Citation
Late Holocene Coastal Evolution and Human Occupation on the Northern Gulf Coast of Florida

Material Information

Title:
Late Holocene Coastal Evolution and Human Occupation on the Northern Gulf Coast of Florida
Creator:
Mcfadden, Paulette M
Place of Publication:
[Gainesville, Fla.]
Florida
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (208 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Anthropology
Committee Chair:
SASSAMAN,KENNETH EDWARD,JR
Committee Co-Chair:
OYUELA-CAYCEDO,AUGUSTO
Committee Members:
WALLIS,NEILL JANSEN
JAEGER,JOHN M
Graduation Date:
5/2/2015

Subjects

Subjects / Keywords:
Birds ( jstor )
Eggshells ( jstor )
Horseshoes ( jstor )
Inlets ( jstor )
Lithofacies ( jstor )
Middens ( jstor )
Sand ( jstor )
Sea level ( jstor )
Sediments ( jstor )
Shorelines ( jstor )
Anthropology -- Dissertations, Academic -- UF
coastal -- environment -- pre-columbian -- sea-level
Gulf of Mexico ( local )
Genre:
bibliography ( marcgt )
theses ( marcgt )
government publication (state, provincial, terriorial, dependent) ( marcgt )
born-digital ( sobekcm )
Electronic Thesis or Dissertation
Anthropology thesis, Ph.D.

Notes

Abstract:
This geoarchaeological investigation of Horseshoe Cove on the Northern Gulf Coast of Florida incorporated analysis of marine sediment cores, terrestrial sediment samples, and excavations at three archaeological sites to determine the impact of sea-level rise on resident communities. The initial flooding of Horseshoe Cove due to sea-level rise occurred between 4,000 and 4,500 years ago, which is consistent with the timing of the flooding of the Suwannee Delta and Waccasassa Bay to the south. Slowing rates of rise allowed for the formation of the resource rich estuarine environment that humans began to exploit by at least 4,400 years ago. Evidence for fluctuations in sea level identified in other areas of the southeastern United States is lacking in the cores collected from Horseshoe Cove, however, changing human practices manifest in the archaeological deposits suggest possible environmental shifts. Additional testing is warranted to address these questions further. Finally, during the Deptford and Early Swift Creek periods, choices about places to settle appear to have been based on environmental setting. Locations that were protected from the high-energy, open-marine environment, but with direct access to the resources available there, appear to have been targeted. By the Weeden Island period, choices about occupation appear to have been based more on social rather than environmental factors. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2015.
Local:
Adviser: SASSAMAN,KENNETH EDWARD,JR.
Local:
Co-adviser: OYUELA-CAYCEDO,AUGUSTO.
Electronic Access:
RESTRICTED TO UF STUDENTS, STAFF, FACULTY, AND ON-CAMPUS USE UNTIL 2016-05-31
Statement of Responsibility:
by Paulette M Mcfadden.

Record Information

Source Institution:
UFRGP
Rights Management:
Applicable rights reserved.
Embargo Date:
5/31/2016
Classification:
LD1780 2015 ( lcc )

Downloads

This item has the following downloads:


Full Text

PAGE 1

LATE HOLOCENE COASTAL EVOLUTION AND HUMAN OCCUPATION ON THE NORTHERN GULF COAST OF FLORIDA By PAULETTE SIZEMORE MCFADDEN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT O F THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2015

PAGE 2

© 2015 Paulette Sizemore McFadden

PAGE 3

To Steve and Mom

PAGE 4

4 ACKNOWLEDGMENTS There are many people who contributed to the succe ssful completion of this dissertation project. First, I want to thank my committee. Ken Sassaman has been a true mentor throughout this process. He has labored beside me in the field excavating test units and stood with me, knee deep, in marsh muck to c ollect sediment cores. I have grown tremendously both academically and professionally under his mentorship an d I am immensely grateful to him for his guidance and support. Neill Wallis was gracious enough to allow me to work with him at the Garden Patch site during a summer field school where we collected data that w ere integral to this project. Beyond that he had continued to be a positive influence, both professionally and personally, and I am so thankful that I have had the opportunity to benefit from his input and encouragement. Augusto Oyuela Caycedo contributed significantly to my perception of the relationship between humans and the landscape and his input positively influenced the way that I interpreted my data. J ohn Jaeger, with the Department of Geological Sciences , contributed greatly to this project, devoting significant time and energy to its completion. from the sediment cores, and I often felt that he devoted as much t ime to me as he did his own graduate students. I have to give credit here to my good friend Chris Moore. I first met Chris when I was an undergraduate and he was working on his dissertation. I worked with him in the field and later as a lab assistant, wh ich is when I became enamored with sediment geoarchaeology and he has continue d to provide moral support and advice throughout my doctoral work.

PAGE 5

5 My fellow graduate stu dents also contributed to the success of this project. Micah Mon é s and I have spent many hours in the archaeology lab pondering our respective study areas on the Gulf Coast, comparing data, and bouncing ideas around . Andrea Palmiotto enthusiastically col lected and analyzed the faunal remains from Bird Island and spent many hours working with me to integrate her data with mine. In addition to Micah and Andrea, many others have given their time and energy in the field, including Elyse Anderson, Melissa Ayv az, Stephanie Boothby , Kristen Hall, Ginessa Mahar, Stephen McFadden , Cristina Oliveira, Ken Sassaman, and Haley Singleton . Mark Brenner and Jason Curtis braved the mysterious pond at Garden Patch to collect the pond core. The incredible Jamie Cona assis ted with the often tedious tasks associated with analysis of the sediment cores and sediment samples in the geology lab. Cristina Oliveira, Dale Torres, Alan Schneider, and Patricia Caldwell provided assistance in the archaeology lab . Permission to excava te on Bird Island was provided by David Nelms, and the Nelms family have enthusiastically supported th is research. Patsy and Warren Nelms are two of the most wonderful people I have ever met. They always made us feel welcome on the island and p rovid ed w o nderful refreshments and meals for the field crews , including homemade peppermint ice cream and boiled peanuts! Shannon Moore, Kathleen Bonany, and Reshmie Punwasi helped sort through the substantial assemblage of surface collected pottery at Bird Island and a s pecial thanks goes to Dr. Elizabeth Wing, who lent her expertise to the collection of faunal materials at Bird Island. Permission to excavate on Butler Island was given by the Dixie County Board of County Commissioners and was greatly facilitated b y the County Manager, Mike

PAGE 6

6 Cassidy. Access to Cotton Island was given by the Ozaki family. Thanks to the Florida Fish and Wildlife Conservation Commission and Randy Havens for their support during the work at the Garden Patch site. T hanks to William Ken ney , with the Land Use and Environmental Change Institute (LUECI) , who provided access to the Paleoenvironmental Research Laboratory . Charles B. Stoer provided luxury accommodations at his wonderful Gulf front home . I spent many hours on his porch, w hic h overlooks nearly my entire study area , and learned a great deal about the environment through observations made from his rocking chairs. I am forever grateful to Chuck for his generosity and kindness. I imagine few archaeologists are lucky enough to ha ve such lavish accommodations while doing their fieldwork. The Horseshoe Beach community enthusiastically supported my work in the area and I always felt welcome when I was in town. Jean nie Jones, the President of the Horseshoe Library, along with Zinnett e Green, continue to work with me to set up displays at the library and to organize the annual Horseshoe Beach Archaeology Day. Mayor Laura Wigglesworth, along with the town council , have been great supporters of my project and the work at Garden Patch co nducted by Neill Wallis. I h ave made great friends at Horseshoe Beach, all of which have enthusiastically supported me and I want to thank Ray and Brenda Rodriguez, Flo Gilchrist, Rob and Donna Ives, Tina Brotherton, Tim and Renee Futch, Jim and Sarah Bet h Stribling , and Sabra and Gary Nave for their friendships and support over these last five years. Of course many other s in Horseshoe Beach have contributed to the successful completion of this project and any omission reflects on ly my shortcomings and no t a lack of appreciation.

PAGE 7

7 One important person that deserves special thanks is Julian Granberry. I met Julian during my first visit to Horseshoe Beach with Ken Sassaman and I have developed a deep affection for him over the past few years. Influenced by the great Franz Boaz, Julian is a true four field anthropologist . He has lived in Horseshoe Beach for five decades and was the first person to excavate at Bird Island . He has a wealth of knowledge about the archaeology in th is area and has long argued th at Horseshoe Beach deserves significant attention because of its archaeological deposits. He e nthusiastically em braced my work when I began this project and has always been anxious to hear about my research as I progressed . My respect for Julian is only surpassed by my love for him and his advice and wisdom have been much appreciated. Perhaps the most important people who have supported me during this project are my family. My incredible husband, Steve McFadden, has been a true partner throughout this en deavor . His love and support have carried me through this entire process and I know that I would not have accomplished this without him. I can never express how important his contribution was to this project. In addition to years of moral support, he al so provided ferry service for me and my crews back and forth to the islands, helped with the collection of marine sediment cores, dug more than a few shovel tests, screened, and backfilled, and was always willing to do more . My mother, Helen Sizemore, als o contributed both in moral and financial support. She enthusiastically embraced my decision to pursue this course and never discouraged me. I can only hope that I will someday have her same strength and character. Finally, archaeology is not free and I thank Hyatt and Cici Brown for their financial support provided through the Hyatt and Cici Brown E ndowment for Florida

PAGE 8

8 Archaeology. Funds provided by their generous endowment enabled me to focus all my energy on this research project and I am truly apprec iative of their support.

PAGE 9

9 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ .......... 12 LIST OF FIGURES ................................ ................................ ................................ ........ 13 ABSTRACT ................................ ................................ ................................ ................... 17 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 19 Temporal Resolution and Spatial Variation ................................ ............................. 21 Regressions, Higher Than Present Sea Level, and the Nature of Sea Level Rise ................................ ................................ ................................ ..................... 25 Factors Influencing Relative Sea Level ................................ ................................ ... 25 Big Bend Region ................................ ................................ ................................ ..... 27 Hypotheses ................................ ................................ ................................ ............. 29 Hypothesis 1 ................................ ................................ ................................ ..... 30 Hypothesis 2 ................................ ................................ ................................ ..... 31 Hypothesis 3 ................................ ................................ ................................ ..... 31 Hypothesis 4 ................................ ................................ ................................ ..... 32 Hypothesis 5 ................................ ................................ ................................ ..... 32 Structure of the Dissertation ................................ ................................ ................... 33 2 MIDDLE TO LATE HOLOCENE COASTAL EVOLUTION OF HORSESHOE COVE ON THE NORTHERN GULF COAST OF FLORIDA ................................ .... 38 Introduction ................................ ................................ ................................ ............. 38 Methods ................................ ................................ ................................ .................. 42 Results ................................ ................................ ................................ .................... 43 Marine Sediment Core Lithofacies ................................ ................................ ... 46 Limestone ................................ ................................ ................................ .. 46 Dark gray medium to fine sand ................................ ................................ .. 48 Brown medium sand ................................ ................................ .................. 50 Gray medium sand to black muddy sand ................................ ................... 50 Light gray to brown medium sand ................................ .............................. 51 Black muddy sand with low carbonate content ................................ .......... 52 Black muddy sand with higher carbonate content ................................ ...... 53 Dark gray to brown sand with shell ................................ ............................ 53 Dark gray to brown fine sand ................................ ................................ ..... 54 Dense shell in sand matrix ................................ ................................ ......... 54 Fresh Water Pond Core Lithofacies ................................ ................................ . 54 Brown sand ................................ ................................ ................................ 56

PAGE 10

10 Brown mottled sand ................................ ................................ ................... 56 Black muddy sand ................................ ................................ ...................... 56 Discussion ................................ ................................ ................................ .............. 57 Conclusions ................................ ................................ ................................ ............ 67 3 ARCHAEOLOGICAL INVESTIGATIONS OF THREATENED STRATIFIED SITES IN HORSESHOE COVE, NORTHERN GULF COAST, FLORIDA .............. 70 Introduction ................................ ................................ ................................ ............. 70 Environmental Setting ................................ ................................ ............................. 72 Bird Island ................................ ................................ ................................ ............... 7 3 Current Research ................................ ................................ ............................. 76 Test unit 1 (TU1) and test unit 3 (TU3) ................................ ...................... 76 Test unit 2 (TU2). ................................ ................................ ....................... 82 Material culture ................................ ................................ .......................... 84 Butler Island ................................ ................................ ................................ ............ 87 Current Research ................................ ................................ ............................. 88 Test unit 1 (TU1) ................................ ................................ ........................ 90 Test unit 2 (TU2) ................................ ................................ ........................ 90 Test units 3 (TU3) and 3N (TU3N) ................................ ............................. 92 Material Culture ................................ ................................ .......................... 94 Other Sites ................................ ................................ ................................ .............. 98 Butler Island ................................ ................................ ................................ ..... 98 Cotton Island ................................ ................................ ................................ .... 99 Environmental Change and Human Occupation ................................ ..................... 99 Conclusions ................................ ................................ ................................ .......... 105 4 COASTAL EVOLUTION AND PRE COLUMBIAN HUMAN OCCUPATION IN HORSESHOE COVE ON THE NORTHERN GULF COAST OF FLORIDA .......... 107 Introduction ................................ ................................ ................................ ........... 107 Study Area ................................ ................................ ................................ ............ 110 Methods ................................ ................................ ................................ ................ 112 Results ................................ ................................ ................................ .................. 114 Terrestrial Sediments ................................ ................................ ..................... 115 Bird Island ................................ ................................ ................................ 115 Butler Island ................................ ................................ ............................. 115 Garden Patch, Mound IV ................................ ................................ .......... 117 Garden Patch Mound V ................................ ................................ ......... 119 Test Unit Excavations ................................ ................................ ..................... 120 Bird Island ................................ ................................ ................................ 120 Butler Island ................................ ................................ ............................. 122 Garden Patch ................................ ................................ ........................... 124 Discussion ................................ ................................ ................................ ............ 124 Conclusion ................................ ................................ ................................ ............ 132

PAGE 11

11 5 ENVIRONMENTAL AND SOCIAL FACTORS IN PRE COLUMBIAN SETTLEMENT ON THE NORTHERN GULF COAST OF FLORIDA .................... 134 Introduction ................................ ................................ ................................ ........... 134 Study Area ................................ ................................ ................................ ............ 137 Methods ................................ ................................ ................................ ................ 140 Results ................................ ................................ ................................ .................. 142 Marine Sediment Cores ................................ ................................ .................. 142 Fresh Water Pond Core ................................ ................................ .................. 147 Terrestrial Sediments ................................ ................................ ..................... 148 Test Unit Excavations ................................ ................................ ..................... 149 Bird Island ................................ ................................ ................................ 150 Butler Island ................................ ................................ ............................. 151 Garden Patch ................................ ................................ ........................... 153 Discussion ................................ ................................ ................................ ............ 158 Conclusion ................................ ................................ ................................ ............ 168 6 CONCLUSION AND RECOMMENDATIONS ................................ ....................... 170 Hypothesis 1 ................................ ................................ ................................ ......... 171 Hypothesis 2 ................................ ................................ ................................ ......... 171 Hypothesis 3 ................................ ................................ ................................ ......... 172 Hypothesis 4 ................................ ................................ ................................ ......... 173 Hypothesis 5 ................................ ................................ ................................ ......... 175 Recommendations for Future Research ................................ ............................... 177 APPENDIX INDIVIDUAL CORE DESCRIPTIONS WITH LITHOFACIES CHARACTERISTICS ................................ ................................ ............................ 181 REFERENCES ................................ ................................ ................................ ............ 198 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 208

PAGE 12

12 LIST OF TABLES Table page 2 1 Summary of Radiocarbon Dates from Sediment Cores. ................................ ..... 44 2 2 Summary of Radiocarbon Dates from Archaeological Contexts. ........................ 45 2 3 Horseshoe Cove lithofacies interpretations, after Wright (1995), with lithology and percentage of fines, carbonate, and organic matter. ................................ ... 58 3 1 Frequencies of Identified Pottery Types by S ite. ................................ ................ 97

PAGE 13

13 LIST OF FIGURES Figure page 1 1 Sea level changes based on Kurtosis values and period of climate change. Note: Figure from Marquardt, Wi lliam H. 2010. Mounds, Middens, and Rapid Climate Change During the Archaic Woodland Transition in the Southeastern United States. In Trend, Tradition and Turmoil: What Happened to the Southeastern Archaic? Vol. 93, edited by David Hurst Thomas and Matthew C. Sanger, pp. 253 272. American Museum of Natural History, New York. ....... 22 1 2 Sea level indicators in the Gulf of Mexico with a 7 point floating average of mean sea level. Note: Figure from B alsillie, James H. and Joseph F. Donoghue. 2004. High Resolution Sea Level History for the Gulf of Mexico since the Last Glacial Maximum. Vol. 103, State of Florida Department of Environmental Protection, Tallahassee, FL. ................................ ....................... 24 1 3 Study area showing locations of archaeological sites at Butler Island (8DI50), Bird Island (8DI52), and Garden Patch (8DI4), and locations of sediment cores. ................................ ................................ ................................ .................. 34 2 1 Schematic representation of lithofacies in each core in the southwest to northeast transect with radiocarbon dates in cal BP. ................................ .......... 47 2 2 Schematic representation of lithofacies in each core in the southwest to northeast transect with radiocarbon dates in cal BP. ................................ .......... 48 2 3 Detailed lithofacies descriptions with percentage of carbonate, organic matter, and fines, and mean grain size a nd sorting statistics for selected marine sediment cores. ................................ ................................ ...................... 49 2 4 Detailed lithofacies descriptions with percentage of carbonate, organic matter, and fines, and mean grain size and sorting stati stics for the fresh water core (HBT19). ................................ ................................ ........................... 55 2 5 Schematic representation of interpreted lithofacies in each core in the southwest to northeast transect with calibrated radiocarbon dates. ................... 59 2 6 Schematic representation of interpreted lithofacies in each core in the southwest to northeast transect with calibrated radiocarbon dates. ................... 60 3 1 Topographic map of Bird Island (8DI52) showing locations of Test Unit 1, Test Unit 2, and Test Unit 3. The house, boardwalk, dock, and seawall are shaded. ................................ ................................ ................................ ............... 77 3 2 East profile of Test Unit 1 with stratigraphic designations and radiocarbon dates, 8DI52. ................................ ................................ ................................ ...... 78

PAGE 14

14 3 3 North profile of Test Unit 3, 8DI52. ................................ ................................ ..... 78 3 4 Gast ropod Adze recovered from Level I of Test Unit 1, 8DI50. .......................... 81 3 5 Whelk cluster in Feature 3 in Test Unit 3, 8DI52. ................................ ............... 83 3 6 North pro file of Test Unit 2, 8DI52. ................................ ................................ ..... 83 3 7 Two views of a hematite bead recovered from TU3, Bird Island. ........................ 86 3 8 Topographic map of Butler Island NE (8DI50) showing locations of structures, augers and test pits, and excavation units. Locus A and B are designated by the by circles. Map produced by K. Sassaman. ........................... 89 3 9 South profile of Test Unit 1, 8DI50. ................................ ................................ ..... 89 3 10 West profile of Test Unit 2, Butler Island (8DI50). ................................ ............... 91 3 11 East profile of Test Unit 3N, 8DI50. ................................ ................................ .... 93 3 12 Three views of a perforated bone tool recovered from Test Unit 3N, Butler Island (8DI50). ................................ ................................ ................................ .... 96 4 1 West profile of Test Unit 3 , Bird Island (8DI52) with sponge spicules per gram of sediment, percentage of fines, and grain size and sorting. ........................... 116 4 2 Stratigraphic profile of the sediment column from TU4, Mound IV, 8DI4, with percentage of fines, grain size, sorting, frequencies of pottery and lithics, and designated soil horizons and zones. ................................ ................................ . 117 4 3 Stratigraphic profile of the sediment column from Profile 4, M ound V, 8DI4, with percentage of fines, grain size, sorting, and designated soil horizons and zones. ................................ ................................ ................................ ............... 118 4 4 Paleoenvironmental reconstructions of northern Horseshoe Cove and dates of occupatio symbol. ................................ ................................ ................................ ............. 126 5 1 Schematic representation of interpreted lithofacies in each core in the southwest to northeast transect with radiocarbon dates. ................................ .. 143 5 2 Schematic representation of interpreted lithofacies in each core in the east to west transect with radiocarbon dates. ................................ .............................. 144 5 3 features designated by area numbers with Area X added. Adapted from Kohler, Timothy. ................................ ........................ 154

PAGE 15

15 5 4 Planview drawing of features in Test Unit 7 (left) and Test Unit 8 (right), Area I, Garden Patch (8DI4). Figure from Wallis, Neill J. and Paulette S. McFadden. 2014. Suwannee Valley Archaeological Field School 2013: The Garden Patch Site (8DI4). Miscellaneous Report No. 64. Division of Anthropology, Florida Museum of Natural History, University of Florida, Gainesville. ................................ ................................ ................................ ....... 156 A 1 Core HBT2 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ....................... 182 A 2 Core HBT3 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ....................... 183 A 3 Core HBT4 with lithofacies desc riptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ....................... 184 A 4 Core HBT6 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, an d grain size and sorting. ................................ ....................... 185 A 5 Core HBT8 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ....................... 186 A 6 Core HBT9 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ....................... 187 A 7 Core HBT10 with lithofacies des criptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ............. 188 A 8 Core HBT11 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ............. 189 A 9 Core HBT12 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ............. 190 A 10 Core HBT13 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ............. 191 A 11 Core HBT14 with lithofaci es descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ............. 192 A 12 Core HBT15 with lithofacies descriptions, percentage of organic matter, fines, and carb onate, and grain size and sorting. ................................ ............. 193 A 13 Core HBT16 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ............. 194 A 14 Core HBT17 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ............. 195

PAGE 16

16 A 15 Core HBT18 with l ithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ............. 196 A 16 Core HBT19 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting. ................................ ............. 197

PAGE 17

17 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy LATE HOLOCENE COASTAL EVOLUTION AND HUMAN OCCUPATION ON THE NORTHERN GULF COAST OF FLORIDA By Paulette Sizemore McFadden May 2015 Chair: Kenneth E. Sassaman Major: Anthropology This geoarchaeological investigation of Horseshoe Cove on the Northern Gulf Coast of Florida incorporated analysis of marine sediment cores, terrestrial sediment samples, and excavations at three archaeological sites to determine the impact of sea level rise on resident communities. The initial flooding of Horsesho e Cove due to sea level rise occurred between 4,000 and 4,500 years ago, which is consistent with the timing of the flooding of the Suwannee Delta and Waccasassa Bay to the south. Slowing rates of rise allowed for the formation of the resource rich estuar ine environment that humans began to exploit by at least 4,400 years ago. Evidence for fluctuations in sea level identified in other areas of the southeastern United States is lacking in the cores collected from Horseshoe Cove, however, changing human pra ctices manifest in the archaeological deposits suggest possible environmental shifts. A dditional testing is warranted to address these questions further. Finally, during the Deptford and Early Swift Creek periods, choices about places to settle appear to have been based on environmental setting. Locations that were protected from the high energy, open marine environment, but with direct access to the resources available

PAGE 18

18 there, appear to have been targeted. By the Weeden Island period, choices about occu pation appear to have been based more on social rath er than environmental factors.

PAGE 19

19 CHAPTER 1 INTRODUCTION The Big Bend region of the northern Gulf Coast of Florida is a large open water, low energy region that supports an extensive estuarine system. Th e pre Columbian residents of Horseshoe Cove in this region have thrived for millennia on the resources provided by this rich environment even as they navigated its changing conditions. Living in this dynamic environment presents unique challenges and nece ssitates the development of adaptive strategies that help communities deal with inevitable environmental change, including sea level rise, alterations in shoreline morphology, changing access to fresh water, and shifting communities of flora and fauna. Tr ansgression in this low gradient area can be significant during periods of high rates of sea level rise, and indeed hundreds of square kilometers of the continental shelf have been inundated since the end of the Last Glacial Maximum (ca. 22,000 years ago). In contrast, the productive marsh environment can keep pace with moderate rates of sea level rise (e.g. Leonard et al. 1995; Jaeger et al. 2005) and offshore oyster bioherms can trap sediment landward of the reefs (e.g. Goodbred 1994; Wright 1995), resul ting in relatively long periods of localized stasis. Archaeologists interested in understanding the relationship between the environment and human practices along the coast necessarily need to understand the nature of the environmental changes experienced by the subjects of their research. However, navigating the literature on sea level change and paleoenvironmental reconstructions can be challenging since the temporal and spatial scales are typically incommensurate with those used by archaeologists.

PAGE 20

20 On a gross scale, sea level was approximately 125 meters below present at the end of the Last Glacial Maximum (Fleming et al. 1998:327) and rose rapidly, between 20 30 meters, over a period of 2,000 years. After 20,000 years ago, rates of rise slowed and remai ned relatively consistent until around 7,000 years ago, at which time sea level was 3 5 meters below present. Over the past 7,000 years, rates have continued to decrease, particularly after 2,000 years ago (Fleming et al. 1998:340). This curve is in gene ral agreement with other eustatic curves, but is of minimal utility when trying to understand the impact of sea level change on human communities. Obviously this millennial scale reconstruction is too coarse grained to be applied to questions of human imp act on the scale of social memory . Another issue with using global reconstructions is that global sea level change has a very different impact on the resilient estuarine environment of the northern Gulf Coast than, for instance, on higher energy wave domi nated coastlines. Additionally, the west coast of Florida is a passive margin where tectonic and isostatic processes have minimal influence. Following is a brief review of studies in the southeastern United States that focus on sea level change and the re lationship between sea level change and changes in human settlement patterns over the past 5,000 years, which encompasses the entire temporal range of the extant archaeological deposits in Horseshoe Cove. These studies are presented in several topical cat egories relevant to this study : Temporal resolution and spatial variation; regressions, higher than present sea level, and the nature of sea level rise; factors influencing relative sea level; and finally, studies of coastal evolution in areas of the Big Bend region.

PAGE 21

21 Temporal Resolution and Spatial Variation Fine grained temporal resolution is necessary to scale paleoenvironmental reconstructions to human experience, however, the higher the temporal resolution, the less generalizable it is to geographic r egions. For instance, Tanner (1993) produced a mean sea level curve using kurtosis values from sediments on beach ridges in Denmark. His reconstruction suggests significant variation in sea level over the last 7500 years, and the 50 year intervals that h e presents make his study of particular interest for understanding environmental change on a human scale. Marquardt (2010) presents Tanner's sea level study to illustrate the necessity of a fine grained curve to better understand the relationship between fluctuating sea levels and shifting pre Columbian settlement patterns along the coasts of the southeastern United States. grained changes in sea level, while arguing tha t similarly fine grained paleoenvironmental reconstructions need to be site specific in order to understand sea level changes along a specific coastline in Denmark, whic h is in a very different tectonic and isostatic setting than gulf coast Florida. Changes in sea level indicated by his data may be the result of processes other than eustatic changes, for instance uplift or subsidence. An additional problem with the grap h presented by Marquardt (2010), is that it provides only the kurtosis values and not measures of actual sea level and the variation in the graph does not reflect the true magnitude of the suggested changes in sea le vel (see Figure 1 1).

PAGE 22

22 Figure 1 1. Sea level changes based on Kurtosis values and period of climate change. Note: Figure from Marquardt, William H. 2 010 . Mounds, Middens, and Rapid Climate Change During the Archaic Woodland Transition in th e Southeastern United States. In Trend, Tradition and Turmoil: What Happened to the Southeastern Archaic? Vol. 93, edited by David Hurst Thomas and Matthew C. Sanger, pp. 253 272. American Museum of Natural History, New York. The changes in sea level prop osed by Tanner (1993) do not mirror changes observed in other regions. A study by Brooks et al. (1979) in the Lower Cooper River Valley in South Carolina merged geological and archaeological evidence to investigate the relationship a mong archaeological si tes and sea level. The Cooper River empties into a large estuary system that is significantly influenced by fluctuating sea levels. Sediment cores provided evidence of alternating fresh, salt, and brackish water marsh deposits and archaeological evidence suggests changing settlement patterns in response to fluctuating sea level and the resulting changes in resource availability. The researchers identified four periods of relative high sea level stands, in the time frames between 4753 4055 cal. BP , 3309 2 982 cal. BP , 2281 1684 cal. BP , and 1532 942 cal. BP Several of these periods of relative high sea level are in disagreement with the

PAGE 23

23 Tanner (1993) curve, suggesting that localized processes in the Lower Cooper River Valley exerted significant control ove r relative sea level and those processes differed from those on the coast in Denmark. Variability in sea level data closer to the present study, in the Gulf of Mexico, underscore the need for local scale reconstructions. A 2004 publication by Balsillie an d Donoghue for the State of Florida Department of Environmental Protection illustrates that even within the relatively restricted geographic area of the Gulf of Mexico, relative sea level is highly variable (Figure 1 2). They compiled all of the available literature concerning sea level in the Gulf of Mexico to create a seven point floating average based on several mean sea level curves. Twenty three data sets from various areas along the coast of the Gulf of Mexico produced 353 dated indicators of sea le vel stands covering 20,000 years. They overlaid these data onto the seven point floating average of global mean sea level to compare global and local changes. They concluded that, overall, sea level changes in the Gulf of Mexico mirrored global sea level changes, but the indicators of localized sea level along the Gulf coast were highly variable over the past 7 , 000 years. While the general trend of data appear to follow the mean sea level curve, there is as much as six meters difference between coeval bu t geographically separated data points. To illustrate, a study by Morton et al. (2000) along the Gulf coast in New Orleans, Louisiana (indicated by an arrow pointing to the green triangle on Figure1 2) put sea level at one meter above present in that loca tion at around 1100 BP A later study by Stapor and Stone (2004) in the western Gulf off the coast of Texas (indicated by an arrow pointing to the blue square on Figure 1 2) put sea level at one

PAGE 24

24 Figure 1 2. Sea level indicators in the Gulf of Mexico wi th a 7 point floating average of mean sea level. Note: Figure from Balsillie, James H. and Joseph F. Donoghue . 2004 . High Resolution Sea Level History for the Gulf of Mexico s ince the Last Glacial Maximum. Vol. 103, State of Florida Department of Environ mental Protection, Tallahassee, FL. meter below present during the same time. These two study areas are relatively close geographically and a two meter difference along the shallow continental shelf in the Gulf of Mexico represents a significant differen ce in the amount of land area that is inundated by ocean water. This high variability is partially a product of the use of differe nt data sets as proxies for sea level; Morton et al. (2000) used wave cut platforms and above water beach ridges as proxy dat a, whereas Stapor and Stone (2004) used sediments collected in cores. Although the proxy data used in the studies cited by Ba l sill i e and Donohue (2004) may account for some of the discrepancies in sea level among the various

PAGE 25

25 stud ies , localized processes ex ert significant control over relative sea level, a point that will be revisited below. Regressions, H igher T han P resent S ea L evel, and the N ature of S ea L evel R ise Several studies have found evidence of regressions and higher than present sea levels over t he past 4,000 years. DePratter and Howard (1981) found evidence of an up to 3 meter drop in sea level along the coast of the Georgia Bight between 3 , 000 and 2 , 400 BP (DePratter and Howard 1981:1287), evidenced by radiocarbon dates on submerged tree stumps and the identification of submerged archaeological deposits. On the South Carolina coast there is evidence of a two meter or more drop in sea level around this same time (Brooks et al. 1989; Gayles 1992). Studies along the Gulf coast have found evidence for higher than present sea level stands during various periods in the past (Blum et al. 2001; Morton et al. 2000; Stapor et al. 1991; Walker et al. 1995) . Morton et al. (2000) found evidence on the Gulf Coast of Louisiana for a highstand at around 1100 B P Blum et al. (2001) argue for a highstand between 6,800 and 4,800 BP on the Gulf Coast of Texas. Stapor et al. (1991) and Walker et al. (1995) suggest a highstand between 2 , 300 and 1 , 450 BP on the southern Gulf Coast of Florida. M any of these studies h ave been criticized for using proxy data to infer marine highstands in the absence of evidence of peat formations above current sea level that are indicative of marine environments (Otvos 2004). Factors Influencing Relative Sea Level Determining the nature of sea level rise in a localized area is also problematic. Several studies in the Gulf have interpreted offshore ridges as paleoshorelines and suggested that there were long periods of shoreline stability punctuated by episodes of rapid sea level rise (F razier 1974; Nelson and Bray 1970; Thomas and Anderson 1994).

