xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim
leader nam 22 Ka 4500
controlfield tag 007 cr-bnu---uuuuu
008 s2010 flu s 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0004750
An assessment of microevolutionary change among prehistoric florida populations through the analysis of craniometric data
h [electronic resource] /
by Samantha Seasons.
[Tampa, Fla] :
b University of South Florida,
Title from PDF of title page.
Document formatted into pages; contains X pages.
Thesis (MA)--University of South Florida, 2010.
Includes bibliographical references.
Text (Electronic thesis) in PDF format.
Mode of access: World Wide Web.
System requirements: World Wide Web browser and PDF reader.
ABSTRACT: The analysis of craniometric data collected from skeletal remains, combined with archaeological data, can provide very valuable information pertaining to biological distance and gene flow among prehistoric populations through space and time. The goal of this study was to examine microevolutionary change among prehistoric populations in south Florida based on the degree of cranial variation among populations at seven prehistoric sites. It was expected that as time progressed, microevolutionary forces caused significant changes in the crania of the various populations based on the relative geographic proximity of the sites and the temporal distance between sites. A Microscribe 3-DX digitizer was used to collect coordinate data using the full protocol of cranial landmarks. Twenty-three interlandmark distances for n=223 skulls from seven sites, ranging in age from 8120 B.P. to 260 B.P., were analyzed using Cluster Analysis, an Analysis of Variance (ANOVA), a Tukey's Pairwise Comparison (post-hoc test), a Multiple Analysis of Variance (MANOVA), and Principal Components Analysis (PCA) in SPSS 18.0. The seven sites were Windover (8Br246), Perico Island (8Ma6), Captiva Island (8Ll57), Belle Glade (8Pb40), Horr's Island (8Cr41), Safety Harbor (8Pi2), and Fuller Mound A (8Br90). Of the 223 crania used, zero (0) skulls were 100% complete. Results of univariate and multivariate statistical analyses indicate that there are significant differences among the seven groups. The agglomerative cluster analysis did not provide significant results. When using Maximum Cranial Length (GOL), the ANOVA (F=5.190, p 0.000) and post-hoc tests indicated that there was a significant amount of variation among the seven populations. In a series of 12 MANOVA tests, it was determined that significant variation existed between Windover and each of the remaining six sites (F > 5, p 0.000). Additionally, the MANOVA tests indicated that significant variation existed between Horr's Island and Safety Harbor (F = 8.151, p 0.000) and between Safety Harbor and Fuller Mound A (F = 5.549, p 0.000). Last, a Principal Components Analysis demonstrated that measurements consistent with length or breadth accounted for the largest percentage of variation among the populations. In conclusion, the data strongly demonstrate a significant amount of variation among prehistoric populations as time progressed from 8120 B.P. to 260 B.P. Specifically, changes in gene flow which can be attributed to significant differences among populations based on craniometric data parallel major time gaps and historical events in Florida. More generally, these results can be applied to other past populations to investigate similar patterns of gene flow and changes that may have occurred due to various social, political, and environmental stressors.
Advisor: Erin Kimmerle, Ph.D.
Prehistoric Florida populations
t USF Electronic Theses and Dissertations.
An Assessment of Microevolutionary Change among Prehistoric Florida Populat ions through the Analysis of Craniometric Data by Samantha M. Seasons A thesis submitted in partial fulfillment of the requirements for the degree of Masters of Arts Department of Anthropology College of Arts and Sciences University of South Florida Major Professor: Erin Kimmerle, Ph.D. David Himmelgreen, Ph.D. Nancy White, Ph.D. Date of Approval: November 5, 2010 Keywords: microevolution, craniometric analysis, gene flow, human variation, prehistoric Florida populations Copyright 2010, Samantha M. Seasons
Dedication To my fianc Bobby You have been my guiding strength the last two years. Without you I would have been lost. You mean everything to me and for that I dedicate this to you. Thank you!
Acknowledgements It is with pleasure that I thank Dr. Erin Kimmerle, Dr. David Himmelgreen and Dr. Nancy White for providing me with guidance during this study. I thank Dr. L orena Madrigal and Charles Dionne for assistance with the statistical analysi s. I thank Dr. David Hunt for allowing me to collect my data at the Natural Museum of Natural History. I thank Dr. Richard Jantz and Donna Freid for providing me with craniometric data collected on the Windover remains. I thank Dr. Brent Weisman, Dr. Michael Russo, Rhonda Coolidge, and Melissa Pope for assisting with background research. I thank the Department of Anthropology and the Graduate and Professional Student Council at the University of South Florida for providing funding for travel to present this resea rch at the 2010 AAPA Annual Meetings.
i Table of Contents List of Tables ..................................................................................................................... iii List of Figures ......................................................................................................................v Abstract ............................................................................................................................. vii Chapter 1 Introduction .........................................................................................................1 Chapter 2 Literature Review ................................................................................................4 Human Variation ......................................................................................................4 Craniometric Analysis .............................................................................................7 Previous Studies of Microevolutionary Change ......................................................9 Archaeological Characterization and Geographic Analysis ..................................12 Chapter 3 Materials and Methods ......................................................................................28 Research Methods ..................................................................................................28 Data Samples .........................................................................................................29 Methods for Univariate Statistical Analysis ..........................................................38 Methods for Multivariate and Canonical Statistical Analyses ...............................40 Chapter 4 Results ...............................................................................................................43 Univariate Statistical Analysis ...............................................................................43 Multivariate and Canonical Statistical Analyses ...................................................47 Chapter 5 Discussion .........................................................................................................78 Archaeological Sites Used for Analysis ................................................................79 Interpretation of Statistical Analyses to Illustrate Variation................................ ..79 Variation among populations of different occupation periods ........................80 Variation among contemporaneous relationships ............................................84 Variation of genetic admixture among populations during European contact ........................................................................................86 Chapter 6 Conclusion .........................................................................................................92 Suggestions for Future Research ...........................................................................93 Bioarchaeology as Applied Anthropology.............................................................94
ii References ..........................................................................................................................95 Appendices .......................................................................................................................100 Appendix A Â– Agglomerative Cluster Analysis ..................................................101
iii List of Tables Table 2.1 Florida Archaeological Sites with Crania Samples Used in this Research (n=223) .................................................................................13 Table 3.1 Sample Sizes by Site ..................................................................................32 Table 3.2 List of Interlandmark Distances .................................................................33 Table 3.3 List of Type 1 Landmarks ..........................................................................34 Table 4.1 Descriptive Statistics of Maximum Cranial Length (GOL) f or all 7 Archaeological Sites ............................................................................45 Table 4.2 ANOVA for all Archaeological Sites Investigated using GOL .................45 Table 4.3 TukeyÂ’s HSD Post-hoc Test Describing Significance of Differe nce between sites for GOL .........................................................................46 Table 4.4 Sample Size for each Archaeological Site .................................................51 Table 4.5 MANOVA for all Archaeological Sites Investigated ................................51 Table 4.6 Sample Sizes for Windover and Perico Island ...........................................53 Table 4.7 MANOVA for Windover and Perico Island ..............................................53 Table 4.8 Sample Sizes for Windover and Captiva Island ........................................55 Table 4.9 MANOVA for Windover and Captiva Island ............................................55 Table 4.10 Sample Sizes for Windover and Belle Glade .............................................57 Table 4.11 MANOVA for Windover and Belle Glade ................................................57 Table 4.12 Sample Sizes for Windover and HorrÂ’s Island ..........................................59 Table 4.13 MANOVA for Windover and HorrÂ’s Island ..............................................59
iv Table 4.14 Sample Sizes for Windover and Safety Harbor .........................................61 Table 4.15 MANOVA for Windover and Safety Harbor.............................................61 Table 4.16 Sample Sizes for Windover and Fuller Mound A ......................................63 Table 4.17 MANOVA for Windover and Fuller Mound A .........................................63 Table 4.18 Sample Sizes for Perico Island and Captiva Island ...................................65 Table 4.19 MANOVA for Perico Island and Captiva Island .......................................65 Table 4.20 Sample Sizes for Captiva Island and Belle Glade .....................................67 Table 4.21 MANOVA for Captiva Island and Belle Glade .........................................67 Table 4.22 Sample Sizes for Belle Glade and HorrÂ’s Island .......................................69 Table 4.23 MANOVA for Belle Glade and HorrÂ’s Island ...........................................69 Table 4.24 Sample Sizes for HorrÂ’s Island and Safety Harbor ....................................71 Table 4.25 MANOVA for HorrÂ’s Island and Safety Harbor .......................................71 Table 4.26 Sample Sizes for Safety Harbor and Fuller Mound A ...............................73 Table 4.27 MANOVA for Safety Harbor and Fuller Mound A...................................73 Table 4.28 KMO and BartlettÂ’s Test for all Archaeological Sites Inve stigated ..........76 Table 4.29 Total Amount of Variance Explained by Each Component ......................76 Table 4.30 Component Matrix for Components with Eigenvalues Greater Than 1 ..................................................................................................77
v List of Figures Figure 2.1 Florida Map Illustrating 7 Prehistoric Archaeological Si tes from which Crania were Examined ..............................................................14 Figure 3.1 Frontal view of Type 1 Landmarks ............................................................35 Figure 3.2 Lateral view of Type 1 Landmarks ............................................................36 Figure 3.3 Basilar view of Type 1 Landmarks ............................................................37 Figure 4.1 Scatter Plot of Component 1 and Component 2 for all Archaeological Sites ............................................................................52 Figure 4.2 Scatter plot of Component 1 and Component 2 for Windover and Perico Island.........................................................................................54 Figure 4.3 Scatter plot of Component 1 and Component 2 for Windover and Captiva Island ..................................................................................... 56 Figure 4.4 Scatter plot of Component 1 and Component 2 for Windover and Belle Glade ..........................................................................................58 Figure 4.5 Scatter plot of Component 1 and Component 2 for Windover and HorrÂ’s Island ........................................................................................60 Figure 4.6 Scatter plot of Component 1 and Component 2 for Windover and Safety Harbor .......................................................................................62 Figure 4.7 Scatter plot of Component 1 and Component 2 for Windover and Fuller Mound A....................................................................................64 Figure 4.8 Scatter plot of Component 1 and Component 2 for Perico Island and Captiva Island................................................................................66 Figure 4.9 Scatter plot of Component 1 and Component 2 for Captiva Island and Belle Glade ....................................................................................68
vi Figure 4.10 Scatter plot of Component 1 and Component 2 for Belle Glade and HorrÂ’s Island ........................................................................................70 Figure 4.11 Scatter plot of Component 1 and Component 2 for HorrÂ’s Island and Safety Harbor .......................................................................................72 Figure 4.12 Scatter plot of Component 1 and Component 2 for Safety Harbor and Fuller Mound A....................................................................................74
vii Abstract The analysis of craniometric data collected from skeletal remains, combine d with archaeological data, can provide very valuable information pertaining to biolog ical distance and gene flow among prehistoric populations through space and time. The goal of this study was to examine microevolutionary change among prehistoric populati ons in south Florida based on the degree of cranial variation among populations at seven prehistoric sites. It was expected that as time progressed, microevolutionar y forces caused significant changes in the crania of the various populations based on the relati ve geographic proximity of the sites and the temporal distance between sites. A Microscribe 3-DX digitizer was used to collect coordinate data using the f ull protocol of cranial landmarks. Twenty-three interlandmark distances for n=223 skull s from seven sites, ranging in age from 8120 B.P. to 260 B.P., were analyzed using Cluster Analysis, an Analysis of Variance (ANOVA), a TukeyÂ’s Pairwise Compari son (post-hoc test), a Multiple Analysis of Variance (MANOVA), and Principal Components Ana lysis (PCA) in SPSS 18.0. The seven sites were Windover (8Br246), Perico Island (8Ma6), Captiva Island (8Ll57), Belle Glade (8Pb40), HorrÂ’s Island (8Cr41), Safe ty Harbor (8Pi2), and Fuller Mound A (8Br90). Of the 223 crania used, zero (0) skulls were 100% complete. Results of univariate and multivariate statistical analyses indicate that there are significant differences among the seven groups. The agglomerative cluster analysis did
viii not provide significant results. When using Maximum Cranial Length (GOL), t he ANOVA (F=5.190, p 0.000) and post-hoc tests indicated that there was a significant amount of variation among the seven populations. In a series of 12 MANOVA tests, it was determined that significant variation existed between Windover and each of the remaining six sites (F > 5, p 0.000). Additionally, the MANOVA tests indicated that significant variation existed between HorrÂ’s Island and Safety Harbor (F = 8.151, p 0.000) and between Safety Harbor and Fuller Mound A (F = 5.549, p 0.000). Last, a Principal Components Analysis demonstrated that measurements consistent wit h length or breadth accounted for the largest percentage of variation among the populations. In conclusion, the data strongly demonstrate a significant amount of variation among prehistoric populations as time progressed from 8120 B.P. to 260 B.P. Specifically, changes in gene flow which can be attributed to significant dif ferences among populations based on craniometric data parallel major time gaps and historic al events in Florida. More generally, these results can be applied to other past populat ions to investigate similar patterns of gene flow and changes that may have occur red due to various social, political, and environmental stressors.
1 Chapter 1 Introduction Within the field of biological anthropology, numerous studies into the different trends of human variation as it relates to the biological distance between populations (e.g. Relethford 2004; Relethford 2002; Relethford 1994; Relethford and Harpending 1994; Jantz 1973; Key and Jantz 1981; Key and Jantz 1990; Jantz and Owsley 2001; Owsley et al. 1982; Stojanowski 2004; Stojanowski 2005) have shown that morphological variation as it pertains to craniometric variation is a valuable tool for examining microe volutionary change within and among populations of restricted geographic and temporal space. Specifically, studies have focused on the phenotypic expression of genetic marke rs that illustrate the amount of variation among and within populations and have found that in Florida and other regions of the United States, much of the variation can be attributed a change in gene flow that results from the introduction of new populations to already settled populations (Stojanowski 2004; Stojanowski 2005). However, preceding genetic studies, biological anthropologists developed and executed methods for assessing huma n variation through craniometric analysis (Key and Jantz 1990; Jantz and Owsley 2 001; Owsley et al. 1982; Jantz 1973). Morphological variation that can be expressed through differences in cranial measurements contributes to theories of human evolution and human variation by demonstrating the various evolutionary forces that act on populat ions over temporal space. According to Pietruswesky (2008:487), the analysis of
2 morphological traits provides both interand intra-group variation. This tool is ofte n utilized among bioarchaeologists whose research aims to examine cranial var iation attributed to microevolutionary processes. Unfortunately, in Florida, little research has been done on craniometric a nalysis to investigate microevolutionary change over time among prehistoric populations. Wit hin the field of biological anthropology, conducting research regarding human varia tion is critical as the presence of biological variation may suggest that evoluti onary forces were acting on populations for a variety of reasons through time and space; specifically among prehistoric Florida populations. Additionally, the scope of this research can contr ibute to archaeology by providing biological support for population interactions that have al ready been suggested in archaeological literature. Due to the lack of research that incorporates biological anthropology with archaeology, this project addresses the issue of micro-evolutionary change among prehistoric Florida populations by employing methods for craniometric analy sis to examine varying degrees of gene flow and biodistance. Interlandmark distances the metric measurement (in mm) between any two landmarks on the skull, were used to establish the degree of variation based on different dimensions of the skull. This study aims to accomplish several goals: (1) establish biodistance among prehistor ic populations in Florida based on differences in cranial morphology; (2) determine which interlandmark distances are contributing to any observed variation; and (3) provi de biological support for various types of movement, trade, exchange, and interactions that were suggested based on the literature reviewed for this project. After revi ewing similar
3 studies that have examined microevolutionary change among populations in different regions of the United States (Jantz 1973; Key and Jantz 1981; Stojanowski 2004; Stojanowski 2005) it is expected that: (1) significant variation will exist am ong populations of varying temporal relationships; (2) variation should decrease as populations become more contemporaneous; and (3) variation will increase due to changes in gene flow among diverse populations as a direct result of European settlement.
