.503 1.559 0.285 1.095 2.814 AMP Female 1 .211 1.319 0.356 0.973 3.635 Male 1.127 1.183 0.187 0.983 2.614 African 1.114 1.192 0.246 0.973 3.279 Asian 1.210 1.311 0.348 0.983 3.635 European 1.163 1.244 0.253 1.013 2.794 Historic 1.143 1.222 0.253 0.973 3.279 Modern 1.172 1.278 0.322 0.9 92 3.635
251 Figure 5 39 Scatterplots of AMTL Index and mean molar wear. The black line indicates the least squares regression line. A) AMTL Index and mean wear score, rho = 0.058; B) AMTL Index and mean posterior wear score, rho = 0.015; C) AMTL In dex and mean anterior wear score, rho = 0.401. 5 10 15 0 20 40 60 80 100 A Mean Wear Score AMTL Index 5 10 15 20 0 20 40 60 80 100 B Mean Posterior Wear Score AMTL Index 1 2 3 4 5 6 7 8 0 20 40 60 80 100 C Mean Anterior Wear Score AMTL Index
252 Figure 5 4 0 Scatterplots of suture complexity and mean wear score. The black line indicates the least squares regression line. A) PMP complexity and mean wear score, rho = 0.144; B) TP complexity an d mean wear score, rho = 0.017; C) AMP complexity and mean wear score, rho = 0.038. 5 10 15 1 2 3 4 5 6 A Mean Wear Score PMP Complexity 5 10 15 1.0 1.5 2.0 2.5 3.0 B Mean Wear Score TP Complexity 5 10 15 1.0 1.5 2.0 2.5 3.0 3.5 C Mean Wear Score AMP Complexity
253 Figure 5 4 1 Scatterplots of suture complexity and mean posterior wear score. The black line indicates the least squares regression line. A) PMP complexity and mean posterior wear score, rho = 0.113; B) TP complexity and mean posterior wear score, rho = 0.012; C) AMP complexity and mean posterior wear score, rho = 0.060. 5 10 15 20 1 2 3 4 5 6 A Mean Posterior Wear Score PMP Complexity 5 10 15 20 1.0 1.5 2.0 2.5 3.0 B Mean Posterior Wear Score TP Complexity 5 10 15 20 1.0 1.5 2.0 2.5 3.0 3.5 C Mean Posterior Wear Score AMP Complexity
254 Figure 5 4 2 Scatterplots of suture complexity and mean anterior wear score. The black li ne indicates the least squares regression line. A) PMP complexity and mean anterior wear score, rho = 0.198; B) TP complexity and mean anterior wear score, rho = 0.068; C) AMP complexity and mean anterior wear score, rho = 0.093. 1 2 3 4 5 6 7 8 1 2 3 4 5 6 A Mean Anterior Wear Score PMP Complexity 1 2 3 4 5 6 7 8 1.0 1.5 2.0 2.5 3.0 B Mean Anterior Wear Score TP Complexity 1 2 3 4 5 6 7 8 1.0 1.5 2.0 2.5 3.0 3.5 C Mean Anterior Wear Score AMP Complexity
255 Figure 5 43. Sca tterplots of suture complexity and AMTL Index. The black line indicates the least squares regression line. A) PMP complexity and AMTL Index, rho = 0. 030; B) TP complexity and AMTL Index, rho = 0.044; C) AMP complexity and AMTL Index, rho = 0.064. 0 20 40 60 80 100 1 2 3 4 5 6 A AMTL Index PMP Complexity 0 20 40 60 80 100 1.0 1.5 2.0 2.5 3.0 B AMTL Index TP Complexity 0 20 40 60 80 100 1.0 1.5 2.0 2.5 3.0 3.5 C AMTL Index AMP Complexity
256 Ta ble 5 28. Relationship of fusion and palatal traits, employing the 15 section/4 phase summary score. Trait rho Left accessory lesser palatine foramina 0.029 Right accessory lesser palatine foramina <0.001 Left marginal crest 0.013 Right marginal c rest 0.037 Left lateral groove bridging 0.038 Right lateral groove bridging 0.054 Left medial groove bridging 0.015 Right medial groove bridging 0.042 Palatine torus 0.105 Left maxillary torus 0.056 Right maxillary torus 0.036 Left maxill ary exostoses 0.011 Right maxillary exostoses 0.011 Maxillary bone quality 0.175 Palatine bone quality 0.231 Palatal porosity 0.025 Palate shape 0.029 Transverse palatine suture shape 0.061 Left zygomaticomaxillary suture shape 0.081 Righ t zygomaticomaxillary suture shape 0.084 Table 5 29. Relationship of age and palatal traits. Trait rho Left accessory lesser palatine foramina 0.023 Right accessory lesser palatine foramina 0.054 Left marginal crest 0.059 Right marginal cres t 0.047 Left lateral groove bridging 0.040 Right lateral groove bridging 0.034 Left medial groove bridging 0.092 Right medial groove bridging 0.115 Palatine torus 0.050 Left maxillary torus 0.232 Right maxillary torus 0.211 Left maxi llary exostoses 0.205 Right maxillary exostoses 0.200 Maxillary bone quality 0.228 Palatine bone quality 0.265 Palatal porosity 0.007 Palate shape 0.077 Transverse palatine suture shape 0.006 Left zygomaticomaxillary suture shape 0.029 Right zygomaticomaxillary suture shape 0.021
257 Table 5 square tests of palatal trait frequency for sex, ancestry, and time period. Sex Ancestry Time Period Trait 2 df p value 2 df p value 2 df p value Left accessory lesser palatine foramina 11.322 5 0.045 a 16.891 10 0.077 6.989 5 0.221 Right accessory lesser palatine foramina 23.133 5 <0.001 a 5.739 10 0.837 2.992 5 0.701 Left marginal crest 2.891 1 0.089 14.695 2 <0.001 a 9.680 1 0.002 a Right marginal crest 3.134 1 0.077 11.828 2 0.003 a 5.335 1 0.021 a Left lateral groove bridging 5.056 3 0.168 98.459 6 <0.001 a 40.509 3 <0.001 a Right lateral groove bridging 7.190 3 0.066 82.871 6 <0.001 a 23.969 3 <0.001 a Left medial groove bridging 2.303 3 0.512 31.561 6 <0.001 a 18.449 3 <0.001 a Right medial groove bridging 3.562 3 0.313 37.686 6 <0.001 a 8.373 3 0.039 a Palatine torus 22.406 3 <0.001 a 26.201 6 <0. 001 a 1.387 3 0.709 Left maxillary torus 12.405 3 0.006 a 72.245 6 <0.001 a 14.678 3 0.002 a Right maxillary torus 21.367 3 <0.001 a 32.806 6 <0.001 a 0.073 3 0.995 Left maxillary exostoses 17.434 3 0.001 a 28.212 6 <0.001 a 20.017 3 < 0.001 a Right maxillary exostoses 20.796 3 <0.001 a 31.537 6 <0.001 a 16.177 3 0.001 a Maxillary bone quality 2.940 2 0.230 35.134 4 <0.001 a 17.343 2 <0.001 a Palatine bone quality 2.220 2 0.330 83.098 4 <0.001 a 65.575 2 <0.001 a Palatal po rosity 25.341 1 <0.001 a 73.696 2 <0.001 a 3.938 1 0.047 a Palate shape 3.498 3 0.321 241.152 6 <0.001 a 28.525 3 <0.001 a Transverse palatine suture shape b 1.831 3 0.608 125.401 6 <0.001 b 1.651 3 0.648 Left zygomaticomaxillary suture sha pe b 10.188 2 0.006 49.272 4 <0.001 a 7.220 2 0.027 a Right zygomaticomaxillary suture shape b 11.012 2 0.004 40.871 4 <0.001 a 5.277 2 0.071 a Significant at p < 0.05 b Did not include unobservable/obliterated values
258 Table 5 3 1 rank order correlations of mean wear and palatal variants. rho Trait All Posterior Anterior Left accessory lesser palatine foramina 0.012 0.004 0.023 Right accessory lesser palatine foramina 0.067 0.079 0.008 Left marginal crest 0.014 0.00 3 0.010 Right marginal crest 0.040 0.041 0.020 Left lateral groove bridging 0.011 0.021 0.022 Right lateral groove bridging 0.031 0.059 0.043 Left medial groove bridging 0.005 0.017 0.013 Right medial groove bridging 0.004 0.015 0.011 Palatine torus 0.002 0.009 0.057 Left maxillary torus 0.013 0.067 0.012 Right maxillary torus 0.009 0.009 0.050 Left maxillary exostoses 0.011 0.025 0.036 Right maxillary exostoses 0.067 0.061 0.016 Maxillary bone quality 0.069 0.049 0.092 Palatine bone quality 0.093 0.085 0.058 Palatal porosity 0.147 0.153 0.010 Palate shape 0.105 0.100 0.054 Transverse palatine suture shape a 0.103 0.117 0.008 Left zygomaticomaxillary suture shape a 0.034 0.072 0.011 Right zygomati comaxillary suture shape a 0.038 0.072 0.009 a Did not include unobservable/obliterated values
259 Table 5 order correlations of AMTL Index and palatal variants. Trait rho Left accessory lesser palatine foramina 0.029 Right acc essory lesser palatine foramina 0.052 Left marginal crest 0.061 Right marginal crest 0.051 Left lateral groove bridging 0.064 Right lateral groove bridging 0.061 Left medial groove bridging 0.092 Right medial groove bridging 0.137 Pala tine torus 0.064 Left maxillary torus 0.292 Right maxillary torus 0.324 Left maxillary exostoses 0.341 Right maxillary exostoses 0.331 Maxillary bone quality 0.314 Palatine bone quality 0.315 Palatal porosity 0.031 Palate shape 0.123 Transverse palatine suture shape 0.008 Left zygomaticomaxillary suture shape 0.003 Right zygomaticomaxillary suture shape 0.020
260 Table 5 order correlations of suture complexity and palatal variants. rho Trait PMP Complex ity TP Complexity AMP Complexity Left accessory lesser palatine foramina 0.067 0.020 0.017 Right accessory lesser palatine foramina 0.076 0.009 0.030 Left marginal crest 0.129 0.019 0.051 Right marginal crest 0.124 0.038 0.036 Left lateral gro ove bridging 0.025 0.077 0.081 Right lateral groove bridging 0.009 0.087 0.068 Left medial groove bridging 0.027 0.009 0.012 Right medial groove bridging 0.029 0.004 0.039 Palatine torus 0.201 0.179 0.264 Left maxillary torus 0.086 0.113 0.132 Right maxillary torus 0.110 0.054 0.094 Left maxillary exostoses 0.040 0.055 0.071 Right maxillary exostoses 0.089 0.026 0.085 Maxillary bone quality 0.059 0.001 0.095 Palatine bone quality 0.033 0.100 0.072 Palatal porosity 0.103 0.083 0.054 Palate shape 0.095 0.081 0.089 Transverse palatine suture shape 0.052 0.006 0.087 Left zygomaticomaxillary suture shape 0.083 0.022 0.083 Right zygomaticomaxillary suture shape 0.058 0.028 0.048 Table 5 34. Results from Pearson square tests of bilateral palatal traits and order correlations for right and left sides per trait Trait 2 df p value rho Accessory lesser palatine foramina 1.940 5 0.857 0.340 Marginal crest 0.043 1 0.836 0.880 Lateral groov e bridging 6.632 3 0.085 0.724 Medial groove bridging 0.327 3 0.955 0.585 M axillary torus 4.122 3 0.249 0.692 M axillary exostoses 0.231 3 0.972 0.833 Zygomaticomaxillary suture shape 0.707 3 0.872 0.855
261 Table 5 order corre lations of palatal traits as compared to each other. Trait LPF a Crista a Lat bridge a Med bridge a Pal torus Max torus a Max exo a Max bone Pal bone Porosity Pal shape TP shape ZM shape a LPF Crista 0.065 Lat bridge 0.023 0.033 Med bridge 0.009 0.013 0.191 Pal torus 0.021 0.001 0.092 0.049 Max torus 0.011 0.008 0.171 0.129 0.139 Max exo 0.005 0.006 0.106 0.134 0.087 0.417 Max bone 0.085 0.095 0.024 0.016 0.133 0.061 0. 151 Pal bone 0.058 0.156 0.040 0.025 0.069 0.037 0.124 0.636 Porosity 0.045 0.060 0.022 0.038 0.135 0.040 0.017 0.047 0.109 Pal shape 0.025 0.005 0.055 0.058 0.043 0.089 0.077 0.015 0.033 0.091 TP shape 0.036 0.014 0.023 0.047 0.013 0.047 0.006 0.007 0.006 0.080 0.029 ZM shape 0.046 0.009 0.031 0.010 0.050 0.085 0.020 0.025 0.032 0.027 0.018 0.047 a Left side used in comparison.
