REEVALUATING THE GULF OF MEXICO SEA-LEVEL

FLORIDA STATE UNIVERSITY COLLEGE OF ARTS AND SCIENCE

THE TROUBLE WITH THE CURVE: REEVALUATING THE GULF OF MEXICO SEA-LEVEL CURVE

By SHAWN JOY

A Thesis submitted to the Department of Anthropology in partial fulfillment of the requirements for the degree of Master of Science

2018

i

Shawn Joy defended this thesis on May 9, 2018. The members of the supervisory committee were:

Jessi Halligan Professor Directing Thesis

Tanya M. Peres Committee Member

Seth Young Committee Member

The Graduate School has verified and approved the above-named committee members and certifies that the thesis has been approved in accordance with university requirements.

ii

This thesis is dedicated to Sarah Sheehan for her support and ability to keep the laughs rolling throughout this whole process, and Michael Hubbard for teaching me that you don’t always have artifacts, but you’ll always have dirt.

iii

ACKNOWLEDGMENTS Dr. Jessi Halligan, Dr. Michael Faught, Dr. David Thulman, Dr. Jessica Cook-Hale, Dr. Seth Young, Dr. Tanya Peres, Sarah Sheehan, Morgan Smith, Brendan Fenerty, Dan Schoonover, Sarah Stanley, Kelly Ledford, Dr. Sandra Brooke, Dr. Amanda Oehlert, Dr. Ali Pourmand, Randall Funderburk, and many others who have helped me along the way...

iv

TABLE OF CONTENTS

List of Tables ................................................................................................................................ vii List of Figures .............................................................................................................................. viii Abstract ............................................................................................................................................x 1. INTRODUCTION.......................................................................................................................1 2. COASTALLY ADAPTED: A GLOBAL PERSPECTIVE ........................................................5 2.1 Coastally Adapted Africa .....................................................................................................7 2.2 Coastally Adapted Oceania ................................................................................................11 2.3 Coastally Adapted Mediterranean......................................................................................14 2.3.1 Klissoura Cave ........................................................................................................15 2.3.2 Franchthi Cave ........................................................................................................17 2.4 Coastally Adapted South America .....................................................................................21 2.4.1 Huaca Prieta ............................................................................................................22 2.4.2 Quebrada Jaguay .....................................................................................................27 2.4.3 Quebrada Tacahuay.................................................................................................30 2.4.4 Ring Site ..................................................................................................................31 2.4.5 Monte Verde............................................................................................................34 2.5 Coastally Adapted North America .....................................................................................36 3.5.1 Channel Islands .......................................................................................................36 2.6 Discussion ..........................................................................................................................40 3. REVIEW OF SUBMERGED GULF OF MEXICO ARCHAEOLOGICAL SITES................43 3.1 Ray Hole Springs (8TA171) ..............................................................................................45 3.2 Econfina Channel Site (8TA139).......................................................................................48 3.3 Fitch Site (8JE739) ............................................................................................................51 3.3.1 Stratigraphy .............................................................................................................52 3.4 J&J Hunt Site (8JE740) .....................................................................................................55 3.4.1 J&J Hunt, Locus L, and Area A, B, C ......................................................................55 3.4.2 Stratigraphy .............................................................................................................57 3.4.3 Site Occupation .......................................................................................................59 3.4.4 Bchron Analysis of J&J Hunt Radiocarbon Samples .............................................60 3.5 Discussion ..........................................................................................................................61 4. COASTAL GEOLOGY AND TAPHONOMIC PROCESSESS AT TRANSGRESSED SITES .............................................................................................................................................64 4.1 Idealized Stratigraphic Description of a Transgression Event ...........................................64 4.2 Stratigraphic Description of J&J Hunt Site ......................................................................65 4.3 Core Descriptions...............................................................................................................66