4 Chapter 2 Literature Review Human Variation The evaluation of population affinity through time and space, which incorporates the analysis of genetic differences, has contributed greatly to studies of hum an variation within anthropology. In many studies, the investigation of genetic admixture and ge netic isolation has supplemented research regarding breeding populations and population isolates (Molnar 2002). In terms of establishing population affinity, the examina tion of gene distribution is vital because, Â“genetic loci are shared by all human popul ations and, with rare exceptions, none are unique to any one groupÂ” (Molar 2002:240). For biological anthropologists and bioarchaeologists, this idea is crucial when inve stigating population distance among prehistoric populations through craniometric analysis due t o the fact that biological distance among populations results from morphological expressions of genetically controlled traits (Griffin et al. 2001:226). If it is accepted that genetic loci are not unique to any one group, it may be assumed that cranial varia tion among populations would be minimal. Specifically, the lack of unique genetic loci would reduce the amount of genetic variation within a specific population. As a result, the amount of variation among phenotypic traits such as the size and shape of the crania would be significantly reduced. Further, the lack of variation due to the absence of
5 unique genetic loci within a group will ultimately contribute to a lack of variation am ong groups. In addition, social relationships and the environment should be considered as influencing factors when utilizing craniometric data for investigating human variation, as those components often play a large role in the expression of morphological trait s. The functions of various social relationships, like reproduction, dictate the expression of morphological traits through genetic admixture. At the same time, environment al factors, such as access to resources and disease, also have the capability to shape the phenot ypic expression of traits through growth and development. According to Molnar (2002:248), Â“except for a superficial identification of the majority of the inhabitants of a continent, Â‘basic stockÂ’ or Â‘geographical raceÂ’ tells us little about biological diver sity of the interrelationships between breeding populations or the effects of the environment which are the dimensions of the selective forces that act on the populations.Â” Â“Geographic al raceÂ”, in this context, refers to a large inclusive group that includes smaller local groups that exhibit diversity (Molnar 2002). Unfortunately, the concept of Â“geographical raceÂ”, as Molnar suggests, neglects the diversity among the local groups that often i ncludes differences in social interaction, which ultimately affects the way i n which the local groups reproduce: inbreeding versus outbreeding. Numerous studies demonstrate the effect that breeding practices, due to varyi ng degrees of social interaction, have on gene flow and genetic admixture (Santos et al. 1999; Glass et al. 1952; Kostyu et al. 1989). In many of these studies, genetic characters such as blood types and haplotypes were examined to demonstrate that the genetic
6 variation between groups increases as social isolation becomes more distinct be tween populations. According to Glass et al. (1952 in Molnar 2002:262) significant differences in blood type were present between the Dunker isolate of 300 persons and U.S. and German populations from which ancestors of the Dunker population had migrated. Although this study emphasizes blood types and specific expressions that are ass ociated, the same concept can be applied to studies that incorporate craniometric anal ysis. This is especially important for bioarchaeologists because it allows for the invest igation of social interaction through the examination of inter-sample craniometric variation. The incorporation of standard methods of craniometric analysis in studying human remains from prehistoric sites has provided bioarchaeologists with opportunitie s to conduct research on prehistoric human variation. By analyzing interlandmark distances of the crania, biological anthropologists can examine the degree of c ranial variation among populations. Relethford (2002:397) argues that, Â“the strong simila rity between genetic and craniometric results suggests that global patterns of cr aniometric variation can be considered, on average, selectively neutral.Â” Therefore, c raniometric data are useful for looking at population distance among contemporaneous populations that live in the same geographic region because when analyzed on a multivariate l evel they are not influenced by natural selection. That is not to say that natural se lection does not influence the morphology of the skull. It simply says that when populations live at the same time in the same place there are not enough environmental differences that would compromise fitness in a way for natural selection to cause significant changes.
7 Craniometric Analysis To analyze morphological variation of the crania among prehistoric populations in Florida, the interlandmark distances must reflect homology (OÂ’Higgins and Str and Vidarsdottir 1999:135; van Valen 1982). The analysis of interlandmark distances collected using a Microscribe 3-DX digitizer on the full protocol of cranial landm arks has served as a prominent tool for exploring human variation between environmentally influenced populations by analyzing differences in size of the crania. With the use of such methods for craniometric analysis, biological anthropologists aim to descr ibe human variation more efficiently and to assess microevolutionary forces such as n atural selection, genetic drift, mutation, and migration, which have acted on specific populations. At the genetic level, all four evolutionary forces can be measured by calculating the changes in gene frequencies from one generation to the next. A t the phenotypic level, evolutionary forces can be measured by determining qualitative and quantitative patterns of various traits that are associated with differe nt populations. Comparisons can be made from both a temporal perspective and a geographic perspe ctive to evaluate microevolutionary changes among specific populations. For example, phenotypic traits of the skull such as interlandmark distances can assist in asse ssing microevolutionary changes by examining the differences in size of the crani a among populations. These differences could be associated with geographic space, tempora l space, or both. Despite recent developments in the analysis of three-dimensional coordinate dat a for craniometric variation (Kimmerle et al. 2008; Ross and Williams 2008; Slice 2005),
8 the multivariate analysis of basic interlandmark distances continues to prove us eful for examining population distance. As stated by Pietrusewsky (2008:488), a theoretic al foundation for metric analysis is supported through the repeatability and heritabil ity of interlandmark distances. It is this theoretical foundation that provides justif ication for the analysis of interlandmark distances as a tool to investigate the degree of gene flow among prehistoric populations in Florida. Further, because multivariate craniometric data reflect gene flow and genetic admixture, craniometric analysis can be exploited to assess specific relationships among groups. According to Jantz (1973:15), Â“the goal of multivariate studies may be either to determine the relationships among several groups, from which may follow interpre tations of an historical or evolutionary nature; or to develop parameters for classifyin g individuals into their proper group, as in the case of sex.Â” In prehistoric Florida populations, the goal is not to group individuals by sex but rather to investigate with craniometric data the group parameters that were developed as a result of microevolutionary forces, in addition to determining if relative relationships among the groups existed based on archaeological context. Fortunately, multivariate and can onical statistical analyses of craniometric data can contribute greatly to differentiating patterns of human biological variation. Additional support for the application of craniometric analysis to population distance, specifically in Florida, is provided by Relethford (2004) through his applica tion of population genetic models for craniometric data. It is in this study that Re lethford (2004:382) suggests that on a global level, Â“[data on classic genetic markers] sugge st that
9 roughly 10% of human genetic diversity is found among geographic regions, 5% among local populations within regions, and 85% within local populations.Â” He later suggests that global craniometric variation is similar to the model of genetic variati on previously discussed (Relethford 2004:382). Based on the model that Relethford presents, it can be hypothesized that at the global level, prehistoric populations within Florida would be considered local populations within a region. Consequently, the cranial variation among the prehistoric populations in Florida should be minimal. Previous Studies of Microevolutionary Change Within the field of biological anthropology, a variety of studies have been executed to examine microevolutionary changes among specific populations in the United States. These studies vary to include investigation of change through time and space both within populations and between populations, and often include cranial and dental variables for analysis. These studies also vary by analyzing microe volutionary change in populations that inhabited various regions of the Unites States before Eur opean contact or shortly after, including the Arikara (Jantz 1973; Key and Jantz 1981; Key and Jantz 1990; Owsley et al. 1982); the Blackfoot, Cheyenne, Omaha, Pawnee, Ponca and Sioux (Jantz and Owsley 2001); and The Guale (Stojanowski 2004; Stojanowski 2005). In studies that investigated microevolutionary change among Arikara populations in South Dakota, cranial metrics were submitted to statistical analysi s. In all three studies (Jantz 1973; Key and Jantz 1981; Owsley et al. 1982), the results indicated that significant microevolutionary changes did occur among the Arikara. In two of the st udies
10 (Jantz 1973; Key and Jantz 1981), the microevolutionary changes were attributed to variation in gene flow among the Arikara who inhabited different sites in South Dakota. According to Key and Jantz (1981:250), variation was mostly attributed to difference s in sex and some differences were attributed to differences in geographic locat ion of the sites inhabited by these populations. Based on the finding of Jantz (1973) and Key and Jantz (1981), it should be expected that analysis of cranial measurements of various sites in Florida will demonstrate heterogeneity because sex should contribute more significant variation than the small geographical space among sites. In a similar study, Owsley et al. (1982) demonstrated that cranial variation among the Arikara may also be attributed to temporal separation. This finding also provi des a reason to believe that Florida sites will also demonstrate microevolutionary changes due to temporal distance among the sites. The sites used in this analysis range i n age from 8120 B.P. to 260 B.P. thus allowing 7860 years between the earliest and latest occupation periods for microevolutionary changes to occur. In 2001, Jantz and Owsley conducted a study in which cranial measurements of several different Native American populations were examined to determine i f significant variability existed among ancient American crania. After submitting 22 crania l measurements from Blackfoot (n=66), Cheyenne (n=22), Omaha (n=16), Pawnee (N=27) Ponca (n=19), and Sioux (n=28) to multivariate statistical analyses, Jantz an d Owsley (2001) concluded that heterogeneity was present among these six populations. Accordin g to Jantz and Owsley (2001:152), Â“high variability among early American f ossil crania may not by itself provide evidence of multiple migrations, but it is consistent wit h an
11 emerging consensus that different populations were involved in the early peopling of North America.Â” When applying this consensus to prehistoric Florida populations, it m ay be expected that distinctly different populations inhabited Florida. Conseque ntly, distinct phenotypic differences among populations would suggest a significant amount of genetic variation among the groups. Unless there was a significant amount of gene flow among the groups that would allow an adequate mixture of genetic variation, heterogeneity would increase among groups. Fortunately, studies have been performed in which Native groups of north Florida and south Georgia have been evaluated for microevolutionary forces including ge ne flow and genetic drift. In these studies, Stojanowski (2004; 2005) measured mesiodistal and buccolingual tooth crown dimensions of late precontact and historic-period populations t o evaluate population history and structure. After the tooth dimensions were subjecte d to statistical analysis, the results indicated that homogeneity was gre atest in the pre-contact period (prior to 1607). As stated by Stojanowski (2004:323), homogeneity among precontact samples is a result of sharing common mating patterns. Conversely, Stojanowski (2004:324) found that this trend was reversed in the Late Mission period (1686-1702) due to European contact decreasing extralocal gene glow. These findi ngs may further suggest that similar to European contact in north Florida and south Georgi a, this study will show that European contact in south Florida will limit gene flow ultimately contributing to an increase in heterogeneity among populations.
12 Archaeological Characterization and Geographic Analysis Currently, in the state of Florida, little research using combined methods of craniometric and archaeological characterization analyses have bee n performed to address the issue of biological or genetic interaction among prehistoric popul ations. The purpose of this literature review was to illustrate various possibilities of i nteraction among peoples from seven prehistoric sites in south Florida (Refer to Table 2.1 and Figure 2.1 for site number, sample size, date, cultural affiliation, and geog raphic location) in order to provide a foundation for investigating the degree of human variation among the populations. Furthermore, the archaeological context of seven archaeological s ites will be examined for any evidence of cultural exchange and contact so that hypothes es regarding the opportunity for gene flow among prehistoric sites in Florida ca n be developed and tested in the future. The analysis of craniometric data collected from human skeletal remains ca n provide very valuable information pertaining to the amount of genetic variation among prehistoric populations through space and time. Specifically, the analysis of morphological variation is useful for aiding in the understanding of cultural dif ferences, populations structure, and the degree of interaction between groups from the archaeological record (Key and Jantz 1990:54). In the state of Florida, a large sample of cranial remains from numerous archaeological sites in the Peninsular reg ion are available for collecting coordinate data to examine intra Â– and intersample variabili ty.