PAGE 26

26 However, further study of some of these offshore ridges found that many had been reworked by the marine environment and had migrated (Rodriguez et al. 1999), making it difficult to use these ri dges to infer still stand positions. Analysis of peat formations in sediment cores in other areas suggest gradual but decelerating sea level rise over the course of the Holocene, albeit with some variability in rates (Goodbred et al. 1998; Robbin 1984; Sc holl et al. 1969; Törnqvist et al. 2002; Toscano and Macintyre 2003; Wright et al. 2005). Marshes are highly productive environments and their response to eustatic changes in sea level is very different from that of wave dominated coastal settings. A stud y of sedimentation rates in the marsh grasses around Cedar Creek, a tidal creek in the Crystal River estuary system about 60 km south of the Suwannee delta, found that accretion in the marsh due to sediment introduction from the creek averaged 12 millimete rs per year, keeping pace with current sea level rise in that particular location (Leonard et al. 1995). A later study by Jaeger et al. (2009) investigated the Loxahatchee River in southeast Florida and found that sediment accumulation rates in this estua ry have kept pace with relative sea level rise throughout the late Holocene. Of importance is that this finding is consistent with those from many other estuary systems (Nichols 1989), suggesting that these environments tend to stay in equilibrium and min or fluctuations in mea n sea level can be mitigated or have minimal impact in estuary systems. Any condition that facilitates an increase in the amount of sediment delivered to the estuary could create a prograding shoreline where accretion outpaces sea le vel rise. In contrast, increased rates of sea level rise can create a tipping point where the ability of the system to equalize can cause significant environmental change.

PAGE 27

27 Anthropogenic impacts can also either amplify or mitigate sea level changes. Land use patterns in the drainage basin of a river system can significantly decrease or increase sediment loads and thus affect the morphology of the estuary system. Deforestation and agricultural activities (e.g., Crumely 2007; Dull 2007; Jaeger et al. 2009; Rawat and Rawat 1994) can increase erosion, which in turn increases sediment loads in rivers and streams. The addition of organic material to the river system can also affect sediment loads in the river, depending on concentrations (Dennett et. al. 1998) . Evidence suggests that horticultural activities started as early as A.D. 600 in the Suwannee Valley (Milanich 1994:215), just prior to a period of regressive seas according to Tanner. If the sediment load carried in the Suwannee River increased due to erosion in the drainage basin that could have acted to amplify falling sea levels at the mouth of the river. A regression of the same magnitude would likely not be observed in areas where less sediment is delivered to the coast. Big Bend Region Because of their close proximity and similar temporal focus, the studies by Wright et al. (2005) and Goodbred et al. (1998) are particularly relevant to this research. Wright et al. (2005) report the results of earlier dissertation work by Wright (1995) that focu sed on relative sea level change in the Suwannee Delta. The data suggest that sea level rose rapidly until 5 , 500 cal. BP , slowed until 2 , 500 cal. BP , then decreased further until around 750 cal. BP. Prior to 8 , 000 cal. BP , the area around the mouth of th e Suwannee River was a flat plane punctuated by eolian dunes that had formed during the late Pleistocene. Shoreline transgression was rapid and significant and by 5 , 400 cal. BP , the shoreline was within eight kilometers of its modern location. Increased sediment input from the Suwannee River allowed for deltaic formation by 4 , 840 cal. BP ,

PAGE 28

28 seaward of its present location. Decreasing rates of sea level rise allowed for the formation of oyster bioherms parallel to the shoreline, which trapped marine and oth er biogenic sediments and slowed shoreline retreat. Accretion of these sediments kept pace with sea level rise until around 4 , 440 cal. BP , when the shoreline moved to within 6 kilometers of its current position and a new inner oyster bioherm formed, becom ing well established by 3 , 630 cal. BP Rates of rise continued to decrease, and the previously retrogradational system transitioned to aggradational. Shoreline retreat slowed significantly, which facilitated the formation of the extensive marsh system nea r its present location by 2 , 350 cal. BP (Wright et al. 2005:635). Transgressive system tract deposits consist of thin fresh to brackish water marsh deposits, salt marsh deposits along the modern coastline, and widespread marine deposits (Wright et al. 200 5). Shoreline ravinement reworks these thin sedimentary units and seaward deposits are limited to marine or reworked coastal sediments. Some preserved stratigraphy was observed in cores collected from topographic lows in the limestone platform. The stud y by Wright et al. (2005) only briefly addresses lowstand system tract deposits, which were observed in some of the cores collected from incised river channels. The deposits are discontinuous and undated, and the authors do not offer an interpretation of these deposits. There is no evidence in the cores of fresh or brackish water marsh deposits overlying salt water marsh sediments, suggesting that these deposits could be the result of localized processes linked to the deltaic environment. The preservati on of relict dune morphologies near current sea level and evidence from analysis of the stratigraphic

PAGE 29

29 units lead the authors to conclude that relative sea level in the Suwannee Delta has never been higher than prese nt during the past 7,000 years. The simil ar work by Goodbred (1994) in nearby Waccasassa Bay suggests the timing of the evolution of the bay is consistent with that of the Suwannee Delta and provides evidence for eventful changes in sea level. Goodbred (1994) collected sediment cores from the Wa ccasassa Bay, just south of the Cedar Keys on the Gulf Coast, to investigate the development of the marsh system in that area. Similar to around 4 , 400 cal. BP , after whi ch, rates of sea level rise slowed, allowing for stability in the system and the formation of an extensive oyster bioherm. Between 1 , 800 1 , 700 cal. BP , there was a rapid transgression that caused a shoreline migration of two to four kilometers in the bay. The transgression was so rapid that it overstepped a significant portion of coastal deposits. The timing of the transgression in Waccasassa Bay coincides with the Roman Warm Period ( indicated by Marquardt in Figure 1 1 ) , when sea level globally is expec ted to have risen. H owever, the magnitude of the transgression could be partially a product of localized processes. A similar period of increased rates of sea level rise was identified by Wright (1995) around 1 , 600 cal. BP After this rapid transgression in Waccasassa Bay, rates of rise again slowed. The deposits that were overstepped during the transgression were reworked and redeposited along the marsh edges by storm energies, creating levees that effectively slow shoreline retreat (Goodbred et al. 1998 ). Hypotheses The above review of the literature on sea level change over the past 5,000 years evokes several premises: the effects of global sea level changes are not uniform across

PAGE 30

30 space, localized processes exert significant control over relative sea le vel and coastline morphology, and paleoenvironmental reconstructions that are specific to a geographic area are necessary when investigating the relationship between environmental change and human practices along the coast. With these premises in mind, se veral hypotheses can be constructed and tested using data collected from Horseshoe Cove. Hypothesis 1 Slowing rates of sea level rise allowed for the formation of the marsh and estuarine environment in the Horseshoe Cove area at approximately the same tim e that these environmental settings developed at the Suwannee Delta and Waccasassa Bay, sometime around 4,500 years ago. The low gradient of the northern Gulf Coast of Florida is particularly vulnerable to high rates of sea level rise and transgression aft er the end of the Last Glacial Maximum was rapid prior to 7,000 years ago. Between 7,000 and 5.500 years ago on the Gulf Coast of Florida, rates decreased to an average of 0.16 cm per year (Wright et al. 2005:631), however, the high rates and continued ra pid transgression likely prohibited the development of the estuarine system (e.g. Russo 2010). After 5,500 cal. BP , rates decreased to 0.07 cm per year, slowing shoreline retreat and allowing for the development of the extensive marshes and oyster bioherm s within (Wright et al. 2005:631) that would have provided sufficient resources to support significant human populations on the coast. At Waccasassa Bay, Goodbred et al. (1998) report the formation of black muds interpreted as the initial formation of swa mpy deposits by about 4,400 cal. BP (240) in the area of the modern salt marsh. Although the working premise is that paleoenvironmental reconstructions from other regions are not applicable to Horseshoe Cove, the close proximity of these two studies sugge st the coastal evolution

PAGE 31

31 of the study area may be similar to that of the these areas to the south. If that is the case, then radiocarbon dates from basal peats in the marine sediment core should be similar to those from the Suwannee Delta and Waccasassa B ay. Hypothesis 2 Lowered rates of sea level rise or a global drop in sea level resulted in a regression that forced the earliest residents in the Horseshoe Cove study area to relocate between about 3,500 and 2,500 years ago. Shifting settlement patterns in many coastal areas of the southeastern United States, including along the northern Gulf Coast of Florida between 1 , 350 550 BC (Sassaman et al. 2014:149), have been observed archaeologically and fall within the timeframe of the regression identified in t he Georgia Bight area (DePratter and Howard 1981) . Although there is evidence of aggrading and prograding surfaces in areas of south Florida, the studies at the nearby Suwannee Delta (Wright 1995; Wright et al. 2005) and Waccasassa Bay (Goodbred 1994; Goo dbred et al. 1998) found no substantial evidence for a localized regression. If the Horseshoe Cove area experienced a period of regression, then one would expect to see facies transitions suggestive of lowered water levels, for instance marsh deposits ove rlying open marine deposits or terrestrial sand strata over marsh or marine deposits in the sediment cores. Hypothesis 3 A period of increased rates of sea level rise caused a significant transgression and environmental change after around AD 200, and res ulted in changes in settlement patterns of the coastal residents of Horseshoe Cove. Goodbred et al. (1998) identify a pulse in sea level rise at Waccasassa Bay that resulted in a 2 4 km transgression in that area. Wright et al. (2005) found evidence of

PAGE 32

32 i ncreased rates of sea level rise at the Suwannee Delta sometime before AD 400 and associates that increase with the same event identified by Goodbred et al. (1998). Hypothesis 4 Specific morphological characteristics and environmental settings were targe ted for occupation as sea level rose and communities were forced to relocate further inland. Coastal residents of the northern Gulf Coast of Florida have a long history of experience with sea level rise and environmental change. As sea level rose and lan dward movement was required, areas that had specific morphological characteristics were likely targeted for colonization, including elevated areas on the landscape, a source of fresh water, protection from the higher energy open marine environment, and eas y access to resources. If coastal communities did intentionally target areas with these characteristics, then we should see initial significant occupation at sites that had those characteristics at the time of settlement. Hypothesis 5 Social and historica l factors played a role in decisions about the continued use or reoccupation of sites that were in different environmental settings than those of initial occupation. If this is the case, then there should be evidence of occupation in areas that were previ ously occupied, these sites should be in different settings than that of the initial occupation, and sites with similar environmental settings that were not previously occupied would remain unoccupied. Data to address the above hypotheses w ere collected f rom Horseshoe Cove on the northern Gulf Coast of Florida. Horseshoe Cove is located approximately 16 k ilometers n orth of the Suwannee River delta on the northern Gulf Coast of Florida (see Chapter 2 for a detailed description of the

PAGE 33

33 environmental and geol ogical setting of the study area). A constellation of three islands situated in the northern portion of this shallow water embayment contains stratified archaeological deposits, and an additional site is located approximately 1 km to the northeast on the mainland. Sedimentary data w ere collected from 13 marine sediment cores, one freshwater core, and sediment samples from selected excavation units. Archaeological data w ere collected from survey and test unit excavations at Bird Island (8DI52), Butler Isl and (8DI50), and Garden Patch (8DI4) (Figure 1 3). The reader is referred to Chapter 2 for a detailed description of the geological and environmental setting of the study area. Structure of the Dissertation This dissertation is composed of four manuscript s (hereafter referred to as chapters) that have been or are in the process of being submitted for publication. Each successive chapter contributes to the overall goal of understanding the human experience of living on the coast . The re is some redundancy among the chapters, albeit in varying levels of detail depending upon the focus of each individual manuscript. To avoid further redundancy, I refer the reader to Chapter 2 for a detailed description of

PAGE 34

34 Figure 1 3. Study area showing locations of arch aeological sites at Butler Island (8DI50), Bird Island (8DI52), and Garden Patch (8DI4), and locations of sediment cores. the study area and methods for analysis of the sediment cores. Chapters 3, 4, and 5 provide descriptions and locations of the archaeo logical sites. Chapter 2 presents the results of analysis and interpretation of the marine and fresh water sediment cores and provides a chronology of the evolution of the coastal environment in northern Horseshoe Cove. This chapter offers the greatest level of detail on the sediment cores. Chapter 3 provides details of survey and excavations on the island sites of Bird Island (8DI52) and Butler Island (8DI50), and also describes several sites that have

PAGE 35

35 either been destroyed or heavily impacted by sea le vel rise and high energy storms. At the time of this writing, Chapter 3 has been accepted for publication in The Florida Anthropologist . Given the evidence of the destroyed and impacted sites discussed in this chapter, it is important to point out here t hat the sites sampled in the study area represent only those that have survived to the present. Although much data ha ve been lost, the remaining stratified archaeological deposits contain significant cultural and environmental information that can be brou ght to bear on the questions addressed by this research project. Chapter 4 combines the data from the sediment cores, terrestrial sediment samples, and archaeological excavations to create a chronology of environmental change and human occupation of the si tes, including the mainland site of Garden Patch (8DI4). This chapter lays out the evidence for my suggestion that areas targeted for initial settlement were chosen for their particular environmental settings. Desirable attributes appear to have been e levated areas that were protected from the high energy, open marine environment by marshes but had relatively easy access to the marine resources via tidal creeks or streams. The island sites appear to have experienced periods of abandonment, but have evi dence of reoccupation even though they had very different morphological characteristics. Chapter 5 puts a humanistic perspective on the data. I suggest that, in addition to environmental factors, social factors played a role in decisions about places to s ettle and how these places would be utilized in the future. I build on the evidence previously presented in Chapter 4 that Deptford period people chose areas for initial settlement based on their morphological attributes at the time of initial occupation, and likely also

PAGE 36

36 for the anticipated future environmental setting of the area. However, decisions by later Weeden Island people to reoccupy these same island sites may have had more to do with social and historical factors than did the environmental setti ng. Although there are debates about the nature of sea level change during the Holocene, the overall trajectory has been toward transgressing seas, and sea level rise has been a normal condition since the end of the Pleistocene. The people who lived along this coast had generations of accumulated knowledge of the coastal environment and certainly had developed adaptive strategies that enabled them to successfully navigate this dynamic setting. Morphological changes on the coast inevitably occur and it is certain that coastal populations had the expectation that landward movement was an eventual necessity. The intimate understanding of the environment and a rich oral history that chronicled experiences of environmental change would have informed decisions about places to colonize when landward movement was necessary, and likely would have provided an expectation of how those places would change over time as the environment changed. This created a rolling mosaic of occupations on the landscape at Horseshoe Cove. More recent places of occupation were situated on elevated protected areas on the mainland that had relatively easy access to marine resources via tidal creeks and streams. Island sites, which had previously been mainland protected sites, continue d to be utilized even though they were isolated from the mainland and in very different environmental settings. Although the sample size is relatively small, consisting of only three test unit excavations at each of the island sites, it appears that pract ices varied among the mainland and island sites. After a period of hiatus, all of the

PAGE 37

37 sites were reoccupied during the Weeden Island period, but areas with similar environmental settings within the study area were not colonized by these later people. Thi s suggests that the evidence of previous occupation could have influenced decisions about the utilization of these places. Admittedly, this project focuses on a very small geographic area. Because one of the main goals of this research was to create a pal eoen vironmental reconstruction that was specific to a constellation of archaeological sites, the study area was necessarily restricted. However, it is important to note here that environmental and historical trajectories in Horseshoe Cove were influenced by much larger regional, and in the case of sea level, global forces. Finally, Chapter 6 provides conclusions from this study and offers suggestions for avenues of additional research that will place Horseshoe Cove within this larger context.

PAGE 38

38 CHAPTER 2 M IDDLE TO LATE HOLOCENE COASTAL EVOLUTION OF HORSESHOE COVE ON THE NORTHERN GULF COAST OF FLORIDA * I ntroduction The Big Bend region of the northern Gulf Coast of Florida is among the largest open marine marsh systems in North America. This 350 km long sh oreline is a low energy, low gradient (1:5000), sediment starved region characterized by wide marshes, numerous tidal creeks, significant oyster bioherms, and mud flats. Hine et al. ( 1988) identified and described four major morphologic sectors in this re gion. The berm ridge coast is characterized by a slowly eroding marsh, cross cut by tidal creeks, with a sandy (siliciclastic and skeletal) berm ridge and/or beach at the water marsh interface. The sands from the beach and berm ridge are transported via storm energy into the marsh, creating washovers that significantly contribute to marsh accretion. Marsh peninsulas form along rocky outcrops that protrude into the shallow marine environment. When flooded, these rocky areas become suitable substrate for oyster bioherm formation. The marsh archipelago is a constellation of marsh islands situated on topographic highs in the bedrock. Finally, shelf embayments are shallow depositional basins that are tide dominated, fresh water influenced, and marsh rimmed. Horseshoe Cove is a large embayment situated 16 km north of the Suwannee River. The cove is bordered by the town of Horseshoe Beach to the northwest, by extensive low lying forested areas and salt marsh to the north, and by Fishbone Creek to the east. T he Eocene limestone Ocala Formation underlies a thin sediment cover in * Reprinted with permission from McFadden and Jaeger (2015)

PAGE 39

39 the cove, outcropping in some areas. High points in the complex karstic topography of the limestone support small islands and the remnants of paleodunes that formed around the end of t he Pleistocene. Butler Island is the remnant of a parabolic shaped paleodune (Wright 1995; Wright et al. 2005) situated 0.8 km to the southeast of the town of Horseshoe Beach in the very shallow marine and marsh area of Horseshoe Cove. The island is bounde d by salt marsh to the north and northeast, and by shallow water and extensive oyster beds to the south and southeast. Lolly Creek flows from the marsh to the south of the main portion of the island and several small tidal creeks are located in the marsh t o the north. The two arms of the u shaped island extend to the southwest and are cut off from the central portion of the island by marsh. Two small islands, Bird Island to the north and Cotton Island to the south are located near the distal ends of the arm s (see Figure 1 3) . This relatively protected area, where little scouring from wave and storm energies occurs, provides an opportunity to investigate environmental changes at a smaller scale in the Horseshoe Cove area during the middle to late Holocene and add to the growing body of knowledge about sea level rise on the northern Gulf Coast of Florida. In additional to the construction of a chronology of marsh formation and shoreline retreat, this study also addresses questions of lowstands and higher than present sea level in this area. Geology of the Modern System . Previous studies along the Big Bend at the Suwannee River delta, and further south, at Waccasassa Bay, have shown that three major factors control shoreline morphology and rates of retreat in t his region: the amount of available sediment in the system, the topography of the underlying Eocene limestone platform, and rates of sea level rise (Goodbred 1994; Hine et al. 1988; Wright

PAGE 40

40 et al. 2005 ). The paucity of relict siliciclastic sediments along this coastline is the result of a lack of sediment input from the southern Appalachians during the late Paleogene, when the open marine seaway of the Suwannee Straits isolated peninsular Florida from areas to the north. During the Quaternary, sea level ch anges allowed for mobilization of sediments from the north that had accumulated in the Suwannee Straits, and additional sediment was transported southward via rivers and streams (Hine et al. 1988). Modern sediment input in the area near Horseshoe Cove, ho wever, is low due to the low relief of the Suwannee drainage basin. Modern accretion in the marshes is a result of the landward transport of pre Holocene sediments that were deposited in northwestern Florida during a period of dry and cool climate around t he end of the Pleistocene. Large dune fields developed throughout the southeastern United States adjacent to and paralleling coastal rivers and s treams (Markewich and Markewich 1994) during this time. Sediments accumulated in floodplains during periods o f variable discharge and seasonal flooding, after which they were mobilized by unidirectional winds and trapped on the building landforms by moderate to heavy vegetation (Ivester and Leigh 2003; Ivester, Leigh, and Godfrey Smith 2001; Moore 2009). Transgr essive seas transported these sediments shoreward, evidenced by a relative lack of sediment cover seaward of the marsh system, modern mineralogy of sediments that is similar to that of pre Holocene basal sediments, and the presence of marine microfossils i n reworked sediments (Goodbred 1994). Because of the sediment starved nature of the northern Gulf Coast of Florida, the carbonate platform is covered by only a thin veneer of siliciclastic carbonate sediment, and like the Suwannee Delta and Waccasassa Bay (Goodbred 1994; Goodbred et al.

PAGE 41

41 1998; Wright 1995; Wright et al . 2005) to the south, the morphology of Horseshoe Cove is largely dictated by the complex karstic topography of the underlying limestone. Fractures caused by dissolution from fresh water sprin gs created rectilinear depressions that became tidal creeks in the marsh, and variable dissolution of the limestone creates an irregular surface, with rocky highs that support terrestrial vegetation and lower elevations that collect sediment and can eventu ally become marshy ( Hine et al . 1988). Reconstructions from the Suwannee Delta and Waccasassa Bay (Goodbred 1994; Goodbred et al. 1998; Wright 1995; Wright et al . 2005) suggest that a decrease in rates of sea level rise facilitated the development of the m odern marsh system in the Big Bend. Prior to 5,500 BP, sea level rose at a rate of about 0.16 cm per year (Wright et al . 2005), and shoreline transgression across the low gradient shelf was relatively rapid. Pre Holocene sediments were scoured from the c arbonate platform and transported landward during shoreline retreat, leaving pockets of sediment only in karstic features (Goodbred 1994). After 5,500 BP, sea level rise slowed to approximately 0.07 cm per year (Wright et al . 2005), and sediments from the shallow carbonate shelf began to accrete vertically rather than being transported horizontally (Goodbred 1994). By 2500 BP, rates had further slowed to around 0.05 cm per year (Wright et al. 2005) and the more stable shoreline allowed time for the develo pment of the near modern marsh system, with continued accretion facilitated by transport of sediments into the marshes via storms (Hine et al. 1988). This study builds on the work by Hine et al. (1988), Goodbred (1994), Wright (1995), and others to reconst ruct the timing and nature of the evolution of the open

PAGE 42

42 marine marsh system in Horseshoe Cove. Wright ( 1995 ) reconstructs shoreline retreat after 5,500 BP; however, the deltaic environment at the mouth of the Suwannee River is very different from that of Horseshoe Cove, where there is no river drainage. The Waccasassa Bay area is more analogous to Horseshoe Cove; however, Goodbred (1994) focused more on the evolution of the marsh system, which entailed collection of vibracores primarily from the marsh and upland areas rather than from the offshore marine areas. M ethods Figure 1 3 shows the study area and locations of sediment core collection. Thirteen marine sediment cores were collected at 200 m intervals along a 2 km southwest to northeast transect usi ng a vibracore with 3 in aluminum barrels. Three additional sediment cores were collected along an east to west transect nearly perpendicular to the southwest to northeast transect, and one final core was collected using a percussion core in a fresh water pond located on the mainland near the marsh system. After penetration, the time and depths from the top of the barrel to the water and to the sediment water interface were recorded. Elevations for each core were corrected to mean low low water (MLLW) us ing the NOAA datum at Horseshoe Cove and recorded water levels at nearby Cedar Key, with adjustments made for geographic distance between Horseshoe Cove and Cedar Key. Bulk density was recorded by gamma density on a Geotek core logger . Each core was split , photographed, and described. A fter which , smear slides were produced and sediment samples were collected from the upper, middle, and lower portions of each lithostratigraphic unit. Sediment samples were dried, weighed, and fine grained materials, small er than 63 , were removed by wet sieving. The dried weight of the

PAGE 43

43 remaining sample was recorded and the percentage of fine grained materials (fines) calculated. The resulting sand fraction of the sample was processed in a settling column and sediment texture s tat istics were produced using the Folk method (1980). Percentage of organic matter was determined by loss on ignition (LOI) in a muffle furnace at 550 C for 2 h and carbonate content by LOI at 1050 C for 2 h. Mineralogy and organic components were identifie d by macroscopic and microscopic observation. Nine radiocarbon dates were obtained from the base of selected lithostratigraphic units in th e marine sediment cores, and an additional three radiocarbon dates were obtained from locations that suggested tran sitions in depositional regimes in the pond core. Nine of the samples were bulk sediment samples collected from the base of facies that were highly organic, but in which larger pieces of organic material were absent. Two radiocarbon dates were obtained o n plant material, and one from charred material. All samples were corrected for isotopic fractionation, using the delta 13C calculation. Radiocarbon dates in both radiocarbon and calibrated years for sediment cores are summarized in Table 2 1 and for arc haeological context in Table 2 2 . R esults The study area is very shallow, with MLLW depths of no more than 3 ft along the southwest to northeast transect. Sediment cover in this area is very thin, less than 150 cm in depth in the majority of the cores, all but four of which terminated at the degraded top of the Ocala limestone. Nevertheless, there was significant preservation of lithofacies with marked transitions in all of the cores.

PAGE 44

44 Table 2 1. Summary of Radiocarbon Dates from Sediment Cores Lab Sa mple Number Core # Depth in Core (cm) Approx. Elevation (cm) a Lithof acies Sample material 13C/12C Ratio 14 C yr BP 2 sigma Calibrated Age Beta 359175 2 81 180 Medium sand with muddy fine sand above oyster bed. Charred material 24.0 340 +/ 30 AD 1450 16 40 (500 310 BP) Beta 381534 3 94 62 Black muddy peat Organic sediment 19.7 1210 +/ 30 AD 600 770 (1290 1180 BP) Beta 388850 19 28 NA Organic rich muddy sand Organic sediment 24.1 1230 +/ 30 AD 680 880 (1270 1070 BP) Beta 381535 9 85 82 Organic ric h muddy sand Organic sediment 19.0 1880 +/ 30 45AD BC AD 75 (1995 1875 BP) Beta 384266* 19 45 NA Organic rich muddy sand Organic sediment 24.7 2320 +/ 30 405 370 BC (2355 2320 BP) Beta 381536* 19 73 NA Bedded sand with mud Organic material 25.7 2520 +/ 30 790 540 BC (2740 2490 BP) Beta 362269 18 93 130 Bedded sand and mud Organic sediment 19.5 2580 +/ 30 890 880 BC (2840 2820 BP) Beta 388851 11 84 80 Black muddy peat Organic sediment 22.7 3520 +/ 30 2005 1780 BC (3955 3730 BP) Beta 388848 1 6 87 112 Bedded sand and mud Organic sediment 22.0 3550 +/ 30 2030 1885 BC (3980 3835 BP) Beta 362268 9 120 113 Black muddy peat Organic sediment 23.2 3880 +/ 30 2470 2290 BC (4420 4240 BP) Beta 388849 17 113 132 Medium sand Organic sediment 22.3 4090 +/ 30 2875 2580 BC (4825 4530 BP) Beta 388847 12 80 76 Black muddy peat Plant material 25.3 4120 +/ 30 2870 2800 BC (4820 4750 BP) a Relative to modern mean low low water. * Radiocarbon dates originally reported in Wallis and McFadden (2014) and Wallis, McFadden, and Singleton (in review).

PAGE 45

45 Table 2 2. Summary of Radiocarbon Dates from Archaeological Contexts Lab Sample Number Site Prov. Material 13C/12C Ratio 14 C yr BP 2 sigma Calibrated Age Beta 388846 8DI50 TU1 Str II Charred Material 25 .3 900 +/ 30 AD 1035 1215 (915 735 BP) Beta 388845 8DI50 TU3N Str V Charred Material 22.9 1 070 +/ 30 AD 885 1015 (1065 935 BP) Beta 301594 8DI52 TU1 Str IIB Charred Material 25.7 1140 +/ 30 AD 810 980 (1140 970 BP) Beta 385477* 8DI4 Area I, TU7/8 Level C Soot from sherd NA NA AD 775 980 (1175 970 BP) Beta 385474* 8DI4 Area I, TU7/8 Fea. 14 Charred Material 26.9 1240 +/ 30 AD 715 890 (1235 1060 BP) Beta 385475* 8DI4 Area I, TU7/8 Fea 16 Charred Material 25.8 1220 +/ 30 AD 715 890 (1235 1060 BP ) Beta 385476* 8DI4 Area I, TU7/8 Fea 20 Charred Material 24.8 1250 +/ 30 AD 675 870 (1275 1080 BP) Beta 384265* 8DI4 Area X, TU1 Str III Charred Hickory Nut 24.7 1450 +/ 30 AD 560 650 (1390 1300 BP) Beta 386087* 8DI4 Mound II Str IV Charred Mater ial 24.6 1510 +/ 30 AD 435 610 (1515 1340 BP) Beta 384264* 8DI4 Area X, TU1 Fea 3 Charred Material 26.3 1560 +/ 30 AD 425 595 (1525 1355 BP) Beta 386088* 8DI4 Mound II Str VI Charred Material 26.0 1600 +/ 30 AD 405 550 (1545 1400 BP) Beta 384569* 8DI4 Mound IV Fea 5 Charred Material 27.8 1650 +/ 30 AD 395 540 (1555 1410 BP) Beta 386089* 8DI4 Mound II Str VI Charred Material 26.0 1660 +/ 30 AD 345 530 (1605 1420 BP) Beta 383175* 8DI4 Mound II Str III Charred Material 26.3 1670 +/ 30 AD 340 425 (1610 1525 BP) Beta 386496* 8DI4 Area X, TU1 Str IV/V Charred Material 24.7 1650 +/ 30 AD 340 425 (1610 1525 BP) Beta 383174* 8DI4 Mound V Fea 8 Charred Material 24.7 1670 +/ 30 AD 265 420 (1685 1530 BP) Beta 382227* 8DI4 Mound IV, TU4 Str II I//IV Soot from Sherd 24.1 1690 +/ 30 AD 255 405 (1695 1545 BP) Beta 383176* 8DI4 Mound II, TU3 Fea 4 Charred Material 25.1 1720 +/ 30 AD 240 395 (1710 1555 BP) Beta 338163* 8DI4 Mound V Fea 1 Charred Material 25.6 1730 +/ 30 AD 240 400 (1710 155 0 BP) Beta 384570* 8DI4 Mound IV, TU6 Str IV Charred Material 25.0 1720 +/ 30 AD 240 395 (1710 1555 BP)

PAGE 46

46 Table 2 2. Continued Lab Sample Number Site Prov. Material 13C/12C Ratio 14 C yr BP 2 sigma Calibrated Age Beta 382228* 8DI4 Mound IV, TU4 Str IV/V Residue from Sherd 25.2 1920 +/ 30 AD 25 130 (1925 1820 BP) Beta 388844 8DI50 TU3N Str VIIB Charred Material 25.1 2060 +/ 30 170 BC AD 5 (2120 1945 BP) Beta 301595 8DI52 TU1 Str IIIB Charred Material 24.5 2180 +/ 30 360 170 BC (2310 2120 BP) Beta 301596 8DI52 TU1 Str VB Charred Material 26.3 3910 +/ 40 2480 2290 BC (4430 4240 BP) Beta 384568* 8DI4 Mound IV Fea 9 Charred Wood 26.1 2840 +/ 30 1045 905 BC (2995 2855 BP) * Radiocarbon dates originally reported in Wallis and McFadden (2014 ) and Wallis, McFadden, and Singleton (in review) . Marine Sediment Core Lithofacies Figure 2 1 provides a schematic representation of lithofacies in each core in the southwest to northeast transect, and Figure 2 2 provides the same for the west to east tr ansect. Figure 2 3 provides detailed descriptions of selected cores. Ten basic lithofacies were identified based on color, content, and sediment texture and given descriptive designations. Following is a brief description of each lithofacies along with basic contextual information. Limestone The limestone bedrock underlying the study area has a very low gradient, rising a maximum of ten cm in elevation landward along the two km southwest to northeast transect. The top of the limestone is covered by a wh ite (10YR 8/1) to gray (10YR 5/1) weathered micritic residuum. The very fine grained sediments are cemented to well indurated and are present at the base of all but four of the marine sediment cores. The contact between the degraded limestone and facies a bove is sharp. In two of the cores,

PAGE 47

47 Figure 2 1. Schematic representation of lithofacies in each core in the southwest to northeast transect with radiocarbon dates in cal BP. HBT10 and HBT13, the upper portion of the limestone transitions to a thin l ens of bluish gray (10B 6/1 and 10B 7/2) fine grained sediments . These sediments are not represented in Figure 2 1.