v

5. DOLOMITE FORMATION AND POTENTIAL AS A SEA-LEVEL PROXY .....................68 5.1 Dolomite ............................................................................................................................69 5.2 Models for Dolomite Formation at Fitch and J&J Hunt sites ............................................70 5.2.1 Hypersaline Model ..................................................................................................70 5.2.2 Mixing-Water Model ..............................................................................................71 5.2.3 Seawater Subtidal Model ........................................................................................71 5.2.4 Additional Models...................................................................................................72 5.3 Dolomite Formation at Fitch and J&J Hunt Sites ..............................................................72 5.4 Thin Section Sample Descriptions .....................................................................................74 5.5 Overview of U/Th Dating ..................................................................................................77 5.6 Dolomite Dating Methods..................................................................................................78 5.7 Results ................................................................................................................................79 6. THE TROUBLE WITH THE CURVE .....................................................................................80 6.1 Review of Balsillie and Donoghue (2004).........................................................................80 6.2 Radiocarbon Calibration Issues .........................................................................................84 6.3 Sampling Issues with Balsillie and Donoghue (2004) .......................................................88 6.4 New Curve Methods ..........................................................................................................89 6.5 Rating System ....................................................................................................................90 6.6 New Curve Results ............................................................................................................97 6.7 High Stands and Low Stands .............................................................................................99 6.8 Bayesian Analysis of Gulf of Mexico Transgression Event ............................................100 6.8.1 Bchron: Age-Depth Modeling...............................................................................100 6.8.2 Methods .................................................................................................................102 6.9 Results ..............................................................................................................................106 7. CONCLUSION .......................................................................................................................110 APPENDICES .............................................................................................................................114 A. BCHRON R CODE.................................................................................................................114 B. SEA-LEVEL CURVE DATASET .........................................................................................115 C. EDITED SEA-LEVEL DATASET.........................................................................................153 References Cited ..........................................................................................................................192 Biographical Sketch .....................................................................................................................206

vi

LIST OF TABLES

1

Sites Discussed in Chapter 2 with Range, Lithic Technology, and Subsistence .....................42

2

Faunal Remains at the Fitch Site (Faught 1990) ......................................................................54

3

Faught and Donoghue (1997) Radiocarbon Samples from J&J Hunt. Samples Were Calibrated Using OxCal 4.3. Wood Samples Were Calibrated Using Incal13 and Oyster Using Marine13 .......................................................................................................................58

4

Eastern Oyster Samples from Balsillie and Donohue Dataset .................................................93

5

Sample Rating System .............................................................................................................94

6

Quantiles of Predicted Ages by Depth ...................................................................................107

vii

LIST OF FIGURES

1

Relative Sea-level Over 550,000 years. Developed from (Lambeck 2002) ..............................2

2

Pinnacle Point Cave South Africa..............................................................................................8

3

Boodie Cave Australia .............................................................................................................11

4

Submerged Coastal Plain Profile Offshore of Boodie Cave ....................................................12

5

Franchthi and Klissoura Cave Sites in Southern Greece .........................................................14

6

Submerged Coastal Plain Profile Offshore of Franchthi Cave ................................................18

7

Peruvian Coastal Sites..............................................................................................................21

8

Submerged Coastal Plain Profile Offshore of Huaca Prieta ....................................................24

9

Southern Peruvian Coastal Sites ..............................................................................................27

10 Submerged Coastal Plain Profile Offshore of Quebrada Jaguay .............................................28 11 Submerged Coastal Plain Profile Offshore of Quebrada Tacahuay.........................................30 12 Submerged Coastal Plain Profile Offshore of Ring Site ..........................................................32 13 . Monte Verde in Southern Chile .............................................................................................34 14 Channel Islands Sites Off the Coast of California ...................................................................37 15 Overview of Apalachee Bay Submerged Sites, Florida (Faught 1996) ...................................44 16 Profile of Ray Hole Springs. Developed from (Anuskiewicz 1988) .......................................46 17 Overview of Submerged Sites in Apalachee Bay, Florida (Faught 1996) ...............................48 18 Site Map of Econfina Channel (Cook-Hale 2016) ...................................................................49 19 Profile of the Fitch Site (Faught 1990) ....................................................................................51 20 Site Map of J&J Hunt (Arbuthnot 2002) .................................................................................56 21 Bchron 95% Chronology of Radiocarbon Samples from J&J Hunt (Faught and Donoghue 1997) ........................................................................................................................................61