13 Table 2.1 Florida Archaeological Sites with Crania Samples Used in this Res earch (n = 223) SiteLocationSite NumberSample Size (n)DateCultural AffiliationSource WindoverBrevard County8BR246668120-6980 B.P.Early A rchaicDoran 2002:11 Perico IslandManatee County8MA6272510-1210 B.P.Mana sotaLuer and Almy 1982 Captiva IslandLee County8LL57121310-810 B.P.Caloosa hatchee IIHutchinson 2004:28 Belle GladePalm Beach County8PB40 371310-760 B.P.Gl ades I and Glades IIWilley 1948:216 Horr's IslandCollier County8CR41 201260497 B.P.Gl ades II and Glades IIIMilanich 1994:301 Fuller Mound ABrevard County8BR90421010-247 B.P.Mal abar IIWilley 1954 Safety HarborPinellas County8PI2191110-260 B.P.Safe ty HarborHutchinson 2006:31 Hutchinson 2004:95Mitchem 1989
14 Figure 2.1 Â– Florida Map Illustrating 7 Prehistoric Archaeological Sites f rom which Crania were Examined
15 For many of the archaeological populations in Florida, the location, type of environment inhabited, and the ability to interact with other groups plays an important role on the morphological variation between groups. As stated by Hutchinson (2006:20), Â“the historic accounts make some of the differences between populations apparent, but i n many descriptions, the distinctions are hidden behind the major themes of the narrati ves Â– conflicts, political structure, food and goldÂ”. For the seven sites that I analy zed for cranial variation, one is attribute to the Manasota Culture Region (Perico Isl and), one site is from the Safety Harbor Culture Region (Safety Harbor), two sites are from the Glades Culture Region (Belle Glade and HorrÂ’s Island), one site is from the Caloos ahatchee Culture region (Captiva Island), one site is from the Malabar II period (Ful ler Mound A), and one from the Early Archaic period (Windover). The Manasota Culture is believed to have begun around 2510 B.P. and ended around 1210 B.P. as part of the Formative and Mississippian cultural development (Lue r and Almy 1982; Milanich 1994). According to Milanich (1994), Manasota developed its name from a combination of Manatee and Sarasota counties and is often recognized by its Weeden Island burial practices. Further, Luer and Almy (1982:37) defined Mana sota culture based on the observable trends in the fishing, hunting, and gathering economy, in addition to primary, flexed burial practices, and the types of pottery and cerami c classification consistent throughout the culture period. Perico Island (8Ma6), as discussed by Willey (1949a), is a site that possess es material cultural and demonstrates economic trends that are consistent with Manasota as outlined by Luer and Almy (1982). During excavation by Marshall Newman in 1933 and
16 1934, a burial mound revealed 185 flexed burials (Janus Research 2002; Willey 1949a); a mortuary practice that is consistent with early Manasota Culture. Accor ding to Milanich (1994:227) burials during early Manasota times, 2510 B.P. Â– 1910 B.P. were primarily flexed. These burials were often primary burials in cemeteries or shell midde ns. Very few grave goods or burial offerings were recovered in the burial mounds. In fact the only evidence of material culture were sherds discovered in the fill of the mound and assumed to be accidental inclusions (Willey 1949a:176). Fortunately, a large abundance of pottery and ceramics were excavated from two shell middens associated w ith the site and were utilized to date the site approximately. The various types of pottery that were discovered in the shell middens at Perico Island, based on the temper and decoration, were of the Glades Plain, the Perico Ser ies, Biscayne Series, Deptford Series, and other miscellaneous types (Wille y 1949a). These same types of pottery are seen at various sites around south Florida including, Saf ety Harbor, Belle Glade and Fuller Mound A. According to Luer and Almy (1982), the Manasota culture, in which Perico Island belongs, dates from approximately 2510 B.P. to 1210 B.P., predating other sites in Florida where similar pottery types are found. In contrast, according to Willey (1949a), the pottery at Perico Island is of the G lades I period and sometimes associated with the Deptford and Santa Rosa-Swift cultures The Glades I period as defined by Willey (1948) begins around 1310 B.P. which would overlap the end of the Manasota culture by 100 years. To add to the confusion, Goggin (1950) redefined the structure of ceramic culture throughout the Glades occupation and claimed that Glades I should begin around 2510
17 B.P. and end around A.D. 1. If this adjustment is accepted then it should be assumed that, based on pottery classification, Perico Island is consistent with the ear ly part of the Manasota period, not the later part as Willey (1949a) suggests. The presence of SantaRosa pottery, according to Goggin, would extend occupation to 1360 B.P., which would carry over in to the Glades II period. While much confusion surrounds the exact time frame in which Perico Island was occupied, it can still be noted that, based on the divers e classification of pottery and ceramics found at the site, Perico Island inha bitants were interacting with different culture groups throughout south Florida. The Safety Harbor site (8Pi2), originally excavated by M.W. Stirling in 1930 (Stirling 1931; Mitchem 1989), is located on the west coast of Florida in Pinellas Count y and was occupied between late prehistoric and postcontact times (1110 B.P. to 285 B.P.) (Mitchem 1989:556; Hutchinson 2006; Hutchinson 2004). According to Mitchem (1989:50; 1989:567), Safety Harbor (8Pi2) is the type site for the Safety Harbor Cultur e and was occupied over four distinct phases: Englewood (1110 B.P.-1010 B.P.), Pinellas (1010 B.P.510 B.P.), Tatham (510 B.P.443 B.P.), and Bayview (443 B.P.-285 B.P.). Similar to Perico Island, Safety Harbor Culture resembles Mississ ippian culture consistent to that of northwest Florida and the greater Southeast. This classif ication was made based on the presence of Pinellas Plain and Pinellas Incised pottery that i s similar to Fort Walton and Lake Jackson types (Willey 1949a:137). According to Mitchem (1989:551), Â“[Willey] noted the strong similarities between decorated Safety Harbor pottery types and those of the Fort Walton culture of the northwest Florida and adja cent areas.Â” In addition to Fort Walton type wares, Glades Plain, Biscayne Pla in, Biscayne
18 Check Stamped, St. Johns Plain, and St. Johns Check Stamped types of sherds were recovered from both the village site area and the burial mound at Safety Harbor (W illey 1949a:138; Mitchem 1989:556). According to Milanich (1994:401) Â“St. Johns Plain and Belle Glade Plain utilitarian ceramics are most common, perhaps a refle ction of ceramic transitions to the pottery assemblages of the Okeechobee Basin-Kissimm ee River region and the lake district of central Florida.Â” The variety of these artifacts m ay suggest several different periods of contact at Safety Harbor along with possible indication of ex change with other native groups in the surrounding area (Mitchem 1989:55). Sociopolitical boundaries among the inhabitants of Safety Harbor and surrounding populations also affected the degree of interaction among various prehi storic sites. According to Bullen (1969:417), the Safety Harbor ceramic complex resem bles that of western Timucua. This assumption is based on the fact that the Â“Â… Timucua made the Indian pottery found with other historical material in known Timucua territor y Â… It is also assumed that the strikingly different Glades Area pottery was m ade by the CalusaÂ” (Bullen 1969:415). This observation is important for understanding the type of interaction between these areas because, as Mitchem (1989) suggests, Safety H arbor was the town of Tocobaga, which was known to be in conflict with the Calusa. Based on the presence of Glades pottery at the Safety Harbor site it can be suggested tha t interaction occurred between Tocobaga and the Calusa either through exchange of pottery types, by the northward expansion of the Calusa power, or both. On the east coast of Florida, also representative of Glades Culture, is a sit e in Palm Beach County known as Belle Glade (8Pb40), which was inhabited by the Calusa
19 society (MacMahon and Marquardt 2004:78). This site is located near the southeast shore of Lake Okeechobee, in the physiographic subdivision of the Everglades-Lake Okeechobee Basin creating part of the northern border of the Everglades (Will ey 1949b:17). Based on the types of pottery that were recovered from the habitation site at Belle Glade, it was determined by Willey (1949b:125) that the site belonged to two cultural periods: Belle Glade I and Belle Glade II. These culture periods date from approximately 1310 B.P. to 760 B.P. (Willey 1948: 216). The types of artifacts that were used to determine the appearance of the Bel le Glade II culture are interesting as they suggest that exchange with Gul f Coast cultures transpired. According to Willey (1949b:125), Belle Glade II is characteri zed by Biscayne Check Stamped, Weeden Island and Englewood Series types of pottery. There is a lso evidence suggesting that the pottery identified as the Weeden Island series were actual trade pieces rather than imitations made by locals in the Belle Glade a rea (Willey 1949b:128). Additional evidence to support hypotheses that the Glades Cultures were interacting and exchanging with surrounding groups in south Florida is provided in the analysis of the burial mound at Belle Glade. According to Willey (1949b:128) the practice of secondary burials and partial cremation may have been adopted from Gulf Coast cultures as well as from St. Johns cultures. In addition, St. Johns-series pot tery (Biscayne pottery series) along with the previously mentioned Weeden Islan d-series pottery was discovered in the burial mound (Willey 1949b:128).
20 As mentioned previously, Belle Glade was populated by members of the Calusa society. Knowing that trade, exchange, and tribute was valued by the Calusa, i t may be suggested that there was large degree of interaction among Belle Glade popul ations and other groups in south Florida. According to MacMahon and Marquardt (2004:80-81) the Calusa often participated in exchange at various political and social occa sions, including diplomatic conferences, rituals, marriage and ceremonies of alliance. Th e large variety of occasions which incorporate trade and exchange imply that many networks of inter action existed among the Calusa society and other culture groups across large regions of south Florida. Also located in the Glades Region is HorrÂ’s Island (8Cr41), which, according to the Florida Master Site File (Florida Division of Historical Resources 2009) has components of the Glades 2 and Glades 3 cultures. Dated from 1260 B.P. to 497 B.P. (Milanich 1994:301), the Glades cultures were known for their interactions with the Be lle Glade and Caloosahatchee cultures. Not only did these cultures demonstrate simil arities in their hunting, fishing, and foraging practices, but there is also strong evide nce of social interaction among the groups. Specific evidence comes from different mixtur es of pottery types found at archaeological sites which spread across south Florida. According to Milanich (1998:113), Â“the people of these cultures exchanged ideas and traded with one another as well as with cultures farther north. They were well aware of their social and natural surroundings.Â” While the site at HorrÂ’s Island was excavated by M. Sterling in 1931 and 1933 (Russo 2009, personal communication) it appears that analysis of burial practices and
21 material culture are lacking in published literature. Therefore, suggestions for social interaction are very limited to general patterns of interaction observed for Gl ades cultures. Research for HorrÂ’s Island is still ongoing and will contribute to f uture research regarding phylogenetic relationships among prehistoric populations in Florida. Slightly farther north from HorrÂ’s Island on the west coast of Florida, Cap tiva Island (8Ll57), located in Lee County, was occupied during the Caloosahatchee II per iod 1310 B.P. and 810 B.P. (Hutchinson 2004:28). The Caloosahatchee region was the historical territory for the Calusa (Hutchinson 2004:22-23). The Calusa were known f or their interactions across south Florida, including their inter-personal confli cts with the Timucua and Tocobaga to the north. According to Widmer (1988:7), warfare was an unremitting practice of the Calusa. The historic record of the Calusa in add ition to the variety of pottery and artifact types found at Captiva Island provide reason t o believe that inhabitants of Caloosahatchee sites were actively interacting with surr ounding sites in south Florida. Specific material culture used to support evidence of interaction was often found in the sand burial mounds at Captiva Island. Wakulla Check Stamped, St. Johns Check Stamped, Weeden Island, and Safety Harbor sherds and vessels were often present in the burials, signifying use over many generations (Milanich 1994:227). Captiva Isl and, as classified by the Florida Master Site File, is considered to have Weeden Island II, Safety Harbor, and Glades cultural components. Similar to other sites that have been discuss ed, Captiva Island contains evidence that trade and exchange occurred with site s from other culture regions in Florida. This observation further supports the idea that prehistoric
22 populations were not isolated from each other and rather experienced a large degre e of social interaction. On the east coast of Florida, Fuller Mound A (8Br90), a site located about one mile south of the town of Artesia in Brevard County, dates to the Malabar II (1010 B.P. to 247 B.P.) period based on the classification of pottery recovered from the sit e (Willey 1954:82, 86). The Malabar II period, as described by Rouse (1951:251-256), was determined based on 57 sites in which patterns of ecology, habitation, burials, food, pottery and other forms of material culture were discovered and analyzed. At this site, a variety of artifacts from different culture regions wer e discovered. These include St. Johns Plain, St. Johns Check Stamped, St. Johns Simple Stamped, Glades Plain, and Belle Glade Plain sherds (Rouse 1951:196). The diversity of artifa cts from various regions around Florida provides strong evidence that the inhabitants who occupied the areas surrounding Fuller Mound A interacted and traded with other groups. According to Rouse (1951:197), Â“from the number and variety of European implements and ornaments, it is inferred that these were obtained during the Period of Friendship [ Â…] a date in the late sixteenth and early seventeenth centuriesÂ”. This Period of Fri endship occurred during the latter three to four centuries of occupation at Fuller Mound A. Additionally, the mortuary practice of building sand mounds and interring human remains either in extended or semi-flexed positions (Willey 1954:83-84) is consis tent with burial practices observed in the Glades region to the south and the Gulf Coast reg ion to the west. The similarity in burial practices and large variety of typed pott ery at numerous sites in South Florida may also suggest communication and exchange of ideas.