262 Table 5 36. Results from multiple regression analyses for 15 section fu sion summary score. Variable Estimate Std Error t value Probability Intercept 39.113 3.207 12.195 <0.001 a Age 0.227 0.018 12.617 <0.001 a Sex 18.669 3.801 4.911 <0.001 a PMP suture complexity 7.178 2.481 2.893 0. 004 a TP suture complexity 13.405 1.389 9.650 <0.001 a Age:Sex 0.062 0.025 2.504 0.013 a Sex:PMP suture complexity 5.757 2.578 2.233 0.026 a Sex:TP suture complexity 5.549 1.834 3.026 0.003 a a Difference is sig nif i cant at the p < 0.05 level. Table 5 37. Results from multiple regression analyses for fusion ratio summary score. Variable Estimate Std Error t value Probability Intercept 0.544 0.052 10.390 <0.001 a Age 0.002 0.000 9.240 <0.001 a Sex 0.517 0.055 9.434 <0.001 a PMP suture complexity 0.234 0.036 6.524 <0.001 a TP suture complexity 0.166 0.025 6.637 <0.001 a Palatine torus 0.077 0.032 2.390 0.017 a Age:Sex 0.001 0.000 3.442 0.001 a Sex:PMP suture complexity 0.206 0.037 5.559 <0.001 a Sex:TP suture complexity 0.165 0.026 6.235 <0.001 a TP suture complexity:Palatine torus 0.051 0.021 2.431 0.015 a a Difference is signif i cant at the p < 0.05 level.
263 Table 5 38. Results from multiple regression analyses for age using the 15 section summar y fusion score. Variable Estimate Std Error t value Probability Intercept 25.841 2.225 11.614 <0.001 a 15 section summary fusion score 0.829 0.091 9.080 <0.001 a Sex 7.068 1.207 5.858 <0.001 a AF (ancestry DV1) 2.690 3.173 0.848 0. 397 AMTL Index 0.179 0.021 8.689 <0.001 a Left maxillary torus 10.533 2.137 4.930 <0.001 a AS (ancestry DV2) 2.022 1.673 1.208 0.227 Right maxillary torus 10.870 2.257 4.817 <0.001 a Left maxillary exostoses 0.954 1.201 0.794 0.427 Summary fusion:DV1 0.341 0.131 2.593 0.010 a DV1:AMTL Index 0.155 0.034 4.621 <0.001 a DV1:Left maxillary torus 12.396 2.523 4.914 <0.001 a DV2:Left maxillary torus 8.605 2.790 3.085 0.002 a DV1:Right maxillary torus 11.407 2.482 4.595 <0.001 a DV2:Right maxillary torus 10.869 2.661 4.084 <0.001 a AMTL Index:R maxillary torus 0.047 0.022 2.080 0.038 a AMTL Index:L maxillary exostoses 0.067 0.030 2.256 0.024 a a Difference is signif i cant at the p < 0.05 level. Table 5 39. Results from multiple regression analyses for age using the fusion ratio summary score. Variable Estimate Std Error t value Probability Intercept 36.579 1.842 19.857 <0.001 a Fusion ratio summary score 36.492 6.458 5.651 <0.001 a Sex 3.356 1.461 2.297 0.022 a AF (ancestry DV1) 4.511 2.540 1.776 0.076 AMTL Index 0.231 0.019 11.888 <0.001 a Left maxillary torus 11.127 2.179 5.106 <0.001 a AS (ancestry DV2) 0.515 1.703 0.302 0. 763 Right maxillary torus 11.922 2.298 5.187 <0.001 a Left maxillary exostoses 1.051 1.225 0.857 0.392 Fusion ratio summary score:Sex 28.363 12.189 2.327 0.020 a Fusion ratio summary score:DV1 19.118 8.807 2.171 0.030 a DV1 :AMTL Index 0.127 0.033 3.852 <0.001 a DV1:Left maxillary torus 12.888 2.575 5.006 <0.001 a DV2:Left maxillary torus 9.285 2.846 3.263 0.001 a DV1:Right maxillary torus 12.056 2.525 4.774 <0.001 a DV2:Right maxillary torus 11.348 2.716 4.179 <0.001 a AMTL Index:R maxillary torus 0.055 0.023 2.380 0.018 a AMTL Index:L maxillary exostoses 0.074 0.031 2.435 0.015 a a Difference is signif i cant at the p < 0.05 level.
264 Table 5 40. Results from multiple regression a nalyses for age using the 15 section summary fusion score and no interactions. Variable Estimate Std Error t value Probability Intercept 27.268 1.909 14.281 <0.001 a 15 section summary fusion score 0.713 0.076 9.434 <0.001 a Sex 7.085 1.232 5.75 2 <0.001 a AF (ancestry DV1) 2.674 1.386 1.930 0.054 AS (ancestry DV2) 0.425 1.304 0.326 0.744 AMTL Index 0.213 0.016 13.601 <0.001 a Left maxillary torus 1.282 0.741 1.729 0.084 a Difference is signif i cant at the p < 0 .05 level.
265 CHAPTER 6 DISCUSSION Research Question 1 The expectation of some degree of association between maxillary suture fusion and age was met. There are positive correlations between known age and fusion scores for all sections of sutures, full sut ures, and summary scores, though the strength of the relationship varies based on what locations are analyzed and how. Overall, the correlation of known age and closure, regardless of system employed, is never more than 0.500. Summary scores show higher correlations with age than single sutures or sections of sutures, but there do not appear to be any appreciable differences in correlation values for age and closure when comparing sections of sutures versus full sutures or 4 phase, 3 phase, and binary sys tems. Because of this, a system akin to Nawrocki (1998) or Wheatley (1996) is preferred due to the inclusion of all palatal sutures, as opposed to examining all sutures but basing the age estimate on the last suture to display any amount of obliteration ( Mann et al. 1991) Galera et al. (1998) also found that the use of a summary score produced lower bias values for age estimation from the Terry Collection. The full suture/binary system, designed for this study to be as close as possible to the descrip tion of palatal suture aging provided by Mann et al. (1991) while also attempting to provide some degree of objectivity to the scoring of the sutures, has generally lower correlations per suture and overall than the other qualitative scoring systems. This attempt at standardization was not effective at devising a system to clarify or improve the revised scoring method procedure, likely due to an oversimplification of the variation seen in palatal suture fusion. Because scores of no
266 fusion are fairly commo n in this sample and have been noted in other studies (e.g., Kokich, 1976; Wang et al., 2006; Beauthier et al., 2010 ) failure to account for subtleties in fusion across sutures may produce lower correlation values for fusion and age. Thus, again the recom mendation is made that a system with multiple phases that encompasses all sutures of the palate is preferred to a fused/non fused categorical system. An additional recommendation by Mann et al. (1991) is to incorporate other traits of the palate; discussi on of these additional traits can be found in sections below. Comparing the quantitative system, which employs measured values from three full length sutures, correlation values are similar to the qualitative full suture/4 phase binary systems for all thre e sutures and similar in summary score to the binary system. Both the quantitative and binary system summary score correlation values are lower than summary scores for 4 phase systems, whether full suture or suture sections are employed. The quantitative system may suffer from the inability to include the IN suture. This suture is the first to fuse, fuses nearly completely in most individuals, and shows the highest correlation values for age in the 15 section and full suture/4 phase systems. Therefore, the inability to quantitatively capture the IN suture likely affects the overall relationship of the quantitative summary score to age. Scoring the sutures as sections does create a marginally higher correlation between known age and suture closure when an overall summary score is then employed. However, condensing categories into a no fusion some fusion complete fusion scoring system does not improve the correlation between known age and suture closure. In fact, the simplest system scoring fusion as pr esent or absent produces the lowest correlation value between age and closure. The reduction of the number of
267 categories for scoring may aid in decreasing interobserver error but it does not markedly improve the relationship between known age at death a nd suture closure. Concerning developmental asymmetry, there are no large scale differences in correlation values of suture fusion and age for left and right sides. Where significance was tested for sides, such as with the TP suture, no significant diffe rences between left and right sides are observed. Either side is therefore appropriate for use in age estimation, but here it is recommended that if a choice must be made then the right side be used for the TP suture due to a slightly higher correlation w ith known age at death. Methods that combine le f t and right sides, such as component methods, have been previously recommended (McCormick and Kenyhercz 2015) and this recommendation is further supported in this research, where the summary score that inc ludes both sides has the highest correlation with age. The correlation values for the control sutures are lower than those for the palatal sutures. Of the three scored, the nasofrontal suture has not only the lowest rho for the control sutures, but also of all sutures examined. Right and left zygomaticomaxillary sutural fusion is nearly identical, indicating symmetry in sides. The difference between values for the nasofrontal and zygomaticomaxillary sutures is also interesting in terms of functional co nsiderations because the nasofrontal suture is likely subjected to less force from mastication than the zygomaticomaxillary sutures or the palatal sutures (Rogers 1984; Wroe et al. 2007) The relationship of biomechanical variables and fusion is discuss ed further below. What is clear is that the palatal sutures alone do not demonstrate correlation values that indicate a large amount of variation in age can be accounted for by variation
268 in suture closure, though they do perform better in terms of age than the other facial sutures tested. In order to be a powerful tool for age estimation, the relationship between indicator (biological age) and chronological age must be fairly strong and age progressive changes should be easily classified, unidirectional, a nd occur at similar times across populations (Milner and Boldsen 2012a) This is not the case when considering palatal sutures in isolation, though as others have found, the fusion of palatal sutures alone does relate to age and may be useful for general categorization of individuals into broad age categories ( Wheatley, 1996; Ginter, 2005; Sakaue and Adachi, 2007; Beauthier et al., 2010; Apostolidou et al. 2011; Siegel and Passalacqua 2012) Based on the lack of a clear linear relationship between fusi on and age and the fact that many of the age ranges per fusion score show overlap, it is not possible to categorize certain combinations of closure into phases. Together the results indicate that other variables are influencing palatal suture closure. Res earch Question 2 While there were several outliers in the female group, overall females have much lower average summary fusion scores than males. This indicates that males undergo fusion earlier than females, consistent with previous studies (Ashley Monta gu 1938; Mann et al. 1991; Nawrocki 1998) Because of this difference in fusion, sex must be considered as a meaningful contributor to variation in palatal suture closure. It is difficult to tie these differences to sexual dimorphism, as overall palat e size was not one of the variables considered in this research. However, more likely than differences in size is an underlying difference between the sexes in aging and the progression of palatal fusion, which may relate to genetics or differing environm ental conditions. If fusion were a solely maturational process, it would be expected that females would
269 undergo fusion earlier than males, as is the case in overall skeletal growth and development. Thus this difference in the sexes points not only to int rinsic and extrinsic differences in males and females concerning suture obliteration, but it also suggests that fusion differs from maturation. Significant differences in summary fusion scores were also observed for ancestral groups. Of the three ancestr al groups examined here, Asian individuals have the lowest mean suture scores, African individuals have the highest, and European individuals fall between these two groups. This differs from the results of Mann et al. (1991) and Meindl and Lovejoy (1985), who found no differences between European and African Americans, but it is consistent with Galera et al. (1998) and Nawrocki (1998), who did find differences. Galera et al. (1998) found a higher correlation with age and fusion for African American indivi duals from the Terry Collection, and Nawrocki (1998) found that palatine sutures played more of a role in equations for individuals of African ancestry than those of European. Neither study tested palatal suture closure in individuals of Asian ancestry no r commented on the amount of fusion as greater in one group over another. Because the Mann et al. (1991) method does not provide for differences among ancestral groups, this likely explains the poor performance for Japanese individuals observed by Sakaue and Adachi (2007), especially considering this group had lower average fusion scores than both African and European groups. The differences in fusion scores among the ancestral groups point to possible differe nces in underlying genetic make up and/or envi ronmental factors. These differences are often why regionally specific age estimation methods are recommended,
270 though large and diverse samples can alleviate the need for these types of modifications (Brooks and Suchey 1990; Konigsberg et al. 2008) F or the palatal sutures both time periods exhibit nearly identical average fusion scores and distributions of fusion scores. The results of this study agree with Zambrano (2005), who found no significant secular trends in vault suture closure between histo ric and modern individuals. However, t hey are contrary to the finding that fusion occurs more slowly in modern individuals (Nawrocki 1998). A decrease in fusion for modern individuals could be related to improved nutrition and decreased dental disease ( i.e., loss of teeth correlated with more fusion). However, there is no systematic trend observed in this sample for decreased fusion in modern over historic individuals, therefore inferences about improved nutrition and decreased dental disease cannot be made. An absence of secular trends in palatal suture fusion is an important finding because it means that the method as presented here can be used for both modern and historic individuals; time period does not need to be taken into account when consideri ng age estimation from the palate. It also suggests that secular trends seen in earlier maturation and epiphyseal fusion in modern individuals are not carried over into the palate (Langley Shirley and Jantz 2010) It is also important to consider that p alatal suture fusion, with the exception of the IN suture, occurs largely in adulthood, at a time period when skeletal maturation has completed in the rest of the skeleton. Thus, as with the results from sex, no difference in time period may further sugge st that fusion is not tied to maturation and instead represents a separate process, related in part to age.