viii

22 Profile from Core 01-02B from J&J Hunt Site Apalachee Bay, Florida .................................66 23 Dolomite Sample 01 from J&J Hunt Site ................................................................................74 24 Dolomite Sample 02 from J&J Hunt Site ................................................................................74 25 Dolomite Sample 01 Under 10x and 20x Power Cross Nichols ..............................................75 26 Dolomite Sample 02 Under 2.5x Power Plain Polar................................................................76 27 Fairbanks (1990) Barbados Overlay Balsillie and Donoghue (2004) ......................................82 28 Bard (1990) Barbados Overlay Balsillie and Donoghue (2004) ..............................................82 29 Lambeck (2014) Eustatic Overlay Balsillie and Donoghue (2004) .........................................83 30 Toscano and Macintyre (2003) Caribbean Overlay Balsillie and Donoghue (2004) ..............83 31 Gulf of Mexico Sea-level Curve Developed from Balsillie and Donoghue (2004) ................87 32 Joy Gulf of Mexico Sea-level Curve with Environmental Indicators......................................96 33 Coastlines and Area Submerged by Millennia (with contributions from Cook-Hale) ............98 34 Bchron 95% Chronology for 290 Unedited Coral Samples...................................................108 35 Joy Gulf of Mexico Sea-level Curve with Bchron Overlay...................................................109

ix

ABSTRACT During the last glacial episode (130,000-11,500 years ago), nearly 5% of the Earth’s water was locked within ice sheets. This caused the lowering of global sea-levels to approximately 134 meters below modern levels. The reintroduction of freshwater into the oceans radically changed global sea-levels and littoral landscapes. Over the last 20,000 years, approximately 15-20 million km2 of coastal landscape has been submerged worldwide, roughly the area of South America. The inundation of these landscapes explains the relative rarity of global glacial period coastal archaeological sites, creating gaps in the history of coastal human activity around the world. Florida’s gently sloping continental shelf causes extreme coastline changes with minor vertical shifts in sea-levels. During the last glacial period, Florida’s landmass was nearly twothirds larger then at present, creating a substantial amount of exposed coastal plain for habitation. Understanding Paleoindians’ interactions with this coastal environment requires an accurate sealevel curve for the Gulf of Mexico. Balsillie and Donoghue’s (2004) sea-level curve has been used as an oceanic transgression model for over a decade. Yet, when compared to global sea-level models, Balsillie and Donoghue’s curve differs as much as 25 meters. This research addresses these issues and introduces new data and methodologies to enhance the Gulf of Mexico sea-level transgression model. A review of global coastally-adapted habitation sites is also conducted to create a model detailing mobility ranges and material culture to improve distribution modeling for locating submerged coastal archaeological sites. Finally, a geoarchaeological assessment was conducted on known submerged archaeological sites on Florida’s western continental shelf to better understand the taphonomic processes of transgressed sites.

x

CHAPTER 1 INTRODUCTION Human interaction with the coastal regions has played a long and pivotal role in the development of culture and modern human behavior (Jerardino and Marean 2010). The first evidence for modern human behavior in the world included modification of ocher, the utilization of coastal resources using lunar patterns and tides, and pyro-technology to heat-treat lithic material. These technological human advances can all be tied to 162,000 calendar years before present (cal BP) coastal environments located on the South African coast (Marean et al. 2007). Paleoenvironment processes and fluctuating sea-levels have dictated both global human coastal occupation and cultural adaptations (Karkanas and Goldberg 2010). Throughout human history, paleoclimates have oscillated between glacial and interglacial episodes, with a majority of time being weighed toward glacial environments (Compton 2011). During the most recent glacial maximum (24,000 cal BP), roughly 66.8x106 km3 (5%) of the Earth’s water was locked within ice sheets in both the Southern and Northern Hemispheres (Pirazzoli 1996). Lower sea-levels during glacial periods would have provided a substantial amount of exposed coastal plain for habitation. The focus of this research is to address the three important criteria for surveying the Gulf of Mexico’s submerged coastal plain for Pleistocene archaeological sites: cultural history, coastal geology, and sea-level rise. This research uses new cultural model and sea-level data to identify potential submerged landscape on Florida’s continental shelf for Paleoindian coastal habitation sites. Minor vertical shifts in sea-levels caused extreme coastline changes in areas with low sloping continental shelf relief. Florida’s western continental shelf has an extremely low gradient