23 This may further provide probable cause to investigate different levels of ph ylogenetic relationships among prehistoric populations in Florida. The geographic location of Fuller Mound A is also important to consider when developing an investigation phylogenetic relationships among prehistoric site s in Florida. Located in the Cape Canaveral area on the east coast of Florida, Fuller Mound A was a crucial transition location between the two distinct societies in Florida at th e time. As demonstrated by Willey (1954:79-80), Fuller Mound A was positioned near the division that sets apart hunting, fishing, and gathering tribes in south Florida from the popula tions in the north who practiced agriculture. It may be reasonable to suggest, based on this observation, that human remains from Fuller Mound A will account for a large part of the genetic variation in this region of Florida due to the cross-cultural interact ions that may have occurred between these two types of groups. In contrast, however, Rouse (1951:256) states that the Malabar II culture period represented the Ais I ndians and that, based on artifact data, there is no evidence to suggest a divergence from t he previous hunting-fishing-gathering economy. If Indians at Fuller Mound A did not adopt agriculture practices from the North, their skeletal remains may consequ ently be more genetically similar to those from other sites in south Florida, rather than demonstrating a mix of northern and southern populations. Although the Windover site (8Br246) is much older than the six other sites discussed previously, it is famously known for the large amount of skeletal remains that were recovered and serves as a useful site for comparison of genetic var iation over time. According to Doran (2002:11) the site dates from approximately 8,120 B.P. to
24 approximately 6,980 B.P., which corresponds with the Early Archaic period. The site is located close to the east coast of Florida in Brevard County, the same county as Fuller Mound A. Unique to this site was the mortuary practice of burying the dead in small ponds (Doran 2002:11). This type of mortuary practice is very different from the burial practices of later prehistoric sites where most burials were re covered from mounds. This difference in burial practice may suggest distinct culture differences be tween Early Archaic populations and later prehistoric populations. Consequently, these cultural a nd temporal differences would provide a reason to suggest that there would also be significant genetic variation between Windover and more recent prehistoric populat ions. Furthermore, Windover differs from more recent sites in regards to the subsistence patterns practiced by the inhabitants of Early Archaic populations According to Doran (2002) and Goggin (1998) diet dependent on marine-based subsistence was not evident at the Early Archaic site. As stated by Doran ( 2002:10), Â“Windover and other early Florida sites reveal a subsistence orientati on focused on the abundant and diverse inland riverine, pond, and marsh resources coupled with the utilization of large and small terrestrial resources.Â” It was not until popul ations increased in complexity between the Middle Archaic and the time of Spanish contact that evide nce of marine exploitation was apparent (Doran 2002:11). This strong evidence of dietary change between Early Archaic and later prehistoric sites in Florida ma y also account for a significant amount of biological distance among populations over time. The location of the site has also presented explanations for trends in population growth based on the exploitation of water sources. According to Dickel and Doran
25 (2002:54), at a site such as Windover where a large cemetery is present, it has bee n proposed that, Â“The reduction of water resources and their availability led to r elative reductions in mobility, increased village size, and consequently the development of l arger formal cemeteries with hundreds of burials.Â” This pattern of population growth woul d support the hypothesis that a large amount of variation existed between the Windover s ite and sites that were occupied around the time of European contact. The reason for such large amounts of variation between the Early Archaic period and the contact period is that limited mobility, in conjunction with increased population sizes, would presumably suggest that the Early Archaic populations were reproducin g within their group and not genetically interacting with other groups. This patte rn ultimately limits gene interaction and results in heterogeneity among populat ions. Differences in mobility among Early Archaic and later prehistoric st ies may be attributed to warfare experienced during later periods of occupation. It may be possible that European settlers as well as native populations like the Calusa displaced native populations during their invasions, forcing them to migrate to different regions of Fl orida. In addition to the analysis of cultural materials that were recovered from these seven archaeological sites, geographical analysis of prehistoric Flori da may provide evidence that interaction was facilitated. According to Willey (1949b:17), Â“t he country is flat, averaging only 20 feet above sea level and characterized by swamps and m arshes.Â” In addition, there were no major natural geographic barriers which would prohibit populations from interacting with each other. While several of the sites discussed i n this research envelop the Everglades, archaeological evidence in and around the Evergl ades
26 suggest that mobility was not restricted for prehistoric populations (Grif fin 1974:343). There is evidence, however, that Â“[Â…] geographic mobility was more restric ted for PaleoIndian groups in Florida than elsewhere in the continental United States. This is probabl y related to the restricted land mass of the peninsula; although due to lower sea levels the state would have been much larger than it is presentlyÂ” (Daniel 1985:264-265). This ris e in sea level would result in the depletion of the Florida coastline, decreasing the ove rall area of Florida, especially compromising the width. For prehistoric peoples who were interacting with each other, the decrease in area would have been beneficial as it would have allowed easier east-west movement across the state. The geographic location of the state not only facilitates interaction for popul ations within Florida, but also for populations from different countries and continents. Florida is a peninsular state on the southeast corner of the United States, surrounded by lar ge bodies of water on three of its four boundaries. This state, as referred to by Cooke (1945:3), is natural barrier that separates the Gulf of Mexico from the Atlan tic Ocean. This direct access to water enables populations during prehistoric and post-contact periods to travel south from the north and west regions of the United States and allows populations from Central America, South America, Europe, and Africa to travel into the United States. Goggin (1940:29) presents this idea in his discussion of pottery distribution in the Glades region by suggesting that South Florida was a reasonable plac e of embarkment for travel to the Bahamas and Cuba. Additionally, direct evidence of such travel patterns can been seen during European contact when European explorers were making landfall on the Atlantic and Gulf Coasts of Florida (Mitchem 1989:53;
27 MacMahon and Marquardt 2004: Worth 2001:5-14). While it does not seem that there are genetic elements from the Caribbean in the crania of the populations discuss ed in this thesis, understanding how populations can migrate into Florida from other countries is important for validating the European admixture. As demonstrated in the literature, there is a considerable amount of evidence demonstrating cultural exchange during prehistoric times in Florida, which would ultimately increase the opportunities for gene flow among the groups. Much of this evidence consists of the presence of specific pottery types either in the buri als, or in surrounding middens, plazas, or associated structures. Additionally, a discussion of the general material evidence at the sites provided a useful demonstration of how s pecific groups interacted and with whom they interacted. It must be noted, however, that not all of the sites are contemporaneous and this limits the ability to make cross-cul tural comparisons. While cross-cultural comparisons may be limited, the investigat ion of differing patterns of cranial variation should still be executed as a mea ns for describing micro-evolutionary change over a period of time. This will be useful in demonstr ating how changes over time affected gene flow and genetic admixture among populat ions in Florida.
28 Chapter 3 Materials and Methods In the past 30 years there has been an increase in research using craniometri cs to investigate variability among and within various populations. Unfortunately, thes e advanced methods for data collection have rarely been applied to prehistoric populations in Florida. This project utilizes craniometric data as a method for investig ating microevolutionary change among the prehistoric Florida populations. This chapter outlines research materials, samples, sample selection criteria, methods for uni variate statistical analyses, and methods for multivariate and canonical statistical analyses. A database containing the raw data collected can be obtained from the Forensic Anthropolog y and Bioarchaeology Laboratory at the University of South Florida. Please cont act the author if you wish to access this data for further analysis. Research Materials Craniometric data were collected using a Microscobe-3DX digitizer and the program ThreeSkull, written by Steve Ousley (2004), from available prehistoric F lorida crania during a bioarchaeology internship at the National Museum of Natural His tory (NMNH) at the Smithsonian Institution in Washington, D.C., during the summer of 2009. The museum houses more than 30,000 sets of human skeletal remains from various locations world-wide. Some of this collection includes the famous Robert J. Terry
29 Anatomical Skeletal Collection and the Huntington collection as well as skel etal remains from prehistoric archaeological sites in Florida. Data Samples Craniometric data were collected on a sample that consisted of crania (n=223) from seven different prehistoric sites in Peninsular Florida: Windover (8Br246), P erico Island (8Ma6), Captiva Island (8Ll57), Belle Glade (8Pb40), HorrÂ’s Island ( 8Cr41), Safety Harbor (8Pi2), and Fuller Mound A (8Br90). The data for Windover were obtained from Dr. Richard Jantz and Donna Freid at the University of Tennessee, Knoxville. I collected the data for crania from all other sites at the NMNH. Skeletal remains from the six of the seven prehistoric Florida sites used in t his study were collected by a variety of researchers in the early 1900s. One sit e, Windover, excavations began at the site in 1984 by faculty from Florida State University, i ncluding Glen H. Doran. George Woodbury was responsible for collecting remains Fuller M ound A (8Br90). Marshall T. Newman collected remains at Perico Island (8Ma6), Henr y B. Collins collected remains at Captiva Island (8Ll57), and Gene M. Stirling col lected remains at Belle Glade (8Pb40). Matthew W. Stirling was responsible for collected remains at both HorrÂ’s Island (8Cr41) and Safety Harbor (8Pi2). Unfortunately the conditions of the crania are very poor. Damage to the many of the crania includes incomplete or missing facial bones, incomplete or missing calvaria, and incomplete or missing mandibles. In addition, post-cranial elements were not st ored
30 with their respective crania and are commingled. The damaged crania and com mingled post-cranial elements rendered the collections very incomplete. The skulls from the seven sites used in this study were selected based on two criteria. The first was that the skull must have been from an adult. Due to ra pid growth changes in juveniles, an accurate analysis of cranial variation would have been impossible. Specifically, the inclusion of juveniles would have potentially demonstr ated more variation among and within populations than would have been observed in samples consisting of only adults. Â“AdultÂ” classifications were determined by the f ull eruption of the third mandibular and maxillary molars (Buikstra and Ubelaker 1994:51). If a nd when the mandible and/or the maxilla were not present to examine to molar, the basilar suture was used to assess age. Â“AdultÂ” classifications were determined based on the pres ence of more than 50.0% of fusion of the basilar suture (Buikstra and Ubelaker 1994:32). The second criterion for selection was that the skull must have been more than 50.0% complete. In order to be considered more than 50.0% complete, the majority of both the face and calvarium had to be present. By ensuring near-completeness of the skull, the number of missing data was reduced and therefore significant sample s izes were present for statistical analyses. Overall there were slightly more males at most of the sites than female s (Table 3.1). Sex was estimated by scoring five non-metric features on the skull: nuchal crest, mastoid process, supra-orbital margin, supra-orbital ridge, and mental eminence (Buikstra and Ubelaker 1994:20). Sex was used to ensure that there was a representative sa mple of males and females in each population. This allowed for a more accurate analysis of
31 variation among sites. Also, sex was used in this analysis for the purpose of repla cing missing data. Each population was separated by sex so that the mean that replac ed missing values would reflect male or female and minimize the risk of skewed da ta. The full protocol of over 90 x, y, z coordinates was used to collect the coordinate data for interlandmark distances using the digitizer. Twenty-three interla ndmark distances (Table 3.2) were selected for analysis based on the inclusion of 10 Type -1 cranial landmarks (Table 3.3; Figure 3.1, Figure 3.2, and Figure 3.3). Type-1 landma rks are those landmarks that are located in discrete locations on the skull and that do not change regardless of cranial morphology. Type-2 and Type-3 landmarks are those landmarks that are dependent on the morphology of the skull and may change from individual to individual due to size or shape of the crania. Type-1 landmarks are preferred over Type-2 and Type-3 landmarks because of their reproducibility and di screte locations on the crania and measurements of coordinates are accurate to appr oximately 0.5mm (Ross and Williams 2008; OÂ’Higgins and Strand Vidarsdottir 1999:136).
32 SiteSite NumbernMalesFemales Windover8BR2466643.9% (29)56.1% (37)Perico Island8MA62740.7% (11)59.3% (16)Captiva Island8LL571241.7% (5)58.3% (7)Belle Glade8PB40 3756.8% (21)43.2% (16)Horr's Island8CR41 2055.0% (11)45.0% (9)Safety Harbor8PI21957.9% (11)42.1% (8)Fuller Mound A8BR904271.4% (30)28.6% (12)Total22356.7% (89)43.3% (68) Table 3.1 Sample Sizes by Site
33 AbbreviationDescriptionGOLGlabello-occipital lengthNOLNasion-occiptal lengthBNLBasion-nasion lengthBBHBasion-bregma heightWFBMinimum frontal breadthBPLBasion-prosthion lengthNPHNasion-prosthion heightNLHNasal heightNLBNasal breadthSSSZygomaxillary subtenseFMBBifrontal breadthNASNasio-frontal subtenseFRCFrontal chordFRSFrontal subtenseFRFFrontal fractionPACParietal chordPASParietal subtensePAFParietal fractionOCCOccipital chordOCSOccipital subtenseOCFOccipital fractionFOLForamen magnum lengthUFBRUpper facial breadth Table 3.2 List of Inter-landmark Distances
34 SideAssociated Inter-Landmark Distances 1. AlareRight/leftNLB2. BasionMidlineBNL, BBH, BPL, FOL3. BregmaMidlineBBH, FRC, FRS, FRF, PAC, PAS, PAF4. Frontomalare anteriorRight/leftFMB5. Frontomalare temporaleRight/leftUFBR6. LambdaMidlinePAC, PAS, PAF, OCC, OCS, OCF7. NasionMidlineNOL, BNL, NPH, NLH, NAS8. OpisthocranionMidlineGOL, NOL9. OpisthionMidlineOCC, OCS, OCF, FOL 10. Frontotemporale Right/leftWFB (Recreated from Kimmerle et al. 2008:55) Landmark Table 3.3 List of Type 1 Landmarks
35 Figure 3.1 Â– Frontal View of Type 1 Landmarks
36 Figure 3.2 Â– Lateral View of Type 1 Landmarks
37 Figure 3.3 Â– Basilar View of Type 1 Landmarks
38 Methods for Univariate Statistical Analysis Upon completion of data collection, Maximum Cranial Length (GOL) was subjected to univariate analyses to determine potential patterns of variation GOL was selected for two reasons: its inclusion of a Type-1 landmark and its potential abi lity to capture genetic variation as a result of gene flow. Initially, the craniometr ic data were analyzed to examine the level of homogeneity among the different populations by si te. Cranial variation attributed to sexual dimorphism was not analyzed in this study a s it was assumed that sexual dimorphism was present in all populations. To control for sample size and to reduce error, all missing values were replaced with the mean for their respective variable and sex within each population. The purpose for replacing the val ue with the mean instead of the median was to control for variation. In this type of study the median does not capture the complete range of variation and has the potential to skew the data to the right or the left. By using the mean, the standard deviation, standard err or and overall mean of a group are not altered. The replacement values simply serve as place holders so that SPSS can still perform the analyses correctly on the ra w data that was collected. Preliminary statistical analyses were performed us ing Agglomerative Cluster Analysis, Analysis of Variance (ANOVA), and TukeyÂ’s Pairwise C omparison (post-hoc test), in SPSS 18.0 for Windows. The first preliminary analysis consisted of an agglomerative cluste r analysis. If the data showed a distinction among the populations due to the absence of gene flow then the null hypothesis (H 0 ) will state that the data are able to be grouped together in six
39 distinct groups according to their likenesses. Consequently, the alternate hypot hesis (H 1 ) states that no cluster distinctions will exist among the six groups. H 0 : The seven groups are distinctively different based on craniometric data. H 1 : There is no distinct difference among the groups based on craniometric data To test this contention, a series of agglomerative hierarchical cluster analyses were performed. The objective of the agglomerative hierarchical cluster analysis is to group the individual cases into larger groups or clusters based on similarities of interlandmark distances. The cluster analysis, along with ANOVA, compared how t he individual cases are similar to and different from cases in other clusters. By demonstrating the level of significant variation, the contention that simil arities or differences resulting from gene flow among prehistoric Florida populati ons can be examined. The ANOVA was used to illustrate the degree of variance between the means of each population for Maximum Cranial Length (GOL). This variable was chos en for preliminary analysis because it is well demonstrated in the literature that cranial length is less plastic than other cranial measurements and is more likely attribute d to genetic factors (Sparks and Jantz 2002). When conducting the ANOVA, the null hypothesis (H 0 ) states that the mean of each interlandmark distance should be equal for all s ites. If the H 0 is rejected based on the ANOVA test, it should be expected that the F value (rati o of the sum of differences of central tendency divided by the average variance) wil l be greater than five and that the P value (probability of error in rejecting H 0 ) should be p .05. The
40 alternate hypothesis (H 1 ) states that crania from any two sites will be different when comparing the means of interlandmark distances. H 0 : Means of the samples are significantly similar among the groups H 1 : Means of interlandmark distances are not significantly similar for any two of the seven sites. Finally, TukeyÂ’s Pairwise Comparison (post-hoc) illustrated which populations are significantly similar and different based on the results from the ANOV A. Post-hoc tests were performed to examine the significance of difference ( p .05) for Maximum Cranial Length (GOL) when the crania from each site were compared to the crania from each of the other six sites. The null and alternate hypotheses for the post-hoc te st are the same for that of the ANOVA, which state: H 0 : means of samples are significantly similar among groups H 1 : means of samples are different between groups. When p > .05 then the null hypothesis cannot be confidently rejected and it can be assumed that the compared groups are similar. Methods for Multivariate and Canonical Statistical Analyses In addition to preliminary statistical analyses, more advanced multivari ate and canonical statistical analyses were performed to describe more accur ately the variance present in the dataset. Multivariate analyses are useful for demonstrati ng inter-group variation by comparing the means of multiple variables simultaneously. The var iables that most commonly provide significant results are those that best differenti ate the groups in the analyses. Additionally, Key and Jantz (1981:247) state that Â“canonical variat es
41 have been considered particularly useful because they parsimoniously describe inte rgroup variation and permit representation in low-dimensional space.Â” For the specific purposes of this research and the type of data that were collected, a Multiple Analysis of Variation (MANOVA) and a Principal Component Analysis (PCA) were performed to ana lyze the variance present among the seven groups in the manner of Key and Jantz 1981 and Kimmerle et al. 2008. First, the MANOVA was used to illustrate the inter-group variation among the seven populations for each variable (interlandmark distance). When conducting the MANOVA, the null hypothesis (H 0 ) states that the mean for m samples and p variables are equal. If the H 0 is rejected based on the MANOVA test, it should be expected that the F value (ratio of the sum of differences of central tendency divided by the average variance) will be greater than five and that the Sig. value (probability of error in rejecting H 0 ) should be less than .050 for all four tests (WilkÂ’s Lambda, RoyÂ’s Largest Root Tes t, PillaiÂ’s Trace Statistic, and Lawes-HotellingÂ’s Trace). All f our of these tests are automatically performed in SPSS 18.0 for Windows as part of the MANOVA. The alternate hypothesis (H 1 ) of the MANOVA states that any two populations will be different when comparing the means of interlandmark distances. H 0 : The mean for all observations within the 7 sites are equal for the 23 variables submitted for analysis. H 1 : The mean for all observations within the 7 sites are not equal for the 23 variables submitted for analysis. After completion of the MANOVA, a Principal Component Analysis (PCA) was performed in order to reduce the dimensionality of the dataset and to maximize an y
42 separation of space among the groups (Jantz 2005). The Principal Component Analysi s results in new factors or principal components that are designed to redistribute the variance of the dataset to reduce the dimensionality of the data. These unde rlying factors, however, are only useful if the original variables are highly correlated e ither positively or negatively (Manly 2005:75). Fortunately, cranial measurements, unless crania have be en culturally modified, are strongly correlated as with many other metric m easurements of the human body. According to Jantz (1973; Key and Jantz 1981), Principal Components Analysis is a common canonical analysis performed to examine microevolutionar y change among populations (Jantz 1973; Key and Jantz 1981). Additionally, the newly produced factors will be employed to examine whether the variation is attribut ed to more genetic factors like gene flow or more environmental factors that lea d to adaptation and natural selection. Due to the strictly exploratory nature of Principal Compone nt Analysis, the presentation of a null and alternate hypothesis is not necessary.