271 When considering ancestry and temporality together, it is seen that differences in fusion are significant for ancestry but not for time period. Be cause differences are seen ancestrally but not temporally, this suggests variation in fusion is due at least in part to genetics, rather than environment. This is because individuals of different ancestral backgrounds but within the same temporally delimi ted skeletal collection are not likely subjected to a large amount of variation in environment (Sparks and Jantz 2002) The absence of temporal differences thus further strengthens the difference as being attributed to intrinsic factors rather than plast icity to environmental factors. Research Question 3 Dental Wear The relationship of mean wear scores and fusion scores is positive, indicating that for an increase in wear score there is also an increase in fusion score. The reverse is also true for low fusion scores individuals exhibit low wear scores. The hypothesis that low wear, indicative of decreased bite force, would be related to increased fusion is not supported by these data. Interestingly, the highest correlation values for fusion and wear a re actually found with mean anterior wear and fusion scores. Based on the contact surfaces of the anterior dentition, it is unexpected that the highest relationship between wear and fusion would be for the anterior teeth since the posterior teeth contribu te more to overall occlusal loading based on their larger contact surfaces. The presence of a relationship between anterior wear and fusion does suggest that anterior wear may be driving posterior palatine suture fusion, if the palate can reasonably be mo deled as a plate, where stress and strain to the anterior portion causes the most force in the posterior aspect. While it could be suggested that a positive relationship between fusion and wear may be explaining fusion as a means to strengthen a palate
272 th at is subjected to higher loading (Herring 1972), the distribution of wear scores is also an important consideration. Wear scores in the entire sample are low, even for molars with large occlusal surfaces, so this variable may not be accurately portrayin g bite force in the maxilla, regardless of the relationship found with fusion. Wear differs by age and demographic group. Age and dental wear show a positive association, which is not surprising considering dental wear has long been employed for age est imation, albeit with varying rates of success (Buikstra and Ubelaker 1994; Mays 2014) The presence of lower female wear as compared to males is not as clear, though this may be explained by sexual dimorphism and an overall smaller masticatory complex i n females as compared to males. With larger muscles, males have greater overall bite force and thus are more likely to have greater wear. Ancestral differences are even less clear, especially since they differ based on which mean wear score is employed. As with sex, higher wear may be attributed to overall palatal size, since Europeans generally have smaller palates and could also have smaller mean wear. However, ancestral differences in wear could also be attributed to diet, and without knowing the foo d consumed by different groups it is not possible to determine if this is the case. In this sample, historic individuals always have higher wear than modern individuals. This result supports the idea that modern individuals are eating foods that are less tough and gritty ( Wescott and Jantz, 2005; Skorpinski 2014) Antemortem Tooth Loss Calculating the AMTL Index was a means of quantifying one of the variables given by Mann et al. (1991) to be considered in conjunction with fusion of the palatal sutures. The recording of complete edentulism versus at least one tooth present was
273 also employed for this purpose. The AMTL Index displays a positive association with fusion, so that an increase in AMTL Index means an increase in fusion score. This result does correspond to the expected outcome, in which individual s with decreased dentition and decreased occlusal loading also disp lay decreased bite force and increased fusion (EngstrÂšm et al. 1986). Wheatley (1996) also found that partial to complete tooth loss was associated with earlier fusion of the sutures. The decrease in fusion could be related to a narrowing of the sutures from a decrease in tension, causing sutural margins to approach and enabling bony bridging. These results are further confirmed when comparing completely edentulous individuals to individuals with at least one tooth, and finding that there is a significant difference in fusion scores between the two groups. AMTL Index differs by age, ancestry, and time period, but not between the sexe s. For age, increased age is associated with increased tooth loss, which can be related to greater loss over time due to use or disease. Mays (2014) also found AMTL, as measured by the height of the posterior mandibular corpus, to be associated with age. Europeans have higher rates of tooth loss than Africans or Asians, though why this difference is present is not clear. Between modern and historic individuals, modern individuals have higher mean AMTL than historic. Given modern dental practices, this result is somewhat surprising since teeth can often be restored rather than fully removed. Dental wear and AMTL Index show a positive relationship, with the strongest correlation between tooth loss and anterior wear. In terms of indicating occlusal loadi ng, these two variables do agree with one another: an increase in wear is related
274 to an increase in tooth loss, and vice versa, though again the highest correlation is with the anterior teeth, which is unexpected. Sutural Complexity Sutures in the palate are not complex, with the most complexity displayed in the TP suture. The complexity values for the AMP and PMP are nearly identical, which is logical because these sutures represent anterior and posterior portions of the same midpalatal suture. Simple s utures are indicative of a largely tensile loading environment ( Rafferty and Herring, 1999; Herring and Ochareon 2005) Since palatal sutures tend to resist large scale fusion, the tensile environment may explain observed fusion patterns. The relations hip of fusion and complexity is negative, indicating that increased sutural complexity is associated with decreased fusion. This means that sutures that are close to straight are more likely to fuse than sutures of a more sinuous nature. As with AMTL, a straight suture may have a proclivity towards bridging if the margins approach one another. Sutural complexity shows little to no relationship with age, indicating that in terms of adulthood, complexity does not increase with age in the palatal sutures (Zo llikofer and Weissmann, 2011). This is also related to the completion of growth and development of the palate and facial skeleton in adulthood, regardless of smaller scale changes in size and shape that may occur during the adult years (Albert et al. 200 7) I f growth is complete, then it may not be possible for sutures to continue to adapt to changing stresses, and the adult form of sutures may in fact be static (Herring 1972; Zollikofer and Weissmann 2011)
275 Males have less complex sutures than females When considered alongside differences in fusion for the sexes, decreased complexity in males corresponds to higher fusion ratios in this group. This suggests that less complex sutures are more likely to fuse than more complex sutures. Asian individual s have higher mean sutural complexities for the PMP and AMP sutures, while for the TP suture Europeans have the highest mean sutural complexity. These results are interesting in light of traditionally employed methods of ancestry determination sutural c omplexity indicating Asian and a jagged/M shaped TP suture indicating European ( Rhine, 1990; Gill 1998; Hefner 2009) For the AMP suture, complexity is slightly higher in modern individuals as compared to historic. There is no relationship between time period and complexity for the PMP and TP sutures. These relationships or lack thereof do not seem to relate to fusion as there are no significant differences in fusion between modern and historic individuals. Biomechanics and Fusion One of the most str iking observations to be made from the three variables examined as biomechanical proxies is that sutural complexity of the PMP, TP, and AMP sutures appears to show very little association with wear or AMTL Index, but it clearly influences fusion. When con sidered alongside sex and ancestry the pattern becomes even clearer. In groups with simpler sutures, such as males and Africans, average fusion summary scores are higher than groups with more complex sutures, such as females and Asians. This indicates th at a more complex suture is less likely to fuse than a simpler suture, though the exact mechanism is not clearly elucidated by this study. Sutural complexity does not exhibit a clear secular trend, which also is logical given the similarity in summary fus ion scores between historic and modern individuals.
276 While complexity may not speak to dietary or health differences in individuals, as it is not related to wear or AMTL Index, it does appear to offer some predictive value in terms of fusion. Hotzman (200 4) also found that mid palatal suture complexity was poorly correlated with food toughness, and she suggested that complexity might be related more to age; this relationship is not supported by the data in this study. The weak relationships between wear a nd complexity and AMTL Index and complexity suggest that sutures are already adapted morphologically to their loading environments lifetime This again supports a lack of plasticity in sutural morphology in adulthood, as all individuals in this study were over the age of 20 years. Control suture fusion could not be compared to complexity since complexity was not measured for these sutures, but it was compared to bo th wear and AMTL Index. Wear and fusion show very low relationships for all mean wear scores and all control sutures except for anterior wear and zygomaticomaxillary sutures. The weak relationship for fusion of the nasofrontal and wear is consistent with lower loading in this region of the facial skeleton, with higher loading in the zygomatic region of the maxilla (Rogers 1984; Wroe et al. 2007) However, fusion for the nasofrontal and zygomaticomaxillary sutures is positively associated with AMTL Inde x, and the correlation values are nearly identical. While these values indicate a moderate positive relationship between control suture fusion and tooth loss (i.e., higher fusion related to more tooth loss), the similarity of nasofrontal and zygomaticomax illary suture values indicates that there is not in fact a clear difference in these regions if AMTL does indeed indicate differences in occlusal loading.
277 Research Question 4 The large number of additional palatal traits scored complicates analyses of re lationships. What is immediately clear when examining correlation values across all traits, age, groups, and biomechanical variables is that very few traits show strong associations with fusion, age, biomechanical variables, or each other. Concerning fre quencies by sex, ancestry, and time period, about half of the traits show significant differences in frequency among sex or time period, while the majority of traits have significantly different frequencies for ancestry. Very few palatal traits show any amount of association with fusion, with the exception of bone quality and, to a small extent, the palatine torus. The relationship of fusion to bone quality more fusion seen with decreased bone quality does support the subjective assessments included in the Mann et al. (1991) method. It is also consistent with fusion occurring alongside bone degradation as a means of strengthening the palate (Herring 1972). However, the relationship of fusion and bone quality, while moderate when considered in isola tion, is not significant when considered with multiple other traits. The slight relationship of fusion and palatal tori is likely related to the fact that these protrusions are located directly on a suture. Therefore a slight positive association indicat es that when mounding is present, more fusion is also observed, perhaps due to the proximity of the sutural margins in these instances. The same relationship does not hold for maxillary tori or exostoses, suggesting that not only are these traits not rela ted to fusion, but also that either fusion or these growths are not correlated with biomechanical processes in the palate. It was thought that overall suture shape might affect sutural fusion since shape alters the general morphology of the suture. Comp arisons of fusion and ordinally
278 scored suture shape for the TP and zygomaticomaxillary sutures indicate that the relationship of shape and fusion, whether fusion was scored ordinally or measured, is very weak (close to 0). While complexity is related to f usion for the TP suture, the ordinal score of the shape of the TP and fusion is not. This suggests that while these shapes may be helpful in classifying observed variation for ancestry or other group differences, they are not informative in summarizing sh ape as it relates to actual complexity and do not influence the presence o r absence of fusion along these given sutures. As with fusion, most traits show very little relationship to age, with the exception of the maxillary tori and exostoses and bone qua lity. The negative relationship of tori and exostoses with age is also related to the positive correlation of tooth loss and age. As teeth are lost, alveolar bone starts to resorb due to the lack of occlusal loading. Based on the relationships of age, A MTL, and tori/exostoses, resorption does seem to affect torus and exostosis expression on the alveolar bone of the maxilla in that there is reduced expression of internal and external bony growth. Additionally, there is a small negative relationship with age and medial palatine groove bridging. Like maxillary tori and exostoses, this suggests a decrease in expression with increased age; the same negative relationship is also seen between medial bridging and the AMTL Index. Together, the results of age, f usion, AMTL Index, and maxillary torus/exostosis expression indicate that decreased masticatory loading from tooth loss results in observable osseous changes in the palate. This is in agreement with Mann et al. (1987) and Bass (2005), who describe smoothe r and less rugose palates with advanced age.