1

Figure 1. Relative Sea-level Over 550,000 years. Developed from (Lambeck 2002).

that ranges from 0.2 to 4.0 meters per kilometer (m/km) (0.67–1.25 ft/mi) overall but is steeper (6–9 m/km) (20–30 ft/mi) near the present-day coastline (Davis 2017). The shelf extends more than 500 km (300 mi) north to south and, at its maximum width, it is 240 km (149 mi) (Hine and Locker 2011). As the last glacial episode ended around 21,000 cal BP, the reintroduction of freshwater into the oceans radically changed the global sea-levels and littoral landscapes. Between the last glacial maximum and modern-day sea-levels, approximately 15-20 million km2 of coastal landscape has been submerged worldwide, roughly the size of South America (Faure et al. 2002). Between 21,000 and 6,000 cal BP, nearly 250,000 km2 was submerged, the equivalent of an area twice the size of the state of Mississippi. Florida’s human occupation extends back to at least 14,550 cal BP (Halligan 2012; Halligan et al. 2016), when the state was nearly double its land mass and the Gulf of Mexico was hundreds of kilometers from the earliest upland sites. As global climates shifted from glacial to interglacial, Florida experienced several periods of oscillating microclimate shifts from wet to

2

arid climates (Dunbar 2007; Faure 2002; Grimm 2006). These arid conditions severely reduced the amount of freshwater within Florida’s aquifer system (Thulman 2009). The scarcity of water forced Florida’s Rancholabrean fauna, including mastodons (M. americanum), bison (B. antiquus), camel (Camelops sp) and horses (Equus sp.) into a condensed landscape. The lowering of the aquifer system limited the expulsion of freshwater to deep inland sinkholes and springs on the continental shelf (Faure 2002). It is reasonable to assume that Paleoindian (>14,550-11,500 cal BP) and Archaic (11,5005,000 cal BP) peoples used the coastal regions for its increased sources of freshwater (Thulman 2009) and its ample resources, including shellfish and marine mammals (Faure et al. 2002). This assumption leads to questions such as: Were Paleoindian and Archaic peoples permanently living in the coastal landscape? If so, were they coastally adapted? Were their annual home and daily foraging ranges comparable to upland peoples? To what extent was this landscape utilized by people during fluctuations in the paleoclimate? How were peoples affected during an environmental period that significantly altered the entire biosphere, including the extinction of 37 species of megafauna, many of which were utilized by human for subsistence (Meltzer 2015; Waters et al. 2015)? The answers to these questions remain elusive due to the relatively small number of identified coastal Paleoindian and Archaic sites. Research into coastal Paleoindian sites requires the identification of paleo-coastlines. As sea-levels transgressed the landscape, human occupation areas would have had to move accordingly with the ever-changing littoral landscape. Identifying the probable locations for coastal Paleoindian regions requires distinguishing where the coastline was throughout time as it transgressed over the continental shelf. Ethnographic studies of hunter-gatherer mobility define

3

ranges in which the groups will forage for food, and limited Pleistocene archaeological evidence suggests that coastal resources are heavily utilized whenever possible. Chapter 2 reviews coastally-adapted sites from around the globe from the Late and Terminal Pleistocene to develop a model for cultural material and expectations for where and how these sites may have been distributed. Chapter 3 and 4 review submerged sites on Florida’s continental shelf to understand the coastal geological processes that take place at an archaeological site during sea-level transgression. Chapter 5 introduces the potential for modern dolomite to be used as a sea-level proxy. Chapter 6 discusses the issues with the Balsillie and Donoghue’s (2004) Gulf of Mexico sea-level curve and presents a new curve that was created using more stringent editing methods and statistical modeling. Chapter 7 presents avenues for future research and a summary of models developed throughout this research to assist in surveys for submerged coastal Paleoindian sites.