43 Chapter 4 Results Univariate Statistical Analysis The agglomerative cluster analysis was not included in the results as it did not provide any significant results with regard to similarities or differences among the seven groups, however it can be accessed in Appendix A. It is possible that the cluster a nalysis did not provide significant results because the analysis compared measurements on a n individual basis. By comparing individuals instead of the group, the true inter-group variation is not captured. Rather, inter-group variation is reflected more whic h is not the focus of this thesis. Due to the lack of clarity provided by the cluster analysis and the individual comparisons, statistical analyses that account for the amount of variation among t he populations as a whole based on the Maximum Cranial Length (GOL) were performed. An analysis of variance (ANOVA) and a TukeyÂ’s HSD Post-hoc tests indicat e that a significant amount of variation existed among the seven populations when time was controlled as the dependent variable. Table 4.1 provides general descriptive statistics for the GOL for all se ven populations. When comparing the means of GOL, it can be seen that from the earliest site (Windover) to the latest site (Fuller Mound A), there was an overall decrea se in maximum length of the crania. However, there seems to be a large amount of fluc tuation
44 throughout temporal space in maximum cranial length, ranging from 173.88 mm to 181.14 mm. To verify the observation that these populations were not homogenous, an ANOVA was performed. Table 4.2 provides the results of the ANOVA based on GOL for all seven groups. As presented in the table, the results demonstrate significant differences among the sites (F = 5.190, p 0.000). Based on the confidence level of 95% ( p .05 ) that there will be error rejecting the null hypothesis, the null hypothesi s can be confidently rejected. Overall Table 4.2 indicates relative heterogeneity a cross temporal space. To explore which populations were contributing to the overall variation demonstrated in the ANOVA, a TukeyÂ’s HSD Post-hoc test was performed. Table 4.3 presents the significance of difference between any two populations in the e ntire sample. Interestingly, when considering GOL, the only population comparisons which demonstrated variation were Windover and Perico Island ( p = .008), and Windover and Safety Harbor ( p 0.000).
45 Table 4.1 Â– Descriptive Statistics of Maximum Cranial Length (GOL) for all sites Std. SiteNMeanDeviationStd. ErrorLower BoundUpper BoundM inimum (mm)Maximum (mm) Windover66181.146.8630.845179.45182.82166195PericoIsland27177.076.7651.302174.4179.75161189Captiva Island12175.426.7351.944171.14179.7169193Belle Glade37178.497.6621.26175.93181.04163198Horr's Island20177.358.9752.007173.15181.55162199Safety Harbor19174.216.3381.454171.16177.27164185Fuller Mound A42173.888.2791.278171.3176.46148188Total223177.607.8410.525176.57178.64148199 95% Confidence Interval for Mean Table 4.2 Â– ANOVA for all Florida Archaeological Sites using GOL Sum of SquaresdfMean SquareFSig. Between Groups1719.580.6286.5975.1900.000Within Groups11927.9021655.222Total13647.48222
46 Table 4.3 Â– TukeyÂ’s HSD Post-hoc test Describing Significance of Differe nce between sites for Maximum Cranial Length (GOL) (I) Site(J) SiteMean Difference (I-J)Std. ErrorSig.Lower BoundUpper Bound WindoverPerico Island4.0621.6980.207-0.999.12 Captiva Island5.7202.3320.182-1.2212.66Belle Glade0.2651.5260.592-1.897.19Horr's Island3.7861.8970.420-1.869.43Safety Harbor6.926*1.9350.0081.1712.68Fuller Mound A7.255*1.4670.0002.8911.62 Perico IslandWindover-4.0621.6980.207-9.120.99 Captiva Island1.6572.5780.995-6.029.33Belle Glade-1.4121.8810.989-7.014.19Horr's Island-0.2762.1921.000-6.806.25Safety Harbor2.8642.2250.857-3.769.49Fuller Mound A3.1931.8330.589-2.268.65 Captiva IslandWindover-5.7202.3320.182-12.661.22 Perico Island-1.6572.5780.995-9.336.02Belle Glade-3.0702.4690.876-10.424.28Horr's Island-1.9332.7130.992-10.016.14Safety Harbor1.2062.7400.999-6.959.36Fuller Mound A1.5362.4320.996-5.708.78 Belle GladeWindover-2.6501.5260.592-7.191.89 Perico Island1.4121.8810.989-4.197.01Captiva Island3.0702.4690.876-4.2810.42Horr's Island1.1362.0620.998-5.007.27Safety Harbor4.2762.0970.393-1.9710.52Fuller Mound A4.6061.6750.091-0.389.59 Horr's IslandWindover-3.7861.8970.420-9.431.86 Perico Island0.2762.1921.000-6.256.80Captiva Island1.9332.7130.992-6.1410.01Belle Glade-1.1362.0620.998-7.275.00Safety Harbor3.1392.3810.843-3.9510.23Fuller Mound A3.4692.0190.605-2.549.48 Safety HarborWindover-6.926*1.9350.008-12.68-1.17 Perico Island-2.8642.2250.857-9.493.76Captiva Island-1.2062.7400.999-9.366.95Belle Glade-4.2762.0970.393-10.521.97Horr's Island-3.1392.3810.843-10.233.95Fuller Mound A0.3302.0551.000-5.796.44 Fuller Mound AWindover-7.255*1.4670.000-11.62-2.89 Perico Island3.1931.8330.589-8.652.26Captiva Island-1.5362.4320.996-8.785.70Belle Glade-4.6061.6750.091-9.590.38Horr's Island-3.4692.0190.605-9.482.54Safety Harbor-0.3302.0551.000-6.445.79 The mean difference is significant at the 0.05 le vel 95% Confidence Interval
47 Multivariate and Canonical Statistical Analyses A series of Multiple Analysis of Variance (MANOVA) tests were conduc ted to examine the degree of variation among the seven prehistoric populations on a multivariate level. By using multiple variables in the analysis, the variat ion was more accurately described. Specifically, different types of variables such as length, width, and height were used simultaneously which allowed the variation to be explained in regard t o both the size and the shape of the crania. A total of twelve MANOVA tests were performed comparing crania from se ven sites in a variety of temporal relationships. Table 4.5 demonstrates that whe n crania from all seven sites were analyzed for variation, significant differences e xisted among the crania based on interlandmark distances ( p 0.000). To investigate which populations were contributing to the variation, two additional series of MANOVA tests wer e performed. The first series of tests analyzed the variation between crani a from Windover (the earliest population) and each of the other six populations (Refer to Tables 4.6, 4.7, 4.9, 4.11, 4.13, 4.15, and 4.17). The second series of tests analyzed the variation successively, in chronological order beginning with Windover and ending with Fuller Mound A (Refer to Tables 4.19, 4.21, 4.23, 4.25, 4.27). The results of the MANOVA test performed comparing the Windover population to all other sites suggest that the Windover population was significantly differe nt in terms of cranial size and shape from all other populations. This comes as no surprise considering that the Windover population was at least 4470 years older than the other populations. In all cases, the F value was greater than 5 and p 0.000. As presented in
48 Table 4.7, Windover and Perico Island were significantly different from each other (F = 16.445, p 0.000). Figure 4.2 provides a visual representation of the variation between these two sites based on the first and second principal components. Table 4.9 demonstrates that Windover and Captiva Island were significantly different fr om each other (F = 8.011, p 0.000). Figure 4.3 demonstrates that when the first two principal components for Windover and Captiva Island were plotted, significant differences wer e present. Table 4.11 shows that Windover and Belle Glade were significantly diffe rent from each other (F = 15.807, p 0.000). Figure 4.4 further demonstrates that Windover and Belle Glade were significantly different based on the principal components one and two. As presented in Table 4.13, Windover and HorrÂ’s Island demonstrated significant differences (F = 6.877, p 0.000). Figure 4.5 represents the differences between Windover and HorrÂ’s Island when principal components one and two were considered. Table 4.15 demonstrates that Windover was significantly different from Safety Harbor (F = 17.223, p 0.000). Figure 4.6 demonstrates the significant differences between Windover and Safety Harbor based on the first two principal components. Last for this series, Table 4.17 demonstrates that significant differences were pre sent between Windover and Fuller Mound A (F = 13.059, p 0.000). Figure 4.7 represents the differences between Windover and Fuller Mound A when considering principal components one and two. These results suggests that the significant temporal separ ation between Windover and the other six sites contributed to significant phenotypic differences between Early Archaic populations and groups who lived closer t o the time of European contact.
49 Interestingly, when the MANOVA tests were performed successively in chronological order, slightly different results were produced than that of the fir st series of MANOVA tests performed. Table 4.6 demonstrated that significant differences occurred between Windover and Perico Island. When Perico Island and Captiva Island are subjected to a MANOVA (Table 4.19), the results indicated that there was not a significant difference (F = 1.789, p = 0.123) in cranial size and shape between these two populations. In addition, these results demonstrate that the significant amount of ti me passed between the time of Windover occupation (8120 B.P.to 6980 B.P.) and Perico Island occupation (2510 B.P. to 1210 B.P.) contributed to a significant amount of variation between the two populations. The next two MANOVA tests performed demonstrate the opposite of that for Windover and Perico Island, which show that significant differences among Captiva Island and Belle Glade as see n in Table 4.21 (F = 1.031, p 0.000) and Belle Glade and HorrÂ’s Island as seen in Table 4.23 (F = 1.421, p 0.174) were not present. Figure 4.9 and Figure 4.10 further support that there were no significant difference between Captiva Island and Belle Glade and betw een Belle Glade and HorrÂ’s Island, respectively. These results indicate that relative homoge neity was present among sites that were occupied from 2510 B.P. to 497 B.P. The last two MANOVA tests suggested that significant differences were present among sites that were occupied from 1260 B.P. to 247 B.P.. Table 4.25 demonstrates that HorrÂ’s Island and Safety Harbor were significantly different ( F = 8.151, p 0.000). Similarly, Table 4.27 indicates that significant differences were prese nt between Safety Harbor and Fuller Mound A (F = 5.549, p 0.000). In Figure 4.11 and Figure 4.12 there
50 are no clear distinctions between HorrÂ’s Island and Safety Harbor and bet ween Safety Harbor and Fuller Mound A, respectively. However, in both figures, the individual elements which were plotted appear to be very spread apart, a pattern very dif ferent from the populations which demonstrated no significant differences. This pattern of vari ation suggests that sometime between 1260 B.P. and 260 B.P., significant events occurred that would have altered the genetic variation among these populations which contributed to significant differences in phenotypic variation.