279 Bone quality also shows the same relationship with age as it does with fusion with increased age poorer bone quality is more common providing further support for a relationship of fusion, age, and overall bo ne quality as mention ed in Mann et al. (1991). Porosity, as measured here (present or absent), shows no association with age. This could indicate that older individuals are not displaying increased porosity as might be expected with bone density loss, or it is indicative that this variable poorly captures bone quality. There is a need to better quantify bone density as maxillary bone in particular shows high degrees of porosity and pitting even for healthy, dense bone in young individuals. Females show greater degrees of expression of palatal tori, while for maxillary tori and exostoses males have greater degrees of expression. For the palatine tori this is in agreement with previous research ( Miller and Roth, 1940; Woo, 1950; Eversole 2011 ) The res ults for maxillary tori and exostoses are less clear since fusion is negatively correlated with these traits. Since males have higher rates of fusion, it would be expected that they would have reduced expression in terms of maxillary bony growth, while in fact the opposite is observed in this sample. However, the AMTL Index does not differ significantly between males and females, so it is possible that the difference in maxillary exostoses and tori is actually related to differences among the sexes versus a solely biomechanical explanation. Additional differences by sex include higher counts of lesser palatine foramina for males, consistent with what is reported in Hauser and De Stefano (1989); increased porosity for males; and a larger occurrence of tra its that could not be scored due to obliteration for males. A zygomaticomaxillary shape of only one angle is more common
280 for males than females but this could be biased by the higher number of males scored as obliterated. It is not clear what biological significance these differences signify, and all other traits do not differ in frequency between the sexes, which is likely why they have not been used in sex determination (Hauser and De Stefano 1989) Since nonmetric traits are often used to aid in ances try assessment, it is not surprising that nearly all palatal traits show significant differences in frequencies by ancestry. This indicates that palatal traits are useful for assessing population affiliation, though, as Hefner (2009) pointed out, trait ex pression varies across all groups; it is just the frequencies that are significantly different. Caution should still be exercised when examining a trait in isolation. Notably, palatal shape and TP suture shape follow expected ancestral trait expressions ( Rhine, 1990; Gill 1998; Hefner 2009; Maier 2013 ) The lesser palatine foramina are the only palatal traits scored that do not show significant differences in frequency by ancestry. Significant differences in frequencies of palatal traits do exist betw een modern and historic individuals, to include these traits: marginal crest, lateral and medial bridging, left maxillary tori and exostoses (but not right), bone quality, porosity, palate shape, and left zygomaticomaxillary shape. The difference in sign ificance for bilateral traits left is significant while right is not is not clear. For the lateral and medial bridging, historic individuals show greater expression of these traits. When considered alongside AMTL and bone quality, which are higher fo r modern individuals, there again appears to be a relationship among bony palatal growth, bone quality, and tooth loss, though modern individuals more frequently display the presence of a marginal crest. The same trend is not observed for tori and exostos es. With palate shape differences,
281 the make up of the historic group should be considered since the majority of African individuals are historic and they were less frequently encountered in modern collections. Thus the difference in palate shape can agai n be interpreted as a difference in ancestry, versus a secular trend. The majority of palatal traits and the three biomechanical variables show low to no association. Wear has a negative association with palate and TP suture shapes and a positive associ ation with porosity in both the overall and posterior mean wear scoring systems. This indicates that the posterior dentition, with larger occlusal surfaces, may be slightly affecting porosity of the palate. This relationship seems tenuous, however, when it is known that porosity is most often visible on the anterior portion of the maxilla. The anterior mean wear score and palatal porosity have no association, indicating that in the overall mean wear score it is the posterior dentition that is causing an association to be seen in the overall mean wear score. Biologically, it is not clear what effect wear alone could have on palate and TP suture shapes but with significant differences in overall and posterior mean wear scores by ancestry, it is more likely that the differences seen here are again related to sample composition in terms of ancestry. The AMTL Index was already discussed above in terms of its relationship to bite force and palatal variants (i.e., more tooth loss associated with poor bone qualit y and decreased alveolar bone growth). When considered alongside wear in terms of palatal variants, it is seen that AMTL may be a better indicator of bite force as wear has little to no relationship to bone quality or maxillary bony growth. In fact, the rho values for AMTL Index when compared to maxillary tori and exostoses and maxillary and palatine
282 bone quality are the highest for any palatal traits among wear, AMTL Index, and sutural complexity. Palatal traits and complexity generally have weak assoc iations with the exception of the palatine torus. This trait is negatively associated to complexity, and with more complexity, there is a tendency to have no or weak expression of the torus. As compared to fusion, this is the opposite effect. With fusio n, there is a positive association with palatine torus expression. Thus the presence of a palatal torus indicates that fusion may be slightly more common, while complexity will be less common. This agrees with more fusion seen in less complex sutures for this sample. Palatal traits show bilateral expression that is not significantly different between left and right sides. This indicates that these traits are not developmentally asymmetrical and either side may be scored with similar results. For ease of comparison the left side was used when comparing all palatal traits so as not to introduce correlations between left and right sides of the same traits. Not surprisingly, similar traits show higher relationships than dissimilar traits (e.g., palatine and maxillary bone quality, maxillary tori and exostoses). This result is not unexpected due to the location of these traits on the palate and the fact that the palate, regardless of being composed of multiple bony elements, is an integrated unit. Additiona l positive associations of traits indicative of osseous growth indicate that tori, exostoses, and bridging are found more commonly in the same individual. Negative associations are largely present among variables of bony growth and bone quality, indicatin g that when bone quality is poor, expression of tori, exostoses, and bridging is reduced. Along with age and fusion, these results further suggest that increased age and fusion are
283 associated with decreased bony growth in the palate. Without a longitudin al study it is not possible to tell if these traits were initially present and resorbed d ue to age or reduced bite force or if these traits were never present. Their association with age, fusion, and each other would appear to suggest the latter, though t heir relationship to ancestry should not be overlooked. The Final Step T he results and above discussion demonstrate that it is not enough to simply look at palatal fusion in terms of age. There are a wide array of factors that contribute to fusion of th e sutures and prediction of age using palatal morphology, which necessitates a multi faceted approach. While nearly all morphological variables were examined in this study, there are more than likely factors that cannot be observed macroscopically, includ ing genetic factors, that contribute to palatal variation in terms of age, group affiliation, biomechanics, and palatal variants. The addition of a variable to estimate size geometric mean of the measured chords per individual was not shown to significantly impact fusio n or age prediction. Fusion, whether scored ordinally or measured by suture, is most influenced by age, sex, sutural complexity, palatine torus exp ression, and interactions of these variables. While other factors individually contribute to palatal fusion, they are not significant effects when considered alongside all factors. With significant terms and interactions, variation in fusion as measured by qualitative and quantitative summary scores is described by approximately 50% of the variation in explanatory variables. These two models leave another 50% of fusion to be explained, indicating that while they have significant effects on palatal suture fusion, part of the picture is still missing.
284 For age, there are far more variables that contribute to variation in age prediction though the coefficient of determination is very close to that of fusion approximately 50% of the variation in age predict ion is accounted for by variation in the explanatory variables. Comparing the equations produced for age to those produced by Nawrocki (1998) that also included information on sex and ancestry, it is seen that standard errors range from 7.0 to 11.0 years, as compared to 14.6 years in this study when using the 15 section summary fusion score. The overall equation using the sum of all sutures provided by Nawrocki (1998) and found to be the best for predicting age regardless of sex or ancestry by Zambrano (2 005) has a standard error of 12.9 years. The higher standard error for the solely palatal suture equations indicates that combining the palatal sutures with other cranial sutures is warranted.
285 CHAPTER 7 CONCLUSIONS This research has shown that of the va rious ways of scoring suture fusion, the most effective is the inclusion of multiple sections in a four phase ordinal scoring system. Measurement of the sutures is less effective at capturing variation in suture fusion. Differences in fusion exist for th e sexes and three ancestral groups examined; no secular differences were noted. Of the biomechanical traits analyzed in relation to fusion, sutural complexity shows the best relationship to fusion. More complex sutures are less likely to fuse than simple sutures. Wear was not shown to be a good indicator o f bite force in terms of fusion; the AMTL Index is preferred. The relationships of the biomechanical proxy variables indicate that complexity is not strongly associated with either wear or AMTL Index, b ut that the latter are associated with one another. Palatal variants showed little relationship to fusion. In terms of age, the most significant effects on age based on this sample included fusion, sex, ancestry, AMTL Index, maxillary tori/exostoses, and interactions of these variables. The resultant equation to generate an age estimate from these variables does so with approximately 15 years of error on either side of the point estimate, which results in a very large and in accurate age estimate. For a m ore accurate age estimate, the interval would need to include two standard errors ( + 30 years), which is clearly lacking in precision. However, the presence of fusion as one of the significant effects on age and age as one of the significant on fusion doe s support the relationship of these two variables, though it is weaker than hoped for use as an age predictor.
286 While the palatal sutures do not appear to perform with enough accuracy or precision on their own, they may contribute to multiple indicator me thods. With limited age estimation methods for the adult skull, it would be beneficial to combine palatal sutures with other cranial sutures, and dental observations such as tooth loss and dental wear. This more comprehensive view of age estimation from the skull is helpful when considering that the cranium and mandible are highly recognizable and thus encountered frequently by the biological anthropologist. A consideration of method usage is also important, since differences do exist between age estimat ion in the forensic realm versus age estimation for a sample of multiple individuals that can be seriated prior to producing age estimates. While this research examined variability in age based on traits of the palate and attempted to account for variation in terms of sex, ancestral/geographic origin, temporality, mechanical loading patterns, and developmental asymmetry, it did not investigate health, nutritional, socioeconomic status or individual rates of senescence. To provide an even more comprehensive view of aging and fusion in the palate, it would be helpful to fully investigate all variables. There are certain limitations, namely the availability of these data in reference collections, but more invasive analyses, such as isotopes, might further elu in which pathological conditions and antemortem trauma for the entire skeleton were noted could serve to contribute to the attribution of fusion to age or other effects. Further research sho uld also consider examining internal features of the palatal sutures to capture more fine scale detail. Fusion generally occurs first internally, and then progresses outwards, so there is some level of detail being missed with a uniquely
287 macroscopic, exte rnal view. This could be accomplished with micro computed tomography scanning. An additional consideration would be to test bone density in different regions of the palate to examine how bone density quantitatively relates to age, fusion, wear, complexit y, and tooth loss. Finally, no age estimation technique is complete without an examination of interobserver error. While the palatal sutures and the palate as a whole are not optimal for age prediction, it is still important to see how different people sc ore the various features of the palate. If the method produces both large age intervals and suffers from poor replicability, it cannot be recommended for age estimation.
288 APPENDIX A DATA COLLECTION WORKSHEET Figure A 1. Data collection worksheet p age 1.
289 Figure A 2. Data collection worksheet page 2.
290 APPENDIX B PALATAL TRAIT FREQUENCIES BY GROUP Table B 1. Left accessory lesser palatine foramina frequencies by sex. 0 1 2 3 4 5 n % n % n % n % n % n % Female 7 46.7 191 55.4 127 43 .1 45 47.9 4 40.0 1 100.0 Male 8 53.3 154 44.6 168 56.9 51 53.1 6 60.0 0 0.0 Total 15 345 295 96 10 1 Table B 2. Right accessory lesser palatine foramina frequencies by sex 0 1 2 3 4 5 n % n % n % n % n % n % Female 15 78.9 189 53.2 1 38 48.1 28 32.6 4 28.6 1 100.0 Male 4 21.1 166 46.8 149 51.9 58 67.4 10 71.4 0 0.0 Total 19 355 287 86 14 1 Table B 3 Left marginal crest frequencies by sex. 0 1 n % n % Female 178 52.8 197 46.4 Male 159 47.2 228 53.6 Total 337 425 Table B 4 Right marginal crest frequencies by sex. 0 1 n % n % Female 176 53.0 199 46.3 Male 156 47.0 231 53.7 Total 332 430
291 Table B 5. Left lateral groove bridging frequencies by sex. 0 1 2 3 n % n % n % n % Female 12 8 52.7 187 46.1 50 56.2 10 41.7 Male 115 47.3 219 53.9 39 43.8 14 58.3 Total 243 406 89 24 Table B 6. Right lateral groove bridging frequencies by sex. 0 1 2 3 n % n % n % n % Female 129 50.8 199 46.5 39 63.9 8 42.1 Male 125 49.2 2 29 53.5 22 36.1 11 57.9 Total 254 428 61 19 Table B 7. Left medial groove bridging frequencies by sex. 0 1 2 3 n % n % n % n % Female 34 58.6 168 49.0 159 47.9 14 48.3 Male 24 41.4 175 51.0 173 52.1 15 51.7 Total 58 343 332 29 Table B 8. Right medial groove bridging frequencies by sex. 0 1 2 3 n % n % n % n % Female 37 59.7 166 47.6 160 49.5 12 Male 25 40.3 183 52.4 163 50.5 16 Total 62 349 323 28
292 Table B 9. Palatine torus frequencies by sex. 0 1 2 3 n % n % n % n % Female 121 53.5 193 43.2 55 67.1 6 85.7 Male 105 46.5 254 56.8 27 32.9 1 14.3 Total 226 447 82 7 Table B 10. Left maxillary torus frequencies by sex. 0 1 2 3 n % n % n % n % Female 221 55.0 111 42.9 31 3 9.7 12 52.2 Male 181 45.0 148 57.1 47 60.3 11 47.8 Total 402 259 78 23 Table B 11. Right maxillary torus frequencies by sex. 0 1 2 3 n % n % n % n % Female 211 56.3 120 43.8 42 43.8 2 11.8 Male 164 43.7 154 56.2 54 56.3 15 88.2 T otal 375 274 96 17 Table B 12. Left maxillary exostoses frequencies by sex. 0 1 2 3 n % n % n % n % Female 275 52.3 88 48.1 11 24.4 1 12.5 Male 251 47.7 95 51.9 34 75.6 7 87.5 Total 526 183 45 8 Table B 13. Right maxillary ex ostoses frequencies by sex. 0 1 2 3 n % n % n % n % Female 279 53.9 82 43.6 13 27.7 1 11.1 Male 239 46.1 106 56.4 34 72.3 8 88.9 Total 518 188 47 9
293 Table B 14. Maxillary bone quality by sex. 0 1 2 n % n % n % Female 247 48 .2 74 47.4 54 57.4 Male 265 51.8 82 52.6 40 42.6 Total 512 156 94 Table B 15. Palatine bone quality by sex. 0 1 2 n % n % n % Female 115 46.0 97 48.5 163 52.2 Male 135 54.0 103 51.5 149 47.8 Total 250 200 312 Table B 16. Pala tal porosity by sex. 0 1 n % n % Female 164 61.9 211 42.5 Male 101 38.1 286 57.5 Total 265 497
294 Table B 17. Palate shape frequencies by sex. 0 1 2 3 n % n % n % n % Female 113 51.8 68 47.2 189 49.3 5 29.4 Male 105 48.2 76 52 .8 194 50.7 12 70.6 Total 218 144 383 17 Table B 18. Transverse palatine suture shape frequencies by sex. 0 1 2 3 Unobs/ OBL a n % n % n % n % n % Female 98 50.3 104 51.0 136 53.8 26 60.5 11 16.4 Male 97 49.7 100 49.0 117 46.2 17 39.5 5 6 83.6 Total 195 204 253 43 67 a Suture unobservable due to obliteration. Table B 19. Left zygomaticomaxillary shape frequencies by sex. 0 1 2 Unobs/ OBL a n % n % n % n % Female 266 54.6 36 37.1 68 49.3 5 12.5 Male 221 45.4 61 62.9 70 50.7 35 87.5 Total 487 97 138 40 a Suture unobservable due to obliteration. Table B 20. Right zygomaticomaxillary shape frequencies by sex. 0 1 2 Unobs/ OBL a n % n % n % n % Female 262 53.9 31 34.8 77 52.0 5 12.8 Male 224 46.1 58 65.2 71 48.0 34 87.2 Total 486 89 148 39 a Suture unobservable due to obliteration.