4

CHAPTER 2 COASTALLY-ADAPTED: A GLOBAL PERSPECTIVE Predicting the cultural material typology of Southeastern coastal Paleoindians is a challenge. Only a handful of unequivocal coastal Paleoindian sites have been identified in the Americas (Erlandson and Jew 2009; Dillehay et al. 2008). These sites are located on the west coast of the Americas where sea-level rise was less dramatically invasive due to the steep coastal topography (Richardson 1998). Creating a model for Paleoindian coastal habitation and subsistence patterns involves examining global coastal sites throughout modern human history with emphasis on time periods with higher to modern sea-level conditions. In this chapter, the cultural material records of coastal sites are compared throughout the Pleistocene from around the globe. Utilizing anthropological uniformitarianism, Pleistocene hunter-gatherer sites with coastal materials will be evaluated from regions including South Africa (Marean 2010), Northern Australia (Veth et al. 2007, 2017; Williams et al. 2018), South America (Dillehay et al. 2008; Dillehay et al. 2017; Sandweiss et al. 1998), Greece (Starkovich, Munro, and Stiner 2017), and the Californian Channel Islands (Erlandson and Jew 2009). These regions contain evidence of coastal subsistence patterns spanning 150,000 years in coastal settings similar to Florida’s oscillating climate, submerged coastal plain profile, and potential subsistence resources. Pleistocene sites are exclusively reviewed due to the similarity in fluctuating climatic and sealevel conditions and potential cultural similarities in the western hemisphere to Paleoindians in the Americas. Investigating sites of such antiquity and varying global locations may rule out patterns within the model that could be strictly confined to a single cultural group. The comparison will specifically address the changes in subsistence patterns, coastal mobility ranges,

5

and lithic tool technology through time as the proximity to the sea changed due to fluctuating sea-levels. Before searching for Terminal Pleistocene (20,000-11,500 cal BP) coastally-adapted sites in the Gulf of Mexico, we must define what it means to be coastally-adapted. The definition of coastal within this chapter will strictly refer to the immediate region in which the land meets the sea. This region will include the 10 to 12 km range in which hunter-gatherers may travel in a day to pursue subsistence resources (Binford 1980; Fisher et al. 2010). This review does not include freshwater rivers or lakes, or areas where sites may have been accessible to the coastline via maritime navigation up rivers or streams. This chapter specifically explores the changes in material culture as coastlines fluctuate in proximity to static archaeological sites throughout the Late Pleistocene. It also presents a review of global coastal sites where hunter-gatherer groups have become systematically adapted to coastal subsistence strategy. Coastally-adapted is defined as societies in which:” …coastal foods are so important that the mobility system is designed to intercept the coast as a planned part of the annual mobility strategy, sometimes moving between the interior and the coast, or even staying there all year. Coastal adaptations have a substantial portion of the diet derived from animals that live along the coastline in and about the inter-tidal zone, where the coast is the zone where sea and land processes intermix” (Marean 2014). The definition should extend further to state that coastallyadapted peoples will 1) solely rely on coastal resources when on the coast, and 2) modify their tool kit to exclude projectile points utilized in terrestrial hunting, while 3) employing utilized flake and micro-blade technology as their main use of lithic material. The difference between coastally-adapted and maritime-adapted must also be defined. Being maritime adapted includes the use of boats and the subsistence strategy may include littoral and benthic species, as well as

6

the tool kit required to hunt in such environments (Marean 2014; Smith 1956). Sites discussed in the Mediterranean will include a shift in subsistence pattern as groups transition from coastally to maritime-adapted during the Terminal Pleistocene, while the California Channel Island sites contain evidence of people that have always been maritime-adapted. The term “coastally-adapted” was first used in the 1970s (Snow 1972). Coastally-adapted was referred to as spending a significant portion of time utilizing the coastline for subsistence economy, collecting shellfish, and processing them at nearby midden sites (Snow 1972). The definition has expanded to include other littoral resources such as sea birds, ocean mammals, and fish (Dillehay et al. 2017). Peoples inhabiting coastal sites throughout the world appear to specialize in unique forms of subsistence strategies in accordance to local and regional littoral resources. Some groups may specialize in fishing for nearshore species while utilizing nets, while others may hunt sea mammals and birds by clubbing them, and still others use compound fishing hooks (Rick et al. 2005; Sandweiss 2008). In all cases, the substance pattern is customized to the local environmental setting. Despite the vast variation in methods, these groups have made a concerted decision to leave the upland terrestrial setting in exchange for a lifestyle that strictly utilizes coastlines for subsistence. 2.1 Coastally Adapted Africa This review begins at the dawn of modern human behavior (ca. 150,000 years ago). Behavioral modernity is categorized by abstract thinking, the use of complex chain technology (recipes), symbolic behavior (art and ornamentation), and lithic blade technology (Henshilwood et al. 2003). Modern human behaviors and technologies are the results of cumulative cultural adaptations being passed on and built upon generation after generation (Hill et al. 2009). Coastal sites, such as Pinnacle Point cave, dated to 162,000 cal BP, include the earliest evidence for