PAGE 48

48 Figure 2 2. Schematic representation of lithofacies in each core in the southwest to northeast transect with radiocarbon dates in cal BP . Dark gray medium to fine sand The dark gray (5Y 4/1 to 10YR 4/1), medium to fine, non fossiliferous siliciclastic sand is present in four cores. The thickest deposits are in cores collected from topographic lows in the limestone in the deepest water be tween the arms of Butler

PAGE 49

49 Figure 2 3. Detailed lithofacies descriptions with percentage of carbonate, organic matter, and fines, and mean grain size and sorting statistics for selected marine sediment cores.

PAGE 50

50 Island. These sediments have less than 2% carbonate content and fines between 2 5%. Woody organic material is present in one core and all have occasional burrows that include sediments from overlying deposits. Despite the burrows, the contact with overlying deposits is sharp. Brown medium sand O nly present in three cores, this facies is composed of dark yellowish brown (10YR 4/4) to light brownish gray (10YR 6/2) medium well sorted siliciclastic sand with no carbonate and less than 1% organic content. Fines constitute less than 5% in the majorit y of the unit where bedding is massive and burrows are common. A radiocarbon date on a bulk sediment sample from just above the transition to this facies in core HBT17 yielded a date of 4825 4530 cal yr BP, one of the oldest dates in the study area. This core is located southwest of Bird Island in an area that appears to be a drowned tidal creek. In the lower portion of the facies in this core, the sands are interbedded with finer sediments that have slightly higher organic content and an increase in fin es. In the second core (HBT10), located at the mouth of a tidal creek that transects the northern arm of Butler Island, the base of the unit is interbedded with sediments from the top of the dark gray medium to fine sand facies. In both cores, this facie s grades upward into the gray medium sand facies. Finally, it is the uppermost facies in the core collected from the mouth of Lolly Creek (HBT4). Gray medium sand to black muddy sand This lithofacies was identified in eight cores and ranges from light gra y (10YR7/2) to black (10YR 2.5/1) in color, and contains up to 15% fines , less than 5% carbonate, and up to 10% organic matter. Shell fragments are rare, and where not heavily

PAGE 51

51 bioturbated , bedding is present. It unconformably overlies the limestone base in four cores , HBT8, HBT15, HBT16, and HBT18 . A bulk sample of organic sediment from the base of this lithofacies in core HBT18, to the southwest of Bird Island, yielded a date of 2840 2750 cal yr BP and a bulk sediment samples from the base of this facie s in core HBT16, between the arms of Butler Island, yielded an age of 3980 3835 cal yr BP. It is present in two cores collected from tidal creeks, HBT4 from Lolly Creek and HBT10 in the parabola of Butler Island. Light gray to brown medium sand These light gray (10YR 7/2) to very pale brown (10YR 8/2) medium sands contain less than 5% each of organic matter and fines, and less than 2% carbonate. This facies was identified in five cores, two of which were collected from the area between the distal port ion of the northern arm of Butler Island and Bird Island, in the relatively unprotected area. In both cores, there is a sharp contact between this lithofacies and the underlying dark gray to brown sand with shell. The approximately ten cm thick deposit i n HBT16 is the only place in which shell is present, predominately tellina spp . The top of the deposit is interbedded with the dark gray to brown fine sand overlying it. In the nearby HBT14, there is no shell in the facies and bioturbation has created a very diffuse contact with the overlying dark gray to brown fine sand. The remaining three cores in which this facies is present are all located at or near the mouths of tidal creeks. In all three cores, this lithofacies overlies black muddy sand with no carbonate, with the exception of HBT9 (see Figure 2 3 ), which has an additional area of similar deposits at a higher elevation overlying dark gray to brown sand with shell. The upper portion of this lithofacies in all three marsh cores exhibits cross and flaser bedding that include sediments from the overlying deposits.

PAGE 52

52 Black muddy sand with low carbonate content This facies is composed of black (10YR2.5/1) slightly muddy organic rich sands with very low (< 3%) to no carbonate content, no shell, and occas ional woody organic material. The percentage of fines and organic content varies in this facies with proximity to tidal creek channels. In two cores collected from areas not directly affected by tidal creek channels contain between 30 49% organic matter and 35 45% fines. In contrast, in two cores collected from areas directly in tidal creek channels, organic content is only 3 4% and fines drop to 4 8%. In a core from the relatively unprotected area between the arms of Butler Island, this facies has 5 8% organic content and 8 10% fines. Where present, this facies overlies either limestone or the dark gray medium fine sand facies on a sharp contact. Radiocarbon dates were obtain from the base of this unit in four cores, three from the area to the west of the main part of Butler Island either in the marsh or from areas near the marsh water interface, and one from the marsh landward of Butler Island. A date obtained on plant material collected from just above the sharp contact with the underlying dark gray medium to fine sand facies in HBT12 yielded an age of 4820 4525 cal yr BP. This date is contemporaneous with the date from the base of the brown medium sand facies near the southwestern shore of Bird Island, but is 40 cm higher in elevation relative to M LLW. Dates on organic sediment from HBT9 and HBT11 (See Figure 2 3 ) yielded ages of 4420 4240 cal yr BP and 3955 3730 cal yr BP respectively. Both dates, while younger, come from elevations below that of the transition observed in HBT12; however, both co res are in or near tidal channels. The facies sits above a sharp contact with the limestone in HBT9, suggesting a possible ravinement surface since the dark gray medium to fine sand facies that underlies these deposits in the other cores is absent. In HBT 11 (see Figure 2 3 ), the

PAGE 53

53 dark gray medium to fine sand facies is thinner (11 cm) than in HBT12 (19 cm) and HBT3 (30 cm), also suggesting the removal of some portion of the underlying facies. The final date comes from the core collected from the marsh land ward of Butler Island where the facies is elevationally higher than in all other cores. Organic sediment collected from the base of the facies yielded an age of 1290 1180 cal yr BP. Black muddy sand with higher carbonate content This black (10YR 2.5/1) to very dark gray (10YR 3/1) slightly muddy organic rich sand facies is present in three cores, stratigraphically above the black muddy sand with low carbonate facies. Carbonate is elevated in these deposits to 4 13%. Fines range from 2 28%, and organic co ntent from 6 31%. In all cases, this facies is interbedded at the base with the underlying light gray to brown medium sand facies, which separates it from the black muddy sand with low carbonate lithofacies. A radiocarbon date obtained on organic sedimen t from the base of this facies in core HBT9 (see Figure 2 3 ) yielded an age of 1995 1875 cal yr BP. Dark gray to brown sand with shell This facies is characterized by very dark gray (10YR 3/1) to grayish brown (10YR 5/2) mottled medium sand that contains n umerous whole and fragmented shells, predominately Tellina spp. and Macoma spp. Sediments contain 3 15% carbonate, 4 8% organic matter, and 4 15% fines. This lithofacies is the most wide spread in the study area, covering nearly the entire transect from east of Bird Island to the marsh water interface near the main portion of Butler Island, and is also the area of deepest water in the transect. These deposits overlie a variety of lithofacies, including unconformably above the limestone in the unprotected area between Bird Island and the arms of Butler Island. A radiocarbon date obtained on charred material at the

PAGE 54

54 transition above the dense shell facies in HBT2 (see Figure 2 3 ), on the northeastern shoreline of Bird Island, returned an age of 500 310 cal yr BP. The elevation of the area from which the date was obtained is the lowest in the study area, suggesting the date could be incorrect, likely due to significant disturbance along this portion of the shoreline and reworking of the sediments. Dark gray to brown fine sand With the exception of cores collected from the modern marsh area or shallow areas that have the brown medium sand facies, these deposits sit stratigraphically above all other facies. It is characterized as dark gray (10YR 4/1) to brown (10YR 3/2) fine sand with 5 15% carbonate, 5 20% fines, and 5 12% organic content. The unit has numerous burrows and few shell fragments, and generally overlies the dark gray to brown sand with shell facies on a grading to sharp contact. Dense shell in s and matrix This facies contains dense whole and crushed shell, predominately Crassostrea virginica , and other mollusks, including Littorina littorea , Mytilidae , Tellina and Macoma spp . Carbonate content is between 15 20%, with organic content and fines be tween 3 5%. Present in three cores, these deposits are overlain by dark gray to brown sand with shell in two cores and dark gray medium sand in the core collected from the mouth of Lolly Creek. Fresh Water Pond Core Lithofacies Figure 2 4 provides a schem atic representation of lithofacies identified in the core collected from the fresh water pond. Three basic lithofacies were described based on color, content, and sediment texture and given descriptive designations. Following is a brief description of ea ch lithofacies along with basic contextual information.

PAGE 55

55 Figure 2 4. Detailed lithofacies descriptions with percentage of carbonate, organic matter, and fines, and mean grain size and sorting statistics for the fresh water core (HBT19).

PAGE 56

56 Brown sand This u nit is characterized by dark grayish brown (10YR 4/2) to brown (10YR 3/2) medium sand that contains woody organic material. The facies grades from compact s ands at the base to sediments that are mottled with darker, more organic, sediments at the top. Ca rbonate content is zero and organic content is less than 2%. Fines in this unit are between 3 4%. It grades into the brown mottled sand facies above. Brown mottled sand The characteristics of this facies are similar to those of the underlying brown sand, with the exception of increased mottling and areas of linear bedding with darker sediments containing increased organic content and fines. A radiocarbon date obtained on organic material that was collected from the lowermost area of linear bedding at the base of the unit yielded an age of 2740 2490 cal yr BP. Black muddy sand The dense, black heavily organic sediments of this unit contain large roots and woody pieces of organic material. Organic content increases with elevation, rising from a low of 4% at the base to a high of 35% at the top. Percentage of fines increases significantly over that of the brown mottled sand below, ranging from 18 37%. Carbonate increases to 2%, with the very top of the facies (top of the core) increasing to 4%. The sand fraction fines upward and sediments become more poorly sorted with elevation. Microscopic observation revealed a very high density of sponge spicules in this unit. A 10 cm thick area of white sand, interbedded with darker sediments from the surrounding u nit, is present near the base of the facies. Two radiocarbon dates were obtained on organic sediment from this facies. One, from just above the transition from

PAGE 57

57 brown mottled sand, returned an age of 2355 2320 cal yr BP. The second date was from just abo ve the white sand deposit and yielded an age of 1270 1070 cal yr BP. D iscussion The evolution of the modern marine and marsh environment in the northern section of Horseshoe Cove follows the same basic sequences as seen in both Waccasassa Bay and the Suwan nee Delta, but this study provides a smaller scale reconstruction of environmental change that is specific to the northern portion of Horseshoe Cove and may shed light on localized environmental changes during the middle to late Holocene. Lithofacies desc riptions and interpretations from Wright ( 1995 ), as adapted from Hine et al. (1988), are similar to lithofacies identified in the Horseshoe Cove cores and are useful models for the interpretation of data from this study, with the exception of the tidal cha nnel and event bedding categories that have been added here (Table 2 3) . Figure 2 5 provides a schematic representation of the interpreted lithofacies for each core in the southwest to northeast transect, and Figure 2 6 provides the same for the west to e ast transect. Figure 4 4 provides a paleoenvironmental reconstruction based on the interpretations of lithofacies transitions, elevations of those transitions, and modern morphology and bathometry of the study area. The limestone bedrock along the 2 km t ransect gently slopes upward toward the mainland. The clayey degraded top of this unit was encountered in the majority of cores, and with the exception of a few karstic lows, is relatively uniform across the study area. Despite the relatively thin sedime nt cover above the limestone bedrock, the protected area around Butler Island allowed for preservation of facies transitions that were suggestive of environmental shifts and shoreline retreat during the Late Holocene.

PAGE 58

58 Table 2 3. Horseshoe Cove lithofacies interpretations, after Wright (1995), with lithology and percentage of fines, carbonate, and organic matter. Interpretation Lithology Fines (%) Carbonate (%) Organic (%) Eocene Limestone Limestone 20 45% 40 45% < 3% Pleistocene Sand Dark gray medium to fine sand 10 15% 0% 2 4% Tidal Channel Deposits Brown medium sand < 5% 0% < 1% Inter to Subtidal Sand /Mud Gray medium sand to black muddy sand 1 15% < 5% 4 10% Event Bedding Light gray to brown medium sand < 5% < 2% < 5% Fresh to Brackish Marsh Black m uddy sand with low carbonate 4 45% < 3% 3 49% Salt Marsh Black muddy sand with high carbonate 2 28% 4 13% 6 31% Restricted Marine Dark gray to brown sand with shell 4 15% 3 15% 4 8% Marine Dark gray to brown fine sand 5 20% 5 15% 5 12% Oyster Reef/Bioh erm Dense shell in sand matrix 3 5% 5 20% 3 5% With the exception of karstic lows in the limestone bedrock and areas underlying tidal channel deposits, the thin veneer of Pleistocene sediments appears to have been scoured during shoreline transgression, leaving a ravinement surface between the degraded top of the Eocene limestone bedrock and overlying sediments. Islands in Horseshoe Cove are composed of these Pleistocene sediments, many of them remnants of paleodunes that formed around the end of the LG M. A single TL date from the parabolic shaped paleodune of Butler Island yielded an age of 20 +/ 4 ka (Wright et al. 2005). The Pleistocene siliciclastic sands that were scoured during shoreline transgression, along with other marine derived clastics ( e .g. , Goodbred et al. 1998), became one of the major sources for sediments that were reworked and deposited over much of the study area. Additional sediment was likely delivered to the area via the tidal creeks and streams that drain into the Gulf of Mexic o at Butler Island. The low carbonate black muddy sand deposits represent the initial transition to freshwater marsh or swamp 400 m to the west of the main portion, and between the

PAGE 59

59 Figure 2 5. Schematic representation of interpreted lithofacies in eac h core in the southwest to northeast transect with calibrated radiocarbon dates. arms of, Butler Island, sometime after about 4820 cal yr BP. Soon after, these deposits were accumulating 200 m to the east near the modern shoreline of the island, after abo ut 4420 cal yr BP. A similar black mud facies at about the same elevation and relative shoreline proximity at Waccasassa Bay postdates the earliest Horseshoe Cove dated deposit by as much as four centuries, or because of overlap in the 2 sigma date

PAGE 60

60 Fi gure 2 6 . Schematic representation of interpreted lithofacies in each core in the southwest to northeast transect with calibrated radiocarbon dates. ranges, is more likely contemporaneous with the advent of freshwater swamp in Horseshoe Cove (Goodbred et al. 1998). Similar results were reported by Wright (1995) for the Suwannee Delta, albeit a bit earlier due to an increase in the Suwannee River discharge as the water table rose. Contemporaneous with the deposition of the freshwater swamp sediments, brown medium sand, interpreted as tidal channel deposits, began to accumulate along the southwestern shoreline of what would become Bird Island. In core HBT17, these

PAGE 61

61 deposits overlie Pleistocene sands. This lithofacies was also identified in core HBT10, which is located in an area where a tidal creek transects the northern arm of Butler Island, and in this core, the facies directly overlies the Eocene limestone. Inter to subtidal muds and sands, the dark gray muddy sands and gray medium sands, unconformably ov erlying the limestone in several cores, suggest likely scouring of any marsh deposits by a transgressive shoreline in the open areas seaward of Bird Island and between Butler Island and Bird Island. A date from HBT16, between the two islands, suggests tha t area had become inter to subtidal sometime after about 3980 cal yr BP. A date from the base of inter to subtidal muds in the most seaward core, HBT18, postdates the deposits from HBT16 by as much as a millennium. There are two possible explanations for the disparity between these two dates. HBT18 is located near what would have likely been the distal arm of the parabolic dune (Butler Island) and likely had a thicker Pleistocene sediment cover than the low lying area in the center of the parabolic landf orm. Thus, this area could have remained terrestrial even after shoreline transgression into the central portion of the study area. Subsequent sea level rise would have scoured the Pleistocene sediments from this area, resulting in the accumulation of in ter to subtidal muds sometime around 2840 to 2750 cal yr BP. This hypothesis is somewhat supported by a date from the freshwater pond in an area of the core that appears to be the onset of periodic filling of the shallow depression after about 2740 cal yr BP, suggesting the water table had risen enough for a spring to become active in the area, and thus suggesting sea level had also risen. A second hypothesis is that this area was scoured by a high energy storm event, after which the muds began to accumul ate again. Possible evidence for this comes from a storm deposit on Bird

PAGE 62

62 Island that underlies archaeological midden dating to sometime between 2310 to 2120 cal yr BP (McFadden et al. 2014). It is unknown how long Bird Island was unoccupied after this po ssible storm event, so the archaeological date must be treated as a terminus ante quem for the storm deposit. Of course, it is possible that both scenarios contributed to the later date of the seaward core. Inter to sub tidal sands and muds are the predom inate lithofacies in the more landward HBT4 core, and overlie a relict oyster bed. The upper 10 cm of this core contains tidal channel deposits. The reversal of the expected transgressive facies transitions is a result of the location of HBT4 at the mout h of Lolly Creek, which has sufficient sediment discharge to create a small delta in this portion of the study area. In the majority of the other cores, the inter to subtidal sands are overlain by restricted marine and marine deposits. The only radiocarb on date from a transition to restricted marine comes from the northeastern shoreline of Bird Island from just above a relict oyster bed. This date suggests that transition occurred between 500 and 310 cal yr BP. Given the close proximity to the shoreline , and the unprotected area from which the core was collected, it is likely that this area has experienced significant reworking by storm and wave energies. In any case, this date appears to be much later than the assumed transitions to restricted marine f acies in other areas of the transect. By 2355 to 2320 cal yr BP, the water table had risen enough for the fresh water pond on the mainland to become a permanent water feature. The transition is marked by a change from medium sands with low percentages of fines and carbonate content to highly organic, muddy sediments that contain chunks of organic material and roots. The white sand deposits that are interbedded with the pond deposits suggest input of

PAGE 63

63 terrigenous materials, and a date obtained from organic sediment just above the white sands yielded an age of 1270 to 1070 cal yr BP. The pond is located in the center of a large pre Columbian archaeological site. Radiocarbon dates from anthropogenic deposits on the shore of the pond place the initial occupat ion at 1925 to 1820 cal yr BP, several centuries after the pond became permanently filled. The latest date for occupation of the site near the pond is 1150 to 970 cal yr BP. It is curious that the dates bracketing the white sands in the pond core encompa ss the entirety of the pre Columbian occupational history of the site. This suggests that the input of sand could be due to human activities, including land clearing and significant mound construction, that mobilized sediments and made them available for transport into the pond. Evidence of salt marsh deposits were identified in HBT9, HBT11, and HBT3. In all three cores, the salt marsh deposits are separated from the underlying freshwater to brackish marsh deposits by a stratum of light gray to brown med ium sand, designated as event bedding. There is a sharp contact between the freshwater deposits and these sediments, which then are interbedded at the top with sediments from the overlying salt marsh deposits. Chronostratigraphic control is lacking for t hese strata, however, it appears that they may be the product of similar processes. HBT9 has two such strata. A radiocarbon date from just above the lowermost sandy stratum places salt marsh deposition between 1995 and 1875 cal yr BP. At Waccasassa Bay, Goodbred, Wright, and Hine (1998) documented a change in lithofacies in several landward cores, which was indicative of the earliest marine influence in this area of the bay. The authors suggest that this was the result of a pulse in sea level rise betwe en 1800 and 1700 cal yr BP, with rates of rise a magnitude of order higher than previous that resulted in a 2 4

PAGE 64

64 km transgression in that area (Goodbred et al. 1998). It is possible that the sandy stratum in the lower portion of HBT9, separating the two ma rsh deposits, is a result of increased sediment input from the mouth of the nearby tidal creek, caused by wetter conditions that coincided with a transgression in Horseshoe Cove around this same time . Even though the basal date from fresh water marsh in H BT11 is later than that from HBT9, the sandy stratum in that core could date to the same period. It is also possible that the sediments could be the result of tidal creek avulsions during periods of flooding or increased discharge. It is curious that the same type of facies separates fresh from salt marsh in HBT3, where the fresh water marsh deposits are much later: 1290 to 1180 cal yr BP. However, given its location in a tidal creek, this could be a product of the same type of increased sediment input fr om the tidal creeks that deposited the sandy stratum in HBT9 and HBT11, and could also account for the upper sandy deposit in HBT9. One final possibility is that the event bedding identified in HBT9 and HBT11 could be indicative of a period of progradation in the marsh areas around Butler Island sometime before the 1800 to 1700 cal yr BP pulse (Goodbred et al. 1998). Leonard et al. (1995) showed that marsh accretion along the west central Florida Gulf coast is keeping pace with current rates of sea level r ise, and any reduction in these rates could allow for accretion to outpace sea level rise. A significant reduction in rates of sea level rise after about 3500 yr BP resulted in aggradational and progradational deposits in other areas along the Florida Gul f Coast (Evans et al. 1985; Parkinson 1989), and this same reduction of rates may have contributed to the event bedding.

PAGE 65

65 Cores HBT14 and HBT16 also have strata of event bedding. Both are located in the relatively unprotect area between Bird Island and But ler Island. The event bedding in HBT16 contains a moderate density of Tellina and Macoma spp. s hells . Grains of 1.5 phi, the median and mean grain size within the sand fraction of these deposits, are present in samples above and below this stratum. Decr eases in frequencies of smaller grains in the sand fraction, and a decrease in percentage of fines, suggest that smaller sediments have been winnowed out by higher energy. This is further supported by the consistency of carbonate content with the facies b elow. In contrast, the event bedding in HBT14 is clean sand with no shell and no carbonate that sits above a sharp contact with the underlying restricted marine deposits at 15 to 25 cm below the top of the core. The lack of shell and carbonate suggests th ese sediment are likely of terrigenous origin; however, this core is not located in the immediate area of a tidal creek, suggesting deposition of these sediments during a high energy storm event. Goodbred and Hine (1993) suggests that storm beds are oblit erated by bioturbation in the offshore environment relatively quickly in Waccasassa Bay. It is likely that storm beds would have the same fate in Horseshoe Cove, which suggests that, if this is a storm deposit, it is relatively recent. Lowstands and Highe r Than Present Sea Level . Sea level oscillations during the late Holocene that resulted in lowstands and higher than present sea level have been suggested along the coasts of the southeastern United States ( e.g. , Brooks et al. 1979; Depratter and Howard 1 981; Marquardt 2010; Tanner 1991; Stapor et al. 1991; Walker et al. 199 5). Wright (1995) suggests that decelerating rates of sea level rise before about 2,500 cal yr BP allowed for a transition to a slightly progradational, or at

PAGE 66

66 least an aggradational, s horeline in the area of the Suwannee delta. Similar progradational sequences have been identified in other areas of the Florida Gulf Coast that have been attributed to these lower rates that allowed for biogenic sediment accumulation to outpace sea level rise after 3,500 yr BP ( Evans et al. 1985; Parkinson 1989). Because of the thin sediment cover on the western Florida shelf, coastal lithosomes are typically destroyed during marine transgressions (Hine 1988), with the exception of sediments that accumula te in topographic lows in the limestone platform. In Horseshoe Cove, cores collected from the protected areas in the marsh provide the best opportunity to identify a regressive sequence. The event bedding that was identified above the fresh water marsh c ould be interpreted as a progradational sequence during a time of decelerating rates of sea level rise that allowed for sediment accumulation near the mouths of the tidal creeks. The subsequent salt marsh then developed after about 1995 cal yr BP, when ra tes of sea level rise increased. No progradational sequences were identified in any of the other cores and better chronostratigraphic control is needed to determine the nature of the event bedding above the fresh water marsh deposits. Several studies alo ng the Gulf coast have found evidence for higher than present sea level stands during various periods in the past ( Blum et al. 2001; Morton et al. 2000 ; Stapor et al. 1991 ; Walker et al. 1995). Many of these studies have been criticized for using proxy da ta to infer marine highstand s in the absence of evidence of upland peat formations that are indicative of marine environments (Otvos 2004). Wright et al. (2005) concluded that there was no evidence of higher than present sea level in the Suwannee Delta ba sed on the preservation of Pleistocene landforms (relict paleodunes) that are

PAGE 67

67 now islands. No cores were collected in the upland areas of Horseshoe Cove, however, archaeological investigations on Bird Island (McFadden and Palmiotto 2012; McFadden et al. 2 014), Butler Island (McFadden 2014), and Garden Patch (Wallis and McFadden 2013 , 2014) included shovel testing, excavation of test units, and the collection of sediment samples from below anthropogenic deposits. Test units at all sites terminated in the c ulturally sterile sand deposits of the Pleistocene landforms. None of the excavation units had evidence of the formation of peaty deposits, despite good stratigraphic preservation of both natural and anthropogenic deposits. Analysis of sediment samples c ollected at 2.5 cm intervals in continuous columns from the profiles of each unit likewise show no evidence of the development of deposits suggestive of marsh or swamp development. The preservation of the relict paleodune, Butler Island, and the lack of ev idence of peaty deposits in nearshore upland areas suggest that sea level has not been higher than present in Horseshoe Cove during the middle to late Holocene. C onclusions In the northern portion of the low energy embayment of Horseshoe Cove, fresh wat er marsh and swamp deposits dating to about 4820 cal yr BP are indicative of the initial flooding of the cove due to sea level rise and the elevated water table. The timing of this initial flooding is consistent with other areas of the coast to the south. Concurrent with the fresh water marsh/swamp development, a tidal creek began to flow near the southwestern shoreline of Bird Island. Continued sea level rise resulted in the development of a tidal flat between the arms of Butler Island sometime before 3 835 cal yr BP, and inter to subtidal muds were accumulating seaward of Bird Island by at least 2750 cal yr BP. Event bedding above fresh water marsh deposits in the cores near

PAGE 68

68 Butler Island suggest some type of environmental shift that resulted in either increased sediment discharge or avulsion of tidal creeks, or decelerating rates of sea level rise that allowed for progradation before 1995 cal yr BP. The transition to salt marsh above the event bedding suggests a rapid flooding event that is likely link ed to the observed accelerated rates of sea level rise in the Suwannee Delta and Waccasassa Bay (Goodbred 1994; Wright 1995). The area landward of Butler Island finally transitioned to fresh marsh or swamp after 1290 cal yr BP, evidenced by the fresh mars h deposits above Pleistocene sands, and sometime later, salt marsh deposits began to accumulate. Lower rates of sea level rise since the 1800 1700 cal yr BP spike have allowed the Horseshoe Cove area to remain remarkably stable into the modern era. The s tudy at Horseshoe Cove adds to the body of work along the northern Gulf Coast of Florida and has the potential to create a more fine grained chronology of the evolution of the coastal environment in this area. The outstanding preservation of lithostratigr aphic units along the 2 km transect allowed for a smaller scale of inquiry and the opportunity to investigate environmental change at the microscale. The study area has experienced significant changes during the middle to late Holocene, transitioning from terrestrial to marine over much of the area. The majority of these changes have occurred during periods of accelerated rates of sea level rise that punctuated the longer periods of system stability. Significant pre Columbian archaeological deposits are l ocated on the islands and the mainland in Horseshoe Cove and the paleoenvironmental reconstruction from this study can address questions of human adaptations to environmental change through time in this specific area. For instance, the period of increased rates of sea level rise

PAGE 69

69 between 1800 1700 cal yr BP, identified by both Goodbred (1994) and Wright (1995), and supported by this study, coincides with a significant shift in aboriginal settlement patterns and cultural expression in Horseshoe Cove. Furthe r research in the area is warranted to address questions of the impacts of modern increasing rates of sea level rise on both the environment and humans who live and work on the coast.

PAGE 70

70 CHAPTER 3 ARCHAEOLOGICAL INVESTIGATIONS OF THREATENED STRATIFIED SITES IN HORSESHOE COVE, NORTHERN GULF COAST, FLORIDA * Introduction Horseshoe Cove is a large shallow water embayment located within the open water marsh system of the Big Bend of Florida. This area, designated as the Horseshoe Beach Tract (Figure 1 3 ), is pa rt of the research area that is the focus of the Lower Suwannee Archaeological Survey (LSAS) that was initiated in 2009 by the Laboratory of Southeastern Archaeology (LSA) (Sassaman et al. 2010). A constellation of three islands in the northern portion of the cove have reported archaeological sites, two of which contain stratified archaeological deposits that are currently threatened by sea level rise and erosion from boat wakes and high energy storms. Survey and test unit excavations on these islands ful fill the LSAS goals of salvaging threatened archaeological data and adding to the poorly understood pre Columbian history of occupation in the Big Bend region of Florida. In addition to the overall LSAS goals, this area has been targeted for investigation as part of a research project that seeks to understand how environmental change, specifically sea level change, affected human/landscape relationships in the past. The geoarchaeological research included the collection of marine sediment cores and the exc avation of directly associated stratified sites. Butler Island, a U shaped paleodune remnant, provided a relatively protected area between the southwesterly * Reprinted with permission from McFadden (2015a) .