295 Table B 21. Left accessory lesser palatine foramina frequencies by ancestry. 0 1 2 3 4 5 n % n % n % n % n % n % African 5 33.3 121 35.1 102 34.6 20 20.8 2 20.0 0 0.0 Asian 4 26.7 116 33.6 108 36.6 31 32.3 5 50.0 0 0.0 European 6 40.0 108 31.3 85 28.8 45 46.9 3 30.0 1 100.0 Total 15 345 295 96 10 1 Table B 22. Right accessory lesser palatine foramina frequencies by ancestry. 0 1 2 3 4 5 n % n % n % n % n % n % African 9 47.4 117 33.0 95 33.1 24 27.9 4 28.6 1 100.0 Asian 5 26.3 118 33.2 103 35.9 33 38.4 5 35.7 0 0.0 European 5 26.3 120 33.8 89 31.0 29 33.7 5 35.7 0 0.0 Total 19 355 287 86 14 1 Table B 23 Left marginal crest f requencies by ancestry. 0 1 n % n % African 128 38.0 122 28.7 Asian 123 36.5 141 33.2 European 86 25.5 162 38.1 Total 337 425 Table B 24 Right marginal crest frequencies by ancestry. 0 1 n % n % African 120 36.1 130 30.2 Asia n 126 38.0 138 32.1 European 86 25.9 162 37.7 Total 332 430
296 Table B 25. Left lateral groove bridging frequencies by ancestry. 0 1 2 3 n % n % n % n % African 66 27.2 121 29.8 44 49.4 19 79.2 Asian 52 21.4 171 42.1 36 40.4 5 20.8 European 125 51.4 114 28.1 9 10.1 0 0.0 Total 243 406 89 24 Table B 26. Right lateral groove bridging frequencies by ancestry. 0 1 2 3 n % n % n % n % African 70 27.6 129 30.1 36 59.0 15 78.9 Asian 61 24.0 184 43.0 17 27.9 2 10.5 E uropean 123 48.4 115 26.9 8 13.1 2 10.5 Total 254 428 61 19 Table B 27. Left medial groove bridging frequencies by ancestry. 0 1 2 3 n % n % n % n % African 19 32.8 114 33.2 100 30.1 17 58.6 Asian 21 36.2 93 27.1 144 43.4 6 20.7 Eu ropean 18 31.0 136 39.7 88 26.5 6 20.7 Total 58 343 332 29 Table B 28. Right medial groove bridging frequencies by ancestry. 0 1 2 3 n % n % n % n % African 20 32.3 120 34.4 92 28.5 18 64.3 Asian 24 38.7 92 26.4 142 44.0 6 21.4 Eur opean 18 29.0 137 39.2 89 27.5 4 14.3 Total 62 349 323 28
297 Table B 29. Palatine torus frequencies by ancestry. 0 1 2 3 n % n % n % n % African 67 29.6 135 30.2 43 52.4 5 71.4 Asian 76 33.6 172 38.5 16 19.5 0 0.0 European 83 36.7 1 40 31.3 23 28.0 2 28.6 Total 226 447 82 7 Table B 30. Left maxillary torus frequencies by ancestry. 0 1 2 3 n % n % n % n % African 87 21.6 100 38.6 45 57.7 18 78.3 Asian 166 41.3 77 29.7 20 25.6 1 4.3 European 149 37.1 82 31.7 13 16.7 4 17.4 Total 402 259 78 23 Table B 31. Right maxillary torus frequencies by ancestry. 0 1 2 3 n % n % n % n % African 105 28.0 87 31.8 44 45.8 14 82.3 Asian 131 34.9 103 37.6 29 30.2 1 5.9 European 139 37.1 84 30.6 23 24.0 2 1 1.8 Total 375 274 96 17 Table B 32. Left maxillary exostoses frequencies by ancestry. 0 1 2 3 n % n % n % n % African 158 30.0 64 35.0 24 53.3 4 50.0 Asian 171 32.5 80 43.7 12 26.7 1 12.5 European 197 37.5 39 21.3 9 20.0 3 37.5 To tal 526 183 45 8 Table B 33. Right maxillary exostoses frequencies by ancestry. 0 1 2 3 n % n % n % n % African 153 29.5 75 39.9 18 38.3 4 44.4 Asian 165 31.9 79 42.0 19 40.4 1 11.1 European 200 38.6 34 18.1 10 21.3 4 44.4 Total 518 188 47 9
298 Table B 34. Maxillary bone quality by ancestry. 0 1 2 n % n % n % African 136 26.6 66 42.3 48 51.1 Asian 186 36.3 57 36.5 21 22.3 European 190 37.1 33 21.2 25 26.6 Total 512 156 94 Table B 35. Palatine bone quality by ancestry. 0 1 2 n % n % n % African 47 18.8 69 34.5 134 42.9 Asian 77 30.8 98 49.0 89 28.5 European 126 50.4 33 16.5 89 28.5 Total 250 200 312 Table B 36. Palatal porosity by ancestry. 0 1 n % n % African 129 48.7 121 24.3 Asian 42 15.8 222 44.7 European 94 35.5 154 31.0 Total 265 497
299 Table B 37. Palate shape frequencies by ancestry. 0 1 2 3 n % n % n % n % African 34 15.6 93 64.6 119 31.1 4 23.5 Asian 154 70.6 14 9.7 86 22.5 10 58.8 European 30 13.8 37 25.7 178 46.5 3 17.6 Total 218 144 383 17 Table B 38. Transverse palatine suture shape frequencies by ancestry. 0 1 2 3 Unobs/ OBL a n % n % n % n % n % African 27 13.8 76 37.3 105 41.5 8 18.6 34 50.7 Asian 121 62.1 58 28.4 41 16.2 28 65.1 16 23.9 European 47 24.1 70 34.3 107 42.3 7 16.3 17 25.4 Total 195 204 253 43 67 a Suture unobservable due to obliteration. Table B 39. Left zygomaticomaxillary shape frequencies by ancestry. 0 1 2 Unobs/ OBL a n % n % n % n % Afric an 145 29.8 19 19.6 64 46.4 22 55.0 Asian 186 38.2 20 20.6 46 33.3 12 30.0 European 156 32.0 58 59.8 28 20.3 6 15.0 Total 487 97 138 40 a Suture unobservable due to obliteration. Table B 40. Right zygomaticomaxillary shape frequencies by ances try. 0 1 2 Unobs/ OBL a n % n % n % n % African 139 28.6 19 21.3 72 48.6 20 51.3 Asian 185 38.1 22 24.7 45 30.4 12 30.8 European 162 33.3 48 53.9 31 20.9 7 17.9 Total 486 89 148 39 a Suture unobservable due to obliteration.
300 Table B 41. Left accessory lesser palatine foramina frequencies by time period. 0 1 2 3 4 5 n % n % n % n % n % n % Historic 5 33.3 167 48.4 160 54.2 42 43.8 4 40.0 0 0.0 Modern 10 66.7 178 51.6 135 45.8 54 56.2 6 60.0 1 100.0 Total 15 345 295 96 10 1 Table B 42. Right accessory lesser palatine foramina frequencies by time period. 0 1 2 3 4 5 n % n % n % n % n % n % Historic 10 52.6 178 50.1 138 48.1 46 53.5 5 35.7 1 100.0 Modern 9 47.4 177 49.9 149 51.9 40 46.5 9 64.3 0 0.0 Total 19 355 287 86 14 1 Table B 43 Left marginal crest frequencies by time period 0 1 n % n % Historic 189 56.1 189 44.5 Modern 148 43.9 236 55.5 Total 337 425 Table B 44 Right marginal crest frequencies by time period 0 1 n % n % Historic 181 54.5 197 45.8 Modern 151 45.5 233 54.2 Total 332 430
301 Table B 45. Left lateral groove bridging frequencies by time period. 0 1 2 3 n % n % n % n % Historic 89 36.6 207 51.0 64 71.9 18 75.0 Modern 154 63.4 199 4 9.0 25 28.1 6 25.0 Total 243 406 89 24 Table B 46. Right lateral groove bridging frequencies by time period. 0 1 2 3 n % n % n % n % Historic 106 41.7 214 50.0 46 75.4 12 63.2 Modern 148 58.3 214 50.0 15 24.6 7 36.8 Total 254 428 61 19 Table B 47. Left medial groove bridging frequencies by time period. 0 1 2 3 n % n % n % n % Historic 28 48.3 146 42.6 182 54.8 22 75.9 Modern 30 51.7 197 57.4 150 45.2 7 24.1 Total 58 343 332 29 Table B 48. Right medial gr oove bridging frequencies by time period. 0 1 2 3 n % n % n % n % Historic 31 50.0 159 45.6 168 52.0 20 71.4 Modern 31 50.0 190 54.4 155 48.0 8 28.6 Total 62 349 323 28
302 Table B 49. Palatine torus frequencies by time period. 0 1 2 3 n % n % n % n % Historic 105 46.5 227 50.8 42 51.2 4 57.1 Modern 121 53.5 220 49.2 40 48.8 3 42.9 Total 226 447 82 7 Table B 50. Left maxillary torus frequencies by time period. 0 1 2 3 n % n % n % n % Historic 183 45.5 128 49.4 50 64.1 17 73.9 Modern 219 54.5 131 50.6 28 35.9 6 26.1 Total 402 259 78 23 Table B 51. Right maxillary torus frequencies by time period. 0 1 2 3 n % n % n % n % Historic 187 49.9 135 49.3 48 50.0 8 47.1 Modern 188 50.1 139 50. 7 48 50.0 9 52.9 Total 375 274 96 17 Table B 52. Left maxillary exostoses frequencies by time period. 0 1 2 3 n % n % n % n % Historic 242 46.0 102 55.7 33 73.3 1 12.5 Modern 284 54.0 81 44.3 12 26.7 7 87.5 Total 526 183 45 8 Table B 53. Right maxillary exostoses frequencies by time period. 0 1 2 3 n % n % n % n % Historic 241 46.5 104 55.3 32 68.1 1 11.1 Modern 277 53.5 84 44.7 15 31.9 8 88.9 Total 518 188 47 9
303 Table B 54. Maxillary bone quality by ti me period. 0 1 2 n % n % n % Historic 227 44.3 94 60.3 57 60.6 Modern 285 55.7 62 39.7 37 39.4 Total 512 156 94 Table B 55. Palatine bone quality by time period. 0 1 2 n % n % n % Historic 76 30.4 136 68.0 166 53.2 Modern 174 69.6 64 32.0 146 46.8 Total 250 200 312 Table B 56. Palatal porosity by time period. 0 1 n % n % Historic 145 54.7 233 46.9 Modern 120 45.3 264 53.1 Total 265 497
304 Table B 57. Palate shape frequencies by time period. 0 1 2 3 n % n % n % n % Historic 110 50.5 98 68.1 161 42.0 9 52.9 Modern 108 49.5 46 31.9 222 58.0 8 47.1 Total 218 144 383 17 Table B 58. Transverse palatine suture shape frequencies by time period. 0 1 2 3 Unobs/ OBL a n % n % n % n % n % Historic 91 46.7 97 47.5 128 50.6 24 55.8 38 56.7 Modern 104 53.3 107 52.5 125 49.4 19 44.2 29 43.3 Total 195 204 253 43 67 a Suture unobservable due to obliteration. Table B 59. Left zygomaticomaxillary shape frequencies by time period. 0 1 2 Unobs/ OBL a n % n % n % n % Historic 243 49.9 36 37.1 75 54.3 24 60.0 Modern 244 50.1 61 62.9 63 45.7 16 40.0 Total 487 97 138 40 a Suture unobservable due to obliteration. Table B 60. Right zygomaticomaxillary shape frequencies b y time period. 0 1 2 Unobs/ OBL a n % n % n % n % Historic 238 49.0 35 39.3 81 54.7 24 61.5 Modern 248 51.0 54 60.7 67 45.4 15 38.5 Total 486 89 148 39 a Suture unobservable due to obliteration.