7

modern human behavior in the world (Marean 2010). This evidence includes the earliest artistic use of ocher, the earliest use of coastal resources that required knowledge of lunar patterns and low tides, and the earliest use of heat-treating lithic material (Marean et al. 2007). This places the dawn of modern human behavior to marine isotope stage six (MIS 6) (195 to 123,000 BP), 100,000 years prior to the Eurasian “Human Revolution” (Marean 2010). Researchers believe that coastal diets rich in omega-3 fatty acids may have afforded early humans the subsistence to mentally develop modern human behavior (Marean 2014). Pinnacle Point Caves were carved into highly folded and faulted exposures of the Skurweberg Formation of the Paleozoic Table Mountain Sandstone Group, which is comprised of coarse-grained, light-gray quartzite sandstone (Fisher et al. 2010). The caves are located on a

Figure 2. Pinnacle Point Cave South Africa.

8

craggy stretch of South African coastline overlooking the Indian ocean and were eroded by wave action during sea-level high stands 430,000 cal BP (Marean et al. 2007; Karkanas and Goldberg 2010). The southern and western shores of South Africa are known as the Cape Floral Region, and have been described as a “Floral Kingdom” with one of the highest ranges of biodiversity on the planet (Marean 2010) (see Figure 2). The area is rich in tubers, shellfish, and the highest diversity of marine fish. These environmental conditions played a vital role in the survival and the development of modern humans (Marean 2010). Changes in sea-level during the Pleistocene impacted the local environment, including human inhabitants, which, in turn, altered the archaeological record (Fisher et al. 2010). A gently sloping continental shelf (the Agulhas Bank) grades southward from Pinnacle Point at a shallow 10 degrees. Lower sea-levels during glacial periods would have provided a substantial amount of newly exposed coastal plain for habitation, transforming the floral and faunal resources in the area. For much of the MIS 6 period, sea-levels were lower and the coastline was as much as 100 km from Pinnacle Point (Marean et al. 2010). Local populations would have utilized the exposed areas and the terrestrial floral and faunal inhabiting the exposed shelf (Fisher et al. 2010). The sea-level model shows a transgression at approximately 167,000 cal BP, placing the coastline within 10 km of the site. At this time, humans began inhabiting the site and the first evidence of coastal resources use is identified in anthropogenic deposits (Marean et al. 2007). Coastal resources at Pinnacle Point reflect a shellfish gathering and coastal scavenging strategy. Faunal remains such as fishbone and barnacles that strictly adhere to whales have been identified at the site. These resources were likely scavenged from the shoreline and returned to the cave for processing and consumption (Marean 2010; Jerardino and Marean 2010). The most prominent resource at the site is shellfish. Shellfish has been identified in four stratigraphic units. The

9

stratigraphic units dating to 174–153,000; 130–120,000; 133–115,000; and 98–91,000 cal BP contain marine resources from when the coastline was within 10 km of the site. Archaeological deposits at Pinnacle Point span between 174–91,000 cal BP, after which the cave is closed off by a sand dune that formed against the mouth of the cave (Fisher et al. 2010; Marean et al. 2007, 2010). The lithic material recovered from Pinnacle Point contains evidence of the first instances of heat-treatment and micro-blade technology (Marean et al. 2007). The local material, silcrete, is a poor lithic resource until it is heat-treated. Multiple lenses of hearth features were identified littered with heat-treated silcrete, which was used as raw lithic material for tool production. Micro-blade flakes produced from the silcrete were then inlaid and adhered to a wooden shaft. This is also the first evidence of using complex recipes to produce the adhesive (Marean et al. 2007). Ethnographic studies of hunter-gatherer mobility define ranges in which the groups will forage for food. Hunter-gatherers subsiding in dry/warm environments typically do not forage long distances from their residential site. The daily range of a hunter-gatherer group is 8-10 km, the distance a person can walk out and back in one day (Binford 1980, 1982; Fisher et al. 2010). The same pattern is identified in the archaeological material at Pinnacle Point as it pertains to coastal subsistence. As the ocean levels transgressed and regressed further and closer to the cave, the overall site habitation changed. When the sea-levels were within 5 km of the cave, evidence of coastal subsistence in the form of shellfish increases. As the sea-levels regressed 10 km or more from the site, shellfish remains began to drop off and lithic material become more prevalent. When the shoreline regressed further than 20 km, the cave was abandoned (Marean 2010).