51 Table 4.4 Â– Sample Size for each Florida Archaeological Site SiteValue LabelN 1Windover662Perico Island273Captiva Island124Belle Glade375Horr's Island206Safety Harbor197Fuller Mound A42 Table 4.5 Â– MANOVA for all Florida Archaeological Sites Investigated EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace0.99912322.555a23.000194.0000.000 Wilk's Lambda0.00112322.555a23.000194.0000.000 Hotelling's Trace1460.92112322.555a23.000194.0000.000 Roy's Largest Root1460.92112322.555a23.000194.0000.000 SitePillai's Trace1.9984.321138.0001194.0000.000 Wilk's Lambda0.0585.204138.0001139.0070.000Hotelling's Trace4.646.467138.0001154.0000.000Roy's Largest Root2.74623.758b23.000199.0000.000 a. Exact Statisticb. The statistic is an upper bound on F that yields a lower bound on the significance level
52 Figure 4.1 Â– Scatter Plot of Component 1 and Component 2 for all Archaeological Sites
53 Table 4.6 Â– Sample Sizes for Windover and Perico Island SiteValue LabelN 1Windover662Perico Island27 Table 4.7 Â– MANOVA for Windover and Perico Island EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace1.0006377.266a23.00069.0000.000 Wilk's Lambda0.0006377.266a23.00069.0000.000 Hotelling's Trace2125.7556377.266a23.00069.0000.000 Roy's Largest Root2125.7556377.266a23.00069.0000.000 SitePillai's Trace0.84616.445a23.00069.0000.000 Wilk's Lambda0.15416.445a23.00069.0000.000 Hotelling's Trace5.48216.445a23.00069.0000.000 Roy's Largest Root5.48216.445a23.00069.0000.000 a. Exact Statistic
54 Figure 4.2 Â– Scatter Plot of Component 1 and Component 2 for Windover and Perico Island
55 Table 4.8 Â– Sample Sizes for Windover and Captiva Island SiteValue LabelN 1Windover663Captiva Island12 Table 4.9 Â– MANOVA for Windover and Captiva Island EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace0.9993174.147a23.00054.0000.000 Wilk's Lambda0.0013174.147a23.00054.0000.000 Hotelling's Trace1351.9513174.147a23.00054.0000.000 Roy's Largest Root1351.9513174.147a23.00054.0000.000 SitePillai's Trace0.7738.011a23.00054.0000.000 Wilk's Lambda0.2278.011a23.00054.0000.000 Hotelling's Trace3.4128.011a23.00054.0000.000 Roy's Largest Root3.4128.011a23.00054.0000.000 a. Exact Statistic
56 Figure 4.3 Â– Scatter Plot of Component 1 and Component 2 for Windover and Captiva Island
57 Table 4.10 Â– Sample Sizes for Windover and Belle Glade SiteValue LabelN 1Windover664Belle Glade37 Table 4.11 Â– MANOVA for Windover and Belle Glade EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace1.0007652.893a23.00079.0000.000 Wilk's Lambda0.0007652.893a23.00079.0000.000 Hotelling's Trace2228.0587652.893a23.00079.0000.000 Roy's Largest Root2228.0587652.893a23.00079.0000.000 SitePillai's Trace0.81915.507a23.00079.0000.000 Wilk's Lambda0.18115.507a23.00079.0000.000 Hotelling's Trace4.51515.507a23.00079.0000.000 Roy's Largest Root4.51515.507a23.00079.0000.000 a. Exact Statistic
58 Figure 4.4 Â– Scatter Plot of Component 1 and Component 2 for Windover and Belle Glade
59 Table 4.12 Â– Sample Sizes for Windover and HorrÂ’s Island SiteValue LabelN 1Windover665Horr's Island20 Table 4.13 Â– MANOVA for Windover and HorrÂ’s Island EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace0.9994988.335a23.00062.0000.000 Wilk's Lambda0.0014988.335a23.00062.0000.000 Hotelling's Trace1850.5124988.335a23.00062.0000.000 Roy's Largest Root1850.5124988.335a23.00062.0000.000 SitePillai's Trace0.7186.877a23.00062.0000.000 Wilk's Lambda0.2826.877a23.00062.0000.000 Hotelling's Trace2.5516.877a23.00062.0000.000 Roy's Largest Root2.5516.877a23.00062.0000.000 a. Exact Statistic
60 Figure 4.5 Â– Scatter Plot of Component 1 and Component 2 for Windover and HorrÂ’s Island
61 Table 4.14 Â– Sample Sizes for Windover and Safety Harbor SiteValue LabelN 1Windover666Safety Harbor19 Table 4.15 Â– MANOVA for Windover and Safety Harbor EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace0.9995262.292a23.00061.0000.000 Wilk's Lambda0.0015262.292a23.00061.0000.000 Hotelling's Trace1984.1435262.292a23.00061.0000.000 Roy's Largest Root1984.1435262.292a23.00061.0000.000 SitePillai's Trace0.86717.223a23.00061.0000.000 Wilk's Lambda0.13317.223a23.00061.0000.000 Hotelling's Trace6.49417.223a23.00061.0000.000 Roy's Largest Root6.49417.223a23.00061.0000.000 a. Exact Statistic
62 Figure 4.6 Â– Scatter Plot of Component 1 and Component 2 for Windover and Safety Harbor
63 Table 4.16 Â– Sample Sizes for Windover and Fuller Mound A SiteValue LabelN 1Windover667Fuller Mound A42 Table 4.17 Â– MANOVA for Windover and Fuller Mound A EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace0.9997287.662a23.00084.0000.000 Wilk's Lambda0.0017287.662a23.00084.0000.000 Hotelling's Trace1995.4317287.662a23.00084.0000.000 Roy's Largest Root1995.4317287.662a23.00084.0000.000 SitePillai's Trace7.81013.059a23.00084.0000.000 Wilk's Lambda0.21913.059a23.00084.0000.000 Hotelling's Trace3.57613.059a23.00084.0000.000 Roy's Largest Root3.57613.059a23.00084.0000.000 a. Exact Statistic
64 Figure 4.7 Â– Scatter Plot of Component 1 and Component 2 for Windover and Fuller Mound A
65 Table 4.18 Â– Sample Sizes for Perico Island and Captiva Island SiteValue LabelN 2Perico Island273Captiva Island12 Table 4.19 Â– MANOVA for Perico Island and Captiva Island EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace1.0002500.412a23.00015.0000.000 Wilk's Lambda0.0002500.412a23.00015.0000.000 Hotelling's Trace3833.9672500.412a23.00015.0000.000 Roy's Largest Root3833.9662500.412a23.00015.0000.000 SitePillai's Trace0.7331.789a23.00015.0000.123 Wilk's Lambda0.2671.789a23.00015.0000.123 Hotelling's Trace2.7431.789a23.00015.0000.123 Roy's Largest Root2.7431.789a23.00015.0000.123 a. Exact Statistic
66 Figure 4.8 Â– Scatter Plot of Component 1 and Component 2 for Perico Island and Captiva Island
67 Table 4.20 Â– Sample Sizes for Captiva Island and Belle Glade SiteValue LabelN 3Captiva Island124Belle Glade37 Table 4.21 Â– MANOVA for Captiva Island and Belle Glade EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace0.9992024.026a23.00025.0000.000 Wilk's Lambda0.0012024.026a23.00025.0000.000 Hotelling's Trace1862.1042024.026a23.00025.0000.000 Roy's Largest Root1862.1042024.026a23.00025.0000.000 SitePillai's Trace0.4871.031a23.00025.0000.468 Wilk's Lambda0.5131.031a23.00025.0000.468 Hotelling's Trace0.9481.031a23.00025.0000.468 Roy's Largest Root0.9481.031a23.00025.0000.468 a. Exact Statistic
68 Figure 4.9 Â– Scatter Plot of Component 1 and Component 2 for Captiva Island and Belle Glade
69 Table 4.22 Â– Sample Sizes for Belle Glade and HorrÂ’s Island SiteValue LabelN 4Belle Glade375Horr's Island20 Table 4.23 Â– MANOVA for Belle Glade and HorrÂ’s Island EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace1.0003533.575a23.00033.0000.000 Wilk's Lambda0.0003533.575a23.00033.0000.000 Hotelling's Trace2462.7943533.575a23.00033.0000.000 Roy's Largest Root2462.7943533.575a23.00033.0000.000 SitePillai's Trace0.4981.421a23.00033.0000.174 Wilk's Lambda0.5021.421a23.00033.0000.174 Hotelling's Trace0.9901.421a23.00033.0000.174 Roy's Largest Root0.9901.421a23.00033.0000.174 a. Exact Statistic
70 Figure 4.10 Â– Scatter Plot of Component 1 and Component 2 for Belle Glade and HorrÂ’s Island
71 Table 4.24 Â– Sample Sizes for HorrÂ’s Island and Safety Harbor SiteValue LabelN 5Horr's Island206Safety Harbor19 Table 4.25 Â– MANOVA for HorrÂ’s Island and Safety Harbor EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace1.0003390.572a23.00015.0000.000 Wilk's Lambda0.0003390.572a23.00015.0000.000 Hotelling's Trace5198.8763390.572a23.00015.0000.000 Roy's Largest Root5198.8763390.572a23.00015.0000.000 SitePillai's Trace0.9268.151a23.00015.0000.000 Wilk's Lambda0.0748.151a23.00015.0000.000 Hotelling's Trace12.4998.151a23.00015.0000.000 Roy's Largest Root12.4998.151a23.00015.0000.000 a. Exact Statistic
72 Figure 4.11 Â– Scatter Plot of Component 1 and Component 2 for HorrÂ’s Island and Safety Harbor
73 Table 4.26 Â– Sample Sizes for Safety Harbor and Fuller Mound A SiteValue LabelN 6Safety Harbor197Fuller Mound A42 Table 4.27 Â– MANOVA for Safety Harbor and Fuller Mound A EffectValueFHypothesis dfError dfSig.InterceptPillai's Trace1.0004235.122a23.00037.0000.000 Wilk's Lambda0.0004235.122a23.00037.0000.000 Hotelling's Trace2632.6434235.122a23.00037.0000.000 Roy's Largest Root2632.6434235.122a23.00037.0000.000 SitePillai's Trace0.7755.549a23.00037.0000.000 Wilk's Lambda0.2255.549a23.00037.0000.000 Hotelling's Trace3.4505.549a23.00037.0000.000 Roy's Largest Root3.4505.549a23.00037.0000.000 a. Exact Statistic
74 Figure 4.12 Â– Scatter Plot of Component 1 and Component 2 for Safety Harbor and Fuller Mound A
75 Following the MANOVA tests, Principal Component Analysis (PCA) was performed to investigate which variables contributed the most amount of variation to t he different populations. Before the results of the PCA were analyzed, KMO and Bar tlettÂ’s tests (Table 4.28) were evaluated to ensure that the PCA was a good model for explai ning the variation of this particular dataset. The results of the KMO test (0.758) indica te that the sample size is adequate for Principal Component Analysis and the BartlettÂ’ s test ( p 0.000) confidently rejects the null hypothesis which states that the correlation ma trix is an identity matrix. Based on these standards, it can be assumed that PCA is an adequa te model for this particular dataset. As presented in Table 4.29, approximately 28.04% of the total variation can be explained by the first eigenvector. Eigenvectors 2-6 contribute an additional 38.95% of the variation for a total of 66.99 % of all variation. While these results are not optima l, they do suggest that economy can be reached by reducing the original twenty-thre e variables to only six new principal components. As can be seen in Table 4.30, GOL, NOL, BNL, BBH, WFB, BPL, FMB, FRC, and UFBR contribute the most amount of variation in the first eigenvector. Eigenvectors 2 and 3 have significantly less variables contributing to variation, and eigenvectors 4, 5, and 6 have virtually no variables contributing a significant amount of variation. It should be noted that the variables that are most significant for contributing variation are associated with the len gth and breadth of the crania.
76 Table 4.28 Â– KMO and BartlettÂ’s Test for all Archaeological Sites Investigate d 0.758 Bartlett's Test of Sphericity3438.748 df253.000Sig.0.000 Kaiser-Meyer-Olkin Measure of Sampling Adequacy Approx. Chi Square Table 4.29 Â– Total Amount of Variance Explained by Each Component Total% of VarianceCumulative % 16.45028.04428.04422.91012.65140.69532.2389.73050.42541.5356.67257.09651.2405.39362.48961.0364.50266.99270.9173.98870.98080.8683.77674.75690.8333.62178.377 100.7643.32181.698110.6782.94884.646120.6252.71987.365130.5522.40289.767140.5082.21091.977150.4571.98893.965160.3581.55795.522170.2921.26996.792180.2721.18497.976190.1640.71198.687200.1400.61199.298210.0750.32899.626220.7400.32199.947230.0120.530100.000 Initial Eigenvalues Extraction Method: Principal Components Component
77 Table 4.30 Â– Component Matrix for Components with Eigenvalues Greater Than 1 123456 GOL0.7740.3340.395-0.034-0.113-0.162NOL0.7710.3000.412-0.039-0.108-0.168BNL0.7800.056-0.1390.1260.263-0.149BBH0.715-0.007-0.091-0.1050.4180.121WFB0.636-0.085-0.157-0.206-0.3990.058BPL0.5690.015-0.0970.2330.1510.078NPH0.353-0.537-0.2010.2950.122-0.086NLH0.466-0.312-0.2300.2650.137-0.362NLB0.456-0.265-0.0680.075-0.1490.320SSS0.3510.3530.0180.4360.1640.042FMB0.750-0.282-0.279-0.043-0.3560.066NAS0.491-0.375-0.2660.036-0.256-0.004FRC0.702-0.2020.011-0.4520.283-0.010FRS0.1240.0330.090-0.8310.063-0.167FRF0.468-0.0460.021-0.1050.4470.353PAC0.4130.829-0.017-0.023-0.0700.085PAS0.1330.782-0.3900.032-0.0210.157PAF0.3210.481-0.0600.046-0.0460.311OCC0.416-0.4320.612-0.0280.1530.151OCS0.2220.0460.8400.100-0.236-0.070OCF0.011-0.2760.5040.244-0.0930.422FOL0.3490.2040.1710.2670.071-0.449UFBR0.756-0.126-0.276-0.019-0.3620.008 Component
78 Chapter 5 Discussion In this study, microevolutionary changes interpreted from human variation among seven prehistoric archaeological sites were assessed using crani ometric data submitted to univariate and multivariate statistical analyses. According to Molnar ( 2002:187), Â“representative skulls from [Â…] extinct ancient populations often possess a uni que combination of characteristics that have been used to suggest relationships betwee n past and present populations. Ancestral origins, migration routes, and ethnic identity have been postulated.Â” This chapter provides an in-depth discussion of the results of the stud y and how they can be interpreted as different forces of microevolutionary change and how that may be used to suggest specific relationships between populations of differe nt occupation periods. This chapter covers specific topics including which archaeolo gical populations were used for the analyses and an interpretation of the statistica l analyses used to investigate human variation. The discussion of human variation is further divided into three categories: variation among populations of different occupation peri ods, variation among contemporaneous populations, and variation of genetic admixture among populations living at the time of European contact.
79 Archaeological Populations Used for Analyses The seven populations used in this study represented the following sites: Windover (8Br246), Perico (Island 8Ma6), Captiva Island (8Ll57), Belle Glade (8P b40), HorrÂ’s Island (8Cr41), Safety Harbor (8Pi2), and Fuller Mound A (8Br90). These sites were selected according to their availability for data collection at t he NMNH in Washington, D.C. In addition, these sites were selected for their sample siz es of crania that met the two criteria established for data collection (refer to Chapte r 3). The geographic location of the sites range east to west from the Atlantic O cean to the Gulf of Mexico and north to south from Brevard County to Palm Beach County; covering almost the entire southern half of the state of Florida. The occupation periods of the populat ions range from as early as 8120 B.P to as late as 260 B.P.; approximately a 7860 year span. Interpretation of Statistical Analyses to Illustrate Variation The statistical analyses consisted of univariate, multivariate, and canonica l methods designed to examine variation among populations using interlandmark distances; and were used to determine if: (1) significant variation existe d among populations of varying temporal relationships; (2) variation decreased as populati ons become more contemporaneous; and (3) variation increased due to changes in ge ne flow among diverse populations as a direct result of European contact. Overall the a nalyses revealed that variation was significant among the groups under two different c onditions: (1) when populations occupied sites at significantly different time periods; and (2) when populations occupied sites during periods of European invasion. As a result,
80 microevolutionary forces contributed to significant cranial variation among the populations over an extended period of time. In addition, through Principal Components Analysis, I was able to differentiate the types of microevolutionary forc es acting on the populations depending on which interlandmark distances provided the most significant contribution to overall variation. Variation among populations of different occupation periods. When assessing human variation among populations that differ with regard to the occupation period, it i s important to examine the interaction between environmental stressors (i.e. a ccess to resources, nutrition, disease) and genetic variation over time. There has been a variety of studies in which researchers argue that crania can exhibit significa nt changes within a single generation due to their high plasticity and the influence of environmental stressors (Boas 1912; Molnar 2002; Gravlee et al. 2003; Relethford 2004). However, in a study that revisited the work of Franz Boas, Sparks and Jantz (2002:14637) argued that, Â“Reanalysis of BoasÂ’ data not only fails to support his contention that cranial plasticity is a primary source of cranial variation but rather supports wha t morphologists and morphometricians have known for a long time: most of the variation is genetic variation.Â” In addition, Sparks and Jantz (2002:14637) argued that facial breadth was greatly influenced by environmental factors and has a lower heritability than length a nd height. Therefore, this indicates that if variation existed and that length and heig ht of the face contributed the largest percentage of variation, then it could be assumed that genetic variation was the main contributor to the observed variation among groups.