305 LIST OF REFERENCES 1993. Daubert v. Merr ell Down Pharmaceuticals, Inc. 509 US 579 ed. fe span and mortality. Budapest Adler J. 2012. R in a nutshell. Beijing: O'Reilly. Albert AM, Ricanek K, Patterson E. 2007. A review of the literature on the aging adult skull and face: implications for forensic science research and appli cations. Forensic Sci Int 172 :1 9. Alesbury HS, Ubelaker DH, Bernstein R. 2013. Utility of the frontonasal suture for estimating age at death in human skeleta l remains. J Forensic Sci 58 :104 108. Angel J. 1984. Variation in estimating age at death of skel etons. Collegium Anthropol 8:163 168. Angel JL. 1976. Colonial modern skeletal change in the U.S.A Am J Phys Anthropol 45 :723 735. Apostolidou C, Eleminiadis I, Koletsa T, Natsis K, Dalampiras S, Psaroulis D, Apostolidi s S, Psifidis A, Tsikaras P, Njau SN. 2011. Application of the maxillary suture obliteration method for estimating age at death in Greek population. Open Forensic Sci J 4:15 19. Ashley Montagu MF 1938. Aging of the s kull. Am J Phys Anthropol 23 :355 375. Avery JK, Devine RK. 1959. The development of ossification centers in the face and palate of normal and cleft palate human embryos. C left Palate Bull 9:25 26. Ayk royd RG, Lucy D, Pollard AM, Rober ts CA. 1999. Nasty, brutis h, but not necessarily short: a reconsideration of the statistical methods used to calculate age at death from adult human skeletal and dental age indicators. Am Antiquity 64 :55 70. Ayk royd RG, Lucy D, Pollard AM, Solheim T. 1997. Technical note: regression analysis in adult age esti mation. Am J Phys Anthropol 104 :259 265. Bacci no E, Ubelaker DH, Hayek LA, Zerilli A. 1999. Evaluation of seven methods of estimating age at death from mature human skeletal remains. J Forensic Sci 44:9 31 936. Barnes E. 1994. Developmental d efects of of the a xial s keleton in p aleopathology. Niwot: University Press of Colorado. Barnes E. 2012. Atlas of d evelopmental f ield a nomalies of the h uman skeleton: a p aleopa thology p erspective. Hoboken : Wiley Blackw ell.
306 Bass WM. 2005. Human osteology : a laboratory and field manual. Columbia: Missouri Archaeological Society. Beauthier J P, Lefevre P, Meunier M, Orb an R, Polet C, Werquin J P, Quatrehomme G. 2010. Palatine s utures as a ge Indica tor: a c ontrolled s tudy in the e lderly. J Forensic Sci 55 :153 158. Bedford M, Russell K, Lov ejoy C, Meindl R, Simpson S, Stuart Macadam P. 1993. Test of the multifactorial aging method using skeletons with known ages at death from the Grant Collection. Am J Phys Anthropol 91 :287 297. Behrents RG. 1985. Growth in the aging craniofacial skeleton. Ann Arbor : Center for Human Growth and Development, University of Michigan. Beresheim AC. 2015. A q ualitative a ssessment of b ilateral p ubic symphyseal f ace a symmetry u sin g the Suchey Brooks method for a dult a ge at d eath e stimation. Presented at the 67th Annual Scientific Meeting of the American Academy of Forensic Sciences Orlando, FL February 2015. Berg GE. 2008. Pubic bone age estimation in adult women. J Forensic Sci 53:569 577. Berry A C, Berry RJ. 1967. Epigenetic variation in the human cranium. Journal Anat 101:361 379. Bertram JE, Swartz SM. 1991. The 'law of bone transformation': a case of crying Wolf f? Biol Rev Camb Philos Soc 66 :245 273. Bethard JD. 2005. A t est of the t ransition a nalysis m ethod for e stimation of a ge at d eath in a dult h uman s keletal r emains: University of Tennessee, Knoxville. Boas, F. 1910. Changes in bodily form of descendants of immigrants. United States Immigration Commission, Senate Document 208, 61 st Congress Washington, DC: Government Printing Office. Bocquet Ap pel J, Masset C. 1982. Farewell to paleodemography. J Hum Evol 11:321 333. Bogin B. 1999. Patterns of human growth. Cambridge : Cambridge University Press. Boldsen J L, Milner GR, Konigsberg LW, Wood J W. 2002. Transition analysis: a new method for estimating age f rom skeletons. In: Hoppa RD, Vaupel JW, editors. Paleodemography: age d istributions from s keletal s amples. Cambridge : Cambridge University Press. p 73 106. Brach RM, Dunn PF. 2004. Uncertainty a nalysis for f orensic s cience. Tucson: Lawyers & Judges Pub. Co.
307 Briggs CA, Martakis M. 1998. Craniofac ial anatomy. In: Clement JG, Ra nson DL, editors. Craniofacial i dentification in f orensic m edicine. New York: Oxford University Press. p 37 47. Broca P. 1 875. Instructions craniologiques et craniomÂŽtriques de la SociÂŽtÂŽ Brodie AG. 1941. Behavior of normal and abnormal facial growth patterns. A m J Orthod Oral Sur 27:633 647. Brooks S. 1955. Skeletal age at death: t he reliability of cranial and pubic age indicators. Am J Phys Anthropol 13 :567 597. Brooks S, and Suchey J. 1990. Skeletal age determi nation based upon the os pubis: a comparison of the Nemesk ri and Suchey Br ooks methods. Human Evol 5 :227 238. Brothwell DR. 1981. Digging up bones : the excavation, treatment, and study of human skeletal remains. Ithaca: Cornell University Press Brown CA. 2009. Uncertainty in skeletal aging: a r et rospective s tudy and t est of m ethods at the Joint POW/MIA Accounting Command Central Identification Laboratory. Chico: California State University, Chico. Brown CA. 2010. Understanding uncertainty in age estimation: e rror associated with the Mann et al. m axillary suture method. Presented at the 62nd Annnual Scientific Meeting of the American Academy of Forensic Sciences, Seattle, WA February 2010. Buckberry JL, Chamberlain AT. 2002. Age estimation from the auricular surface of the ilium: a revised m ethod. Am J Phys Anthropol 119 :231 239. Buikstra J, Ubelaker D, editors. 1994. Standards for d ata c ollection from h uman s keletal r emains. Fayetteville: Arkansas Archeological Survey Research Series. Burdi AR, Faist K. 1967. Morphogenesis of the palate in normal human embryos with special emphasis on the mechanisms involved Am J Anat 120:149 159. Carter DR. 1984. Mechanical loading histories and cortical bone remodeling. Calcif Tissue Int 36 Suppl 1:S19 24. Chohayeb AA, Volpe AR. 2001. Occurrence of torus palatin us and mandibularis among women of different ethnic groups. Am J Dent 14:278 280. Christensen AM, Passalacqua NV, and Bartelink EJ. 2014. Forensic anthropology : current methods and practice. San Diego : Elsevier Academic Press. Cohen J. 1992. A power prim e r. Psychol Bull 112 :155 159.
308 Crawley MJ. 2013. The R book. Chichester : Wiley. Crews DE. 1993. Biological anthropology and human aging: s ome current directions in aging research. Annu Rev Anthropol 22:395 423. Daegling DJ, Hylander WL. 1997. Occlusal force s and mandibular bone strain: is the primate jaw "overdesigned"? J Hum Evol 33:705 717. Devlin H, Horner K, Ledgerton D. 1998. A comparison of maxillary and mandibular bone mineral densities. J Prosthet Dent 79 :323 327. Diewert VM. 1983. A morphometric ana lysis of craniofacial growth and changes in spatial relations during secondary palatal development in human embryos and fet uses. Am J Anat 167 :495 522. Dorandeu A, Coulibaly B, Piercecchi Marti MD, Barto li C, Gaudart J, Baccino E, Leonetti G. 2008. Age at death estimation based on the study of frontosphenoidal sutures. Forensic Sci Int 177 :47 51. Doual JM, Ferri J, Laude M. 1997. The influence of senescence on craniofacial and cervical morphology in humans. Surg Radiol Anat 19:175 183. EngstrÂšm C, Kiliaridi s S, Thilander B. 1986. The relationship between masticatory function and craniofacial morphology. II. A histological study in the growing rat fe d a soft diet. Eur J Orthod 8 :271 279. Enlow DH. 1966. A morphogenetic analysis o f facial growth. Am J Orthod 5 2 :283 299. Enlow DH, Hans MG. 1996. Essentials of facial growth. Philadelphia: Saunders. Eversole LR. 2011. Clinical outline of oral pathology: diagnosis and t reatment. Shelton : People's Medical Publishing House. Falys CG, Lewis ME. 2011. Proposing a way f orward: a review of standardisation in the use of age categories and ageing techniques in osteological analysis (2004 2009). Int J Osteoarchaeol 21 :704 716. Feik SA, Glover JE. 1998. Growth of child ren's faces. In: Clement JG, Ra nson DL, editors. Craniofac ial identification in f orens ic edicine. New York: Oxford University Press. p 203 224. Ferembach D, Schwidetzky I, Stloukal M. 1980. Recommendations for age and sex diagnoses of skeletons. J Hum Evol 9:517 549. Ferguson MW. 1987. Palate development: mechani sms and malformations. Ir J Med Sci 156:309 315.
309 Francillon Vieillot H, de Buffrenil V, Castanet J, Geraudie J, Meun ier F, Sire J, Zylberberg L, de Ricqles A. 1990. Microstructure and mineralization of vertebrate skeletal tissues. In: Carter JG, edito r. Sk eletal biomineralization: p atterns, p rocesses and evolutionary t rends. New York: Van Nostrand Reinhold. p 471 530. Galera V, Ubelaker DH, Hayek LA. 1998. Comparison of macroscopic cranial methods of age estimation applied to skeletons from the Terry Collec tion. J Forensic Sci 43 :933 939. Garman R, Gaud ette G, Donahue LR, Rubin C, Judex S. 2007. Low level accelerations applied in the absence of weight bearing can enhance trabecular bone formation. J Orthop Res 25 :732 740. Garvin HM, Passalacqua NV. 2012. Cur rent practices by forensic anthropologists in adult skeletal age estimation. J Forensic Sci 57:427 433. Garvin HM, Passalacqua NV, Uhl NM, Gipson DR, Overbury RS, Cabo LL. 2012. Develop ments in forensic anthropology: a ge at death estimation. In: Dirkmaat D editor. A c ompanion to f orensic a nthropology. Malden : Wiley Blackwell Gilbert B, McKern T. 1973. A method fo r aging the female os pubis. Am J Phys Anthropol 38:31 38. Gill G. 1998. Craniofacial criteria in the skeletal attribution of race. In: Reichs K, editor. Forensic osteology: advances in the identification of h uman remains. Second ed. Springfield : Charles C Thomas. p 293 315. Ginter JK. 2005. A test of the effectiveness of the revised maxillary suture obliteration method in estimating adult age at d eath. J Forensic Sci 50:1303 1309. Grau N, Daw JL, Patel R, Evans C, Lewis N, Mao JJ. 2006. Nanostructural and nanomechanical properties of synostosed postnatal human crani al sutures. J Craniofac Surg 17 :91 98 Gravlee CC, Bernard HR, Leonard WR. 2003 a He redity, environment, and cranial form: 138. Anthropol 105:326 332. Gruspier KL, Mullen GJ. 1991. M axillary suture obliteration: a test of the Mann method. J Forensic Sci 36 :512 519. Hassett B. 2006. Torus mandibularis: e tiology and b ioarchaeological utility. Dent Anthropol 19 :1 14.
310 Hauser G, De Stef ano GF. 1989. Epigenetic variants of the human skull. Stuttgart: Schweizerbart. Hefner J. 2009. Cranial nonmetric variation and estimating ancestry. J Forensic Sci 54:985 995. Heller JB, Gabbay JS, Wasson K, Mitchell S, Heller MM, Zuk P, Bradley JP. 2007. Cranial suture response to stress: expression patterns of Noggin and Runx2. Plast Reconstr Surg 119 :2037 2045. Henderson JH, Chang LY, Song HM, Longaker MT, Carter DR. 2005. Age dependent properties and quasi static strain in the rat sagittal suture. J Bio mech 38:2294 2301. Henderson JH, Longaker MT, Carter DR. 2004. Sutural bone deposition rate and strain magnitude during cranial development. Bone 34:271 280. Herring S. 1993. Epigenetic and functional influences on skull growth. In: Hanken J, Hal B, editor s. The skull: v olume 1. Chicago: University of Chicago Press. p 153 206. Herring SW. 1972. Sutures -a tool in functional cranial a nalysis. Acta Anat (Basel) 83 :222 247. Herring SW. 2007. Masticatory muscles and the skull: a comparative p erspective. Arch Or al Biol 52 :296 299. Herring SW. 2008. Mechanical influences on suture development and patency. Front Oral Biol 12:41 56. Herring SW, Ochareon P. 2005. Bone -special problems of the craniofacial r egion. Orthod Craniofac Res 8 :174 182. Herr ing SW, Rafferty K L, Liu ZJ, Marshall CD. 2001. Jaw muscles and the skull in mammals: the biomechanics of mastication. Comp Biochem Phy siol A Mol Integr Physiol 131 :207 219. Herring SW, Teng S. 2000. Strain in the braincase and its sutures during function. Am J Phys Anthrop ol 112:575 593. Hershkovitz I, Latimer B, Dutour O, Jellema LM, Wish Baratz S, Rothschild C Rothschild BM. 1997. Why do we fail in aging the skull from the sagittal suture? Am J Phys Anthropol 103:393 399. Hillson S. 1996. Dental anthropology. Cambridge: Cambridge University Press. Hinrichsen GJ, Storey E. 1968. The effect of force on bone and bones. Angle Orthod 38:155 165.