10

2.2 Coastally Adapted Oceania The Oceania region contains evidence of coastal adaptation with great antiquity. Evidence of human occupation at Boodie Cave on Barrow Island extends to 51,000 cal BP. The earliest occupations contain faunal remains indicating a reliance on coastal resources by 42,500 cal BP (Veth et al. 2017). Barrow Island is in the arid northwest karst region of Australia, known

Figure 3. Boodie Cave Australia. as the Montebello Islands. The islands are part of the Trealla Limestone formation (Veth et al. 2017). The island is situated at the edge of Australia’s continental shelf and represents the last vestiges of land to be submerged since the last glacial maximum (LGM) (see Figure 3). The northern coast of Australia underwent the submergence of 2.12 million km2 of landscape since the LGM (Williams et al. 2018). Human occupation of the site began when the island was still part of mainland Australia, though as sea-level increased, the landmass became disconnected 11

from the mainland by 7,000 cal BP (Veth et al. 2007). The sites were abandoned shortly after the landmass was disconnected from the mainland in 6,800 cal BP (Veth et al. 2017) (see Figure 4). The earliest evidence of coastal fauna identified at Boodie Cave is a gastropod (Nerita sp.) species of mollusk that directly dates to 42,600-40,200 cal BP (Veth et al. 2017). A total number of four taxa were identified at the site during this period including Terebralia, Tellina and Nerita. These species have a high meat to shell ratio and were likely transported to the site for consumption. The fourth mollusk type was a Melos species ethnographically known to be used as ornaments and water containers (Veth et al. 2017). Transportation efficiency to the site

Figure 4. Submerged Coastal Plain Profile Offshore of Boodie Cave. was paramount for this period, as the coastline was over 24 km away, and the environmental conditions were extremely arid. The aridity of the local environment may have made Boodie Cave an attractive site at which to camp to avoid the exposure to the harsh elements, making the

12

distance traveled from the coast worth the effort (Veth et al. 2017). The aboriginals that inhabited the cave when the coastline was over 24 km away (Lambeck et al. 2002) relied on a terrestrial subsistence strategy including wallaby (Lagorchestes conspicillatus) and euro (Macropus robustus). As the coastline grew closer in proximity, the subsistence remains changed to include 40 molluscan species, but marine resources were still offset by terrestrial species. By 8,000 cal BP, sea-levels had increased to 13 m bmsl and the coastline was within 3 km of the cave. The subsistence strategy changed to fully marine fauna including: shark (Selachimorpha sp), trigger fish (Balistidae), parrot fish (Scaridae), wrasse (Labridae), hard-shelled sea turtle (Cheloniidae), dolphin (Cetacea), mud crab (Scylla serrata), and over 40 species of marine mollusks (Veth et al. 2017). The lithic assemblage consisted of 6,002 pieces of flakes and stone tools. A majority of the lithic material consisted of local limestone and calcrete with 113 non-local materials including quartz and chert (Veth et al. 2017). Only 10% of the local material show evidence of retouch or edge damage, where 40% of the non-local material has been retouched and/or utilized. Shells were also modified as scrapers and ornamental beads. A Dentalium bead recovered at the site was directly radiocarbon dated to 12,000 cal BP, indicating the importance of shellfish at the site for subsistence, tools, and ornaments (Veth et al. 2017). The lack of projectile points or complex lithic tools reflects the stone tool assemblage of coastally-adapted communities. Boodie Cave represents the oldest evidence for the utilization of coastal resources in Australia (Veth et al. 2017). The coastal material culture was transported over 20 km from the coastline to the site, indicating the importance of coastal resources in the arid environment. The site fits into the emerging model for coastally-adapted sites such as those discussed in Africa and other global sites to be examined. Undoubtedly contemporary and, potentially, older sites