PAGE 71

71 reaching arms where transitions indicative of environmental shifts were preserved in sediment core s and the data could be directly related to archaeological data from Butler Island NE (8DI50) and the nearby site at Bird Island (8DI52). An additional core was collected from a fresh water pond at the mainland site of Garden Patch (8DI4). Bird Island con tains stratified archaeological deposits that span the last 4000 years, with the oldest deposits dating to the Late Archaic Period and subsequent occupations during the Deptford and Weeden Island Periods. The earliest deposits on Butler Island date to the Late Deptford/Early Swift Creek period, with Weeden Island period deposits also present. During the occupational history of these islands, the surrounding environment transitioned from terrestrial to marsh, and finally to open marine, and the occupied ar eas of Bird and Butler that were once connected to the mainland became islands. On the mainland, the village mound complex of Garden Patch borders a large salt marsh and tidal creek, and contains stratified archaeological deposits that span the Deptford t hrough Weeden Island periods (Wallis and McFadden 2014). This report provides the results of survey and test unit excavations at Bird Island (8DI52) and Butler Island NE (8DI50) between 2011 and 2014, along with a brief overview of survey results for addit ional reported sites in northern Horseshoe Cove. Table 2 2 provides a summary of radiocarbon dates for archaeological contexts in Horseshoe Cove . Technical reports are available for Bird Island and Butler Island, and they detail the methods of excavation and analysis for each (McFadden 2014; McFadden and Palmiotto 2012; McFadden et al. 2014). Results from analysis of the marine sediment cores (McFadden and Jaeger 2015) and sediment samples from

PAGE 72

72 archaeological context (McFadden 2014; McFadden et al. 2014) , when applicable, will inform the discussion of the site investigations here. A brief discussion of results of archaeological investigations at the mainland site of Garden Patch (Wallis and McFadden 2014), which was listed on the National Register of His toric Places in 1991, will also be included. Environmental Setting Horseshoe Cove is situated approximately 16 km north of the Suwannee River delta in the Big Bend region of the northern Gulf Coast of Florida. The cove is bordered on the northwest by a pe ninsula on which the mainland town of Horseshoe Beach is located, to the north and east by extensive salt marsh and low lying forested areas, and to the south by Fishbone Creek. Located in the northern portion, the horseshoe shaped Butler Island, along wi th Bird and Cotton islands are the most prominent island cluster in the cove. To the south, numerous other smaller islands and hammocks are encompassed by this in this marshy cove. The limestone substrate in this area underlies a relatively thin sedimen t cover. Diss olution and collapse of the limestone has produced complex karst topography that plays a major role in the morphology of the shoreline. In addition to the karst topography of the area, many of the small islands protruding from the shallow wa ters along the coastline are remnants of relict paleodunes and sand deposits that accumulated around the end of the Pleistocene when the climate was cooler and drier. Compositionally, these inland dunes are accumulations of aeolian (wind borne) quartz san ds that were transported to nearby floodplains by rivers and streams and later mobilized by unidirectional winds and trapped on the elevated landforms by moderate to

PAGE 73

73 dense vegetation (Ivester et al. 2001; Ivester and Leigh 2003; Markewich and Markewich 199 4). Oyster beds and larger reefs characterize the offshore areas of Horseshoe Cove. The eastern oyster ( Crassostrea virginica ) thrives in the estuarine environment of the cove in both subtidal and intertidal zones, and it is not uncommon for oyster reefs to be exposed during periods of low tide since they tend to cluster in depths of less than 3 meters of water. Outcrops of limestone exposed by the scouring of the thin sediment veneer during shoreline transgression provide optimal substrate for oyster col onization, particularly in areas of faster moving nutrient rich currents (Hine et al. 1988; Kilgen and Dugas 1989). Dixie County owned Butler Island, the largest of the three islands, is the remnant of a parabolic paleodune (see Figure 1 3 ) . A thermolumin escence date reported by Wright et al. (2005:626) placed the formation of Butler Island at 20 ± 4 ka, a date that is consistent with dates reported for other paleodunes throughout the southeastern United States (Markewich and Markewich 1994; Ivester et al. 2001). The distinctive U shape of the landform and the southwesterly reaching arms of the island are consistent with other similar landforms observed to the south near the mouth of the Suwannee River and at Cedar Key (Wright et al. 2005). Bird and Cotto n islands, both privately owned, are situated at the distal ends of the northern and southern arms of Butler Island, respectively. Bird Island Of the three islands, Bird Island has received the most archaeological attention. It was originally reported by Goggin in 1954 and designated as site number 8DI52. Because Goggin (1954) described the site as a small shell midden containing pottery

PAGE 74

74 sherds, it did not attract much attention until human and cultural remains began eroding from the southwestern shorelin began after construction of a channel that interrupted the longshore currents that additional erosion, exposing the mid den and human interments. One of these burials was excavated in 1986 by Julian Granberry , a local archeologist who lives in the mainland town of Horseshoe Beach. In his letter dated July 17, 1986, to Mr. Warren Nelms, Granberry writes : I excavated an Ora nge Period burial just below the high tide line on the 55 gentleman. The Norwood type pottery in association with the burial led Granberry to assume that the burials likely dated to the Orange Period. An uncalibrated radiocarbon date on human bone, obtained by him, yielded an age o f 4570 ± 100 BP (Stojanowski and Doran 1998:139). This date is problematic because it was not corrected for 12/13C fractionation or local reservoir effect, but here I used the IntCal13 4.2 database (Reimer et al. 2013) to roughly calibrate the date to 363 0 to 2942 BC. Despite the Norwood pottery recovered by Granberry, this date suggests that the burials likely date to the preceramic Archaic on the northern Gulf Coast of Florida. Cultural materials continued to erode at the shoreline over the next decade, prompting a multidisciplinary team from Florida State University (FSU) to survey the island and the exposed midden in 1993. Within weeks of the survey, and prior to

PAGE 75

75 Cove, causing substantial damage to Bird Island and destroying a significant portion of the midden. In late 1993, FSU archaeologists returned to excavat e into the small portion of remaining midden and recovered human remains and associated artifacts (Stojanowsk i and Doren 1998). salvaged by FSU archaeologists, which supports the early radiocarbon date and a preceramic Archaic age. Analysis of the recovered remains of 36 individuals suggests that they were robust people who enjoyed relatively good health and relied on a marine based diet (Stojanowski and Doran 1998). impact on the archaeological depos its on the southwestern shoreline. He concluded that the scoured midden from the southwestern shoreline, along with a significant amount of sand from nearby offshore bars, had been redeposited at the higher elevations of the island. After placing random hand cores in the upland portion of the island near the southwestern shoreline, he reported up to 1.5 meters of new sand in some areas, topped by a thin layer of redeposited midden materials, and no intact archaeological deposits on this area of the island . A substantial amount of soapstone was recovered from the eroding midden on the southwestern shoreline by the owners of the island and also during excavations by Florida State University. Dasovich (1999) initially reported that 66 kg of soapstone was rec overed, making this one of the largest reported soapstone assemblages in Florida. However, subsequent research revealed that this is a typographical error and the correct weight of the soapstone is 6.6 kg. Yates (2000) analyzed several sherds from

PAGE 76

76 the Bi rd Island assemblage using inductively coupled plasma mass spectroscopy (ICP MS) to identify and compare the signature s of rare earth elements (REEs) in an attempt to identify possible sources for the soapstone. His results were somewhat ambiguous, but he found enough similarities to suggest a possible match with soapstone sources in Spartanburg, South Carolina (Yates 2000:62). A radiocarbon date obtained by Sassaman and reported by Yates (2000:125) from soot on one of the sherds yielded an age range of 4 143 3722 cal BP. Current Research A crew from the LSA conducted systematic shovel testing and excavated two 1 x 2 m test units at Bird Island in May, 2011. Shovel testing at 20 m intervals in a transect across the highest elevation of the island revealed shell deposits of varying depths, with intact deposits identified at the highest elevation of the island, near a modern house. Two areas were targeted for additional subsurface testing. Test Unit 1 was placed at the highest elevation of the island, where deposits extended to a depth of 110 cm below surface. Test Unit 2 was placed in an area where soapstone and a Newnan like biface were recovered in a shovel test from below what appeared to be a buried A horizon. In March, 2013, an additional 1 x 2 m te st unit, Test Unit 3, was excavated near the location of Test Unit 1 to extend the profile from the previous excavation and collect additional data. Figure 3 1 provides a topographic map of Bird Island showing the locations of the test units. Test u nit 1 (TU1) and t est u nit 3 (TU3) Because TU3 extended the excavated area of TU1, these two 1 x 2 m units are reported as one analytical unit. TU1 was placed immediately to the east of the house and TU3 was placed to the east of, and aligned with the southeast corner of TU1. The

PAGE 77

77 Figure 3 1. Topographic map of Bird Island (8DI52) showing locations of Test Unit 1, Test Unit 2, and Test Unit 3. The house, boardwalk, dock, and seawall are shaded. placement of this unit extended the profiles from TU1 to the so uth and east, resulting in a 4 meter long continuous profile from east to west and a 2 meter long profile from north to south. Three main strata of intact midden deposits dating to the Weeden Island, Deptford, and Late Archaic periods, based on diagnosti c materials and radiocarbon dates, were identified in the units (Figures 3 2 and 3 3 ). The Late Archaic midden is separated from the upper Deptford and later Weeden Island deposits by a culturally sterile sand stratum of variable thickness. The uppermost Weeden Island midden

PAGE 78

78 Figure 3 2. East profile of Test Unit 1 with stratigraphic designations and radiocarbon dates , 8DI52. Figure 3 3. North profile of Test Unit 3, 8DI52.

PAGE 79

79 contained a moderate density of oyster shell with lesser densities of oth er bivalves and gastropods, faunal remains, lithics, and pottery. The upper portion of this stratum yielded modern debris in addition to a mix of pottery types diagnostic of different temporal periods, including Deptford Linear Check Stamped, Pasco Plain, Swift Creek Complicated Stamped, St. Johns Plain, Tucker Ridge Pinched, and Weeden Island Incised. The intact Weeden Island deposits below the disturbed portion of the midden contained predominately sand ed sherd and a Carrabelle Punctated sherd were also recovered from this stratum. An AMS assay obtained on charred material collected from a bulk sample at the base of the deposits yielded an age estimate of 1140 ± 30 BP (Beta 301594; charred material; 13 C = 25.7 ), or a two sigma calibrated date range of AD 810 980. An increase in shell density and a change in content marks the transition to the underlying Deptford Period midden stratum. This midden contained predominately oyster shell, with lesser frequ encies of other bivalves and gastropods, and faunal material. There is an increase in shell tools and a decrease in lithics in this stratum. In additional to sand tempered plain sherds, a Deptford Linear Check Stamped sherd and a fiber tempered sherd wer e recovered from this stratum. Charcoal retrieved from a bulk sample at the base of this stratum yielded an AMS age estimate of 2180 ± 30 BP (Beta 301595; charred material; 13 C = 24.5 ), or a two sigma calibrated date range of 360 170 BC. The culturally sterile sand stratum that underlies the Deptford midden was thicker in TU3, allowing for the collection of sediment samples at 2.5 cm intervals in a continuous 30 cm column. Analysis of these samples included, percentage of fine -

PAGE 80

80 grained materials (< 63 m), sediment texture, and microscopy. The identification of microstratigraphy, and a high frequency of sponge spicules observed in the samples, suggests that the sediments in this stratum were transported from an aquatic environment and may be storm surg e or flood deposit s, although more testing is needed to verify this interpretation since other factors may have contributed to the deposition of this stratum . This lowermost shell bearing stratum contained predominately periwinkle ( Littorina sp. ), but also had a high density of oyster shell and a marked increase in gastropod shells and faunal materials. Lithics, all flakes, increased in frequency and a large adze made on a lightning whelk ( Busycon contrarium ) was recovered (Figure 3 4 ). With the exception of three fiber tempered sherds and one crumb sherd, this stratum yielded no pottery. Charcoal recovered from a basal bulk sample in this stratum returned an AMS age estimate of 3910 ± 40 BP (Beta 301596; charred material, 13 C = 26.3 ), with a two sigma calibrated date range of 2480 2290 BC. The estimated date from Granberry predates these deposits by at least 500 years, and because the date was not corrected for fractionation, the antiquity of the cemetery could be even greater. The date from the soaps tone reported by Yates (2000) slightly post dates the initial deposition of midden materials, suggesting that the soapstone was being imported to the island by the individuals who deposited those materials. Three features were identified in TU3. Features 1 and 2 appeared to emanate from near the top of the culturally sterile sand stratum and did not contain materials from the midden above, suggesting they were dug and filled after the event that deposited the sediments but prior to the deposition of the De ptford Period midden. Both were pit

PAGE 81

81 Figure 3 4. Gastropod Adze recovered from Level I of Test Unit 1, 8DI50. features that contained vertebrate faunal remains and sparse shell. Feature 1, identified at 83 cm below surface, was a round pit that measur ed 45 x 39 cm in plan and extended to a depth of 108 cm below surface. It contained no artifacts and did not intersect with the Late Archaic deposits. Feature 2, identified at 82 cm below surface, was also a round pit that measured 50 x 50 cm in plan and extended to a depth of 91 cm below surface. This feature contained a cluster of Orange Incised fiber tempered pot sherds at the base, where it intruded into the emerging Late Archaic midden stratum and intersected with Feature 3. Analysis of the content s of Feature 2 suggest that the fill is a combination of the culturally sterile sediments and materials from the Late Archaic midden that the feature intersects, which could suggest that the Orange Incised pottery was displaced from the earlier deposits. Alternatively, the pottery may be

PAGE 82

82 associated with the digging of the pit, which would then suggest that the pit dates to the Late Archaic and that this area remained unoccupied for some period of time after the deposition of the sterile sand and before the Deptford occupation began. Feature 3 was initially identified at 91 cm below surface as an area of dense shell that contained predominately oyster with lesser quantities of scallop. The feature also had a higher density of crown conch than in the upper m idden strata. The amorphous feature did not emanate from the culturally sterile sand stratum above, but appeared to emerge from the Late Archaic deposits below. At its center, the feature contained 15 large unmodified lightning whelk shells that were tig htly clustered together (Figure 3 5 ), suggesting that they had been in some type of bag or basket. The whelk shell cluster measured 38 x 30 cm and extended to a depth of 101 cm below surface. The surrounding fill also contained a chert flake, a small bif ace fragment, and three fiber tempered sherds. Upon further excavation, it appears that the shell bearing deposits around the gastropod cluster feature were likely the top of the Late Archaic deposits rather than being part of the feature. Test u nit 2 (TU 2). This 1 x 2 m test unit was located approximately 55 meters to the west of TU1 at a lower elevation on the island. It had two distinct macrostrata (Figure 3 6 ). The upper stratum consisted of redeposited sands capped by a veneer of redeposited midden materials. The lower stratum consists of a buried A horizon and underlying shell free sand that contained sparse pottery and lithics, predominately flakes. The recovery of Ruskin Dentate and Carrabelle Punctated sherds from stratigraphically below Deptfo rd Linear Check Stamped sherds, and the lack of intact midden deposits, suggests that this lower unit has also experienced disturbance and reworking. A large area of dense

PAGE 83

83 Figure 3 5. Whelk cluster in Feature 3 in Test Unit 3, 8DI52. Figure 3 6. North profile of Test Unit 2, 8DI52.

PAGE 84

84 charcoal was identified in this lower stratum, but it is interpreted as a burned stump. The stratigraphy revealed in this test unit matches the description by Dasovich (1999) of redeposited sands and midden materials and it is likely that this was the area that he targeted when investigating the 1993 Because of the lack of intact deposits, no radiocarbon dates were obtained from this unit. Material c ulture Because the area o f TU2 has experienced significant disturbance and did not contain stratigraphically intact deposits, the material culture recovered during excavation of the unit is not addressed in this discussion. Frequencies of identifiable pottery types for TU1 and TU 3 combined and TU2 are provided in Table 2. A more detailed listing of pottery by level is available in McFadden and Palmiotto (2012) and McFadden et al. (2014). Three hundred ten sherds were recovered from the two test units combined, 153 of which were crumb sherds, leaving 157 sherds available for analysis. Sand tempered plain sherds dominate, constituting 34 percent of the assemblage. Diagnostic pottery types recovered from the midden strata are consistent with the radiocarbon dates obtained from eac h. Both test units combined yielded a total of 11 fiber tempered sherds. One was recovered from near the base of the Deptford midden in TU1, above the culturally sterile sand stratum, and one from the disturbed materials at the surface in TU3. An additi onal three sherds were recovered from the top of the Late Archaic deposits and an additional six sherds from Feature 2 and 3 that are associated with the Late Archaic deposits.

PAGE 85

85 Diagnostic types in the Deptford Period midden included Deptford Linear Check S tamped, Pasco Plain, and Swift Creek Complicated Stamped sherds. Only three St. Johns sherds were recovered from this stratum, one plain and two unidentifiable (UID) sherds. The Weeden Island Period midden contained a variety of diagnostic sherds, includ Carrabelle Punctated. Pasco Plain and St. Johns Plain and Check Stamped sherds were also present in this stratum. Lithics for both test units combined totaled 129 artifacts, the majority of which were chert flakes and burned limestone. Level A from both units, which contained modern debris and redeposited midden materials, yielded a modified flake, a spokeshave, two biface fragments, and a Hernando type biface. A broken hematit e bead (Figure 3 7 ) was recovered from near the base of the Weeden Island midden, a biface fragment was retrieved from the Deptford midden, and an additional biface fragment was recovered from Feature 3. A total of 55 chert flakes were recovered during ex cavation, the majority of which came from the Late Archaic stratum. Although not collected during excavation of TU1, the bulk of the lithic assemblage was burned limestone. Sixty two pieces (591.6 g) of burned limestone were recovered from TU3, all of wh ich came from anthropogenic deposits. Modified shell artifacts included seven gastropod hammers, seven modified columellae, and a large lightning whelk adze. The adze (Figure 3 4 ) was recovered from the Late Archaic deposits along with four of the gastrop od hammers, with the remaining three coming from the Deptford midden. With the exception of two from the near surface deposits, all of the modified columellae were recovered from the Deptford

PAGE 86

86 Figure 3 7. Two views of a hematite bead recovered from TU3 , Bird Island . midden deposits. Finally, a bone pin was recovered from the Late Archaic midden deposits in TU3. Analysis of faunal remains recovered in bulk samples and selected excavation levels in TU1 and TU3 indicates that the majority of the exploited resources came from the surrounding brackish to marine environment, although the particular species targeted and the methods of processing appear to be different. In addition to oyster, sea catfish, toadfish, killifish, pinfish, and silver perch were com mon in all three midden strata. Periwinkle ( Littoraria sp. ) was heavily exploited during the Late Archaic, with the MNI of these small littoral snails surpassing the MNI of oyster in the Late Archaic midden. The species distribution in the midden deposit s suggests that r esource procurement was selective and occupation was multiseasonal. Processing of fish

PAGE 87

87 differed from later occupations with apparently whole fish skeletons being deposited in the midden (McFadden and Palmiotto 2012; McFadden et al. 2014). The Deptford assemblage suggests multiseasonal occupation and an increased emphasis on killifish. Periwinkle is rare and it is obviously not a targeted resource during the Deptford period. Periwinkle remains rare in the Weeden Island midden deposits and oysters appear to have been processed in some way to remove incidental species, such as barnacles and odostomes. Unlike the earlier occupations, the Weeden Island faunal assemblage suggests primarily warmer weather occupation of the site. Butler Island S ite 8DI50, originally named the Lolly Creek site by Goggin in 1954, was described on the site report as a 200 by 300 ft (61 by 91 m) oyster shell midden on an old dune that extended to a depth of 12 ft in some areas. The midden was eroding along the shore line exposing black, consolidated soil that contained shells and potsherds. Based on pottery types observed in the eroding midden, the site was determined to be Deptford in age. No map of the site location was produced, but Goggin describes the site as bei ng on the right bank at the mouth of Lolly Creek, south of Horseshoe, on the west side of Horseshoe Cove. In 1986, as part of the Dixie County Archaeological Reconnaissance project, Timothy A. Kohler and G. Michael Johnson ( 1986 ) revisited the site in the company of and collected Deptford Simple Stamped, Cross Simple Stamped, and Linear Check Stamped sherds on the surface, which supported the Deptford classification. Th ey also recovered Swift Creek Complicated Stamped and St. Andrews Complicated Stamped

PAGE 88

88 sherds, along with Pasco Plain sherds, prompting them to extend the temporal range of the site into the Swift Creek period. Lithic artifacts included a limestone plummet, a biface, and debitage. The report by Kohler and Johnson (1986) also mentions two structures that were located in the area south of the site. Because a second site was identified on the southern arm of the island, Kohler and Johnson recommended that the name be changed to Butler Island NE. Current Research In March of 2014, a LSA crew conducted survey and test unit excavations at Butler Island NE (8DI50). The site was mapped using a total station and two established datums (Figure 3 8 ). Two main area s of midden were identified by augering and shovel test pit excavation. Locus A is located in the eastern portion of the site. One of the two structures mentioned by Kohler and Johnson (1986) is located at the highest elevation in this locus and is situa ted above an area of significant stratified midden deposits. These deposits are truncated by the eroding shoreline and marsh area to the south and east. Shallower intact deposits were located in the western portion of Locus A. Locus B is in the western portion of the site and has a very shallow, but intact, area of midden. The second structure is located to the north of the midden in this area. Overall, the site parallels the shoreline for about 70 meters, and from the eroded escarpment, it extends int o the upland unit of the island approximately 10 15 meters. Three test units were excavated in areas that augering had revealed intact midden deposits, two in Locus A and one in Locus B. The following is a brief description of the results from each test unit.

PAGE 89

89 Figure 3 8. Topographic map of Butler Island NE (8DI50) showing locations of structures, augers and test pits, and excavation units. Locus A and B are designated by the by circles. Map produced by K. Sassaman. Figure 3 9. South profile of Test Unit 1, 8DI50.

PAGE 90

90 Test unit 1 (TU1) This 1 x 2 m unit was located in Locus A approximately 10 m to the southwest of the structure (see Figure 3 8). It contained two strata of intact midden deposits between the disturbed and redeposited materials near th e surface and the underlying culturally sterile natural subsoil (Figure 3 9). The upper midden stratum was a 30 50 cm thick layer of moderately dense oyster shell that contained vertebrate fauna, pottery, and lithics, including a Bradford type biface and a chunk of hematite. Pottery consisted of mainly Deptford and Swift Creek types, with occasional Weeden Island types near the top of the stratum. An AMS assay obtained on charred wood recovered in a bulk sample from this stratum yielded an age estimate o f 900 ± 30 BP (Beta 388846; charred material; 13 C = 25.3 ) , with a two sigma calibrated date range of AD 1035 to 1215. The date appears to be late compared to the majority of the pottery assemblage, however, a cluster of New River Complicated Stamped and one Weeden Island Punctated sherd was recovered from this stratum, suggesting a Weeden Island occupation. This Late Weeden Island/Early Mississippian date is consistent with dates from elsewhere in the LSAS study area so it is not assumed to be anomalous (K. Sassaman, personal communication 2014). Below this midden stratum was a 10 cm thick stratum of black fine, highly organic, sand with no shell and reduced frequencies of vertebrate fauna and sparse pottery, predominately crumb sherds. Two features, F eatures 1 and 4, were identified in plan during excavation, both of which are interpreted as postholes, suggesting perhaps a domestic structure. Test u nit 2 (TU2) This 1 x 2 meter test unit was placed to the south of the structure in Locus B and contained a very thin upper stratum of midden deposits extending to a maximum depth

PAGE 91

91 Figure 3 10. West profile of Test Unit 2, Butler Island (8DI50). of only 24 cm below surface (Figure 3 10). The predominately oyster shell midden contained vertebrate fauna, pot tery, and lithics. Deptford Linear Check Stamped and Pasco Plain sherds were recovered from the midden stratum along with a large chert core that also appears to have been used as a hammerstone. The midden deposits appear to be intact; however, the locat ion of the unit near the shoreline escarpment and the shallowness of the deposits suggest that the top portion of the midden may have been scoured. A Woodland stemmed biface was recovered from the organically stained sand at the contact between the natura l subsoil and the midden deposits. Beneath the midden stratum, two overlapping pits were identified in plan. Feature 3 appears to have been dug first and contained sparse vertebrate fauna and a moderate density of shell, mostly concentrated near the top of the feature. With the exception of one crumb sherd, no artifacts were recovered from the feature. Feature 2 was dug later, intruding into the earlier pit (Feature 3). It contained sparse shell and vertebrate fauna. A concentration of fragmentary cha rred wood was observed in the center of the feature. Artifact content

PAGE 92

92 was sparse and included one Pasco Plain sherd, a chert flake, and a small chunk of burned limestone. T est u nits 3 (TU3) and 3N (TU3N) TU3 was a 1 x 2 meter unit that was placed approxim ately 5 m to the east of the structure in Locus A, where augering and excavation of a shovel test pit revealed deep and stratified midden deposits. Large roots from a nearby cedar tree intruded into the unit, forcing termination of the excavation at 30 cm below surface. On the northern side of the unit, a smaller 1 x 0.5 m unit was opened adjacent to TU3 and designated TU3N. The inclusion of modern debris and a mix of temporally diagnostic pottery types made it obvious that the excavated materials from T U3 were redeposited, therefore, the first level of TU3N was excavated to 35 cm below surface with a subsequent shift back to 10 cm levels. TU3N had substantially deeper deposits, with stratified midden to a depth of 139 cm below surface (Figure 3 1 1 ). Fie ld observations identified seven distinct midden strata with varying densities of shell between the upper root mat and underlying culturally sterile sandy subsoil. Later laboratory analysis revealed that the upper four strata were layers of redeposited mi dden materials, likely the result of episodic scouring and redeposition of midden from the shoreline, cross cut with layers of ashy deposits that were presumably associate d with activities related to the nearby structure. Modern materials, including glass and metal were recovered from all of these upper midden strata to a depth of approximately 65 cm below surface and pottery types were jumbled, with diagnostically Weeden Island pottery stratigraphically below Deptford and Swift Creek types. There was a m arked transition to the upper portion of the intact midden at about 65 cm below surface, with the sediment matrix containing significantly more

PAGE 93

93 Figure 3 1 1 . East profile of Test Unit 3N, 8DI50. organic matter, no modern materials, and expected stratigr aphic relationships among the differing pottery types in the remainder of the unit. The intact midden contains three distinct strata. The upper portion, designated as Stratum V, is a layer of dense crushed shell containing a Weeden Island Incised sherd, along with Pasco Plain sherds. An

PAGE 94

94 AMS assay on charcoal from a bulk sample yielded and age estimate of 1070 ± 30 BP (Beta 388845; charred material; 13 C = 22.9 ) , with a two sigma calibrated date range of AD 885 1015. Beneath the crushed shell is a stra tum (Stratum VI) of whole oyster shell that decreased in density with depth in the stratum. Vertebrate faunal density was higher than in the overlying deposits, but there was a decrease in pottery, mainly in the frequency of crumb sherds. A bone tool impl ement that had been shaped and rounded on both ends, possibly a fish gorge, was also recovered. A very marked transition to black, organic rich sediments and decreased shell density, mostly whole oyster shell, characterized Stratum VII, the lowermost mid den stratum. There was an increase in pottery frequency in this stratum, including Deptford Linear Check Stamped and Swift Creek Complicated Stamped pottery, and a perforated bone tool (Figure 3 1 2 ) was recovered from the base of the midden. An AMS assay obtained on charcoal collected in a bulk sample from this stratum returned an age estimate of 2060 ± 30 BP (Beta 388844; charred material; 13 C = 25.1 ) , with a two sigma calibrated date range of 170 BC to AD 5. Below this stratum, there is a thin layer of organically stained sand that is a result of the leaching of the overlying midden deposits into the culturally sterile sands of the underlying landform. This stratum contained sparse shell, little vertebrate fauna, and only two crumb sherds. Material C ulture Frequencies of identifiable pottery types for Bird Island Test Units 1 and 3 combined, all excavation units at Butler Island, and by area at Garden Patch are provided in Table 3 1 . A more detailed listing of pottery by level is available in the But ler Island technical report (McFadden 2014). A total of 851 sherds was recovered during excavations at

PAGE 95

95 Butler Island NE, over half of which were crumb sherds, resulting in 373 sherds available for analysis. The pottery assemblage was consistent across al l three of the units, with each having variable frequencies of Deptford, Swift Creek, and Weeden Island types. Nearly half of the assemblage consisted of sand tempered plain and sand tempered check stamped sherds. Of the diagnostic types, Pasco Plain she rds were 10 percent of the assemblage, followed by Deptford Linear Check Stamped and Swift Creek Complicated Stamped sherds, each comprising four percent of the assemblage. Other types included Weeden Island and St. Johns sherds. In Test Unit 1, New Rive r Complicated Stamped sherds were a significant contribution to the assemblage in terms of number of sherds; however, the sherds were concentrated in the western half of the unit and appear to be from the same vessel, making that type a minority in the ove rall assemblage. Although no types that are diagnostic of the later period suggested by the radiocarbon date from this unit, sand tempered check stamped and plain sherds are ubiquitous well into the Late Weeden Island Period and both of these are present in higher frequencies in the upper portions of all of the test units. Forty one lithic artifacts were recovered from all of the test units combined, 58 percent of which were chert flakes. Three chunks of burned limestone were collected from each of the th ree test units in which intact deposits were encountered. In addition to flakes and burned limestone, a Bradford type biface, a biface fragment and a small chunk of hematite was recovered from Test Unit 1. Test Unit 2 yielded a large core/hammerstone an d a stemmed biface, and no stone tools were recovered from Test Units 3 and 3N.

PAGE 96

96 Figure 3 1 2 . Three views of a perforated bone tool recovered from Test Unit 3N, Butler Island (8DI50). Seven modified bones were recovered during excavations, one each from T est Units 1 and 2, two from Test Unit 3, and three from Test Unit 3N. With the exception of two bone tools recovered from the intact midden deposits in Test Unit 3N, all of the modified pieces are fragments of broken bone tools that exhibit shaping and sm oothing. A small bone tool, possibly a fish gorge, was recovered from the intact Weeden Island midden deposits. The second bone tool (see Figure 3 11) recovered from this unit came from the base of the anthropogenic deposits, at 139 cm below surface. Th e tool is made on cortical bone, likely deer. The 1.7 cm wide by 4.7 cm long tool has a nearly perfectly round perforation of 1.0 cm in diameter at one end and is broken at the other. The perforation and the outer edges of the bone are beveled from shapi ng and/or

PAGE 97

97 Table 3 1. Frequencies of Identified Pottery Types by Site Bird Is. Butler Island Garden Patch Type TU1/3 TU1 TU2 TU3N Area I M. II M. IV Area X Total Fiber Tempered Plain 8 8 Orange Incised 3 3 Deptford Linear Check Stamped 5 10 1 1 2 19 Deptford Simple Stamped 6 3 2 32 10 5 58 St. Johns Check Stamped 1 2 1 4 St. Johns Plain 7 13 16 36 Dunns Creek Red 1 2 3 Pasco Plain 14 11 9 3 28 354 30 64 513 Pasco Zoned Check Stamped 1 1 Swift Creek Com p Stamped 4 7 1 4 11 52 20 24 123 New River Comp Stamped 14 14 St. Andrews Comp Stamped 1 1 Crooked River Comp Stamped 4 4 Weeden Island Plain 22 1 23 Weeden Island Punctated 3 1 4 Weeden Island Incised 1 1 2 W eeden Island Red 22 22 Carrabelle Punctated 1 43 44 Carrabelle Incised 2 2 Tucker Ridge Finger Pinched 1 4 5 1 1 Ruskin Dentate 1 6 1 8 Lochloosa Punctated 2 2 Alachua Cob Marked 12 12 Sand Tempered Plain 54 62 6 25 696 541 245 374 2003 Sand Tempered Incised 1 5 16 23 2 47 Sand Tempered Punctated 1 1 7 5 18 27 59 Sand Tempered Cord Marked 93 56 2 151 Sand Tempered Fabric Impressed 8 6 16 30 Sand Tempered Impressed 63 63 Sand Tempered Check Stamped 19 29 12 20 19 2 7 1 109 Total 121 150 29 59 893 1130 475 517 3374 usewear and a small notch has been carved into one side of the tool just below the perforation. Examples of similar bone tools have been found on the Gulf Coast of Texas (Ricklis and Weinstein 2005), however, the function of this tool is unknown. The perforation and associated notch suggest that this is not a pendant, but rather a tool designed for a specific task, possibly an impleme nt for making and maintaining nets.