311 Hinton RJ. 1988. Response of the intermaxillary suture cartilage to alterations in masticatory function. Anat Rec 220:376 387. Hooton EA. 1946. Anthropometrics and orthodontics. Am J Orthod 32:673 681. Hoppa RD. 2000. Population variation in osteological aging criteria: an example from the pubic symphysis. Am J Phys Anthropol 111:185 191. H oppa RD. 2002. Paleodemography: l ooking back an d thinking ahead. In: Hoppa RD, Vaupel JW, editors. Paleodemography: a ge d istributions from s keletal s amples. Cambridge: Cambridge University Press. p 9 28. Hoppa RD, and Vaupel JW. 2002a. Paleodemography: age distribution s from skeletal samples. Cambridge : Cambridge University Press. Hoppa RD, Vaupel JW. 2002b. The Rostock Manifesto for paleodemography: t he way from stage to age. In: Hoppa RD, Vaupe l JW, editors. Paleodemography: a ge d istributions from s keletal s amples. Cambridge: Cambridge University Pres s. p 1 8. Hotzman JL. 2004. Dietary c onsistency and sutural m orphology: t he c omplexity of the m id m alatal s uture in Procolobus badius and Colobus polykomo Gainesville : University of Florida. Hotzman JL. 2010. Mechanical c haracterization of the h ard p alat e Gainesville : University of Florida. Houck MM, Ubelaker D, Owsley D, Craig E, Grant W, Fram R, Woltanski T, Sandness K. 1996. The role of forensic anthropology in the recovery and analysis of Branch Davidian compound victims: assessing the accuracy of age e stimations. J Forensic Sci 41 :796 801. Humphrey T. 1969. The relation between human fetal mouth opening reflexes and closure of the palate. Am J Anat 125:317 344. Hunt DR, Albanese J. 2005. History and demographic composition of the Robert J. Terry an atomical collection. Am J Phys Anthropol 127:406 417. Hylander WL, Johnson KR. 1997. In vivo bone strain patterns in the zygomatic arch of macaques and the significance of these patterns for functional interpretations of craniofacial form. Am J Phys Anthro pol 102:203 232. M, Loth S, Wright R. 1984. Age estimation from the rib by phase analysis: White males. J Forensic Sci 29:1094 1104. M, Loth S, Wright R. 1985. Age estimation from the rib by phas e analysis: White females. J Forensic Sci 30:853 863.
312 Jantz LM Jantz RL. 1999. Secular change in long bone length and proportion in the United States, 1800 1970. Am J Phys Anthropol 110:57 67. Jantz RL. 2001. Cranial change in Americans: 1850 1975. J Forensic Sci 46:784 787. J antz RL, Meadows Jantz L. 2000. Secular change in craniofacial morphology. Am J Hum Biol 12:327 338. Jasinoski SC, Reddy BD. 2012. Mechanics of cranial sutures during simulated cyclic loading. J Biomech 45:2050 2054. Jasinoski SC, Reddy BD, Louw KK, Chinsamy A. 2010. Mechanics of cranial sutures using the finite element method. J Biomech 43:3104 3111. Jaslow CR. 1990. Mechanical properties of cranial sutures. J Biomech 23:313 321. Jefferys WH, Berger JO. 1992. Ockham's razor and Bayesian analysis. Am Sci 80:64 72. Katz D, Suchey J. 1986. Age dete rmination of the male os pubis. Am J Phys Anthropo l 69:427 435. Kemkes Grottenthaler A 2002. Aging through the ages: h istorical perspectives on age indicator methods. In: Hoppa RD, Vaupe l JW, editors. Paleodemography: a ge distributions from skeletal s ampl es. Cambridge: Cambridge University Press. p 48 72. Kenyon CJ. 2010. The genetics of ageing. Nature 464:504 512. Kiliaridis S, EngstrÂšm C, Thilander B. 1985. The relationship between masticatory function and craniofacial morphology. I. A cephalometric long itudinal analysis in the growing rat fed a soft diet. Eur J Orthod 7:273 283. Kimmerle EH, Prince DA, Berg GE. 2008. Inter observer variation in methodologies involving the pubic symphysis, sternal ribs, and teeth. J Forensic Sci 53:594 600. Kirkwood TB. 1 977. Evolution of ageing. Nature 270:301 304. Kirkwood TB, Austad SN. 2000. Why do we age? Nature 408:233 238. Kirschner RE, LaRossa D. 2000. Cleft lip and palate. Otolaryngol Clin North Am 33:1191 1215 Klepinger L, Giles E. 1998. Classification or confus ion: s tatistical interpretation in forensic anthropology. In: Reichs K, editor. Fo rensic o steology: advances in the identification of human r emains. Springfield: Charles C. Thomas. p 427 440. Koc D, Dogan A, Bek B. 2010. Bite force and influential factors on bite force measurements: a literature review. Eur J Dent 4:223 232.
313 Kokich VG. 1976. Age changes in the human frontozygomatic suture from 20 to 95 years. Am J Orthod 69:411 430. Komar DA, Buikstra JE. 2008. Forensic a nthropology: c ontemporary t heory and p ractice. Oxford: Oxford University Press. Komar DA, Grivas C. 2008. Manufactured populations: what do contemporary reference skeletal collections represent? A comparative study using the Maxwell Museum documented collection. Am J Phys Anthropol 137:224 2 33. Konigsberg LW, Frankenberg SR. 1992. Estimation of age structure in anthropological demography. Am J Phys Anthropol 89:235 256. Konigsberg LW, Frankenberg SR. 1994. Paleodemography: n ot quite dead". Evol Anthropol 3:92 105. Konigsberg LW, Herrmann NP, Wescott DJ, Kimmerle EH. 2008. Estimation and evidence in forensic anthropology: age at death. J Forensic Sci 53:541 557. Koolstra JH. 2002. Dynamics of the human masticatory system. Crit Rev Oral Biol Med 13:366 376. Kopher RA, Mao JJ. 2003. Suture growt h modulated by the oscillatory component of micromechanical strain. J Bone Miner Res 18:521 528. Krogman WM, C.C. Thomas. Kroman AM, Thompson GA. 2009. Cranial suture closure as a ref le ction of somatic dysfunction: l essons from osteopathic medicine applied to physical anthropology. Presented at the 61st Annual Scientific Meeting of the American Academy of Forensic Sciences Denver, CO February 2009. Lamendin H, Baccino E, Humbert JF, Ta vernier JC, Nossintchouk RM, Zerilli A. 1992. A simple technique for age estimation in adult corpses: the two criteria dental method. J Forensic Sci 37:1373 1379. Langley Shirley N, Jantz RL. 2010. A Bayesian approach to age estimation in modern Americans from the clavicle. J Forensic Sci 55:571 583. Lanyon LE. 1984. Functional strain as a determinant for bone remodeling. Calcif Tissue Int 36 Suppl 1:S56 61. Lan yon LE, Goodship AE, Pye CJ, MacFie JH. 1982. Mechanically adaptive bone remodelling. J Biomech 1 5:141 154. Lanyon LE, Rubin CT. 1984. Static vs dynamic loads as an influence on bone remodelling. J Biomech 17:897 905.
314 Lanyon LE, Rubin CT. 1985. Functional adaptation in skeletal s tructures. In: Hildebrand M, Bramble D, editors. Functional v ertebrate m o rphology. Cambridge: Harvard University Press. p 1 25. Latham RA. 1971. The development, structure and growth pattern of the human mid palatal suture. J Anat 108:31 41. Lenton KA, Na camuli RP, Wan DC, Helms JA, Longaker MT. 2005. Cranial suture biology. Cu rr Top Dev Biol 66:287 328. Linge L. 1970. A technique for the study of morphology in facial sutures under mechanical influence. Rep Congr Eur Orthod Soc:553 567. Listi GA, Manhein MH. 2012. The use of vertebral osteoarthritis and osteophytosis in age esti mation. J Forensic Sci 57:1537 1540. Little JW, Jakobsen J. 1973. Origin of the globulomaxillary cyst. J Oral Surg 31:188 195. Lovej oy C, Meindl R, Prysbeck T, Mensforth R. 1985a. Chronological metamorphosis of the auricular surface of the ili um: a new met hod for the determination of adult skeletal age at death. Am J Phys Anthropol 68 :15 28. Lovejoy CO, Meindl RS, Mensforth RP, Barton TJ. 1985b. Multifactorial determination of skeletal age at death: a method and blind tests of its accuracy. Am J Phys Anthro pol 68:1 14. Lucas PW, Prinz JF, Agrawal KR, Bruce IC. 2002. Food physics a nd oral physiology. Food Qual and Prefer 13:203 213. Maier C. 2013. Palate s hape and depth: a s hape m atching and m achine l earning m ethod for a ssessment of a ncestry from s keletal r em ains Baton Rouge : Louisiana State University. Maier CA, Zhang K, Manhein MH, Li X. 2015. Palate s hape and d epth: a s hape matching and machine learning m ethod for e stimating a ncestry from h uman s keletal r emains. J Forensic Sci 60:1129 1134. Mann RW. 1987. Maxillary s uture obliteration: a m ethod for e stimating s keletal a ge. Knoxville: University of Tennessee, Knoxville. Mann RW, Jantz RL, Bass WM, W illey PS. 1991. Maxillary suture obliteration: a visual method for estimating skeletal age. J Forensic Sci 36: 781 791. Mann RW, Symes SA, Bass WM. 1987. Maxillary suture obliteration: aging the human skeleton based on intact or fragmentary maxilla. J Forensic Sci 32:148 157.
315 Martrille L, Ubelaker DH, Cattan eo C, Seguret F, Tremblay M, Baccino E. 2007. Comparison o f four skeletal methods for the estimation of age at death on white and black adults. J Forensic Sci 52:302 307. Masset C. 1989. Age estimation on the basis of cranial sutures. In: M, editor. Age m arkers in the h uman s keleton. Springfield: Charles C Thomas. p 71 103. Mays S. 2014. Resorption of mandibular alveolar bone following loss of molar teeth and its relationship to age at death in a human skeletal population. Am J Phys Anthropol 153:643 652. McCormick KA, Kenyhercz MW. 2015. Bilateral asymmet ry in a ge r elated t rait e xpression of the p ubic s ymphysis and a uricular surface of the i lium. Presented at the 67th Annual Meeting of the Americ an Academy of Forensic Sciences, Orlando, FL February 2015 McDonald JH. 2014. Handbook of biological statistic s. Baltimore : Sparky House Publishing. McKern T, Stewart T. 1957. Skeletal a ge c hanges in y oung American males. Natick : Quartermaster Research and Development Command Technical Report EP 45. Meade JB, Cowin SC, Klawitter JJ, Van Buskirk WC, Skinner HB. 198 4. Bone remodeling due to continuously applied loads. Calcif Tissue Int 36 Suppl 1:S25 30. Medawar PB. 1952. An unsolved problem of biology. London: H. K. Lewis. Meindl RS, Lovejoy CO. 198 5. Ectocranial suture closure: a revised method for the determinatio n of skeletal age at death based on the lateral anterior sutures. Am J Phys Anthropol 68 :57 66. Melsen B. 1975. Palatal growth studied on human autopsy material. A histologic microradiographic study. Am J Orthod 68:42 54. Miller SC, Roth H. 1940. Torus pal atinus: a statistical study. J A m Dent Assoc 27:1950 1957. Milner GR, Boldsen JL. 2012a. Skeletal age estimation: w here we are and where we should go. In: Dirkmaat D, editor. A c ompanion to f orensic a nthropology. Malden : Wiley Blackwell. p 224 238. Milner GR, Boldsen JL. 2012b. Transition analysis: a validation study with known age modern American skeletons. Am J Phys Anthropol 148:98 110. Mitteldorf J. 2004. Ageing selected for its own sake. Evol Ecol Res 6:1 17.
316 Mooney MP, Siegel MI, Kimes KR, Todhunter J 1989. A test of two midfacial growth models using path analysis of normal human fetal material. Cleft Palate J 26 :93 99 More ira RS, Sgrott EA, Seiji F, Smith RL. 2006. Biometry of hard p alate on digital photographs: a methodology for quantitative measur ements. In t J Morphol 24:19 23. Moss ML. 1968. A theoretical analysis of the functional matrix. Acta Biotheor 18:195 202. Moss ML. 1969. The differential roles of periosteal and capsular functional matrices in oro facial growth. Rep Congr Eur Orthod Soc:19 3 205. Moss ML, Salentijn L. 1969. The primary role of functional matrices in facial growth. Am J Orthod 55:566 577. Mulhern DM, Jones EB. 2005. Test of revised method of age estimation from the auricular surface of the ilium. Am J Phys Anthropol 126 :61 65 Murray KA Murray T. 1991. A test of the auricular surface aging technique. J Forensic Sci 36:1162 1169. Nawrocki SP. 1995. Race, sex, and age differences in human cranial vault thickness. Presented at the 47th Annual Meeting of the Americ an Academy of Fo rensic Sciences, Seattle, WA, February 1995 Nawrocki SP. 1998. Regression formulae for estimating age at death from cranial suture closure. In: Reichs K, editor. Forensic osteology: advances in the i dentific ation of human remains. Springfield : Charles C. Thomas. p 276 292. Nawrocki SP. 2010. The nature and sources of error in the estimation of age at death fro m the skeleton. In: Latham KE, Finnegan M, editors. Age e stimation of the h uman s keleton. Springfield: Charles C. Thomas. p 79 101. Nawrocki SP. N.D. Prediction intervals for estimates of age at death from the sternal extremity of the rib. Unpublished manuscript. National Research Council 2009. Strengthening f orensic s cience in the United States: a p ath f orward. Washington: National Academies Press. O'Connor JA, Lanyon LE, MacFie H. 1982. The influence of strain rate on adaptive bone remodelling. J Biomech 15:767 781. Osborne D, Simmons T, Nawrocki S. 2004. Reconsidering the auricular surface as an indicator of age at death. J Forensic Sci 49:905 911.