81 In another study that revisited BoasÂ’s data, Gravlee et al. (2003) argue that their findings corroborated BoasÂ’ work and that cranial measurements were shown t o be highly plastic. Interestingly, their conclusions were based on the relationship betwee n the cephalic index, a measure of the ration between the breadth and length of the crania and temporal distance separating the mother and her offspring. Based on the analysis by Sparks and Jantz (2002) there are several problems with using this measurement. As argued by Sparks and Jantz (2002), the length of the skull, one of the two measurements used in BoasÂ’s cephalic index, is greatly influence d by heritability than breadth of the skull. However, facial breadth, the other component of the cephalic index is more likely influenced by the environment. This ratio creat es a conflict in that it uses two different measurements that appear to be influenc ed by different components. With that said, according to the argument presented by Spar ks and Jantz (2002), the re-analysis of BoasÂ’s data by Gravlee et al. (2003) does not indisputably support cranial plasticity as a significant source of cranial variation. Despite the fact that there appears to be a strong debate between genetic and environmental influence on cranial plasticity, recent studies have demonstrate d that the statistical models used for data analysis have a greater influence on how va riation is interpreted (Holloway 2002; Relethford 2004). For example, Sparks and Jantz (2002) used age specific ttests which were standardized by sex (very similar to the methodology used in this thesis) whereas Gravlee et al. (2003) used age as a covariate when comparing U.S.-born to foreign-born individuals (Relethford 2004:380). After revisiting the two studies, Relethford argues that similar results were achieved despite di fferences in
82 methodology and that Â“developmental plasticity does not alter the major sources of differences among the ethnic groups, but does affect the fine detail of differ ences within the two major clustersÂ” (Relethford 2004:381). In other words, Relethford has been abl e to demonstrate that neither genetics nor environment exclusively controls cra nial variation Â– they always interact with one another. After performing a series of multivariate statistical analyses, s ignificant variation was observed between Windover and Perico Island (Table 4.7), Windover and Captiva Island (Table 4.9), Windover and Belle Glade (Table 4.11), Windover and HorrÂ’s Island (Table 4.13), Windover and Safety Harbor (table 4.15), and Windover and Fuller Mound A (Table 4.17). In all comparisons, it is unlikely that geographic distance cont ributed to higher levels of variation among these groups. This contention is supported by the Principal Components Analysis that was performed. In the first two eigenvect ors which accounted for 40.70% of all variation (Table 4.29), all variables contributing sig nificant amount of variation were associated with either cranial length or cranial br eadth (Table 4.30). Therefore, the results of this study are consistent with the findings of Spar ks and Jantz (2002) in their craniometric analysis. Because facial breadth did not c ontribute a significant amount of variation, it can be reasoned that the observed cranial variation among prehistoric Florida populations does not completely reflect environment al factors such as geographic distance or changes in environmental conditions over time. Consequently, it can be suggested that microevolutionary forces such as migrat ion and gene flow were significant contributing factors to the observed genetic vari ation between the occupation periods of Windover and the other sites.
83 With regard to changes over time that could have contributed to cranial variati on, Owsley et al. (1982:182) reported that Â“the differences in cranial morphology could be accounted for if the burial areas were used by temporally distinct populations.Â” As noted in Table 3.1, Windover was occupied between 8120 B.P. and 6980 B.P. (Doran 2002:11) whereas Perico Island was occupied between 2510 B.P. to 1210 B.P. (Luer and Almy 1982), creating a minimum of 4470 years between occupation of these two sites. A tim e gap of this magnitude is crucial for understanding why significant differen ces in cranial morphology were present. Over the course of 4470 years significant changes in soc ial relationships, mating patterns, and environmental stressors can occur that would influence genetic admixture among prehistoric sites. Changes in genetic admixture could potentially result in significantly different morphological features of t he crania. Although such changes in social relationships, mating patterns, nutrition, disease and other environmental stressors may over time become the largest contributor s of variation between the Windover populations and the other archaeological populati ons, there is evidence of specific changes during Windover occupation that may furt her explain the variation. Archaeological evidence of reduced mobility and diet at Wi ndover provides support for significant differences between Windover and later sites. Archaeological evidence at Windover has demonstrated that inhabitants were significantly less mobile than later populations. According to Dickel and Doran ( 2002), the Windover population was extremely limited in movement due to reduced water sources. As a result, the Windover population grew in size as reproduction continued
84 within the group. The limited mobility led to an increase in heterogeneity betwee n Windover and other sites as a result of reduced genetic interaction between gr oups. Windover also differed from later sites with regard to the subsistence pat terns practiced. According to Doran (2002:10) subsistence patterns at Windover were bas ed on diverse inland riverine, pond, and marsh resources. It was not until populations increased in complexity between the Middle Archaic and the time of Spanish contac t that evidence of marine exploitation became apparent (Doran 2002:11). The introduction of marine life to the everyday diet may potentially have affected the growt h and development of individuals at all stages of life. By introducing a diet much higher in protein and more diverse in nutrients than Early Archaic diets, cranial size may have increased around the time of Spanish contact as a result of stimulated growth and increased nutritional health. Variation among contemporaneous populations. When investigating chronological variation among prehistoric Florida populations, interesting patt erns were observed. As discussed previously, there were significant differences between W indover and Perico Island (Table 4.7). This variation was likely due changes in gene flow that occurred during the approximate 4000 years between occupation periods for these two sites. Interestingly, as populations became more contemporaneous in occupati on, the level of variation decreased significantly and with the exception of two pairs of populations, all populations that occupied sites between 2510 B.P. and 247 B.P. experienced increased homogeneity and decreased variation. Specificall y, Perico Island
85 and Captiva Island (Table 4.19), Captiva Island and Belle Glade (Table 4.21), and Bell e Glade and HorrÂ’s Island (Table 4.23) did not demonstrate any significant varia tion. While still occupied during the same relative time frame as Captiva Isla nd and Belle Glade, HorrÂ’s Island and Safety Harbor (Table 4.25) and Safety Harbor and F uller Mound A (Table 4.27) demonstrated significant variation between populations. The increase in homogeneity among the populations can be supported by archaeological evidence suggesting a large degree of social interacti on among prehistoric Florida populations between 2510 B.P. and 497 B.P. With a large degree of social interaction among populations, groups are less likely to reproduce within the group. By reducing reproduction within the group and increasing gene flow among the groups, homogeneity is increased. Specific evidence of social interaction among prehistoric Florida populations includes a wide distribution of pottery types across south Florida, t he adoption of similar burial practices, and the identification of culture groups that were known for their social interactions in south Florida. For example, the various types of pottery that were discovered in the shell middens at Perico Island, based on the temper and decoration, were classified as Glades Plain, the Perico Series, Biscayne Series, Deptford Series, and other misc ellaneous types (Willey 1949a). Interestingly, these same types of pottery are seen at va rious sites around south Florida including Safety Harbor, Belle Glade, and Fuller Mound A. At Captiva Island, a site occupied by the Calusa, various types of pottery were discovered includi ng Wakulla Check Stamped, St. Johns Check Stamp sherds, and Weeden Island vessels (Milanich 1994:227). This may be due to the fact that the Calusa, who were known for
86 their strong influence on social and cultural affairs, often participated in t rade and exchange for various political and social occasions including diplomatic conferenc es, rituals, marriage, and ceremonies of alliance (Marquardt 2004:80-81). Belle Glade and HorrÂ’s Island were also influenced by the social aff airs of the Calusa. Belle Glade, a site occupied by the Calusa, contained archaeological evidence of adopting cultural practices from Glades cultures, the type of cultures inhabi ted HorrÂ’s Island. According to Willey (1949b:128) the practice of secondary burials and parti al cremation at Belle Glade may have been adopted from Gulf coast cultures as well as from St. Johns cultures. And while HorrÂ’s Island was considered a Glades culture si te, Glades cultures were known for their interactions with the Belle Glade and Caloosaha tchee cultures, therefore putting them in direct contact with the Calusa. According to Milanich (1998:113), Â“throughout their histories, the people of these cultures exchanges ideas a nd traded with one another as well as with cultures farther north. They were well a ware of their social and natural surroundings.Â” While there is no direct evidence of genetic interaction among Perico Island, Captiva Island, Belle Glade, and HorrÂ’s Island, the archaeological evidence of social interaction among these sites is stron g enough to provide support for the biological interaction suggested by the results of this stud y. Variation of genetic admixture among populations during European contact Fortunately, more can be said about the differences observed between HorrÂ’s Isla nd and Safety Harbor and between Safety Harbor and Fuller Mound A. All three of thes e sites were occupied during roughly the same time though some variation existed betw een 1260
87 B.P. and 247 B.P. Specifically, as noted in Table 3.1, HorrÂ’s Island was occupied between 1260 B.P. and 497 B.P. (Milanich 1994:301), Safety Harbor was occupied between 1110 B.P. and 260 B.P. (Hutchinson 2006:31; Hutchinson 2004:95; Mitchem 1989), and Fuller Mound A was occupied between 1010 B.P. and 247 B.P. (Willey 1954). Due to the close temporal relationships among these three sites, the possibili ty of cranial variation being attributed to temporal distance can be ruled out. According to Jantz (1973:20), Â“Inasmuch as genetic drift would create random rather than directional chang e, and selection could not be expected to change the gene pool much in the 200-250 years under consideration, gene flow is the most likely candidate for the evolutionary process responsible.Â” While Jantz was referring to the analysis of gene flow in a study that a nalyzed microevolutionary change in Arikara crania, his conclusions are relevant to the significant variation among Florida populations observed in this study and used to suggest that gene flow was the most likely contributing agent for the variat ion. Specifically, based on the relatively short amount of time separating Ho rrÂ’s Island, Safety Harbor, and Fuller Mound A, it can be argued that microevolutionary forces other than gene flow would not have had enough time to make significant changes on a population, therefore genetic drift and natural selection can be eliminated as possible contributors. In all cases, there was no more than 150-250 years of separation between the beginnin g or the end occupation for any two sites. In actuality, the majority of the time of occupation at HorrÂ’s Island, Safety Harbor, and Fuller Mound A overlapped between 1110 B.P. and 497 B.P. Therefore, we must assume that the variation is due to some other source of genetic influence, likely a combination of migration and changes in gene flow.
88 Interestingly, the variation present among HorrÂ’s Island, Safety Ha rbor, and Fuller Mound A populations can be observed in a time period that coincides with European contact in various areas of Florida. According to Worth (2001:7), the Spanish missiona ry interaction with Native Americans between 1565 and 1587 Â“Â… undoubtedly had biological consequences for the indigenous coastal populations.Â” These biological consequences have possibly been confirmed by the significant amount of cranial variation among HorrÂ’s Island, Safety Harbor, and Fuller Mound A sites. Archaeological evidence for contact at Safety Harbor may have indicate d the catalyst for the significant differences observed between HorrÂ’s Isl and and Safety Harbor. According to Mitchem (1989) Safety Harbor was located in the town of Tocobaga which was known to be in conflict with the Calusa. HorrÂ’s Island was located in the Caloosahatchee region which was the historical territory for the Calusa (H utchinson 2004). Unfortunately, Â“Â…the disease introduced by the early expeditions wiped out or significantly weakened some of the Safety Harbor groups, allowing the Calusa t o expand northward and to increase their powerÂ” (Mitchem 1989:575). The invasion of Europeans into Florida, in combination with hostile social relationships, led to a change in gene flo w within Safety Harbor and may ultimately have contributed to significant di fferences between HorrÂ’s Island and other sites such as Fuller Mound A. At the same time that Safety Harbor was experiencing changes in ge ne flow due to diseases brought by European settlers, Fuller Mound A may have been experie ncing changes in gene flow patterns as a result of slightly different soci al interactions. As demonstrated by Willey (1954:79-80), Fuller Mound A was positioned near the division
89 that sets apart hunting, fishing, and gathering cultures in south Florida from the northern populations practicing agriculture. It may be reasonable to suggest that human re mains from Fuller Mound A account for a large part of the genetic variation in this region of Florida due to the cross-cultural interactions that may have occurred betwee n these two groups. This variation in combination with the weakened Safety Harbor groups may be a plausible explanation for the significant amount of the variation between Safety H arbor and Fuller Mound A. In addition to considering differences in social interactions between Safety Harbor and Fuller Mound A as agents for significant genetic variation, attention should be drawn to the distribution of males and females in the Fuller Mound A sample (refer to Table 3.1). Unlike the other populations, there is a significantly larger sample of male s than there are females at Fuller Mound A. If this sample is truly representati ve of Fuller Mound A, it may be suggested that burial practices at Fuller Mound A included differential treatment of males and females in which males were favored. While there is no mention of this type of differential treatment in the archaeological li terature, this could account for the significant variation observed between Safety Harbor and Fulle r Mound A. The data from Fuller Mound A would be more skewed towards males and would make the overall population appear to be larger as a whole. While archaeological context may provide support for changes in gene flow due to European contact and other social interactions among HorrÂ’s Island, Safety H arbor, and Fuller Mound A, it is important to note that the same trend of decreased homogeneity can be seen in the Guale populations of north Florida and south Georgia. According to
90 Stojanowski (2004) there was a distinct shift in homogeneity between the pre-1680 Sa nta Catalina sample and the late mission period (1686-1702). Stojanowski (2004:324) explains this change by stating that Â“Â… the mechanisms that previously catalyzed aggregation or intermarriage were interrupted due in large part of English agg ression and SpainÂ’s realization that the mission system was indefensible and collapsing.Â” By reducing the opportunity for gene flow, populations are more susceptible to geneti c isolation and increased genetic variation between any two groups. Therefore, it is likely that European contact in Florida negatively impacted Native American populations at HorrÂ’s Island, Safety Harbor, and Fuller Mound A. Specifically, it is possible that the variation among these sites was a result of decreased extralocal gene flow for the same reasons Stojanwoksi (2004) suggested decreased homogeneity in the late mission per iod. In addition to investigating various levels of interaction that may have altere d gene flow among HorrÂ’s Island, Safety Harbor, and Fuller Mound A, it is also im portant to examine which variables were most significant for contributing to the observe d variation. In a study conducted by Sparks and Jantz (2002), BoasÂ’ data was revisited t o reevaluate cranial plasticity as a dominant force in cranial size and shap e. As previously stated, Sparks and Jantz demonstrated that most of cranial variation is genetic variation. Specifically, Â“both head-length and Â–breadth measurements show heritibilites g reater than 0.5 indicating that most of phenotypic variation in these traits can be attribute d to genetic factorsÂ” (Sparks and Jantz 2002:14637). Similar to the study performed by Sparks and Jantz (2002), this study demonstrated that measurements consistent with head-length and Â–breadth cont ributed
91 the greatest amount of variation. In Table 4.29, the first two eigenvectors expl ain 40.70% of the total variation. In addition, Table 4.30 demonstrates that the specific variable s that are most significant for contributing variation occur in the first two eigenvec tors. In the first eigenvector which accounted for 28.04% of variation, all variables were a ssociated with either length or breadth. These variables include Glabello-Occipital L ength (GOL), Nasion-Occipital Length (NOL), Basion-Nasion Length (BNL), Minimum F rontal Breadth (WFB), Basion-Prosthion Length (BPL), Bifrontal Breadth (FMB ), Frontal Chord (FRC), and Upper Facial Breadth (UFBR). In the second eigenvector, only three variables contributed significant variation and were consistent with length and hei ght. Those variables were Nasion-Prosthion Height (NPH), Parietal Chord (PAC) and Parietal Subtense (PAS). While it is still possible that environmental fa ctors may have contributed to some of the variation, the results indicate that most of the variation was caused by variables that are attributed to genetic factors; specificall y, gene flow.