PAGE 98

98 Only one modified shell, a gastropod hammer, was recovered from Test Unit 3N, in the upper portion of the intact deposits. Other Sites Butler Island An archaeological site located on the southern arm of the island was r eported by Kohler and Johnson in 1986 and designated as Butler Island South (8DI97) (see Figure 1 3 ). They described the site as a 40 m wide shell midden that paralleled the shoreline along a 400 m stretch and contained dense cultural materials. Pottery collected on the surface included Deptford series Simple Stamped, Cross Simple Stamped, and Linear Check Stamped sherds. Pasco Plain pottery was also recovered. No Swift Creek Complicated Stamped sherds were found, and the midden was assigned to the Deptfo rd Period. Foot survey was not conducted in the area of Butler Island South (8DI97), and no eroding midden was observed during visual inspection from a boat. It is likely that this area of the island has experienced significant disturbance, not only from higher energy storms and boat wakes, but from human activities associated with the established camping area on this portion of the island. A third possible site on Butler Island was reported to the LSA by a local resident during the initial stages of reco nnaissance in the Horseshoe Cove area and was described as flakes and bifaces eroding from the distal end of the northern arm of the island. The lack of pottery in association with the lithics suggested this may be a site that dates to the pre pottery Arc haic, perhaps contemporaneous with early deposits on Bird Island. During the summer of 2013, a group of students from the Florida Museum of Natural History Lower Suwannee Archaeological Field School conducted a foot survey along the shoreline at low tide. Numerous chert flakes were observed among

PAGE 99

99 the seagrass and no pottery was found. Later f oot survey in the area failed to locate eroding lithic materials, however, the survey was conducted during high tide. Two shovel test pits were excavated in the upl and area adjacent to the eroding shoreline, but no archaeological deposits were located (McFadden 2014). It is assumed that the lithic materials observed along the shoreline are the remnants of a destroyed site. Cotton Island The Cotton Island (8DI51) sit e was recorded by Goggin in 1954 and described as shell midden with scarce potsherds. In October, 2011, with permission of the owners, foot survey, augering, and discretionary shovel testing was conducted on the island to locate intact archaeological depo sits. Unfortunately, the 1993 storm eroded a significant portion of the Gulf facing shoreline of the island. A nearly meter high escarpment along the shoreline appeared to reveal eroding midden materials; however, upon inspection it was determined that t hese were scoured midden materials that had been redeposited by the storm, creating a veneer of shell and cultural materials along the escarpment that did not extend into the upland unit. Survey failed to identify intact archaeological deposits elsewhere on the island, and it is assumed that the site reported by Goggin has been destroyed. Environmental Change and Human Occupation Survey and test unit excavations in Horseshoe Cove found evidence of human occupation as early as 2480 BC on Bird Island. The e nvironment inhabited by the Late Archaic residents was very different from the modern setting. Paleoenvironmental reconstruction (Figure 4 4 ) based on marine sediment cores collected in Horseshoe Cove suggest that Bird Island was still part of the mainlan d at the time of initial occupation (McFadden and Jaeger 2015) and the shoreline was likely still kilometers to

PAGE 100

100 the west (e.g., Goodbred 1994; Wright 1995; Wright et al. 2005). A radiocarbon date from the base of tidal channel deposits identified in a cor e collected from near the southwestern shoreline of Bird Island suggest a fresh/brackish tidal creek flowed past the site by at least 2580 BC, and the surrounding low lying areas had transitioned to fresh/brackish marsh (McFadden and Jaeger 2015). The hig h ratio of aquatic to terrestrial snails in the Late Archaic midden, cited by McFadden et al. (2014) as evidence of close proximity to water, and the absence of truncatella , a niche snail that inhabits shorelines of salt water bodies, supports the core dat a that suggest a nearby fresh/brackish tidal creek and marsh. The close proximity of the marsh is also supported by the presence of significant numbers of marsh periwinkle in the Late Archaic midden. The presence of soapstone in associat ion with the Late Archaic burials on the southwestern shoreline suggests engagement with long range trade networks and links to source areas. Petrographic analysis of the assemblage revealed that the sherds represent 18 different types of soapstone and two to four differe nt lithologies (Roberson 2013) . These differ ences suggest multiple sources and likely multiple influxes of the soapstone over time (McFadden et al. 2014; Roberson 2013) . In other areas to the north , the social influence of alliances with the Poverty Point culture stalled the adoption of fiber tempered pottery where soapstone was utilized (Sassaman 1993). The presence of fiber tempered pottery found in the Late Archaic midden deposits and the lack of Poverty Point type artifacts suggest that engagement with the Poverty Point culture may have been minimal.

PAGE 101

101 Determination of site function at the Bird and Butler Island sites is hampered by the limited sampling, but some inferences can be made based on the artifact assemblages and features encounter in the excavat ion units. No evidence of architecture was observed during test unit excavation at Bird Island, however, the artifact assemblage suggests activities associated with domestic occupation. The Late Archaic midden yielded shell tools, including the large lig htning whelk adze, and a cache of unmodified lightning whelk shell that may have been destined to become tools. The only modified bone found in context was in the Late Archaic stratum and there was an increased frequency of flakes as well. Analysis of sed iment samples from the culturally sterile sand stratum that overlies the Late Archaic deposits found high frequencies of sponge spicules in several of the sediment samples, suggesting that at least some portion of the sediments in the stratum were transpor ted from an aquatic environment and could be storm surge or flood deposit s (see Figure 4 1) . Similar characteristic s were observed in storm deposits Although , a high energy storm could have scoured some portion of the underlying Late Archaic midden, the nearly two millennium gap in basal dates between the Late Archaic and Deptford occupations overlaps with a period of apparent abandonment of sites on the northern Gulf Coast of Florida. This hiatus of occupation lasts for a period of about 800 years beginning after 1350 BC (Sassaman et al. 2014:149). Given the evidence of abandonment from other nearby sites, it is likely that Bird Island was not occupied for some period of t ime either before or after the event that contributed sediments to the culturally sterile sand stratum, and it is not until the Deptford Period that the site is

PAGE 102

102 reoccupied. Since it is unclear if occupation resumed at the site immediately after these sedi ments were deposited, the radiocarbon date of 360 170 BC from the Deptford deposits in TU1 provides only a terminus ante quem for this event. Soon after the Deptford occupation begins at Bird Island, Butler Island NE was occupied for the first time. Radio carbon dates from inter to subtidal sands and marine sediments in the cores, and the presence of truncatella in the Bird Island faunal assemblage (McFadden et al. 2014), indicates that the shoreline had transgressed and salt marsh had developed around the islands. Bird Island was still likely attached to the mainland to the north, with a large tidal flat to the west and possibly open water to the east and south. The site on Butler Island would have been bordered to the north and west by upland terrestria l areas and to east by a large area of salt marsh. Tidal channel deposits in a core collected to the northeast of the site indicate that Lolly Creek already flowed past the Butler Island NE site when occupation began (McFadden and Jaeger 2015). The Deptf ord Period environment at the Butler Island NE site likely mirrored that of the Late Archaic environment around Bird Island. Oyster continued to be an important resource during the Deptford Period occupations, as were numerous fish species, including perch , killifish, and sea catfishes; however, periwinkle (which would still be available in the marshy areas around Butler Island) were no longer a targeted resource. The Butler Island faunal remains are awaiting analysis, but field observations by a zooarchae ologist suggest the assemblage is much the same as that of Bird Island. The Deptford period pottery assemblages at both sites appears contemporaneous and increases in Pasco Plain and

PAGE 103

103 Swift Creek Complicated Stamped sherds at both sites suggest occupation into the Early Swift Creek period. As transgression continued and occupation areas at Bird and Butler Islands became more vulnerable, new areas were colonized inland. By AD 25 to 120, Deptford occupation began along the shore of a fresh water pond at the Garden Patch site (Wallis and McFadden 2014:70), which is located on the mainland approximately 1.5 km to the north. Much like the environment during the Late Archaic occupation on Bird Island and the Deptford occupation at Butler Island, Garden Patch was situated along the banks of Lolly Creek and bordered by an extensive low lying area that was likely transitioning to swamp. Radiocarbon dates and the presence of diagnostic pottery types suggest a Weeden Island Period occupation at both Bird and Butler Is lands, although that occupation is ephemeral and these sites are presumably linked to the large village mound complex that had developed by this time at Garden Patch. Studies in nearby Waccasassa Bay (Goodbred 1994; Goodbred et al. 1998) and the Suwannee Delta (Wright 1995; Wright et al. 2005) found evidence of accelerated rates of sea level rise

PAGE 104

104 The setting at the time of initial habitation at Bird Island, Butler Island, and finally Garden Patch suggests that areas protected by marshes but with relatively easy access to marine resources via tidal creeks were targeted for initial occupation. On Bird Island, the Late Archaic occupation was situated near a tidal creek and bordered to the south by fresh to brackish marsh. By the time of the initial Deptford occupation at Butler Island, and slightly later at Garden Patch, th e same conditions existed in those areas. Lolly Creek flowed past each site and extensive marsh had developed to the east. The deposits on the islands are temporally discrete and the nature of the occupations appear to vary based on environmental conditio ns, with marked differences between the deposits on the islands and the deposits from Garden Patch. There is limited evidence of Deptford Period domestic occupation at Bird Island. The identification of two post holes at Butler Island suggest a possible Deptford period domestic structure, although dating of the postholes is necessary to support this inference. In direct contrast, the Deptford deposits at Garden Patch include evidence of substantial architecture and marked differences in the artifact asse mblages, including

PAGE 105

105 exotic lithic materials that are absent from the island assemblages (Wallis and McFadden 2014). Weeden Island midden deposits on Bird and Butler Islands are similar in that they both lack evidence of domestic architecture and tools are scarce in these later deposits. The island deposits differ from those of the Weeden Island village area at Garden Patch where domestic areas, evidenced by postholes and pits, are located in areas devoid of shell and there is a substantial increase in pott ery frequencies (Wallis and McFadden 2014). Conclusions The surveys and test unit excavations performed on Cotton, Butler, and Bird Islands fulfills the LSAS goal of salvaging archaeological data from threatened sites. The data are also an important eleme nt in ongoing problem oriented research centered on the poorly understood pre Columbian history of the northern Gulf Coast of Florida. The stratified deposits on Bird and Butler Islands are currently threatened by sea level rise and erosion from storms. A lthough the marshes along the northern Gulf Coast of Florida are remarkably resilient and tend to stay in relative equilibrium with sea level rise (e.g., Leonard et al. 1995), current and projected increases in the rates of sea level rise pose a real threa t to the modern system. Unfortunately, archaeological deposits on both islands and on nearby Cotton Island have already been impacted by environmental change. Survey on Cotton Island and on the northern arm of Butler Island found only redeposited cultura l materials and the sites at both locations have been completely Archaic cemetery on Bird Island, and despite the construction of a seawall and the emplacement of fill dirt, fragmented human remains and pottery continue to erode into

PAGE 106

106 the Gulf of Mexico. The owners of the island continue to work closely with the LSA to monitor the site. The geoarchaeological research shows that paleoenvironmental reconstruction that is spe cific to the area of focus can be integral to understanding the archaeological record. The results from the project suggest that areas were targeted for initial occupation based on particular characteristics; in Horseshoe Cove these are protected areas wi th ease of access to marine resources. As the environment shifted, and the sites at Bird and Butler Islands were cut off from the mainland, the activities that occurred at these sites also shifted. Despite temporal gaps, the same areas on the islands wer e reoccupied even though new, more protected areas, had been colonized. Perhaps the presence of the middens on the islands provided a perception of deep history and stability in an otherwise changeable environment, and although the practices were differen t, people were continually drawn back to those places.

PAGE 107

107 CHAPTER 4 COASTAL EVOLUTION AND PRE COLUMBIAN HUMAN OCCUPATION IN HORSESHOE COVE ON THE NORTHERN GULF COAST OF FLORIDA * Introduction The Big Bend region of Florida is a low gradient (1:5000), open water marsh system that can be particularly sensitive to changes in sea level, and archaeological research focused on the pre Columbian coastal residents in this region necessarily includes an understanding of sea level change and the resulting human respo nses to those changes. Navigating the literature on sea level and paleoenvironmental reconstructions of coastal areas, however, is challenging since the results are often only applicable to specific geographic areas. This interdisciplinary study seeks to provide a paleoenvironmental reconstruction and chronology of shoreline change that is specific to a group of archaeological sites as a means to understand the effect of environmental shifts on human landscape relationships. First, I briefly review a few studies in which paleoenvironmental reconstructions have shown a relationship between changes in sea level and changes in human settlement patterns and discuss the problems with using these environmental data in other geographic areas. After which, I wil l present the results of this research project that included the collection of archaeological data, marine and fresh water sediment cores, and terrestrial sediment samples from Horseshoe Cove on the northern Gulf Coast of Florida. * Reprinted with permission from McFadden (2015b).

PAGE 108

108 DePratter and Howard (198 1) document ed an up to 3 meter drop in sea level along the Georgia Bight between 3000 and 2400 BP ( DePratter and Howard 1981: 1287), evidenced by submerged archaeological deposits and dates collected from drowned tree stumps. On the South Carolina coast th ere is evidence of a 2 meter or higher regression around this same time (Brooks et al. 1989; Gayles 1992). Not surprisingly, a major shift in human settlement along these coasts between 3300 and 2500 BP coincides with this regression (Thompson and Turck 2 009; Thompson and Worth 2011), with people moving eastward to follow the retreating marine resources (DePratter and Howard 1981). Shifting settlement patterns and human practices have been observed archaeologically throughout the southeastern United States during that same period, abandonment of sites along the northern Gulf Coast of Florida (Sassaman et al. 2014). Interestingly this hiatus of coastal occupation does not extend to south Florida, where sites like Reed Shell Ring (Schwandron 2008) and House Hammock shell ring were constructed and occupied during this time (Russo 2010). This nearly region wide shift in settlement patterns correlates with a global cooling trend betwee n 2500 and 3500 years ago that is believed to have resulted in a global drop in sea level (Marquardt 2010) . This has lead researchers to hypothesize that the regressive event identified along the Atlantic coast similarly affected coastal areas throughout the Southeastern United States (e.g., Marquardt 2010; Russo 2010 ; Widmer 2005 ). The use of sea level data from the Georgia and South Carolina coasts, however, cannot be applied to archaeological sites to the south and along the Gulf

PAGE 109

109 Coast. Isostatic upli ft due to glacial unloading and subsequent forebulge migration and collapse created uplift along the Atlantic coastline from Georgia northward during the Late Holocene ( e.g., Engelhart 2010; Gornitz and Seeber 1990; Pardi and Newman 1987), which certainly would have amplified any drop in eustatic sea level resulting in a significant localized regression. Subsequent subsidence resulted in a transgression that moved the shoreline back to near its previous position. Although in southern Florida, aggrading and /or prograding surfaces that date to between 3400 and 2400 BP appear to support a global drop in sea level, these minor regressions are attributed to a still stand or reduced rates of sea level rise during this period ( Gelsanliter and Wanless 1995; Parkins on 1989; Savarese et al. 2006). On the Northern Gulf Coast of Florida deposits suggesting progradation are in deltaic or tidal channel deposits and could be the result of increased sediment input from the rivers and creeks rather than being eustatically c ontrolled (Goodbred 1994; Goodbred et al. 1998; McFadden and Jaeger 2015; Wright 1995; Wright et al. 2005). Liu and Fearn in Western Lake, where regressive deposits are a bsent. A proposed period of higher than present sea level has been correlated with changing human practices during the Roman Warm Period, between 2300 and 1450 BP. Walker, Stapor, and Marquardt (1995) found geomorphic and archaeological evidence on the so uthern Gulf Coast that suggests a high stand of 1.5 m between 1750 and 1450 BP. Work by Tanner (1991) on beach ridges at St. Vincent Island in the Florida panhandle appears to support the data from south Florida. However, at Waccasassa Bay (Goodbred 1994 ; Goodbred et al. 1998) and the Suwannee Delta

PAGE 110

110 (Wright 1995; Wright et al. 2005) there was no evidence for higher than present sea level, but there was evidence of a period of increased rates of sea level rise between 1800 1700 BP . These rates were an ord er of magnitude higher than the previous 2,000 years, and in Waccasassa Bay resulted in a 2 4 km shoreline retreat (Goodbred et al. 1998). Although it is obvious that global climate change during the Roman Wa rm Period affected eustatic sea level, the expr ession of that effect on specific coastlines was variable and as with the sea level reconstruction from the Atlantic coast, the sea level data from south Florida should not be uncritically extrapolated to other coastal archaeological sites since localized processes were major control factors for relative sea level in that area. The variable expression of changes in sea level among coastal areas throughout the southeastern United States points out the inadequacies of the use of generalized sea level curves i n conjunction with specific archaeological data, particularly in highly productive marsh and estuarine environments or at river deltas where sediment accumulation can remain in equilibrium or even outpace sea level rise ( Jaeger et al. 2009; Leonard et al. 1995). Although significant environmental change obviously affects coastal communities, like those on the Atlantic Coast and at Pineland, using changing settlement patterns or cultural shifts seen elsewhere as proxies for similar environmental shifts may lead to faulty conclusions and can ignore larger social and cultural forces that may be at work. Study Area The Big Bend of Florida is one of the largest open marine marsh systems in North America. This 350 km long shoreline is a low energy, low gradient, sediment starved region that is characterized by wide marshes and tidal flats. These marshes

PAGE 111

111 and tidal flats are transected by numerous tidal creeks that follow fractures in the underlying limestone bedrock and are punctuated by small islands and hammock s composed of Pleistocene accumulations of sand atop topographic highs in the limestone. In addition to mainland sites, many of these small islands and hammocks contain archaeological deposits that span at least the last 4000 years of human occupation alo ng this coastline (Sassaman et al. 2010). Located approximately 16 km to the north of the Suwannee River, a constellation of three islands in the large embayment of Horseshoe Cove and a mainland site bordering the marsh contain stratified archaeological d eposits that offer important information about late Holocene coastal evolution and shifting human practices in this region. Butler Island, the largest of the three, is a remnant of a parabolic shaped paleodune with southwesterly reaching arms (Wright 1995; Wright et al. 2005) that is situated 0.8 km to the southeast of the town of Horseshoe Beach (see Figure 1 3 ). It is bounded by salt marsh to the north and northeast, and by extensive oyster beds in the shallow water to the south and southeast. The Butle r Island NE (8DI50) site is located on the main portion of the island along the northeastern shoreline. Bird Island (8DI52) is located near the distal end of the northernmost arm of Butler Island, and the smaller Cotton Island is situated at the distal en d of the southernmost arm. To the northeast, the large village mound complex of Garden Patch (8DI4) is situated on the mainland bordering an extensive salt and brackish marsh. Lolly Creek flows between the Garden Patch site and the marsh and continues se award, flowing past Butler Island to the south.

PAGE 112

112 Methods Figure 1 3 shows the study area with locations of sediment cores and tested archaeological sites. Although this area of the Big Bend region is a low energy coastline, the thin veneer of sediments abo ve the limestone platform is easily scoured and reworked. The protected area between the arms of the U shaped Butler Island was targeted for testing since it provided the best potential for stratigraphic integrity in the sediment cores. Thirteen marine s ediment cores were collected using a vibracore with 3 in aluminum barrels at 200 m intervals along a 2 km transect through the center of the Butler Island parabola. Three additional cores were collected along a perpendicular transect and one core was coll ected from a fresh water pond located at the inland site of Garden Patch. Elevations for each core were corrected to mean low low water (MLLW) using the NOAA datum at Horseshoe Cove and recorded water levels from nearby Cedar Key, with corrections for geo graphic distance. Each core was split, photographed, and described. Smear slides were produced and sediment samples were collected from the top, middle, and base of each identified lithostratigraphic unit. Smear slides were inspected to determine overall mineralogy and content of the samples. Percentages of organic matter and carbonate were determined by loss on ignition (LOI) in a muffle furnace for 2 hours at 550 C a nd 1050 C, respectively. After drying and weighing, fine grained materials (fines) sma ller than 63 were removed by wet sieving to determine the percentage of fines. The remaining sand fraction was processed in a settling column and sediment texture statistics were produced using the Folk method (1980) and expressed in phi units, calcula ted using the formula:

PAGE 113

113 = Log 2 d Where = phi size and d = grain diameter in millimeters (Boggs 2005 : 52 ). Characterization of grain size classes uses the Wentworth size scale. Archaeological data were collected from shovel testing and survey at all thre e archaeological sites, three 1 x 2 m test excavation units at Bird Island, three 1 x 2 m test excavation units at Butler Island, and seven 1 x 2 m test excavation units at Garden Patch. Finally, the backfill from a trench excavated into Mound V at Garden Patch was removed to recover stratigraphic information and data from a 1960s era excavation that was never reported. Standard archaeological methods were used in all excavations, and open access technical reports provided detailed methods of excavation f or Bird Island (McFadden and Palmiotto 2012; McFadden et al. 2014), Butler Island (McFadden 2014) and Garden Patch (Wallis and McFadden 2013 , 2014). Unless otherwise noted, terrestrial sediment samples were collected at 2.5 cm intervals in continuous colum ns from the west profile of TU3 at Bird Island, the east profile of TU1 at Butler Island, the west profile of TU4 at Garden Patch, and the west profile of the trench in Mound V at Garden Patch. Subsamples from each were processed using the same methods as sediments collected from the marine and fresh water sediment cores to determine the percentage of fines and obtain sediment texture statistics. Sponge spicules were quantified by systematic visual observation of 0.5 gm subsamples using a microscope with a graduated mechanical stage. Thirty seven radiocarbon dates were obtained from various contexts within the study area. Nine dates came from the base of selected lithostratigraphic units in the marine sediment cores collected offshore, and three dates cam e from areas indicative

PAGE 114

114 of transitions in depositional regimes in the pond core. Of these samples, nine dates were obtained on bulk sediment samples, two on plant material, and one from charred material. A suite of radiocarbon dates from the archaeologic al test units include three from Bird Island and three from Butler Island. An additional 19 radiocarbon dates from Garden Patch were previously reported in Wallis, McFadden and Singleton (in review). All samples were corrected for isotopic fractionation using the delta 13C calculation and calibrated using the INTCAL13 database (Reimer et al. 2013). All radiocarbon dates in the text of this article are reported as calibrated years BC/AD. Table 2 1 lists the radiocarbon assays in both radiocarbon and cali brated years for the sediment cores and Table 2 2 lists the radiocarbon assays from archaeological contexts. Results Detailed results of analysis of the marine sediment cores and the freshwater pond core are presented in Chapter 2. This section outlines t he results of analysis of sediment samples collected from terrestrial units and test unit excavations at Bird Island and Butler Island. With the exception of a basic chronology, results from excavations at Garden Patch are not presented here and the reade r is referred to Wallis and McFadden (2013, 2014). MLLW depths are no more than 1 m in the study area, and the sediment cover above the limestone bedrock is less than 1.5m in most of the cores. Despite the shallow water and thin sediment cover, there was significant preservation of lithofacies with marked transitions in all of the cores. Archaeological deposits in higher elevation areas of Bird and Butler Islands similarly had preserved, intact stratified anthropogenic deposits.

PAGE 115

115 Terrestrial Sediments Ove rall, the sediments collected from excavation units at Bird Island, Butler Island, and Garden Patch are moderately well to well sorted, medium (1.98 to 1.46 quartz sand . Samples observed microscopically are all composed predominately of well rounded, frosted quartz grains, suggesting heavy weathering and reworking as well as significant distance from the parent rock. Microfossils are absent in the sediment s underlying anthropogenic deposits, with the exception of Bird Island. Bird Island Two major non anthropogenic depositional units were identified in the sampled test unit at Bird Island (see Figure 4 1 ). The lower unit, which underlies an anthropogenic shell midden deposit is composed of medium well sorted massively bedded quartz sand with fines ranging from 1.5 to 3 percent. Sorting and mean grain size are consistent up column, with virtually no variation among samples, and with the exception of the up per 15 cm of the unit, no sponge spicules were observed in the samples. The upper unit, from 70 to 100 cmbs, is situated between two anthropogenic shell midden deposits. Like the unit below, it is composed of predominately quartz sand, however sorting an d grain size are more erratic among these samples, and the back and forth shifts in both are suggestive of microbedding. Frequencies of sponge spicules in samples from this stratum increase significantly over that of the lower sand unit, ranging from a lo w of 28 spicules per gram to a high of 460 spicules per gram. Rare diatoms and forams were also observed in the samples. Butler Island There are two main zones identifiable within the sediment column from Butler Island, distinguished by sediment texture, consistency of fine grain material percentage,

PAGE 116

116 Figure 4 1. West profile of Test Unit 3, Bird Island (8DI52) with sponge spicules per gram of sediment, percentage of fines, and grain size and sorting. and presence of anthropogenic deposits. Below 60 cmb s, grain size and sorting are consistent down column, with three slight coarsening upward sequences evidenced by increased percentages of fines. At 60 cmbs there is a transition to erratic shifts in percentage of fines and an overall trend toward coarser s ediments with elevation, which coincides with the apparent initial deposition of cultural materials. Sorting is consistent throughout, and does not shift significantly in conjunction with the coarsening of the sand fraction or changes in percentage of fine grained materials. Sponge spicules were observed in the sediment samples collected from the shell midden, the highest frequency of which was 12 spicules per gram. Below the midden, one sample contained two spicules and none were observed in the remainin g samples.

PAGE 117

117 Figure 4 2. Stratigraphic profile of the sediment column from TU4, Mound IV, 8DI4, with percentage of fines, grain size, sorting, frequencies of pottery and lithics, and designated soil horizons and zones. Garden Patch, Mound IV Sediment samp les were collected from the surface to a depth of 200 cmbs in TU4 of Mound IV. Eight zones were designated based on patterns of grain size, sorting, percentage of fines, and frequencies of artifacts (see Figure 4 2). Zone 8 at the base of the unit contai ns packages of fining and coarsening upward sequences with an overall trend toward a higher percentage of fines with elevation. Zone 7, which extends to a depth of 155 cmbs is characterized by minor shifts in grain size and consistent sorting up column. Rare artifacts, including crumb sherds and flakes, are present in the upper portion of this unit.

PAGE 118

118 Figure 4 3. Stratigraphic profile of the sediment column from Profile 4, Mound V, 8DI4, with percentage of fines, grain size, sorting, and designated soil horizons and zones. Zone 6, from 102.5 to 115 cmbs, has erratic shifts in grain size with a trend toward better sorted sediments with elevation. Artifact frequencies increase significantly with elevation as well. Zone 5 is characterized by a fining upwa rd sequence, an increase in percentage of fine grained materials, and a trend toward more well sorted sediments. There is a slight decrease in pottery but an increase in lithic in this zone. Zone 4 extends to a depth of 95 cmbs. Erratic grain size shift s occur in this zone, along with rather erratic shifts in

PAGE 119

119 percentages of fines, albeit with an overall trend toward higher percentages overall. Artifact frequencies increase substantially and peak in this zone. Zone 3, from 32.5 to 60 cmbs, contains pack ages of coarsening upward sequences with continued erratic shifts in the percentages of fines and a decrease in artifact frequencies with elevation. Zone 2 contains similar coarsening upward sequences but sorting becomes uniform and the percentage of fine s begins to decrease, as do artifact frequencies. Finally, Zone 1, the upper 12.5 cmbs of the column, is characterized by a fining upward sequences and erratic shifts in sorting and percentage of fines. A few scattered artifacts were recovered from the l ower portion of this zone. Garden Patch Mound V Figure 4 3 provides a stratigraphic profile of the sediment column from Mound V with percentage of fines, grain size, sorting, and designated soil horizons and zones. Sediment samples were collected in a column from 97 cmbs to 222 cmbs beneath the anthropogenically deposited materials on the mound. Five discrete zones that cross cut the soil horizons were identified in the sediment column based on patterns of grain size, sorting, percentage of fines, and frequencies of artifacts. Zone 5, at the base of the column, has erratic and significant shifts in grains size even as sediments overall are better sorted than those of the overlying Zone 4. Zone 4, which extends to a depth of 190 cmbs is characterized b y fairly consistent grain size up column with gradual fining and coarsening upward trends and uniform sorting. Zone 3, from 127.5 to 142.5 cmbs exhibits an overall pattern of coarsening upward in several packages. The pattern of coarsening occurs in both the percentage of fines and the sand fraction. Sparse vertebrate fauna in this zone is the first evidence of anthropogenic activity on the elevated landform. Zone 2 contains a pattern of coarsening upward followed by a fining upward sequence with sortin g varying in association with these sequences. Finally, Zone 1extends to a depth of approximately 110

PAGE 120

120 cmbs and exhibits erratic shifts in grainsize and sorting. Vertebrate fauna and lithic artifacts are present in this zone, which is at the base of the c onstructed portion of the mound. Test Unit Excavations Detailed results from test unit excavations are available elsew here for Bird Island (McFadden and Palmiotto 2012; McFadden et al. 2014), Butler Island (McFadden 2014), and Garden Patch (Wallis and McF adden 2013 , 2014 ) , and only a basic summary of the results and chronology of occupation is presented here. The earliest cultural deposits recovered from excavation units date to the Late Archaic Period (ca. 3000 1000 BC), however the bulk of the intact de posits date to the Deptford (ca. 500 BC AD 250), Swift Creek (ca. AD 150 350), and Weeden Island (ca. AD 200 900) periods. Bird Island Three test units were excavated at Bird Island (see Figure 3 1) . Test Unit 2, located at a lower elevation on the sout hwestern portion of the island contained two main strata of disturbed deposits and no intact archaeological deposits. Test Units 1 and 3 (considered as a single analytic unit) contained five stratigraphic units, three of which were intact midden strata da ting to the Late Archaic, Deptford, and Weeden Island periods (see Figure 3 2 ). The Late Archaic midden was composed predominately of Periwinkle ( Littorina sp.) shells, in terms of MNI. The midden also included a moderate density of oyster ( Crassostrea virginica ) and gastropod shells, vertebrate fauna, and lithics. Lithic debitage was common in this stratum, and with the exception of three fiber tempered sherds recovered from the upper portion of the deposit, the midden contained no pottery. Charcoal r ecovered from a basal bulk sample in this stratum returned an age

PAGE 121

121 2480 2290 cal yr BC. A culturally sterile sand stratum of variable thickness overlies the Late Archaic deposits and separated it from the Deptford midden above. The Deptford Period midden was predominately composed of oyster shell and also included vertebrate fauna and cultural materials. In contract to the Late Archaic midden, periwinkle was rare. Sand tempered plain sherds dominated the pottery assemblage, which also included Deptford Linear Check Stamped sherds. Several shell tools were recovered from this stratum and lithic debitage was also recovered but in lesser frequencies that from the Late Archaic deposit. Charcoal retrieved from a bulk sample at the base of this stratum yield ed an age of 360 170 cal yr BC. The uppermost Weeden Island midden stratum was composed of mostly oyster shell with vertebrate fauna and cultural materials. Pottery recovered from this stratum was predominately sand tempered plain, but also included a Pap and a Carrabelle Punctated sherd, both temporally associated with the late Weeden Island Period. An AMS assay obtained on charred material collected from a bulk sample at the base of this stratum yielded an age of 810 980 cal yr AD. T hree features were identified in TU1/TU3, including two pit features that emanated from near the top of the sterile sand stratum. Neither feature contained materials from the Deptford period midden above, suggesting they were dug prior to deposition of th ose materials. Only one of the two pits contained artifacts, a cluster of Orange Incised fiber tempered pottery at the base, where it intruded into the Late Archaic midden stratum and intersected a third feature. The third feature was identified at the e merging top of the Late Archaic stratum and contained a cluster of 15 large unmodified lightning whelk shells that appear to have been in some type of bag or

PAGE 122

122 basket at the time of deposition (Figure 3 5) . The fill surrounding the whelk cluster contained n umerous other gastropod shells, predominately crown conch, a chert flake, a small biface fragment, and three fiber tempered sherds, although it is likely that these materials are associated with the Late Archaic midden rather than the whelk cluster. Butler Island Two loci of occupation were identified at the Butler Island site (see Figure 3 8) . Three test units were excavated, two in Locus A in the northeastern portion of the site, and one in Locus B in the southwestern portion of the site. Test Unit 1 (F igure 3 9) was placed in Locus A and had two strata of anthropogenic midden deposits below a thin stratum of modern debris and redeposited materials. The lowermost midden stratum contained pottery and vertebrate fauna in a relatively shell free, dark, org anically stained sand matrix. Two posthole features were identified in this stratum and extended into the sterile subsoil beneath. The upper midden stratum consisted of moderately dense oyster shell midden that contained vertebrate fauna, modified bone, p ottery, and lithics, including a Bradford type biface and a small chunk of hematite. A radiocarbon date obtained on charred wood from a bulk sample in this stratum yielded an age of 1035 to 1215 cal yr AD. The radiocarbon date appear to be late given tha t the bulk of the diagnostic pottery recovered from this midden consisted of Deptford and Swift Creek types. However, it is consistent with a similarly late date from the upper portion of the intact deposits at Test Unit 3N, located approximately 30 m to the southwest of Test Unit 1, and suggests that the latest occupation in this area of the site was less intensive than during the Deptford/Swift Creek period.