317 Overbury RS, Cabo LL, Dirkmaat DC, Symes SA. 2009. Asymmetry of the os pubis: implications for the Suchey Brooks method. Am J Phys Anthropol 139:261 268. Ozcivici E, Garman R, Judex S. 2007. High frequency oscillatory motions enhance the simulated mechani cal properties of non weight bearing trabecular bone. J Biomech 40:3404 3411. Perizonius WRK. 1984. Closing and non closing sutures in 256 crania of known age and sex from Amsterdam (A.D. 1883 1909). J of Hum Evol 13:201 216. Persson M, Thilander B. 1977. Palatal suture closure in man from 15 to 35 years of age. Am J Orthod 72:42 52. Peterson J, Wang Q, Dechow PC. 2006. Material properties of the dentate maxilla. Anat Rec A Discov Mol Cell Evol Biol 288:962 972. Powers R. 1962. The disparity between known a ge and age as estimated by cranial suture closure. Man 62 :52 54. Prader A, Tanner JM, Von Harnack G. 1963. Catch up growth following illness or starvation. An example of developmental canalization in man. J Pediatr 62:646 659. Raff erty KL, Herring SW. 1999 Craniofacial sutures: morphology, growth, and in vivo masticatory strains. J Morphol 242:167 179. Rasband W S. 1997 2014. ImageJ. Bethesda: U.S. National Institutes of Health. Rasskin Gutman D, IzpisÂœa Belmonte JC. 2004. Theoretical morphology of developm ental asymmetries. Bioessays 26:405 412. RCoreTeam. 2014. R: a language and environment fo r statistical computing. Vienna : R Foundation for Statistical Computing. Rhine S. 1990. Non metric skull racing. In: Gill GW, Rhine S, editors. Skeletal a ttribution o f race: m ethods for f orensic a nthropology. Albuquerque: Maxwell Museum of Anthropology. p 9 20. Richtsmeier JT, Corner BD, Grausz HM, Cheverud JM, Danahey SE. 1993. The role of postnatal growth pattern in the production of facia l morphology. Systematic Bio l 42:307 330. Rogers SL. 1984. The human skull: its mechanics, measurement s, and variations. Springfield: Charles C. Thomas. Rubin CT, Lanyon LE. 1985. Regulation of bone mass by mechanical strain magnitude. Calcif Tissue Int 37:411 417.
318 Ruff C, Holt B, Tr inkaus E. 2006. Who's afraid of the big bad Wolff?: "Wolff's law" and bone functional adaptation. Am J Phys Anthropol 129:484 498. Saito K, Shimizu Y, Ooya K. 2002. Age related morphological changes in squamos and parietomastoid sutures of human cranium. C ells Tissues Organs 170:266 273. Sakaue K, Adachi N. 2007. Verification of the method for estimating age at death using maxillary suture obliteration in Japanese. Nihon Hoigaku Zasshi 61:121 128. Saunders SR, Fitzgerald C, Rogers T, Dudar C, McKillop H. 19 92. A test of several methods of skeletal age estimation using a docum ented archeological sample. Can Soc Forensic Sci J 25:97 118. Scheuer L, Black SM. 2000. Developmental j uvenile o steology. San Diego: Academic Press. Schmitt A. 2002. Estimations de l'ag e au deces des sujets adultes a partir du squelette: d es raisons d'espÂŽrer. B ull Mem Soc Anthrop ol Paris 14:51 73. Scott E. 1979. Dental wear scoring technique. Am J Phys Anthropol 51:213 218. Scott JH. 1953. The c artilage of the nasal septum: a contributi on to the study of facial growth. Brit Dent J 95:37 43. Scott JH. 1954. The growth of th e human face. Proc R Soc Med 47 :91 100. Sejrsen B, Kjaer I, Jakobsen J. 1993. The human incisal suture and premaxillary area studied on archaeologic material. Acta Odon tol Scand 51:143 151. Siegel ND, Passalacqua NV. 2012. A test of the Mann maxillary suture aging method. Presented at the 64th Annual Scientific Meeting of the American Academy of Forensic Sciences Annual Scientific Meeting Atlanta, GA February 2012 Sin ger R. 1953. Estimation of age from cranial suture closure. J Forensic Med 1:52 59. Skorpinski KE. 2014. Investigating the role of diet in U.S. craniofacial secular trends. Gainesville: University of Florida. Skrzat J, Holiat D, Walocha J. 2003. A morphome trical study of the human pala tine sutures. Folia Morphol 62:1 9. Smith BH. 1984. Patterns of molar wear in hunter gatherers and agriculturalists. Am J of Phys Anthropol 63:39 56. Sparks CS, Jantz RL. 2002. A reassessment of human cranial plasticity: Boas revisited. P Natl A Sc i USA 99 :14636 14639. Stafne E. 1969. Oral r oentgeno graphic diagnosis. Philadelphia : W B Saunders.
319 Steele DG, Bramblett CA. 1988. The anatomy and biology of the human skeleton. College Station: Texas A&M University Press. Stieda L. 18 91. Der Gaumenwulst (Torus Palatinus). Ein Beitrag zur Anatomie des knÂšchernen Gaumens. Internat Beitrage z wissensch Medicin 1. Suzuki M, Sakai T. 1960. A familial study of torus palatinus and torus mandibularis. Am J Phys Anthropol 18 :263 272. Swartz SM. 1991. Strain analysis as a tool for functional morphology. Am Zool 31:655 669. Tanner JM. 1963. The regulation of human growth. Child Dev 34:817 847. Thilander B. 1995. Basic mechanisms in craniofacial growth. Acta Odontol Scand 53:144 151. Thomas R, Rei f W. 1993. The skeleton space: a finite set of organic designs. Evolution 47:341 360. Thomason JJ, Russell AP. 1986. Mechanical factors in the evolution of the mammalian secondary palate: a theo retical analysis. J Morphol 189 :199 213. Todd TW, Lyon DW, Jr. 1924. Endocranial suture closure, its progress and age relatio nship. Part I: a dult males of white stock. Am J Phys Anthropol 7:326 384. Todd TW, Lyon DW, Jr. 1925a. Cranial suture closure, its progress and age relationship. Part II: e ctocranial closure in adult males of white stock. Am J Phys Anthropol 8 :23 45. Todd TW, Lyon DW, Jr. 1925b. Cranial suture closure, its progress and age relationship. Part III: e ndocranial closure in adult males of negro stock. Am J Phys Anthropol 8:47 71. Todd TW, Lyon DW, Jr. 1925c. Suture closure, its progress and age relationship. Part IV: e ctocranial closure in adult males of negro stock. Am J Phys Anthropol 8:149 168. Uhl NM, Nawrocki SP. 2010. Multifactorial estimatio n of age at death from the human skeleton. In: Latham KE, Finnegan M, editors. Age e stimation of the h uman s keleton. Springfield: Charles C. Thomas. p 243 261. University of Tennessee, Knoxville. N.D. WM Bass d onated s keletal c ollection. http://web.utk.edu/~fac/collection.html Vacha Haase T, Thompson B. 2004. How to e stimate and i nterpret v arious e ffect s izes. J Coun s Psychol 51(4):473 481.
320 Van Gerven DP, Armelagos GJ. 1983. "Farewell to paleodemogra phy?" Rumors of its death have been greatly exaggerated. J Hum Evol 12:353 360. Vaupel JW. 2010. Biodemography of human ageing. Nature 464:536 542. 2011. Age estimation in archaeological skeletal remains: evaluation of four non destructive age calculation methods. J Forensic Odontostomatol 29:14 21. von Wower n N, Stoltze K. 1978. Sex and age differences in bone morphology of mandibles. Scand J Dent Res 86:478 485. Wainwri ght SA. 1988. Form and function in o rganisms. Am Zool 28 :671 680. Wang Q, Dechow PC. 2006. Elastic properties of external cortical bone in the craniofacial skeleton of the rhesus monkey. Am J Phys Anthropol 131:402 415. Wang Q, Smith AL, Strait DS, Wright BW, Richmond BG, Grosse IR, Byron CD, Zapata U. 2010. The global impact of sutures assessed in a finite element model of a macaque cranium. Anat Rec 293:1477 1491. Wang Q, Strait DS, Dechow PC. 2006. Fusion Patterns of Craniofacial Sutures in Rhesus Monkey Skulls of Known Age and Sex from Cayo Santiago. Am Jof Phys Anthropo l 131:469 485. Wescott DJ, Jantz RL. 2005. Assessing craniofacial secular change in American Blacks and Whites using geometric morphometry. In: Slice DE, editor. Modern m orphometrics in p hysical a nthropology. New York: Springer. p 231 246. Wheatley FM. 1996. Standardization of the Mann method: nonmetric assessment of age using palatal suture closure Knoxville: The University of Tennessee, Knoxville. White TD, Black MT, Folkens PA. 2012. Human osteology. San Diego: Academic Press. Williams G. 1957. Pleiotropy, natural selection, and the evolution of senescence. Evolution 11:398 411. Woo JK. 1949. Ossification and growth of the human maxilla, premaxilla and palate bone. Anat Rec 105:737 7 61. Woo JK. 1950. Torus palatinus. Am J Phys Anthropol 8:81 111. Wroe S, Mor eno K, Clausen P, McHenry C, Curnoe D. 2007. High resolution three dimensional computer simulation of homin id cranial mechanics. Anat Rec 290:1248 1255.
321 Yacoub N, Ismail YH, Mao JJ 2002. Transmission of bone strain in the craniofacial bones of edentulous human skulls upon dental implant loading. J Prosthet Dent 88:192 199. Youden WJ. 1998. Experimentation and measurement. Mineola: Dover Publications. Yu JC, Borke JL, Zhang G. 2004. Brief synopsis of cranial sutures: optimization by adaptation. Semin Pediatr Neurol 11:249 255. Zambrano CJ. 2005. Evaluation of regression equations used to estimate age at death from cranial suture closure Indianapolis : University of Indianapolis. Ziva 147:161 167. Zollikofer CP, Weissmann JD. 2011. A bidirectional interface growth model for cranial interosseous suture morphogenesis. J Anat 219:100 114.
322 BIOGRAPHICAL SKETC H Carrie Brown is originally from Wycombe, Pennsylvania. She attended Lawrence University in Appleton, Wisconsin from 2000 to 2004, where she earned a Bachelor of Arts degree in anthropology and French and graduated cum laude in course and magna cum laude in independent study in June 2004. Her undergraduate research focused on understanding community building through the use of public space in urban environments in London, England and Nantes, France. From Wisconsin, Carrie moved to France to teach elemen tary school English for two years. Carrie attended California State University, Chico (CSU C) from 2006 to 2009, earning a Master of Arts in anthropology with distinction in May 2009. In 2008, she was awarded a student research fellowship with the Joint P OW/MIA Accounting Command Central Identification Laboratory (JPAC CIL) on Hickam Air Force Base, Hawaii. During this fellowship she was a member of the inaugural Forensic Science Academy and completed a large scale research project that investigated the u ncertainty associated with age estimation at the JPAC CIL. This research formed the thesis, which was awarded the 2009 2010 C SU C School of Graduate, International, Fo llowing graduation from Chico State, Carrie was awarded a postgraduate research fellowship at the JPAC CIL and in fall of 2009 was hired as a forensic anthropologist. In August 2011 Carrie began her doctoral coursework at the University of Florida while c ontinuing to work at the JPAC CIL during winter and summer breaks. During her time at the University of Florida she was first a Graduate Analyst and then the Senior Graduate Analyst in the C. A. Pound Human Identification Laboratory, taught undergraduate courses in forensic anthropology and human osteology, and served as
323 the Vice President of the Florida Anthropology Student Association. After advancing to candidacy in April 2013, she relocated to the new JPAC CIL laboratory on Offutt Air Force Base, Nebr aska, where she has worked since May 2013; the JPAC CIL was renamed the Department of Defense POW/MIA Accounting Agency Laboratory in January 2015. Carrie is also an adjunct instructor in the graduate program in forensic sciences at Nebraska Wesleyan Univ ersity, Lincoln, Nebraska. She received her Ph.D. from the University of Florida in the spring of 2016.