92 Chapter 6 Conclusion The results of this study demonstrated that, in this specific sample of prehis toric Florida populations, two major factors may have influenced microevolutionary change of cranial interlandmark distances: changes in gene flow over many genera tions and European admixture. When populations were subjected to statistical analyses, s ignificant variation was found among populations that were separated by time periods gre ater than 4470 years. Specifically, comparison of all other populations with the Windover populations, which was more than 4000 years older in occupation than the rest demonstrated significant variation. In contrast, populations from those sites t hat overlapped in occupation periods experienced very little variation. After a long period of homogeneity among prehistoric populations in Florida, European contact significant ly impacted phenotypic variation of crania and increased heterogeneity among later populations. Specifically, these phenotypic differences resulted from gene tic variation attributed to the introduction of new populations. It is extremely likely that due to soci al pressures of European contact and the introduction of disease, gene flow was significantly reduced in south Florida populations as mating practices and interma rriage among groups were drastically limited. This reduction in gene flow led to i ncreased genetic isolation and heterogeneity among groups. As a result, these populati ons experienced microevolutionary changes in a relatively short period of time.
93 Suggestions for Future Research To gain a complete understanding of the microevolutionary pressures that transformed the cranial morphology of prehistoric populations in Florida, further research is required. Specifically, more in-depth analyses of cranial variation should be performed, including larger samples with more geographic representation a nd less temporal distance among the sites. While this study and similar studies (Ows ley et al. 1982; Key and Jantz 1981; Konigsberg 1990; Jantz and Owsley 2001) have demonstrated the microevolutionary change can be examined with relatively small sample siz es, a more comprehensive understanding would be facilitated with a larger sample of popul ations. It would be ideal for each population to have an adequate representative sample (n 30). Better results may be achieved if the populations occupied the sites continuousl y throughout time. This would simply reduce the error and prevent using a biased sample for statistical analysis. In addition, a larger geographic representa tion throughout Florida would provide a more thorough understanding of the impact that environment and geography may have had on prehistoric populations. A better understanding of when microevolutionary changes occurred would be facilitated by comparing sites on a continuous timeline. Lastly, it may be suggested that analyses be perform ed using 3-D coordinate data instead of 2-D interlandmark distances. By doing so, a better understanding of morphological changes may assist in understanding how the inter action between genetics and environmental factor contributes to variation.
94 Bioarchaeology as Applied Anthropology This research contributes to the field of anthropology by describing localize d human variation before and after European contact and strengthens new methods of research in skeletal biology. This research also contributes to Florida archa eology by providing biological support for the hypothesis that there were varying degrees of social interaction among the populations through space and time in addition to already documented archaeological interpretations regarding prehistoric Florida popul ations.
95 References Boas F. 1912. Changes in the bodily form of descendents of immigrants. New York: Columbia University Press. Buikstra JE, Ubelaker DH. 1994. Standards for Data Collection from Human Skeletal Remains. Fayetteville: Arkansas Archaeological Survey. Bullen RP. 1969. Southern Limit of Timucua Territory. The Florida Historical Quart erly 47(4):414-419. Cooke CW. 1945. Geology of Florida. Geological Bulletin No. 29. Tallahassee: Florida Geographical Survey. Daniel, Jr. IR. 1985. A Preliminary Model of Hunter-Gatherer Settlement in Ce ntral Florida. The Florida Anthropologist 38(4):261-273. Devor EJ. 1987. Transmission of human craniofacial dimensions. Journal of Craniofacial Genetics and Developmental Biology 7:95-106. Dickel DN, Doran GH. 2002. An Environmental and Chronological Overview of the Region. In: Doran, GH, editor. Windover: Multidisciplinary Investigations of an Ea rly Archaic Florida Cemetery. Gainesville: University Press of Flori da. p 39-58. Doran GH. 2002. Introduction to Wet Sites and Windover (8BR246) Investigations. In: Doran, GH, editor. Windover: Multidisciplinary Investigations of an Early Arc haic Florida Cemetery. Gainesville: University Press of Florida. p 1-38. Florida Division of Historical Resources. 2009. HorrÂ’s Island 5. Tallahassee: Flor ida Master Site File, Division of Historical Resources, Department of State. Glass B, Sacks MS, Jahn EF, Hess C. 1952. Genetic drift in a religious isolate: An analysis of the causes of variations in blood group and other gene frequencies in a s mall population. American Naturalist 86:145-159. Goggin JM. 1998. Space and Time Perspective in Northern St. Johns Archeology, Florida. Gainesville: University Press of Florida. Goggin JM. 1950. Florida Archaeology. The Florida Anthropologist 3(1-2):9-20.
96 Goggin JM. 1940. The Distribution of Pottery Wares in the Glades Archaeological Are a of South Florida. New Mexico Anthropologist 4(2):22-33. Gravlee CC, Bernard HR, Leonard WR. 2003. Heredity, Environment, and Cranial Form: A Reanalysis of BoasÂ’s Immigrant Data. American Anthropologist 105(1):125-138. Griffin JW. 1974. Archaeology and Environment in South Florida. In: Gleason PJ, editor. Environments of South Florida: Present and Past, second edition. Coral Gables: Miami Geological Society. p 342-346. Griffin MC, Lambert PM, Monahan Driscoll E. 2001. Biological Relationship and Population History of Native Peoples in Spanish Florida and the American Southeast. In: Larsen CS, editor. Bioarchaeology of Spanish Florida: The Impact of Coloniali sm. Gainesville: University Press of Florida. p 226-266. Holloway RL. 2002. Head to Head with Boas: Did He Err on the Plasticity of Head Form? Proceedings of the National Academy of Sciences of the United States of America 99:14622-14623. Hutchinson DL. 2004. Bioarchaeology of the Florida Gulf Coast: Adaptation, Conflict, and Change. Gainesville: University Press of Florida. Hutchinson DL. 2006. Tatham Mound and the Bioarchaeology of European Contact. Gainesville: University Press of Florida. Jantz RL. 1973. Microevolutionary Change in Arikara Crania: A Multivariate Analys is. American Journal of Physical Anthropology 38:15-26. Jantz RL, Owsley DW. 2001. Variation Among Early North American Crania. Amer ican Journal of Physical Anthropology 114:146-155. Janus Research. 2002. Cultural Resource Assessment of Survey of the Perico Harbor Marine Basin Project Area, Manatee County, FL. Tallahassee: Florida Mas ter Site File. Key PJ, Jantz RL. 1981. A Multivariate Analysis of Temporal Change in Arikar a Craniometrics: A Methodological Approach. American Journal of Physical Ant hropology 55:247-259. Key PJ, Jantz RL. 1990. Statistical Assessment of Population Variability: A Methodological Approach. American Journal of Physical Anthropology 82:53-59.
97 Kimmerle EH, Ross A, Slice D. 2008. Sexual Dimorphism in American: Geometric Morphometric Analysis of the Craniofacial Region. Journal of Forensic Science s 53(1):54-57. Konigsberg LW. 1990. Temporal Aspects of Biological Distance; Serial Correla tion and Trend in a Prehistoric Skeletal Lineage. American Journal of Physical A nthropology 82:45-52. Luer GM, Almy MM. 1982. A Definition of the Manasota Culture. The Florida Anthropologist 35(1):34-58. Kostyu DD, Ober CL, Dawson DV, Ghanayem M, Elias S, Martin AO. 1989. Genetic analysis of HLA in the I.S. Schmiedenleut Hutterites. American Journal of H uman Genetics 45:261-269. MacMahon DA, Marquardt. 2004. The Calusa and Their Legacy. Gainesville: Universit y Press of Florida. Manly BFJ. 2005. Multivariate Statistical Methods: A Primer (3 rd Ed.). Boca Raton: CRC Press. McGoun WE. 1993. Prehistoric Peoples of South Florida. Tuscaloosa: The University of Alabama Press. Milanich JT. 1994. Archaeology of Precolumbian Florida. Gainesville: University Pr ess of Florida. Milanich JT. 1998. FloridaÂ’s Indians from Ancient Times to the Present. Gainesvill e: University Press of Florida. Milanich JT, Fairbanks CH. 1980. Florida Archaeology. New York: Academic Press. Mitchem JM. 1989. Redefining Safety Harbor: Late Prehistoric/Protohistori c Archaeology in West Peninsular Florida. Unpublished Ph.D. dissertation. Gainesville: University of Florida, Department of Anthropology. Molnar S. 2002. Human Variation: Races, Types and Ethnic Groups, Fifth Edition. Upper Saddle River: Prentice Hall. OÂ’Higgins P, Strand Vidarsdottir U. 1999. New approaches to the quantitative analysis of craniofacial growth and variation. In: Hoppa RD, Fitzgerald CM, editors. Huma n growth in the past: studies from bones and teeth. Cambridge: Cambridge University Pres s. p 128160.
98 Ousely S. 2004. Threeskull 2.0.77 [Computer Program], http://www.mercyhurst.edu/departments/applied_forensic_sciences/facul ty_staff.html. Owsley DW, Bennett SM, Jantz RL. 1982. Intercemetary Morphological Variat ion in Arikara Crania From the Mobridge Site (39WW1). American Journal of Physical Anthropology 58:179-185. Pietrusewsky M. 2008. Metric Analysis of Skeletal Remains: Methods and Applica tions. In: Katzenberg MA, Saunders SR, editors. Biological Anthropology of the Human Skeleton. Hoboken: Wiley-Liss. p 487-532. Relethford JH. 1994. Craniometric Variation, Genetic Theory, and Modern Human Origins. American Journal of Physical Anthropology 95(3):249. Relethford JH. 2002. Apportionment of Global Human Genetic Diversity Based on Craniometrics and Skin Color. American Journal of Physical Anthropology 118(4):393. Relethford JH. 2004. Boas and beyond: Migration and Craniometric Variation. American Journal of Human Biology 16:379-386. Relethford JH, Harpending HC. 1994. Craniometric Variation Among Modern Human Populations. American Journal of Physical Anthropology 95(1):53. Ross AH, Williams S. 2008. Testing Repeatability and Error of Coordinate Landma rk Data Acquired from Crania. Journal of Forensic Sciences 53(4):782-785. Rouse, I. 1951. A Survey of Indian River Archaeology, Florida. Yale University Publications in Anthropology, no. 44. New Haven: Yale University Press. Santos SEB, Rodrigues JD, Ribeiro-Dos Santos AKC, Zago MA. 1999. Differential contribution of indigenous men and women to the formation of an urban population in the Amazon region as revealed by mtDNA and Y-DNA. American Journal of Physica l Anthropology 109:175-180. Sjvold T. 1984. A report on the heritability of some cranial measurements and non-metric traits. In: Van Vark GN, Howells WW, editors. Multivariate Statis tics in Physical Anthropology. Dordrecht: D. Reidel Publishing Co. p 223-246. Slice DE, editor. 2005. Modern Morphometrics in Physical Anthropology. New York: Kluwer Academic/Plenum Publishers. Sparks CS, Jantz RL. 2002. A reassessment of human cranial plasticity: Boas re visited. Proceedings of the National Academy of Sciences of the United States of Am erica 99(23):1463614639.
99 Stirling MW. 1931. Mounds of the vanished Calusa Indians of Florida. Explorations and Fieldwork of the Smithsonian Institution in 1930. Washington: Smithsonian Institution. Stojanowski CM. 2004. Populations History of Native Groups in Preand Postcontact Spanish Florida: Aggregations, Gene Flow, and Genetic Drift on the Southeastern U.S. Atlantic Coast. American Journal of Physical Anthropology 123:316-332. Stojanowski CM. 2005. Spanish Colonial Effects on Native American Mating Structure and Genetic Variability in Northern and Central Florida: Evidence From Apalac hee and Western Timucua. American Journal of Physical Anthropology 128:273-286. van Valen L. 1982. Homology and Causes. Journal of Morphology 173:305-312. Widmer RJ. 1988. The Evolution of the Calusa: A Nonagricultural Chiefdom on the Southwest Florida Coast. Tuscaloosa: University of Alabama Press. Willey GR. 1948. Culture Sequence in the Manatee Region of West Florida. American Antiquity 13(3):209-218. Willey GR. 1949a. Archaeology of the Florida Gulf Coast. Smithsonian Miscell aneous Collections Volume 113. Washington: Smithsonian Institution. Willey GR. 1949b. Excavations in Southeast Florida. New Haven: Yale University Pres s. Willey GR. 1954. Burial Patterns in the Burns and Fuller Mounds, Cape Canaveral, Florida. The Florida Anthropologist 7(3):79-90. Worth JE. 2001. The Ethnohistorical Context of Bioarchaeology in Spanish Florida. In: Larsen CS, editor. Bioarchaeology of Spanish Florida: The Impact of Colonialis m. Gainesville: University Press of Florida. p 1-21.
101 Appendix A Â– Agglomerative Cluster Analysis