PAGE 123

123 Test Unit 2 was placed in Locus B and contained only a very thin (20 cm) midden deposit that ap pears to have been truncated by scouring (Figure 3 10) . The Crassostrea virginica shell midden contained Deptford Linear Check Stamped and Pasco Plain sherds, a large chert core/hammerstone, and a Woodland stemmed biface. Two overlapping pit features tha t contained shell and vertebrate fauna were identified in this unit. Test Unit 3N was also located in Locus A (Figure 3 11) . The upper 60 cm of the 1 x 1 m unit contained disturbed and mixed deposits, likely the result of redeposition of scoured midden ma terials by storms, and significant amounts of modern debris that were associated with a nearby abandoned structure. Intact stratified midden deposits dating to the Deptford through Weeden Island Periods were found below the disturbed deposits. The lowerm ost midden deposit consisted of black, organic rich sands contain ing a moderate density of shell with pottery and faunal material . Diagnostic pottery types in this stratum included Deptford Linear Check Stamped and Swift Creek Complicated Stamped sherds. A charcoal sample recovered from the bulk sample collected in this strata returned a n age of 1 70 cal yr BC to 5 cal yr A D. There was a very marked transition to the upper midden deposits, which contained very dense crushed shell and vertebrate fauna. Pot tery frequency was lower in these deposits than in the underlying midden and included a Weeden Island Incised sherd. With depth, there is a transition to whole oyster shell , increased vertebrate fauna density , and decreased pottery frequencies in this dep osit . Charcoal recovered from the bulk sample collected in this strata yielded an age of 885 1015 cal yr AD.

PAGE 124

124 Garden Patch The earliest radiocarbon dates come from Mound IV, located near the fresh water pond at the site, suggesting an initial Deptford oc cupation began between 25 to 120 cal yr AD. Occupation continued in this area of the site into the Swift Creek period, by at least 395 to 540 cal yr AD. During the latter stages of occupation at this mound, activities began on two other mounds, Mounds II and V, with dates ranging from 240 to 395 cal yr AD to 5 35 to 610 cal yr AD. Finally, the latest radiocarbon dates suggest a Weeden Island period occupation began to the northwest of the mounds around 675 to 870 cal yr AD and continued through 775 to 980 cal yr AD. This single component Weeden Island occupation, and the lack of evidence of contemporaneous occupation elsewhere at the site, suggests the later period residents chose to remain near the mounds but avoided areas of previous occupation (Wallis and McFadden 2013 , 2014). Discussion Based on similar characteristics, lithofacies descriptions and interpretations from Wright ( 1995 ), as adapted from Hine et al . (1988), provide a useful model for the interpretation of lithofacies identified in the marin e sediment cores (see Table 2 3). Two additional lithofacies that were identified in the Horseshoe Cove cores, tidal channel and event bedding, have been added to those provided by Wright ( 1995 ). Figure 4 4 provides a schematic representation of the pale oenvironmental reconstruction, along with periods of human occupation at each archaeological site. This environmental reconstruction is based on the interpretations of lithofacies transitions, elevations of those transitions, modern shoreline morphology, bathometry of the study area, and radiocarbon dates from cores and archaeological contexts.

PAGE 125

125 The geomorphic attributes of the landforms in the study area are consistent with similar parabolic dunes and sandsheets that accreted throughout the southeastern U nited States around the end of the Pleistocene (Ivester and Leigh 2003; Markewich and Markewich 1994). Butler Island is the remnant of a paleodune that has survived shoreline transgression and inundation as sea level rose because of its relatively high el evation. The mainland site of Garden Patch is situated on a large Pleistocene sand sheet. Sediments collected from beneath anthropogenic deposits in the archaeological units are all medium, moderately well to well sorted, massively bedded quartz sands, c haracteristics that are consistent with those defined by Markewich and Markewich (1994) and Ivester and Leigh (2003) for inland paleodunes and sandsheets. A single thermoluminescence date of 20 +/ 4 ka from Butler Island, reported by Wright et al. (2005) , puts the formation of this landform within the chronological range of similar dunes that formed in the Carolinas (Moore and Daniel 2011) and Georgia (Ivester et al. 2001). Although the morphology of the area is largely controlled by the underlying Ocala limestone, the accumulations of Pleistocene sands created topographic highs that survived shoreline transgression and preserved evidence of human occupation. In lower elevation areas, the gently sloping, karstic limestone bedrock in Horseshoe Cove sits be neath a relatively thin Pleistocene sediment cover. With the exception of topographic lows in the limestone and beneath tidal channel deposits, the thin veneer of sediment was scoured by shoreline transgression, leaving a ravinement surface between the li mestone and overlying lithofacies. The scoured and reworked sands,

PAGE 126

126 Figure 4 4. Paleoenvironmental reconstructions of northern Horseshoe Cove and dates symbol. along with m arine derived clastic materials ( e.g , . Goodbred et al. 1998), are the major source of the available sediments in Horseshoe Cove, with additional minor sediment input from the tidal creeks and streams that drain into the Gulf of Mexico in this area. The in itial flooding of the study area in Horseshoe Cove is indicated by the development of fresh to brackish swamp/marsh near the main portion of Butler Island, in the protected area between the arms, sometime after about 2870 2800 cal yr BC.

PAGE 127

127 Around the same t ime, a tidal creek, likely an extension of the modern Lolly Creek, began to flow along the southwestern shoreline of Bird Island, evidenced by the date of 2875 2580 cal yr BC from the base of tidal channel deposits that overly Pleistocene sands. Initial L ate Archaic period occupation at Bird Island began sometime around 2480 2290 cal yr BC along the southwestern shoreline, near the tidal creek. At this time, the island was still part of the mainland to the north and west, but extensive marsh had developed to the south and east. The tidal creek likely provided a convenient avenue through the marsh to the open marine environment where oysters and other resources could be collected. The high density of periwinkle in the midden dating to this period suggests these small gastropods were readily available in the surrounding marsh grasses. The presence of incidental aquatic snails, but the absence of truncatella, a salt water shoreline niche gastropod, suggests close proximity to a fresh to brackish aquatic env ironment (McFadden et al. 2014), supporting the evidence of the presence of a tidal creek identified in the marine sediment core. By 2030 1885 cal yr BC, inter to subtidal sands began to accumulate in the lower elevations between the northern arm of Butler Island and Bird Island and additional fresh to brackish marsh had developed by 2005 1780 cal yr BC near the main portion of Butler Island. The high and variable frequencies of sponge spicules observed in the culturally sterile sand stratum between the Lat e Archaic and Deptford period middens at Bird Island, along with the presence of diatoms and forams , suggests that some portion of this stratum contains sediment that were transported from an aquatic environment and could be a storm or flood deposit. Sim ilar characteristics were observed in storm deposits at Waccasassa Bay (Goodbred and Hine 1993) and at Cat and Little Bradford

PAGE 128

128 Islands at the mouth of the Suwannee River (Sassaman et al. 2010) after the so called n Gulf Coast of Florida in 1993. The nearly two millennia gap in basal midden dates that sandwich this stratum suggests that some portion of the Late Archaic midden deposits may have been scoured prior to deposition of the sands. Since it is unclear if o ccupation resumed at the site immediately after these sediments were deposited, the radiocarbon date of 360 170 cal yr BC from the Deptford midden provides only a terminus ante quem for this event. Given the evidence of abandonment of other nearby sites du ring the period between 1350 BC to 550 BC (Sassaman et al. 2014), it is likely that Bird Island was also abandoned for some period of time either before or after the event that deposited the sterile sand stratum. Abandonments seen elsewhere have been attr ibuted to a drop in sea level in response to a period of global cooling (e.g. , Marquardt 2010; Sassaman et al. 2014). Although there is evidence for localized regressions elsewhere along Evans et al. 1985 ; Parkinson 1989) where reduc ed rates of sea level rise resulted in aggrading or prograding shorelines, no regressive sequences were identified in the marine sediment cores at Horseshoe Cove. However, subsequent transgression after a relative drop in sea level would likely scour regr essive sequences and with the exception of the tidal channel deposits on the southwestern shoreline of Bird Island, all of the seaward most cores have ravinement surfaces above the limestone base. Two sediment cores, both in the marshy area near the main portion of Butler Island, have event bedding between fresh to brackish marsh deposits and overlying salt marsh deposits. It is possible that these are prograding sequences that occurred during a time of reduced rates of sea level rise; however, both cores are

PAGE 129

129 located near the mouths of tidal creeks and could also be the result of increased sediment discharge from the creeks. Additionally, the date from the transition to salt marsh above the event bedding in one of the cores postdates the period of coastal site abandonment by over 500 years, suggesting that at least that event bed is not linked to the period of coastal site abandonments. If a period of reduced rates of sea level rise, or possibly a global drop in sea level, resulted in a localized regressio n at Horseshoe Cove, a dated transition to inter to subtidal muds seaward of Bird Island, in core HBT18, may provide a temporal marker for the subsequent transgression around 890 880 cal yr BC. Soon after, between 790 540 cal yr BC, the freshwater pond a t Garden Patch became periodically active, evidenced by laminar bedding indicative of periods of anoxic conditions observed in core HBT19 . Sea level rise continued and by 405 370 cal yr BC, the pond was permanently filled. The pond appears to be a perman ent water feature by 4 05 370 cal yr BC , and given the shallowness of the pond, it is likely that it was r eplenished by one of the many springs that would have become active as ground water rose along with sea level. Human occupation at Bird Island, Butler Island, and finally Garden Patch occurred after and during environmental shifts that would have significantly changed the morphology of the coastline. Occupation resumed on Bird Island between 360 170 cal yr BC and by this time, the site would have been s ituated on the tip of a peninsula that extended southward from the mainland and would have been bordered by tidal flats to the west and south, and by open water to the east. The tidal creek likely continued to flow to the southwest of the island, transect ing the tidal flat, for some period of time

PAGE 130

130 before inter to subtidal sands began to accumulate above the drowned creek. A subsequent transition to salt marsh between 45 cal yr BC and 75 cal yr AD in the area near the main portion of Butler Island overlap s with the first substantial human occupation of the island, between 170 cal yr BC and 5 cal yr AD. The environment at the site on Butler Island likely closely resembled that of Bird Island during the Late Archaic, with terrestrial areas to the north and west, extensive marsh to the east and nearby Lolly Creek, which flowed past the site and provided easy access to marine resources. By 25 to 130 cal yr AD, a Deptford Period occupation began on the shore of the freshwater pond at Garden Patch, within sight of Lolly Creek. An increase in rates of sea level rise around AD 200, identified by Goodbred et al. (1998) at Waccasassa Bay, likely also caused rapid environmental change in Horseshoe Cove, resulting in substantial marsh development to the north and east of the study area and a shift in human settlement patterns. No transitions from this transgression have been identified in the marine sediment cores since most dates were obtained on basal marsh deposits, but there is evidence of transitions to restricte d marine and finally marine sediments over much of the core transect that still need to be dated. No evidence of higher than present sea level was observed in the sediment cores or in the archaeological units in the study area, suggesting the high stand i dentified in south Florida (Walker et al. 1995) was the product of localized processes that did not extend northward to Horseshoe Cove. Archaeologically, there is evidence of intensified occupation and a period of rapid and substantial mound construction at Garden Patch that begins a s early as 240 cal yr AD (Wallis and McFadden 2014), which one can hypothesize is linked to the rapid transgression and environmental shift. No

PAGE 131

131 dates from that time period come from Bird and Butler Islands, and it is possible that these islands were not occupied during the florescence of Garden Patch. After this rapid transgression, the Garden Patch site likely had similar morphological attributes to those of Bird Island and Butler Island during their initial occupations. Sti ll connected to the mainland, the site would have bordered extensive swampy areas to the east and south and Lolly Creek would have provided access to open water and marine resources. Another pulse in sea level rise during the Medieval Warm Period could be related to a nearly century long abandonment at Garden Patch. The site appears to have been largely abandoned sometime before AD 600 (Wallis and McFadden 2014). Soon after, peat formation indicative of fresh to brackish marsh development began on the lan dward side of Butler Island, between 660 and 770 cal yr AD. Although it is difficult to assign a cause and effect relationship between this pulse in sea level rise and the abandonment of Garden Patch, it is likely that a rapid transgression would have aff ected resource availability and that could be a contributing factor in the temporary decline of the site. A pulse in sea level rise would increase salinity around the oyster reefs, exposing them to predatory organisms that prefer the higher salinity envir onment. The increased stress on the reefs can make them vulnerable to collapse, which can be significantly accelerated by high energy storms. This could cause significant resource shortages if these reefs collapse before new reefs have time to build in t he lower salinity areas nearer the shoreline. By the time of the Weeden Island period occupation at Garden Patch, the morphology of Horseshoe Cove was very similar to that of the modern coastline, and

PAGE 132

132 this new environment may have resulted in shifting set tlement patterns across the entire study area. The initial reoccupation of Garden Patch by Weeden Island villagers occurred between 675 and 780 cal yr AD and continued through at least 775 980 cal yr AD. A date from above the white sand deposit of the po nd core, which is interpreted as the input of terrestrial sands due to proximate human activity, appears to support a shift in occupation away from that area of the site after about 680 cal yr AD. Weeden Island midden deposits began to accumulate on Bird Island by 810 980 cal yr AD, and concurrently, on Butler Island by 888 1015 cal yr AD, suggesting that both of the islands were reoccupied after Garden Patch is finally abandoned. It is important to note here that this statement applies only to t he Garden Patch site. There are several other reported sites nearby that have not been surveyed and it is certainly possible that these sites could contain later deposits. In any case, the date of 1035 to 1215 cal yr AD from Butler Island suggests that occupation continued, at least on that island, into the early Mississippian Period. Conclusion Environmental change certainly had a significant impact on the pre Columbian coastal communities in Horseshoe Cove. Morphological changes in the shoreline and the loss of terrestrial areas due to rising sea level forced individuals and groups to make informed decisions about where to settle. In Horseshoe Cove, the pre Columbian residents appear to have chosen areas based on fairly specific attributes, namely, protected ar eas bordering the marsh but with easy access, via tidal creeks, to marine resources. However, even though places appear to have been sporadically abandoned in Horseshoe Cove, sites that were exposed to the higher energy open marine environment due to sea level rise and shoreline retreat were targeted for re occupation

PAGE 133

133 during later periods and the decisions to reoccupy these areas may not have been related to environmental conditions alone. The combination of data from archaeological excavations, marine sed iment cores, and terrestrial sediment samples in this study enabled the construction of a chronology of environmental change that is specific to Horseshoe Cove and is directly relatable to the archaeological sites at Bird Island, Butler Island, and Garden Patch. The results of this study show that the environmental shifts observed elsewhere, for instance the significant drop in sea level on the Atlantic Coast and a period of higher than present sea level on the southern Gulf Coast of Florida, either did no t occur or were not of the same magnitude in Horseshoe Cove. This finding supports the argument that incorporating paleoenvironmental reconstructions created in one coastal area into archaeological data from another may create faulty assumption of human ac tions as being a response to environmental conditions rather than being the product of larger social and cultural forces.

PAGE 134

134 CHAPTER 5 ENVIRONMENTAL AND SOCIAL FACTORS IN PRE COLUMBIAN SETTLEMENT ON THE NORTHERN GULF COAST OF FLORIDA * I ntroduction Changes in land use and settlement patterns along the coasts of the southeastern United States have been attributed to shifts in environmental conditions, particularly changes in sea level. Research along the Atlantic coasts in Georgia ( Depratter and Howard 1981 ; Thompson and Worth 2011 ) and South Carolina (Brooks et al. 1989 ), and on the Gulf Coast in South Florida (Walker et al. 1995 ) has incorporated environmental data with archeological data to understand how sea level changes affected coastal pre Columbian s ettlement . In these areas, changes in sea level resulted in the abandonment of areas either because of a regression that forced communities to follow marine resources seaward (Depratter and Howard 1981; Thompson and Worth 2011 ) or because they were inunda ted by a period of higher than present sea level (Walker et al. 1995 ). In both cases, abandoned areas were sometimes reoccupied when the shoreline returned to near its previous relative position, and certainly some of the reoccupied areas were in differen t environmental settings that those at initial occupation. Sea level rise since the end of the L ast G lacial M aximum (20,000 BP) necessitated movement of communities as the shoreline retreated and the existing sites on the coast r epresent only those that r emain above current sea level. Although the archaeological record is truncated by the flooding and destruction of o lder * Reprinted with permission from McFadden (2015c).

PAGE 135

135 sites, the surviving sites provide insight into past decisions about the appropriate places to occupy along the coast. The normalcy of sea level rise since the end of the Pleistocene likely resulted in flexible coastal communities that had the expectation of eventual relocation. The experience of environmental shifts that forced landward movement may not have me, but periodic relocations would certainly be a part of the In Horseshoe Cove, on the northern Gulf Coast of Florida, n ewly colonized areas were located on elevated landforms that were protected by extensive marshes and had access, via tidal creeks and streams, to marine resources in the nearby open water environment . But, previously colonized areas that were isolated from the mainland by sea level rise continued to be utilized, albeit in different ways. This strategy would have created a rolling mosaic of occupations along the coast. Newer settlements would be founded in protected areas inland, likely with the expectation that these areas would be retasked as sea level rose and the environment changed. This would have extended, not only the occupational history, but also the social history of these sites. , as used by Barrett (1999) , social factors played a larger role in Weeden Island peoples choices about places to occupy than did environmental setting. Although the morphology of the islands in Horseshoe Cove is dependent upon natural processes, such as sea level change, patterns of sediment movement, storms, etc., they are also the product of human interaction. Indeed, the very survival of the islands as the shoreline transgressed may have been dependent on the anthropogenic

PAGE 136

136 deposits that effectively raised the elevation of the landform. But, these islands were not (re )occupied just by virtu e of their existence. These islands had a social history that was based in the interactions between humans and the environment that became materialized in the landscape through time. Shell is a particularly durable material and peo ple interacted with it on a daily basis in Horseshoe Cove. The routine activities of collecting, processing, and discarding the remains of shellfish created landscapes that were topographically altered by the accumulations of shell. Because of its durabl e nature (e.g. , Joyce 2004), shell was a continual reminder of the past. The past was always present , and because humans interacted with the shell in the past, it was the human past that was distributed across the landscape. I do not suggest here that she ll, and other durable features on the landscape, were literal representations of past activities, rather the landscapes created by these features were laden with histories that actively brought the past into the present , with the meanings and interpretatio ns of places changing as circumstances change d . Original meanings and specific histories may have been lost, but later occupants of the sites at Horseshoe Cove would have constructed ideas about the past, interpreted meanings, and acted on their own under standings of the anthropogenically altered landscape. Although these perceptions and understandings of the landscape are a result of localized historical trajectories, these occur within a wider sphere of regional interactions that are neither static nor stable. Generational memory fades, interpretations change, and expectations alter to reflect changing priorities and conditions, be they social or environmental. It is possible that these histories provided a

PAGE 137

137 perception of permanence in an otherwise chan geable environment, and that perception anchored people to those places (Bender 2002) . This article outlines the results of a geoarchaeological research project that reconstructed a chronology of environmental change, specifically sea level rise, which dir ectly impacted the pre Columbian residents of Horseshoe Cove. Correlation of the environmental data with archaeological data from three sites in Horseshoe Cove suggest that environmental change had a significant impact on human decision making, but social factors were also an important element in choices concerning places of occupation. S tudy A rea Horseshoe Cove is a large shallow water embayment located approximately 16 km north of the Suwannee River Delta in the Big Bend region of Florida. The cove is b ordered to the northwest by the mainland town of Horseshoe Beach, to the north and east by extensive salt and brackish marsh and low lying forested areas, and to the south by Fishbone Creek. Wide marshes and tidal flats that are transected by numerous tid al creeks fringe the mainland in the low energy, low gradient embayment. A thin veneer of siliciclastic sediments overlie the limestone bedrock, and diss olution and collapse of that limestone has produced complex topography that is a major control over th e morphology of the shoreline. Many of the small islands protruding from the shallow waters along the coastline are remnants of relict paleodunes and sand deposits that accumulated atop topographic highs in the limestone around the end of the Pleistocene, when the climate was cooler and drier ( Markewich and Markewich 1994; Wright et al. 2005 ). Offshore, the eastern oyster ( Crassostrea virginica ) thrives in the shallow, low salinity environment. Outcrops of limestone exposed by the scouring of

PAGE 138

138 sediments du ring shoreline transgression provide optimal substrate for oyster colonization, and they tend to cluster in areas of faster moving nutrient rich waters (Hine et al. 1988 ; Kilgen and Dugas 1989), particularly in area of fresh water input near the mouths of the tidal creeks. There are three archaeological sites in the northern portion of Horseshoe Cove that contain intact stratified deposits span ning at least 4,000 years (see Figure 1 3 ) . Butler Island, the largest of the three, is the remnant of a parabolic shaped paleodune that formed around the end of the Pleistocene (Wright 1995; Wright et al . 2005). A thermoluminescence date from the base of the landform yielded an age of 20 ± 4 ka (Wright et al. 2005 ), which puts formation of the Butler Island paleodun e within the range of similar landforms throughout the southeast ( Ivester and Leigh 2003; Ivester et al. 2001; Markewich and Markewich 1994; Moore and Daniel 2011 ) . Butler Island is situated 0.8 km to the southeast of the mainland town of Horseshoe Beach (see Figure 1 3 ) and i s bounded by marsh to the north and northeast and by extensive oyster beds in the shallow water to the south and southeast. The Butler Island NE (8DI50) site is located on the northeastern shoreline of the main portion of the island and was initially reported by John Goggin in 1954. Originally named the Lolly Creek site, the name was later changed to Butler Island NE after the identification of a second site, Butler Island South (8DI97), on the southern arm of the island (Kohler and Johnson 1986). No archaeological excavations were performed on Butler Island prior to those of the Laboratory of Southeastern Archaeology in 2014. Th e Bird Island (8DI52) site , also reported by Goggin in 1954, is situated on a small island located near th e distal end of the northern arm of Butler Island. This small

PAGE 139

139 island is surrounded by open water on three sides and a large tidal flat that is only exposed during extreme low tides to the north. In 1985, Julian Granberry excavated an exposed burial from the southwestern shoreline and obtained a radiocarbon date on a portion of the human bone (Reported by Stojenowski and Doren 1998), which yielded an age of 4570 +/ 100 BP. This date was not corrected for 12/13C fractionation or local reservoir effect, bu t using Intcal13 4.2 (Reimer et al. 2013) this date roughly calibrates to 3630 to 2942 BC. Over the next decade, continued erosion along the shoreline exposed additional fragmented human remains, prompting investigations by Florida State University in 199 3. Human remains and cultural materials were collected from the shoreline seaward of a 2 m thick shell midden that was eroding due to wave action. ed virtually the entire midden and only a few burials beneath an area of thin midden deposits survived. This area was excavated later in 1993 by FSU. Subsequent random coring in the area of the southwestern shoreline midden failed to identify significant intact deposits that had survived the storm (Dasovich 1999), other areas of the island were not tested and no further archaeological work was performed until 2012. To the northeast of the islands, the large village mound complex of Garden Patch (8DI4) is situated on the mainland , bordering an extensive area of marsh to the south and east. Lolly Creek flows between the archaeological s ite and the marsh and continues seaward, flowing past the Butler Island NE site before emptying into the Gulf of Mexico. T his site was originally investigated by Clarence B. Moore (1902), who reported on excavations in three mounds and described a small fresh water pond that is

PAGE 140

140 situated near the mounds. The site was later visited by Goggin in 1948 and additional areas of the site were described. In the late 1960s, Timothy Thompson conducted excavations at the site, but these were never reported. In 1975, Timothy Kohler Weeden Island Ceremonia the site. Decades later, in 2010, the Florida Division of Historical Resources, Bureau of Archaeological Research, Public Lands Archaeology returned to the site to map the mounds and excavate tw o small test units. Despite the multiple investigations, the site was not systematically tested or extensively excavated until the Florida Museum of Natural History began work in 2012. M ethods Thirteen marine sediment cores were collected using a vibracor e at 200 m intervals along a 2 km transect through the center of the U shaped Butler Island (see Figure 1 3 ). Three additional marine sediment cores were collected along a perpendicular transect near the shoreline of the main portion of Butler Island. On e additional core was collected using a percussion core from the fresh water pond located at the Garden Patch archaeological site. Elevations for each core were corrected to mean low low water (MLLW) using the NOAA datum at Horseshoe Cove and recorded wat er levels from nearby Cedar Key, with corrections for geographic distance. Each core was initially split, photographed, and described. After which, smear slides were produced and sediment samples were collected from the top, middle, and base of each litho stratigraphic unit. Smear slides were inspected to determine overall mineralogy and content of the samples. Percentages of organic matter and carbonate

PAGE 141

141 ively. After drying and weighing, fine grained materials (fines) smaller remaining sand fraction was processed in a settling column and sediment texture statistics were prod uced using the Folk method (1980) and expressed in phi units. The Wentworth size scale was used for characterization of grain size classes. Survey, shovel testing, and test unit excavation was conducted at Bird Island (8DI52), Butler Island (8DI50), and Ga rden Patch (8DI4), and included three 1 x 2 m test excavation units at Bird Island, two 1 x 2 m and one 1 x .05 m test excavation units at Butler Island, and seven 1 x 2 m test excavation units at Garden Patch. In addition to test unit excavations at Gard en Patch, a trench excavated into Mound V in the late 1960s by Thompson was re excavated to recover unreported stratigraphic information and data. Standard archaeological methods were used in all excavations, and open access technical reports provide deta iled methods of excavation for all three sites ( McFadden 2014; McFadden and Palmiotto 2012; McFadden et al. 2014 , Wallis and McFadden 2013 , 2014). Terrestrial sediment samples were collected at 2.5 cm intervals in continuous columns from selected profiles at the three sites. Percentage of fines and sediment texture statistics were obtained using the same methods as described above for samples collected from the sediment cores. Subsamples of 0.5 gm were systematically observed using a microscope with a gra duated mechanical stage to quantify sponge spicule frequencies. Thirty seven radiocarbon dates were obtained from various contexts within the study area. Nine dates came from the base of selected lithostratigraphic units in the

PAGE 142

142 marine sediment cores colle cted offshore, and three dates came from areas indicative of transitions in depositional regimes in the pond core (Table 2 1) . Of these samples, nine dates were obtained on bulk sediment samples, two on plant material, and one from charred material. A su ite of radiocarbon dates from the archaeological test units (Table 2 2) include three from Bird Island and three from Butler Island. An additional 19 radiocarbon dates from Garden Patch were previously reported in Wallis and McFadden (2014). All samples were corrected for isotopic fractionation, us ing the delta 13C calculation and calibrated using the INTCAL13 database (Reimer et al. 2013). All radiocarbon dates in the text of this article are reported as calibrated years BC/AD. R esults M ean low low wate r depths are no more than 1 m in the study area, and the sediment cover above the limestone bedrock is less than 1.5 m in most of the cores. Despite the shallow water and thin sediment cover, there was significant preservation of lithofacies with marked t ransitions in all of the cores. Results from analysis of the fresh water pond core and terrestrial sediments is also provided here. Archaeological excavation reports are available for all three site elsewhere, so the results from each are only briefly ou tlined here. Marine Sediment Cores Figure 5 1 provides a schematic representation of interpreted facies in each marine sediment core in the southwest to northeast transect, and Figure 5 2 provides the same for the west to east transect. Ten lithofacies were identified based on color, content, and sediment texture. The limestone bedrock underlying the study area

PAGE 143

143 Figure 5 1. Schematic representation of interpreted lithofacies in each core in the southwest to northeast transect with radiocarbon dates . rises only about 10cm in elevation across the 2 km transect. The degraded top of the limestone consists of a partially cemented white to gray weathered fine grained sediments that sit below a sharp contact with overlying sediments. In all but four, this lithofacies was encountered at the base of each core. A unit of dark gray, medium to fine quartz sand with occasional burrows was identified either at the base of or above

PAGE 144

144 Figure 5 2. Schematic representation of interpreted lithofacies in each core i n the east to west tr ansect with radiocarbon dates . the limestone in four cores. In one core, t hese sediments contain w oody organic material and the contact with overlying deposits in all four cores was sharp. The Brown Medium Sand unit was identified in three cores and consists of medium quartz sands that are well sorted, have less than five percent fines, one percent organic content, and no carbonate. B urrows are common in this unit. One of the oldest dates in the study area came from just above the tr ansition to this facies in

PAGE 145

145 core HBT17, which was collected in an area that appears to be a drowned tidal creek to the southwest of Bird Island. The bulk sediment sample yielded an age of cal 2875 2580 BC. The remaining two cores that contain this lithofa cies, HBT10 and HBT4, were collected from the mouths of existing tidal creeks. Identified in eight cores, a unit of Gray Medium Sand to Black Muddy Sand is characterized by up to 15% fines, less than 5% carbonate, and up to 10% organic matter. Shell fragm ents are rare, and where not heavily bioturbated, bedding is present. It unconformably overlies the limestone base in four cores, HBT8, HBT15, HBT16, and HBT18. A bulk sample of organic sediment from the base of this lithofacies in core HBT18, to the sou thwest of Bird Island, yielded a date of 890 880 cal yr BC and a bulk sediment samples from the base of this facies in core HBT16, between the arms of Butler Island, yielded an age of 2030 1885 cal yr BC. It is present in two cores collected from tidal cr eeks, HBT4 from Lolly Creek and HBT10 in the parabola of Butler Island. Containing less than 5 percent each organic matter and fines, and less than 2 percent carbonate, the Light Gray to Brown Medium Sand lithofacies was identified in five cores. In cor es HBT3, HBT9, and HBT11, all of which were collected near the main portion of Butler Island, this unit is sandwiched between the underlying Black Muddy Sand with no Carbonate and the overlying Black Muddy Sand with Carbonate units. T he remaining two core s that contain this unit were collected from the relatively unprotected area between the distal end of the northern arm of Butler Island and Bird Island, one of which contains a moderate density of shell fragments . The Black Muddy Sand with Low Carbonate l ithofacies is characterized by black, slightly muddy organic rich sand that contains less than three percent carbonate, no

PAGE 146

146 shell, occasional woody organic material, and variable organic content. This unit overlies either the limestone or the Dark Gray Med ium to Fine Sand facies and was present only in cores collected in close proximity to Butler Island. Four samples of wood charcoal from the base of this unit yield AMS age estimates of cal 2870 2800 B C in HBT12, cal 2470 2290 BC in HBT9, cal 2005 1780 BC in HBT11, and cal AD 600 770 in HBT3. Similar to this facies, the Black Muddy Sand with Higher Carbonate Content lithofacies consists of black muddy sand with carbonate content that ranges from 4 13 percent. In the three cores in which it was identified, it overlies the Light Gray to Brown medium Sand lithofacies and is interbedded with these sediments at the contact. Wood charcoal from the base of the facies in HBT9, near the main portion of Butler Island, yielded an age of cal 45 BC AD 75. The most dis tinctive characteristic of the Dark Gray to Brown Sand with Shell lithofacies is the inclusion of numerous whole and fragmented shell, primarily Tellina and Macoma species. This is the most wide spread lithofacies in the study area, present in all cores f rom landward of Bird Island to the marsh water interface at the main portion of Butler Island, and overlies a variety of other lithofacies. One AMS age estimate was obtained from this unit where it overlies dense oyster shell in HBT2, which yielded a two sigma range of cal AD 1450 1640 . The elevation from which this dated material was collected is below that of the oldest dates in the study area, suggesting that the material may have been displaced from its original context, likely due to disturbance and reworking of sediments in this area. With few exceptions, t he Dark Gray to Brown Fine Sand lithofacies sits stratigraphically above all other facies. Ohiomorpha burrows are common and shell

PAGE 147

147 fragments are sparse in these sediments, and t his facies generall y overlies the Dark Gray to Brown Sand with Shell facies on a grading contact. Finally, t he Dense Shell in Sand Matrix lithofacies, identified in three cores, is characterized by dense whole and fragmented shell. Predominately oyster ( Crassostrea virginic a) , it also contains lesser densities of other mollusks, including Littorina sp. , Mytilidae, Tellina sp . and Macoma sp. Fresh Water Pond Core Three main lithofacies were identified in the core collected from the fresh water pond at the Garden Patch site ( see Figure 2 4 ). The Brown Sand lithofacies consists of compact sands that grade to mottled organic sediments with elevation. These sediments contain no carbonate and less than 2 percent organic matter, with the exception of occasional woody and rooty or ganic materials. Overlying this facies is the Brown Mottled Sand unit, which is characteristically similar to that of the underlying Brown Sand. A marked increase in mottling and areas of linear bedding with darker, more organic, fine sediments distingui sh this unit from the one below. A bulk sample of organic material collected from the lowermost area of linear bedding at the base of this facies yielded an AMS age estimate of cal 790 540 BC. Finally, the dense Black Muddy Sand lithofacies is characteri zed by a significant increase in percentages of organic material and fines, both of which continue to increase with elevation. Large roots and woody pieces of organic material are common and there is a very high density of sponge spicules in these sedimen ts. Carbonate content remains low, however it increases slightly with elevation to a maximum of 4% at the very top of the core. Near the base of this facies, a 10 cm thick lens of white sand is interbedded with the darker sediments from the surrounding u nit. Two AMS age estimates were obtained on

PAGE 148

148 organic sediment from the Black Muddy Sand facies. The first from the base of the facies, returned an AMS age estimate of cal 405 370 BC , and the second f rom just above the white sand deposit yielded an age est imate of cal AD 680 880. Terrestrial Sediments Overall, the sediments collected from the excavation units at the three archaeological sites are moderately well to well sorted, medium sand that is composed of well rounded, frosted qu artz grains. With the exception of Bird Island, microfossils are absent from samples collected below anthropogenic deposits. Sediment samples were collected from 70 to 195 cmbs from the test unit at Bird Island and two major non anthropogenic depositiona l units were identified (see Figure 4 1 . The lower unit, underlying a stratum of shell midden, is massively bedded quartz sand that is consistent in sorting and grain size up the column, with virtually no variation among samples . Sponge spicules in dimin ishing frequencies were observed in the upper 12 cm of this unit, directly beneath the anthropogenic deposits. The upper unit is situated between two strata of shell midden deposits and is primarily composed of quartz sand. Grain size and sorting are err atic among the samples, with back and forth shifts in both that is suggestive of microbedding. There are increased and variable frequencies of sponge spicules in this unit, ranging from 28 to 460 spicules per gram in the samples. Rare diatoms and forams were also observed in these samples. At Butler Island, sediment samples were collected from the surface to a depth of 100 cmbs. Two main zones were identified in the sediment column, distinguished by sediment texture, consistency of percentages of fines, and the presence of anthropogenic deposits. In the lowermost zone, below 60 cmbs, sorting and grain size are consistent up column, with three coarsening upward sequences evidenced by

PAGE 149

149 decreased percentages of fines. Above 60 cmbs the samples become more variable with erratic shifts in percentages of fines and an overall coarsening of the sediments with elevation. This transition coincides with the initial deposition of cultural materials. Sediment remains consistently well sorted in all of the samples a nd does not vary with shifts in grain size and percentage of fines. Sponge spicules are present in very low frequencies, with a maximum of 12 spicules per gram in the midden samples. Below the midden, one sample contained two spicules and none were obser ved in the remaining samples. Sediment samples were collected from the surface to a depth of 200 cmbs from the test unit at Mound IV at Garden Patch and from a depth of 92 to 222 cmbs in Mound V. Multiple zones were identified based on grain size, sortin g, percentage of fines, and artifact frequencies. The two lowermost zones in both mounds are non anthropogenic depositional units composed of quartz sands. In both locations, packages of fining and coarsening upward sequences in the deepest zone transiti ons to a pattern of minor shifts in grain size and consistent sorting up column, suggesting a shift in depositional regime (see Figures 4 2 and 4 3) . Test Unit Excavations Detailed results from test unit excavations are available elsewhere for Bird Island (McFadden and Palmiotto 2012; McFadden et al. 2014), Butler Island (McFadden 2014), and Garden Patch (Wallis and McFadden 2013 , 2014), and only a basic summary of the results of excavation is presented here . Table 3 1 provides frequencies of identified ty pes by site for the study area.

PAGE 150

150 Bird Island Three 1 x 2 m test units were excavated at Bird Island. Test Unit 2 was excavated at a lower elevation of the island and revealed two main strata of disturbed deposits and no intact archaeological deposits, and materials collected from that unit are not included in the discussion here (see Figure 3 6 ) . Test Units 1 and 3 were excavated at the highest elevation of the island and are considered as a single analytic unit . The units contained three intact midden st rata dating to the Weeden Island, Deptford, and Late Archaic periods (see Figure 3 2 ). In terms of MNI, the Weeden Island and Deptford middens were composed predominately of oyster ( Crassostrea virginica ) and the Late Archaic midden contained mostly periw inkle ( Littorina sp.). All three middens also contained faunal materials and lithics, with the majority of lithic artifacts being burned limestone. Pottery was recovered from all three middens, although the fiber tempered pottery recovered from the Late Archaic midden was sparse and likely associated with features that intruded into the otherwise aceramic midden deposits. Recovered pottery was predominately sand tempered plain. Bayou Punctated and Carrabelle Punctated sherds. Charred material from the base of this midden yielded an AMS age estimate of cal AD 810 980 . An increase in shell density and a change in material culture marked the transition to the underlying Deptford P eriod midden materials. Shell tools were more numerous in this midden and Deptford Linear Check Stamped sherds were recovered from level excavations. Charred material from the base of the Deptford deposit yielded an AMS age estimate of cal 360 170 BC. A culturally sterile sand stratum of variable thickness separated the overlying Weeden Island and Deptford deposits from the underlying Late Archaic midden . There

PAGE 151

151 was a marked increase in gastropod shells and faunal material in these earlier deposits, as we ll as an increase in lithic artifacts. Only three fiber tempered sherds were recovered from the upper portion of the midden. Charcoal recovered from the base of the deposits returned an age estimate of cal 2480 2290 BC. Two pit features that emanated fro m near the top of the culturally sterile sand stratum contained no deposits from the middens above, suggesting they were dug prior to the deposition of the Deptford Period midden materials. Only one of the pits contained artifacts, a cluster of Orange Inc ised fiber tempered pottery at the base of the pit where it intersected the Late Archaic midden. The other feature contained no artifacts and did not extend into the Late Archaic deposits. A third feature was identified at the emerging surface of the Lat e Archaic midden and contained 15 large unmodified lightning whelk ( Busycon contrarium ) shells that were tightly clustered, suggesting they had been contained in some type of bag or basket at the time of deposition (see Figure 3 5) . Butler Island The Butle r Island NE site had two main areas of occupation (see Figure 3 8) . Test Units 1 and 3N, were excavated in the northeastern portion of the site and Test Unit 2 was placed in the southwestern portion of the site. Test Unit 1 contained two major midden dep osits below a thin stratum of redeposited materials that included modern debris (Figure 3 9) . The upper midden stratum consisted on oyster shell with vertebrate fauna and modified bone. Pottery was predominately sand tempered plan, but also included Pasc o Plain, Deptford Linear Check Stamped, and Swift Creek types. A large cluster of New River Complicated Stamped sherds was discovered near the base of the midden. A Bradford type biface and a small chunk of hematite were also

PAGE 152

152 recovered from this midden. Charred material from the base of this stratum yielded an AMS age estimate of cal AD 1035 1215. The underlying, relatively shell free, midden stratum extended to a depth of 63 cmbs and contained significantly reduced frequencies of pottery and vertebrate fauna in a matrix of dark, organically stained sand. Two posthole features extended beneath this stratum into the subsoil beneath. Test Unit 3N was excavated approximately 30 m to the northwest of Test Unit 1. This 1 x 0.5 m unit was offset from a defu nct 1 x 2 m unit and had significantly deeper deposits than Test Unit 1, with cultural materials extending to a depth of 135 cmbs (see Figure 3 11 ). The upper 60 cm contained disturbed and mixed deposits, likely a combination of redeposited midden materia ls and activities associated with a nearby abandoned structure. The uppermost intact midden materials consisted of very dense crushed shell that included Weeden Island Incised and Pasco Plain sherds. Charred material from the base of this midden yielded an AMS age estimate of cal AD 885 1015. Beneath the crushed shell, a stratum of whole shell with increased vertebrate fauna density and decreased pottery frequencies extended to a depth of 102 cmbs. The lowermost midden stratum was characterized by decre ased shell density in black, organic rich sand. Pottery frequency increased and included Deptford Linear Check Stamped and Swift Creek Complicated Stamped sherds, and a perforated bone tool was recovered from the base of the deposit. Charred material fro m the base of this midden yielded an AMS age estimate of cal 170 BC to AD 5. Test Unit 2 was placed in southwestern portion of the site and contained only a very thin (20 cm) stratum of midden materials (Figure 3 10) . Deptford Linear Check Stamped and Pa sco Plain sherds were recovered, along with a large chert

PAGE 153

153 core/hammerstone and a Woodland stemmed biface, from the shell midden. Two overlapping pit features that contained shell and vertebrate fauna extended into the culturally sterile subsoil beneath th e midden stratum. Given the shallow nature of the deposits, and the excavation units close proximity to the shoreline, it is likely that the upper portion of the midden has been truncated. Garden Patch Nine visible and topographic features at the Garden Patch site were initially identified by Kohler (1975), and includes Mounds II, IV, V, VI, VII, and VIII. The remaining features include two areas of midden, Area I and Area III, and a shallow freshwater pond, Area IX (Figure 5 3 ) . Shovel testing in the f all of 2012 identified an additional area of midden, designated Area X (Wallis and McFadden 2013). During the summer of 2013, two contiguous test units were excavated in Area I, two test units in Mound II, two test units in Mound IV, and one test unit in Area X. The two contiguous test units in Area I, Test Units 7 and 8, are considered one analytical unit. This excavation block had two strata of intact cultural deposits that were recognized by changes in color and artifact density. These two shell free midden strata extended to a depth of 60 cmbs. Sand tempered plain sherds dominate the pottery assemblage, but diagnostic types, including Swift Creek Complicated Stamped, Carrabelle Punctated, Pasco Plain, and Weeden Island Red sherds were recovered from both strata. Lithics were also present throughout the unit and Pinellas type bifaces were recovered from both strata, but lithic artifacts in the lower stratum were more frequent and also included a Florida Spike type biface. A radiocarbon date obtained on

PAGE 154

154 Figure 5 3 features designated by area numbers with Area X added . Adapted from Kohler, Timothy. soot from a Swift Creek sherd recovered from the lower anthropogenic stratum yielded an age of cal AD 775 980. Below the midden deposits, numerous features extended into the culturally sterile subsoil and included 12 postholes, three pits, and a squarish shaped feature of unknown function (Figure 5 4 ). Radiocarbon dates on charred material from one of the posthole features and one of the pit features returned identic al AMS age estimates of cal AD 715 890. C harred material from the square shaped feature yielded and AMS age estimate of cal AD 675 870 . Test Units 2 and 3 in Mound II had more complex stratigraphy than that of Area I. Multiple strata of anthropogenic de posits in Test Unit 2 contain various densities of shell, with intermittent lenses of shell free sand suggesting multiple episodes of deposition. The two uppermost anthropogenic strata consisted of shell free sand that Midden X

PAGE 155

155 was emplaced over an underlying slop ing deposit of moderately dense shell in an organic rich black sandy matrix. Charred material from the upper sandy deposit yielded an AMS age estimate of cal AD 340 425 and char red material from the shell bearing deposit yielded an AMS age estimate of cal AD 435 610. Although cultural materials were present throughout the unit, the two upper strata contained higher frequencies of artifacts than that of the underlying midden deposits that extend to the culturally sterile subsoil. Two radiocarbon dates fro m this basal stratum include one from the upper portion on charred material that yielded an AMS age estimate of cal AD 345 430 and a second from char red material from the bottom that yielded an AMS age estimate of cal AD 405 550. Cultural materials recove red from the unit include fragments of hematite and mica, a soapstone abrader, a limestone plummet, chert flakes, a Leon hafted biface, and a Columbia hafted biface. In addition to the dominant sand tempered plain, pottery types were consistent throughout the unit and included Pasco Plain, Swift Creek Complicated Stamped, Deptford Simple Stamped, and cord and fabric marked wares. Although the dates at the base appear to be older than those above, the consistency of diagnostic artifacts throughout the uni t suggests the overlapping range of the dates may reflect the correct construction age for this mound around AD 400 425. Test Unit 3 was significantly different, as far as stratigraphy, even though the two units were only 2 m apart. The artifact assemblag e was consistent among the Mound II units, but this second unit had very little shell and four, possibly five, postholes visible in the profile. A large post feature that was excavated in the unit floor contained charred material, one oyster shell, and a large limestone cobble at the base. Charred

PAGE 156

156 Figure 5 4 . Planview drawing of features in Test Unit 7 (left) and Test Unit 8 (right), Area I, Garden Patch (8DI4). Figure from Wallis , Neill J. and Paulette S. McFadden . 2014. S uwannee Valley Archaeologica l Field School 2013: The Garden Patch Site ( 8DI4). Miscellaneous Report No. 64. Division of Anthropology, Florida Museum of Natural History, University of Florida, Gainesville. material from the large post feature yielded an AMS age estimate of cal AD 240 395. Test Units 4 and 6 were excavated near the fresh water pond on Mound IV. Test Unit 4 had one main anthropogenic stratum that consisted of vertebrate fauna, pottery, and lithics in a shell free sandy matrix that extended to a depth of 118 cmbs. S oot from a Swift Creek Complicated Sherd recovered from near the top of the stratum yielded an AMS age estimate of cal AD 255 405. Soot from a second sand tempered cord marked sherd recovered from near the base of the stratum yielded an AMS age estimate of c al

PAGE 157

157 AD 25 120, the earliest date for human occupation at the site. At least four posthole features extend below the anthropogenic stratum. Test Unit 6 was placed approximately 30 m to the west of Test Unit 4 and had two anthropogenic strata. The upper st ratum resembled the deposit in the other test unit and consisted of vertebrate fauna, pottery, and lithics in a sandy matrix. Charred material from this stratum returned an AMS age estimate of cal AD 240 395. Underlying this deposit was a stratum of blac k sand and oyster shell of varying density with lower frequencies of pottery. Exotic lithic debitage, including one quartz and five crystal quartz flakes, was recovered from this deposit. Three features were identified below the shell deposit, two post ho les and a pit. C harred material from the pit returned an AMS age estimate of cal AD 395 540. Substantially more Pasco Plain pottery was recovered from Test Unit 6, and this unit also contained Alachua Cob Marked sherds. These later types are consistent with the date from the pit feature and suggests that the occupation in the area of Test Unit 6 persisted for longer than at Test Unit 4. Test Unit 1, was excavated into the dense midden at Area X and had two anthropogenic strata, both of which contained v ertebrate fauna, and other cultural materials. The pottery assemblage was dominated by sand tempered plain sherds, but also included Swift Creek Complicated Stamped and Pasco Plain sherds, along with lesser frequencies of other types. Other artifacts rec overed from this unit include a Simple Stamped disk, mica fragments, a sandstone plummet, a chert core, a Pinellas type biface, a Leon type biface, and several biface fragments. Char red material from the upper stratum yielded an AMS age estimate of cal AD 560 650, and charred material from the lower stratum returned an AMS age estimate of cal AD 340 425. A

PAGE 158

158 feature, which contained dense shell, bone, and charcoal in black greasy sand, appears to be a pile of material that was deposited in a single episode. Charred material from this feature yielded an AMS age estimate of cal AD 425 595. Additional features included three postholes identified in the profile. Finally, the backfill from a trench excavated into Mound V by Timothy Thompson in the late 1960s w as re excavated to retrieve unrecorded stratigraphic data. The complex stratigraphy suggests multiple episodes of activity on this portion of the mound. A burial was encountered at the contact between the ground surface and the mound deposits, with multi ple associated posthole s nearby. Charred material from two of these features yielded ages of cal AD 265 420 and cal 1045 905 BC. The latter date is obviously anomalous and may suggest that the wood used to make this post was salvaged from the nearby pond or marsh. Charred material from a large pit feature situated near the postholes yielded an AMS age estimate on charred material of cal AD 240 400. D iscussion Data from the range of sampling strategies employed in northern Horseshoe Cove enable the constr uction of a chronology of environmental change and human occupations in this region of the northern Gulf Coast of Florida (see Figure 4 4 ). Models for lithofacies interpretations (Table 2 3 ) come from Wright (1995) and Hine et al. (1988), and include the addition of the tidal channel and event bedding categories. The sediments underlying anthropogenic deposits on the islands and mainland exhibit the expected suite of characteristics of dunes and sandsheets that accreted at the end of the Late Glacial Maxi mum. A transition to more consistent grain size and sorting characteristics above the lowermost zones in the sediment columns collected at Garden

PAGE 159

159 Patch suggests a shift in depositional regime that likely correlates with a climate shift. No dates were obt ained from these deep deposits, but it is likely that this shift represents accumulation of aeolian sediments at the end of the Pleistocene, around the same time that the Butler Island parabolic dune began to form, 20 ± 4 ka (Wright et al. 2005). Thousand s of years later, when the shoreline was in relatively close proximity, these landforms would have been elevated well above sea level. in the study area, cal 3630 2942 BC. Th e lack of fractionation and adjustment for reservoir effects make this date problematic, but it suggests that the interments on the southwestern shoreline predate the development of the marsh system by some four centuries. This means that the burials coul d have been emplaced when the surrounding area was still terrestrial. Excavations of the burials by FSU note that they were contained within an approximately 55 cm thick, dark earth, shell free, aceramic midden that also contained associated artifacts (Da sovich 1999). Up to 2 m of shell midden above the burial stratum contained stratified deposits with temporally diagnostic ceramics ranging from Late Archaic fiber tempered wares through Weeden Island types. Dasovich (1999) reports that soapstone sherds w ere recovered from the shoreline, presumably eroding from the midden. It is unknown if the soapstone was associated with the burials or the overlying shell midden, but soot from one of the sherds yielded an AMS age estimate of cal 2193 1772 BC (Yates 200 0). Fresh to brackish marsh deposits formed in the parabola of Butler Island and a tidal creek, likely an extension of Lolly Creek, flowed to the south of Bird Island by at least cal 2870 BC. With the exception of the Granberry date, the earliest signific ant

PAGE 160

160 human occupation in northern Horseshoe Cove occurred at Bird Island after the initial flooding of the area, as early as cal. 2480 BC. This date is contemporaneous with the earliest occupation at coastal sites to the south in the Cedar Keys area. Shel l Mound was occupied by about cal 2480 BC (Sassaman et al. 2013:82) and slightly later, Cat Island was occupied between cal 2480 2330 BC (Sassaman et al. 2010:192) and Ehrbar between cal 2560 2350 BC (McFadden and Palmiotto 2013:52). At the time of this o ccupation, the island was still connected to the mainland to the north and extensive marsh had developed to the south and east. At the time of initial occupation, the Bird Island site was situated in an area that was protected from the high energy, open w ater environment by the marshes, with the tidal creek providing access to marine resources. The predominance of marsh periwinkle in the midden suggests these littoral snails were easily accessible, but the absence of truncatella , a salt water shoreline ni che gastropod, suggests the environment was still predominantly freshwater (McFadden et al. 2014). However, the inclusion of marine species, including sea trout, mullet, crown conch, and whelks, suggests close proximity to these resources. Gastropod hamme rs and a large whelk adze were recovered from the Late Archaic midden stratum, which suggests woodworking or other activities requiring tools, and it is possible that the unmodified whelk shells discovered in Feature 3 were destined to become tools. Lithi c tool production and/or maintenance was performed at the site, evidenced by the higher frequencies of debitage in this stratum over the later deposits. Given the regional history of the movement of soapstone (e.g., Sassaman 2006), and the date reported b y Yates (2000), it is assumed that this exotic material is associated with the Late Archaic occupation of the site, and the presence of those

PAGE 161

161 materials suggests that this population participated in extralocal exchange networks with materials ultimately sta ying at the site. Sometime after cal 2030 BC, the topographic low between the arms of Butler Island transitioned to inter to subtidal sands, suggesting the formation of a fresh to brackish marsh tidal pond to the northeast of Bird Island. By cal 2005 BC, the marsh overstepped a topographic high and peat deposits began to accumulate on the lower elevations of the western edge of the main portion of Butler Island. There is a gap of nearly two millennia in basal radiocarbon dates between the Deptford Midden a nd the Late Archaic midden at Bird Island, although assuming that it was not emplaced after the end of the occupation, the date on the soapstone sherd suggests the Late Archaic occupation could have continued on the island until as late as cal 1772 BC (Yat es 2000), narrowing the gap to closer to 1400 years. High and variable frequencies of sponge spicules, and the sporadic presence of diatoms and foraminifera in the sediment samples from the culturally sterile sand deposit that separates the two middens, a long with the identification of microstratigraphy, suggests that at least some portion of the stratum is a storm deposit. Similar sedimentary characteristics were observed in storm deposits at Little Bradford Island (Sassaman et al. 2010) and at Waccasass a Bay (Goodbred and Hine 1993) after the so began to accumulate by sometime after cal 360 BC, and since it is unknown if the island was reoccupied immediately, this date provides only a terminus ante quem for the event.

PAGE 162

162 Some portion of the sandy stratum overlying the Late Archaic midden could have accreted in the absence of human occupation. Evidence of site abandonment elsewhere along the northern Gulf Coast of Florida suggests a region wide hiatus in occupation between 1350 BC and 550 BC (Sassaman et al. 2014), and abandonments seen elsewhere in the southeastern United States have been attributed to a period of global cooling that resulted in a drop in sea le vel (e.g., Marquardt 2010). Along have been attributed to reduced rates of sea level rise. This reduction allowed for marsh accretion or deltaic formation to outpace rates of sea level rise, creating aggrading or prograding sequences ( Evans et al. 1985; Parkinson 1989; Savarese et al. 2006). No regressive sequences were identified in the cores from Horseshoe Cove, although any subsequent transgression would likely sco ur regressive sequences and, with the exception of the tidal channel deposits on the southwestern shoreline of Bird Island, all of the seaward most cores have ravinement surfaces above the limestone base. If a relative drop in sea level occurred at Horses hoe Cove, a date from the base of inter to subtidal muds above a ravinement surface in the seaward most core suggests subsequent shoreline transgression occurred as early as cal 890 BC. By at least cal 790 BC, the shallow freshwater pond at Garden Patch was periodically filled with water, evidenced by laminar bedding indicative of periods of wet and dry conditions, and by cal 370 BC, groundwater levels had risen enough to make the spring fed pond a permanent feature.

PAGE 163

163 On Bird Island, Deptford Period midde n materials began to accumulate as early as cal 360 BC. By this time, the morphology of the area was significantly different. The Bird Island site would have been located on the distal portion of a peninsula that extended southward from the mainland. To the west and south the island was bounded by a large tidal flat that was either partially or totally exposed during low tides. The tidal creek continued to flow to the south of the site, between the island and the tidal flat. In the lower elevation area s east and to the northeast between the arms of Butler Island, the area had transitioned to very shallow restricted marine waters. Moderate to dense tellina and macoma shells are present in these deposits that were identified in all of the cores between B ird Island and the main portion of Butler Island. These small bivalves prefer shallow marine environments with sandy bottoms where energy is high enough to deter deposition of muds and silts (Bouchet 2014; Gofas 2014). In contrast to the Late Archaic mid den, oyster shell dominated and virtually no periwinkles were present in the later deposits. This organism would certainly still be readily available in the marsh areas fringing Butler Island and possibly the mainland, so it is more likely that these smal l snails were no longer a targeted resource. Shell tools continued to be utilized during this time, but lithic artifacts diminished significantly, suggesting either less reliance on lithic tools or less frequent manufacture and maintenance activities in t his area. The initial occupation of Butler Island postdates the Deptford occupation of Bird Island by as little as a decade, and the environmental setting of the former closely resembled that of the latter in the centuries prior. Butler Island was stil l largely connected to the mainland, possibly separated by only the small and easily crossed

PAGE 164

164 Tripod Creek that flows to the west. Extensive marsh had develop in the low lying areas to the east, with Lolly Creek providing access from the site to the open m arine environmental and the resources that it provided. Admittedly, determining site function is difficult given the sample size of only artifact assemblages suggest di fferent types of activities occurring at each site. Although the Deptford period pottery types are mostly consistent between the two sites, with the exception of the absence of Deptford Simple Stamped pottery at Bird Island, pottery frequencies were highe r at Butler Island (see Table 3 1 ). The density of vertebrate faunal remains in the midden is lower at Butler Island and only one possibly modified shell was recovered at the site. Nine modified shells were recovered from Bird Island, including three gas tropod hammers. Finally, the units at Bird Island yielded significantly more burned limestone. By as early as cal AD 25, a Deptford Period domestic occupation began near the freshwater pond on Mound IV at Garden Patch and continued activities on the islan ds were likely not related to domestic occupation. Still possibly up to a kilometer inland from the marshes at the time of initial occupation, the site was located along the shore of Lolly Creek. The creek would have been a main avenue to the marine reso urces that were the predominant species in the faunal assemblage and fresh water was readily available in the pond. The artifact assemblage at Garden Patch was significantly different from the Deptford deposits on the islands. Multiple postholes suggest domestic structures, pottery frequencies were substantially higher, and midden deposits were thicker than at the island sites. A notable difference is the presence of modified

PAGE 165

165 flakes, ground stone artifacts, and exotic lithic materials, none of which were recovered from the islands. The exotic lithic materials included crystal quartz, quartz, and quartzite, all of which would have been imported from areas to the north. Occupation at Garden Patch intensified significantly after AD 200, coinciding with an e nvironmental shift that likely created a setting that mirrored that of Butler Island and Bird Island at their initial occupation. Between AD 200 and AD 300, rates of sea level rise increased by an order of magnitude, causing a transgression of between two and four kilometers at Waccasassa Bay (Goodbred et al. 1998). Environmental change was most certainly also occurring at Horseshoe Cove, and likely resulted in the development of extensive areas of marsh to the east of Garden Patch. Bird Island was compl etely isolated from the mainland by this time and Butler Island was surrounded by marsh. After cal AD 240 at Garden Patch, domestic activities continued at Mound IV, construction began at Mounds II and V, and there was significant midden accretion at Area X. By about AD 500, construction was completed on the mounds and a circular village had been established that encompassed Areas III and X. Differences in artifact assemblages among the sites suggest that any continuing occupation on the islands during the fluorescence of Garden Patch was light and likely waning. No radiocarbon dates for this period come from the islands, although this could be a product of bias in the sampling strategy since dates were obtained from only the basal portions of midden st rata. However, to the south, construction began at Shell Mound by about cal. AD 430 and continued for the next two centuries while occupation appears to have ceased at nearby island sites like Deer Island (Monés et al. 2012; Sassaman et al. 2013). At Gar den Patch, Deptford Simple Stamped and Swift Creek

PAGE 166

166 types were common and co occurred. At Bird Island, only two Swift Creek Complicated Stamped sherds were recovered from intact deposits and no Deptford Simple Stamped sherds were recovered (see Table 3 1 ). At Butler Island, intact deposits from all three excavation units combined yielded only six Swift Creek Complicated Stamped sherds and three Deptford Simple Stamped sherds. Exotic materials, including mica fragments and hematite were recovered from Moun d II, Mound IV, and Area X. Other artifacts included a limestone plummet, a sandstone plummet, a Simple Stamped drilled and ground disk, and multiple bifaces and biface fragments, all recovered from deposits dating to the period after cal AD 240 at Garden Patch. At the island sites, the artifact assemblage consisted of only potsherds, sparse locally available lithic debitage, occasional modified shell and bone, and no exotic materials. By about cal. AD 600, the majority of activity at Garden Patch ceased and it is likely that the site was abandoned for nearly a century before the establishment of a Weeden Island village. Presumably, the islands were also abandoned during this time. Between cal AD 660 and 770, the elevated area on the landward side of Bu tler Island transitioned to marsh and marsh encroachment likely continued to the northeast of Garden Patch. A radiocarbon date from a feature at Garden Patch Area I suggests the establishment of a Weeden Island village to the west of the previously occupi ed areas by as early as cal AD 675. But slightly later dates from other associated features in the excavation unit suggest occupation began closer to cal AD 715. It is curious that the Weeden Island residents do not appear to have utilized the previously occupied areas, with the exception of the construction of Mound VIII, which contained whole Weeden Island period vessels and burials (Moore 1902). Otherwise, virtually no Weeden Island

PAGE 167

167 sherds were recovered in the Deptford and Swift Creek period occupati on areas of the site. Bird and Butler Islands were occupied again about a century after the establishment of the Weeden Island village at Garden Patch, by at least cal AD 810 at Bird Island and cal AD 885 at Butler Island. These later occupations on the islands continued into the late Weeden Island period and possibly into the early Mississippian, with the latest radiocarbon age estimate of cal AD 1035 1215 coming from Butler Island. A similar pattern is seen to the south, where basal dates from midden d eposits at Cat Island and Deer Island suggest those sites were reoccupied after cal. AD 600 around the same time that occupation began inside the ring at Shell Mound (Monés 2012; Sassaman et al. 2010; Sassaman et al. 2013). These later occupations in Hor seshoe Cove appear to be ephemeral, with virtually no diagnostically Weeden Island pottery recovered at Butler Island, and very little at Bird Island. Postholes identified in Test Unit 1 at Butler Island extend from the upper midden strata in Test Unit 1, suggesting the structure dates to this later occupation. The presence of hematite, a bead from Bird Island and a small chunk from Butler Island, suggest exotic materials were making their way to the island at this time. Although Bird and Butler Islands were reoccupied, despite the significantly altered environmental setting, it appears that new areas in northern Horseshoe Cove were not settled. By the time of these later occupations, the morphology of the area would likely have been very similar to the modern setting. Survey of Butler Island revealed no archaeological deposits in other areas of the main portion of the island, even though access to those areas would be comparable to that of the Butler Island NE

PAGE 168

168 site. Similarly, no other single component Weeden Island period sites, other than that of the western portion of Garden Patch, have been identified in this area of Horseshoe Cove, or to the south in the Cedar Keys, suggesting this pattern is not unique to Horseshoe Cove. C onclusion Places targeted for occupation by the Late Archaic and Deptford period residents of Horseshoe Cove were most certainly informed by generations of experience with the dynamic and changeable coast. Beyond immediate occupation, it is likely that there was the expectation t hat those places would persist into the future, albeit to be utilized in different ways. This strategy would have created a rolling mosaic of occupations along the coast with newly founded settlements landward and site function changing over time at seawa rd sites as sea level rose. In contrast, the durable evidence of previous habitation appears to have informed decisions by the later Weeden Island p eriod residents about places to occupy, and new areas were not colonized even though they may have had simi lar morphological characteristics. Although the Weeden Island people at Garden Patch did not establish areas of domestic occupation atop previously occupied areas at the site, they did continue to interact with those areas. 999) Bronze Age Britain example, communal ideas about the role of these places may have shifted. During the Deptford and Swift Creek periods the mounds were the product of routine everyday individual and communal activities, including rituals and burials. These constructions were a central part of the community, both physically and socially. In contrast, even though the Weeden Island people were obviously aware of these ancient mounds, they chose to put their village on the periphery of the mound complex . Perhaps the ancient landscape was reserved for

PAGE 169

169 sacred activities, such as the placement of the Weeden Island burial mound in the central portion of the previously occupied area. It is unknown if the Weeden Island people had genetic ancestral connection s with the Deptford and Swift Creek period residents, but the durable remains of ancient practices on the landscape would invite interpretation. The evidence of past habitation likely provided a perception of deep time, creating a history that drew and an chored people to these places.

PAGE 170

170 CHAPTER 6 CONCLUSION AND RECOMMENDATIONS Slowing rates of sea level rise after about 5,000 years ago allowed for the development of extensive oyster reefs and marshes in the Big Bend of Florida and this resource rich env ironment supported aboriginal communities in Horseshoe Cove for at least the last 4, 5 00 years. These communities successfully navigated environmental changes that altered shoreline morphologies and impacted resource distribution and availability. However changes in human practices in this area were not dictated solely by the environment, they were also contingent upon social and cultural factors. Following on the premise that research focused on the relationship between environmental change and shifting h uman practices along the coast need s to rely on local scale paleoenvironmental reconstructions, geological and archaeological data were collected from northern Horseshoe Cove and used to test five hypotheses. The paleoenvironmental reconstruction created from these data (see Figure 2 4) provide an overall figures that illustrates the temporal and spatial aspects of the results. This figure symbol on the figure, provide inf ormation about that transect only and this schematic is a representation of the environment based on an extrapolation of that information into the surrounding area using modern elevations and bathometry. For instance, no cores were collected around Cotton Island and the exact nature of the changes to that island during the evolution of this coastal area is unknown. Although the preserved stratigraphy in both the marine sediment cores and the archaeological sites, along with environmental data collected fr om the archaeological sites, provides some level of

PAGE 171

171 confidence in this reconstruction, it should be viewed as only a representation of the likely changes that occurred in this area. Hypothesis 1 Slowing rates of sea level rise allowed for the formation of the marsh and estuarine environment in the Horseshoe Cove area at approximately the same time that these environmental settings developed at the Suwannee Delta and Waccasassa Bay, sometime around 4,500 years ago. The hypothesis cannot be rejected by the re sults of this study. In northern Horseshoe Cove, the initial transition to fresh to brackish marsh occurred sometime between 4,800 to 4,500 years ago. At Waccasassa Bay (Goodbred 1994; Goodbred et al. 1998), a similar transition was documented at about the same elevation and relative shoreline proximity. The initial flooding of the area in Waccasassa Bay postdates the transition in Horseshoe Cove by as much as four centuries, but the overlap in the 2 sigma date ranges among the two locations suggest the transition was likely contemporaneous. The transition at the Suwannee Delta (Wright 1994; Wright et al. 2005) was a bit earlier due to increase river discharge as the water table rose. Hypothesis 2 Lowered rates of sea level rise or a global drop in sea level resulted in a regression that forced the earliest residents in the Horseshoe Cove study area to relocate between about 3,500 and 2,500 years ago. The hypothesis cannot be rejected; however, the results are ambiguous. Regressive deposits were not i dentified in the marine sediment cores collected from Horseshoe Cove , therefore a dditional data are needed to further test this hypothesis. Similar to the Suwannee Delta (Wright 1995; Wright 2005), seaward sediments were

PAGE 172

172 scoured during transgression creat ing ravinement surfaces and sharp contacts between the limestone bedrock and overlying nearshore or marine deposits . This scouring likely obliterat ed any regressive sequences. No evidence for a transition from marsh to terrestrial environment was identif ied in the cores. However, with the exception of a substantial drop in sea level, erosion during a regression would likely lower the elevation of the marsh area and create accommodation space, thus keeping the marsh in equilibrium with falling sea level. The only radiocarbon date that falls within the range of the hypothesized regression comes from a core located seaward of Bird Island where inter to subtidal sand/mud began to accumulate between 2840 2750 years ago . If a regression occurred in Horseshoe Cove, the date from this core could suggest a subsequent transgression at that time. However, given the location of this core in the shallow open marine area, and the lack of dated transitions elsewhere, I am hesita nt to offer a conclusion on the hypothe sized regression. There is circumstantial evidence from the archaeological record that may suggest some type of environmental shift that resulted in the abandonment of Bird Island. Although it appears that some portion of the sediments in the culturally s terile sand stratum was transported from an aquatic environment, the temporal gap created by this stratum is consistent with coastal abandonments seen elsewhere in the southeastern United States and falls within the timeframe of the regression identified a long the Georgia Bight and the aggregating and prograding deposits observed in South Florida. Hypothesis 3 A period of increased rates of sea level rise caused a significant transgression and environmental change after around AD 200, and resulted in chan ges in settlement patterns of the coastal residents of Horseshoe Cove .

PAGE 173

173 The hypothesis cannot be rejected; however, the results are ambiguous. None of the facies transitions in the marine sediment cores that were targeted for radiocarbon dating date to aro und AD 200. Much like the first hypothesis, archaeological evidence suggests the possibility of some type of environmental shift after AD 200. A lack of radiocarbon dates from Bird and Butler Islands suggest that those sites may not have been extensively occupied during this period, although the sampling strategy of obtaining only basal dates for the midden deposits could create some bias. Occupation at Garden Patch began to intensify as early as AD 240 and significant mound construction and midden accum ulation was occurring by AD 400 (Wallis and McFadden 2014). Similar shifts have been observed at Shell Mound near Cedar Key where radiocarbon dates suggest rapid construction of the mound soon after AD 400 ( K. E. Sassaman , personal co mmunication, 2014). T he lack of evidence in the sediment cores of an environmental shift around AD 200 is likely more a product of a lack of dates rather than a lack of data. Only the most basal transitions were initially targeted for dating in this study. Two cores were pla ced landward of Butler Island in hopes of identifying the most recent transgressions. However, neither of these cores provided viable data to address this hypothesis. Pleistocene sands in one of the cores, likely the elevated edge of Butler Island, remai ned above sea level until at least 1290 B.P., postdating the hypothesized transgression. The second core was collected near the mouth of Lolly Creek, where sediments transported from the creek created a small delta. Hypothesis 4 Specific morphological cha racteristics and environmental settings were targeted for occupation as sea level rose and communities were forced to relocate further inland.

PAGE 174

174 The hypothesis cannot be rejected. Areas targeted for settlement in Horseshoe Cove had specific characteristic s . They were elevated areas on the landscape that were near a source of fresh water, protection from the higher energy open marine environment, and had easy access to marine resources via a tidal creek . Pre Columbian c oastal communities o n the northern G ulf Coast of Florida ha d a long history of experience with sea level rise and environmental change and would have used that knowledge to guide decisions about the appropriate places to resettle when landward movement became necessary. T he paleoenvironment al reconstruction offered in Figure 2 4 shows that the settings at all three sites at the time of initial settlement were very similar. The earliest occupation in the study area occurred at Bird Island at a time when the shoreline was probably still over a kilometer to the west (Wright et al. 2005). The site was situated along a tidal creek that likely transported fresh water to the Gulf of Mexico. Still part of the mainland and bordered by extensive marsh, the area was protected from the higher energy m arine environment, but had easy access to marine resources via the tidal creek. The setting was very similar at the time of initial occupation at Butler Island, although the marsh inside the parabola of the island appears to have been transitioning from f resh or brackish marsh to salt marsh. It is likely that Lolly Creek, which flowed into the Gulf of Mexico on the eastern edge of the site, had variable salinity depending on the tides and was likely not the source of fresh water for the residents. Howev er, given the karstic nature of the landscape, it is possible that there were shallow spring fed fresh water ponds nearby, much like the fresh water pond that still exists at Garden Patch.

PAGE 175

175 T he initial occupation at Garden Patch occur red along the shore of the fresh water pond on an elevated landform that borders Lolly Creek. The surrounding area appears to have been mainly terrestrial during this initial occupation, but by AD 200, it likely had begun transitioning to swamp and marsh. The documented sites at Cotton Island (8DI51) and the southern arm of Butler Island (8DI97) appear to have been heavily impacted or destroyed by storms and sea level rise. No chronological data are available from Cotton Island so it is unknown when that site was initially occ upied. Kohler and Johnson (1986) suggest a Deptford age for the site on the southern arm of Butler Island, so it may be assumed that it was occupied contemporaneously with either or both Bird Island and Butler Island NE , i n which case, the setting was ver y similar to that of Butler Island NE during its early occupation. Hypothesis 5 Social factors played a role in decisions about the continued use or reoccupation of sites that were in different environmental settings tha n those of initial occupation. The hypothesis cannot be rejected. The evidence suggests that social factors likely played a significant role in decisions about places to occupy during the Weeden Island Period, after about AD 800, in Horseshoe Cove. The lack of radiocarbon dates from Bird and Butler Island during the period of substantial occupation at Garden Patch suggests little activity at these sites. After AD 600, Garden Patch itself appears to have been abandoned for a period of time. Sometime after AD 675, a Weeden Island village w as established on the western periphery of the site. Although Weeden Island residents do not appear to have utilized the areas of earlier occupation for domestic activities, they did construct a burial mound within the older mound complex. On Bird and Bu tler Islands, Weeden Island period midden deposits began to accumulate atop

PAGE 176

176 earlier Deptford and Swift Creek period middens. These occupations appear to be ephemeral; virtually no Weeden Island pottery was recovered at Butler Island, and very little was r ecovered from Bird Island, and these occupations continued for up to four centuries. Compared to that of initial settlement, the environmental setting at the sites on Bird and Butler Islands was very different during the Weeden Island period, particularly in terms of proximity to fresh water resources and ease of access to the mainland. Although other areas in northern Horseshoe Cove had similar settings during the Weeden Island period, these areas do not appear to have been targeted for occupation. Inde ed, to date, no single component Weeden Island sites have been located in the study area. Perhaps one of the most important conclusions from this study was that the productive marsh and estuarine environment along the northern Gulf Coast of Florida can, an d does, stay in equilibrium with moderate changes in sea level, and the system can remain in near stasis for centuries. Because of its resilience, significant environmental change in northern Horseshoe Cove likely occurred mainly during periods of increas ed rates of sea level rise that outpaced the ability of the system to stay in equilibrium. The prolonged periods of stasis in Horseshoe Cove allowed time for places of occupation to accumulate durable material remains of past human practices. These place s would have become embedded with histories because of these remains, these references to the past, and would have continue to draw people to them even though the environmental setting may be very different from the setting at initial occupation.

PAGE 177

177 Recommend ations for Fut ure Research This research project in northern Horseshoe Cove is still very preliminary and further work is needed to address many more questions. For instance, this study is geographically very narrow in its focus and the data need to be pu t into a larger regional context since the residents of northern Horseshoe cove lived in a much larger social and geographical context. Similar paleoenvironmental reconstructions from other areas along the Gulf Coast could reveal if the settlement strateg ies employed in northern Horseshoe Cove were similar to those of other coastal communities, or if perhaps those strategies were specific to this area and contingent upon particular morphological attributes of this system or individual experiences of the co mmunities. Perhaps one of the most intriguing mysteries on the northern Gulf Coast of Florida is why were coastal areas apparently abandoned between 1350 BC and 550 BC (Sassaman et al. 2014)? Was it a period of cooling that resulted in a regression at Hor seshoe Cove? If so, evidence for that regression is lacking in the cores, although regressive deposits would likely be scoured by a subsequent transgression. Sandy strata identified between fresh to brackish marsh and salt marsh deposits in several of th e cores could be suggestive of a prograding surface during a period of lowered rates of sea level rise, however these deposits are not well dated and could also be the result of multiple other processes. Further coring in other areas of the marsh or more seaward may help to address questions of a possible localized regression. Additionally, coring in nearby areas w h ere abandonments have been identified, for instance at Shell Mound or near Cedar Key (Sassaman et al. 2014), could provide data that can addre ss this question.

PAGE 178

178 The magnitude of transgression caused by the pulse in sea level rise around AD 200, identified by Goodbred et al. (1998) at Waccasassa Bay, could have caused significant environmental change in Horseshoe Cove as well. Evidence of extens ive to this environmental shift. Radiocarbon dates are needed from lithofacies transitions at higher elevations in the marine sediment cores to determine if, and to wha t extent, a significant transgression affected Horseshoe Cove . Additional cores collected in the marsh landward of Butler Island may also provide data that can be used to further test this hypothesis. The chronological data provided by the core collected from the freshwater pond at the Garden Patch site was invaluable to understanding the timing of environmental change. However, there is a 10 cm thick sandy deposit that crosscuts the organic rich pond deposits that needs further investigation. It is unkn own if the deposit is restricted to just the area from which the core was collected, or if it represents an episode of deposition that affected the entire pond. If not restricted to the core collection area, it is possible that this deposit is the result of flooding that transported the sands into the pond. Curiously, radiocarbon dates from above and below the deposit encompass nearly the entire pre Columbian occupational history of the site, suggesting that the deposit could represent an accumulation of sands that were mobilized over time by nearby anthropogenic activities and trapped in the pond. Additional coring in the pond could help clarify the nature of this deposit. In addition, pollen analysis needs to be conducted on this core, and any subseque nt cores collected from the pond, to address questions about the vegetation at the site during human occupation.

PAGE 179

179 Because the archaeological data are limited to only those deposits that remain above sea level and remain intact, there is an inherent bias in the data. For this reason, archaeological survey needs to be expanded, not just to the north and south along the coast, but also to underwater sites that may provide additional data. I was lucky enough to have unlimited access to a Gulf front home in Hors eshoe Beach and spent a great deal of time there while working on this project. In addition to learning to fish, I also had the opportunity to interact with the local community and learned a great deal from people who had spent their entire life on the co ast. Rather than being a barrier, the sea is simply an extension of the routine landscape and people who live along the coast navigate that landscape in much the way that they do the terrestrial landscape, albeit using different tools and vehicles. The u nderwater landscape has a topography, and that topography attracts certain types of activities. For instance, elevated areas on the sea bottom become targets for fishermen, with the orientation of the fisherman to the elevated area dictated by the behavio rs of fish that congregate there. Different offshore areas may be exploited based on conditions, but many fishermen have areas that they favor over others that may be equally productive. In any case, neither the fish nor the fisherman would be there if n ot for the presence of the topographic high. These previously terrestrial areas may have once been places of human occupation and the social significance of those places may not have been lost just because they are now under water. These places could have remained as landmarks on the aquatic landscape, and likely retained economic importance since fish or other marine organisms would be attracted to this topographic high. Could it even be

PAGE 180

180 possible that the pre Columbian coastal residents understood the ev entual evolution of these areas to underwater habitats when they initially targeted an area for settlement? Obviously, research that targets inundated archaeological sites would be necessary to delve into these types of questions. Initially, the collecti on of sediment cores from topographic highs offshore could identify areas for further testing. Continued archaeological research is necessary on the mainland near Garden Patch. No less than 11 other sites have been reported near Garden Patch, many situate d on an elevated ridge that extends to the north. Research so far has been limited by many factors, not the least of which is access to the areas since much of the land is privately owned. It is important to note here that these site boundaries are arbit rary modern constructions and in no way reflect boundaries created or observed by the pre Columbian residents of Garden Patch and the surrounding area. It is probable that these sites are all small parts of a larger whole and expansion into these other si tes is vital to put Garden Patch in context within this larger area. Finally, faunal analysis from archaeological contexts is vital to understanding the environmental conditions during occupation of the sites. Comparisons of the assemblages among the midd en deposits at Bird Island were an integral part of understanding the evolution of the environmental setting of the island and the same types of analyses need to be conducted on assemblages from Butler Island and Garden Patch.

PAGE 181

181 A PPENDIX INDIVIDUAL CORE DE SCRIPTIONS WITH LITHOFACIES CHARACTERISTICS

PAGE 182

182 Figure A 1. Core HBT2 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 183

183 Figure A 2. Core HBT3 with lithofacies descriptions, percentage of orga nic matter, fines, and carbon ate , and grain size and sorting.

PAGE 184

184 Figure A 3. Core HBT4 with lithofacies descriptions, percentage of organic matter, fines, and carbon at e, and grain size and sorting.

PAGE 185

185 Figure A 4. Core HBT6 with lithofacies descriptions, p ercentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 186

186 Figure A 5. Core HBT8 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 187

187 Figure A 6. Core HBT9 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 188

188 Figure A 7 . Core HBT 10 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 189

189 Figure A 8. Core HBT11 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 190

190 Figure A 9. Core HBT12 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 191

191 Figu re A 10. Core HBT13 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 192

192 Figure A 11. Core HBT14 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 193

193 Figure A 12. Core HBT15 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 194

194 Figure A 13. Core HBT16 with lithofacies descriptions, percentage of organic matter, fines, and carb onate, and grain size and sorting.

PAGE 195

195 Figure A 14. Core HBT17 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 196

196 Figure A 15. Core HBT18 with lithofacies descriptions, percentage of organic ma tter, fines, and carbonate, and grain size and sorting.

PAGE 197

197 Figure A 16. Core HBT19 with lithofacies descriptions, percentage of organic matter, fines, and carbonate, and grain size and sorting.

PAGE 198

198 REFERENCES Balsillie, James H. and Joseph F. Donoghue 2004 High Resolution Sea Level History for the Gulf of Mexico since the Last Glacial Maximum . Vol. 103, State of Florida Department of Environmental Protection, Tallahassee, FL. Barrett, John C. 1999 The Mythical Landscapes of the British Iron Age. In Archa eologies of Landscape: Contemporary Perspectives , edited by Wendy Ashmore and A. Bernard Knapp, 253 265. Blackwell, MA. Bender, Barbara 2002 Time and Landscape. Current Anthropology 43:S103 S112. Blum, Michael D., Tamara J. Misner, Eric S. Collins, Dav id B. Scott, Robert A. Morton, and Andres Aslan 2001 Middle Holocene S ea L evel R ise and Highstand at +2M, C entral Texas Coast. Journal of Sedimentary Research 71 : 581 588. Boggs, S am, Jr. 2005 Principles of Sedimentology and Stratigraphy . Prentice Hal l, Cranbury. Bouchet, P. 2014 Tellinidae Blainville, 1814. World Register of Marine Species. Electronic document, http://www.marinespecies.org/aphia.php?p=taxdetails&id=235 , accessed 01/20/2015. Brooks, Mark J., Donald J. Colquhoun, Richard R. Pardi, W alter Newman, and W.H. Abbott 1979 Preliminary Archaeological and Geological Evidence for Holocene Sea Level Fluctuations in the Lower Cooper River Valley, S.C. The Florida Anthropologist 32:85 103. Brooks, Mark J., Peter A. Stone, Donald J. Colq uhoun, Jan G. Brown 1989 Sea level change, estuarine development, and temporal variability in Woodland period subsistence settlement patterning on the lower coastal plain of South Carolina. In Studies in S outh Carolina Archaeology: essays in honor o f Robert L. Stephenson , edited by A.C. Goodyear & G.T. Hanson, 91 100. South Carolina Institute of Archaeology and Anthropology, University of South Carolina, Columbia.

PAGE 199

199 Sea level changes along the U.S. Atlantic cost: implications for glacial isostatic adjustment model s and current rates of sea level change. Doctoral dissertation, University of Pennsylvania, Philadelphia. Publicly accessible Penn Dissertations. Paper 407. Evans, Mark W., Albert C. Hine, Daniel F. Belknap, and Richard A. Davis, Jr. 1985 Bedrock C ontrols on Barrier Island Development: West Central Florida Coast. Marine Geology 63:263 283. Folk, R.L. 1980 Petrology of sedimentary rocks . Hemphill Publishing Company, Austin. Fleming, Kevin, Paul Johnston, Dan Zwartz, Yusuke Yokoyama, Kurt Lambe ck, and John Chappell 1998 Refining the Eustatic Sea Level Curve Since the Last Glacial Maximum using Far and Intermediate Field Sites. Earth and Planetary Science Letters 163:327 342. Florida Fish and Wildlife Conservation Commission n.d. Saltw ater Fish . Electronic document, http://myfwc.com/wildlifehabitats/ profiles/saltwater /, accessed February 4, 2015. Frazier, David E. 1974 Depositional Episodes: The Relationship to the Quaternary Stratigraphic Framework in the Northwestern Portion of the Gulf Basin . 1 st Ed. Vol. 74, Bureau of Economic Geology, The University of Texas at Austin , Austin.

PAGE 200

200 Gayles, Paul T., David B. Scott, Eric S. Collins, and Douglas D. Nelson 1992 A Late Holocene Sea Level Fluctuation in South Carolina. In Quatern ary Coasts of the United States: Marine and Lacustrine Systems , edited by C.J. Fletcher, III and J.F. Wehmiller, pp. 155 160. Special Publication No. 48, SEPM Society for Sedimentary Geology, Tulsa. Gelsanliter, S., and H.W. Wanless 1995 High F requency Sea Level Oscillations in the Late Holocene of South Florida: A Dominating Control of Facies Initiation and Dynamics. 1 st SEPM Congress on Sedimentary Geology , Congress Program and Abstracts 1. Gofas, S. 2014 Macoma Balthica (Linnaeus, 17 58). World Register of Marine Species. Electronic Document. http://marinespecies.org/aphia.php?p=taxdetails&id=141579, accessed 01/20/2015. Goodbred, S teven L., A lbert C. Hine, and E ric E. Wright 1998 Sea Level Change and Storm Surge Deposition in a Late Holocene Florida Salt Marsh. Journal of Sedimentary Research 68:2 40 252. Goodbred, S teven L. 1994 Geologic Controls on the Holocene Evolution of an Open Marine Marsh System Fronting a Shallow Water Embayment: Waccasassa Bay, West Central Florida . Gornitz, Vivien, and Leonardo Seeber 1990 Verticle Crustal Movements along the East Coast, North America, from Historic and Late Holocene Sea Level Data. Tectonophysics 178:127 150.

PAGE 201

2 01 Jaeger, J ohn M., Ashish Mehta, Richard Fass, and Michael Grella 2009 Anthropogenic I mpacts on S edimentary S ources and P rocesses in a S mall U rbanized S ubtropical E stuary, Florida. Journal of Coastal Research 25 :3 0 47.

PAGE 202

202 E nvironmental and Social Factors in Pre Columbian Settlement on the Northern Gulf Coast of Florida. (In prep).

PAGE 203

203 Monés, Micah P., Neill J. Wallis, and Kenneth E. Sassaman 2012 Archaeological Investigations at Deer Island, Levy County, F lorida . Technical Report 15.

PAGE 204

204 Reimer, Paula J., Edouard Bard, Alex Bayliss, J Warren Beck, Paul G Blackwell, Christopher Bronk Ramsey, Caitlin E Buck, Hai Cheng, R Lawrence Edwards, Michael Friedrich, Pieter M Grootes, Thomas P Guilderson, Haflidi Haflidason, Irka Hajdas, Christine Hatté, Timothy J Heaton, Dirk L Hoffmann, Alan G Ho gg, Konrad A Hughen, K Felix Kaiser, Bernd Kromer, Sturt W Manning, Mu Niu, Ron W Reimer, David A Richards, E Marian Scott, John R Southon, Richard A Staff, Christian S M Turney, Johannes van der Plicht 2013 IntCal13 and Marine13 Radiocarbon Age Ca libration Curves 0 50,000 Years cal BP. Radiocarbon 55:1869 1887. Trend, Tradition, and Turmoil: What Happened to the Southeastern Archaic? Vol. 93, edited by David Hurst Thomas and Matthew C. Sanger, pp. 149 172 . American Museum of Natural History, New York.

PAGE 205

205 Sassaman, Kenneth E., Andrea Palmiotto, Ginessa J. Mahar, Micah P. Monés, and Paulette S. McFadden 2013 Archaeo logical Investigations at Shell Mound (8LV42), Levy County, Florida: 2012 Testing. Technical Report 16. Laboratory of Southeastern Archaeology, Department of Anthropology, University of Florida, Gainesville. Sassaman, Kenneth E., Paulette S. McFad den, Micah P. Monés, Andrea Palmiotto, and Asa R. Randall 2014 Northern Gulf Coastal Archaeology of the Here and Now. In New Histories of Precolumbian Florida , edited by Neill J. Wallis and Asa R. Randall. University Press of Florida, Gainesville. Sa varese, Michael, Ai Ning Loh, and John H. Trefry 2006 Appendix 12 5: Environmental and H ydrologic H istory of Estero Bay: Implications For W atershed M anagement and R estoration . South Florida Environmental Report, South Florida Water Management Distr ict. Scholl, David W., Frank C. Craighead, and Minze Stuiver 1969 Florida Submergence Curve Revised: Its Relation to Coastal Sedimentation Ra tes. Science 163:562 564. Schwadron, Margo 2008 Shell Work Landscapes and Emergent Complexity in the Ten Tho usand Islands, Florida . Paper presented at the 73 rd Annual meeting of the Society for American Archaeology, Vancouver, BC. Stapor, Frank W., Jr., Thomas D. Mathews, and Fonda E. Lindfors Kearns 1991 Barrier Island Progradation and Holocene Sea Lev el History in Southwest Florida. Journal of Coastal Research 7:815 838.

PAGE 206

206 Stapor, Frank W., Jr., and Gregory W. Stone 2004 A New Depositional Model for the Buried 4000 BP New Orleans Barrier: Implications for Sea Level Fluctuations and Onshore Tran sport from a Nearshore Shelf Source. Marine Geology 204:215 234. Thompson, Victor D., and John E. Worth 2011 Dwellers by the sea: Native American adaptations along the south ern coasts of eastern North America. Journal of Archaeological Research 19:51 101. Toscano, Marguerite A., and Ian G. Macintyre 2003 Corrected W estern Atlantic S ea L evel C urve for the last 11,000 Y ears B ased on C alibrated 14C C ates on Acropora Palmata F ramework and I ntertidal M angrove P eat. Coral Reefs , 22: 257 270. Tornqvist, Torbjorn E., Juan L Gonzalez, Lee A. Newsom, Klaas van der Borg, Arie F.M. de Jong and Charles W. Kurnik 2004 Deciphering Holocene Sea Level History on the U .S. Gulf Coast: A High Resolution Record from t he Mississippi Delta. Geological Society of America Bulletin 116 :1026 1039. Walker, Karen J., Frank W. Stapor, Jr., and William H. Marquardt 1995 Archaeological Evidence for a 1750 1450 BP Higher Than Present Sea Level Journal of Coastal Research Special Issue No. 17: Holocene cycles: climate, sea levels, and sedimentation : 205 218.

PAGE 207

207 Wright, Eric E. 1995 Sedimentation and Stratigraphy of the Suwannee River Marsh Coastline. Unp ublished Ph.D. dissertation, University of South Florida, St. Petersburg, FL. Wright, Eric E., Albert C. Hine, Steven L. Goodbred, Jr., and Stanley D. Locker 2005 The Effect of Sea Level and Climate Change on the Development of a Mixed Siliciclasti c Carbonate, Deltaic Coastline: Suwannee River, Florida, U.S.A. Journal of Sedimentary Research 75:621 635. Tallahassee.

PAGE 208

208 BIOGRAPHICAL SKETCH Paulette S izemore McFadden is a native of Charlotte, North Carolina. She earned her Bachelor of Arts with a double major in anthropology and religious studies from East Carolina University and her Master of Arts in anthropology from the same. She was accepted to the University of Florida in 2009 where she focused her research on the Gulf Coast of F lorida. Paulette earned her Doctor of Philosophy in anthropology in 2015, and currently holds a postdoctoral position with the Florida Museum of Natural History at the University of Florida.