13

containing evidence for coastal adaptions are now drowned on the north Australian continental shelf. 2.3 Coastally Adapted Mediterranean Hunter-gather groups within the Mediterranean region underwent dramatic cultural and technological changes during the LGM (Starkovich et al. 2017). Subsistence patterns reflect the adaptation to the changing environments as sea-levels increased, prey species went extinct, and hunting technologies advanced. The effects of rising sea-levels on subsistence patterns were twofold. The submergence of coastal plains reduced the terrestrial landscape and biosphere, while increasing the availability of littoral species. The changes in the material culture at the Franchthi

Figure 5. Franchthi and Klissoura Cave Sites in Southern Greece. and Klissoura sites in the Argolid region of Peloponnese in Southern Greece allow us to compare

14

and contrast adaptations from upland to coastal sites throughout the Terminal Pleistocene, into the Holocene. The Franchthi and Klissoura Cave sites are within a 50 km range of one another with varying distances from the modern Mediterranean coastline (see Figure 5). The Franchthi Cave site is the closest to the coastline throughout the fluctuations in sea-levels. Currently, the site is roughly 11 m from the modern coastline and was no more than 7 km from the sea during low stands (Lambeck et al. 2014; Starkovich et al. 2017). Klissoura would have been an upland environment for early people as it is 12 km from the modern coastline, yet it would have been over 30 km from the sea during the LGM. The two sites contain substantial archaeological deposits spanning 39,000 to 8,500 cal BP (Perlès 2016; Starkovich et al. 2017; Stiner et al. 2012). Examining the faunal remains reveals an evolution in subsistence patterns from terrestrial prey, to an increase in shellfish procurement, then, finally, to pelagic species with the advent of maritime fishing. The changes in subsistence strategies coevolve as sea-levels and the proximity of the sea increase. These patterns allow us to define the coastal mobility ranges and compare subsistence patterns of upland and coastal Mediterranean Paleolithic peoples. 2.3.1 Klissoura Cave Klissoura Cave is a relatively shallow limestone cave located on the banks of the Berbatias River. The site contains 5 m of cultural stratigraphy spanning the Late Pleistocene though early Holocene (39,000-8,500 cal BP). An extensive number of hearth features have been identified spanning the upper Paleolithic (39,000-10,000 cal BP), with the highest concentrations from 39,000-17,000 cal BP. The ample features include clay-lined hearths, an abundance of lithics, fauna, and non-marine shell ornaments. These cultural materials indicate that the location served as a residential site for a majority of its occupation (Stiner et al. 2012). After the LGM,

15

the occupation of the site decreased dramatically until its abandonment by 8,500 cal BP (Starkovich et al. 2017). Throughout the occupation of the site, a large variety of species were utilized for subsistence including European fallow deer (Dama dama), ibex (Capra cf. ibex), European wild ass (Equus hydruntinus), aurochs (Bos primigenius), red deer (Cervus elaphus),wild pig (Sus scrofa), European roe deer (Capreolus capreolus) and chamois (Rupicapra rupicapra) (Stiner et al. 2012). At the LGM, open woodlands were replaced with open grassland environments associated with a drier climate (Starkovich et al. 2017). The subsistence patterns at the site reflect the environmental change as remains of small game animals increased including: European hare (Lepus europaeus), rock partridge, (Alectoris graeca), great bustard (Otis tarda); and spur-thighed tortoises (Testudo sp.) (Bochenski and Tomek 2010). Land snails (Helix figulina) became an important subsistence species during the later occupation period (Stiner et al. 2012). The overall subsistence pattern at Klissoura Cave included a heavy reliance on large ungulates, especially fallow deer, during early occupations (39,000 to 24,000 cal BP). As the climate became drier, the remains of smaller, faster species were more prevalent at the site. The increased utilization of smaller game was accompanied with the intensification of land snail harvesting from 24,000 cal BP to the abandonment of the site at 8,500 cal BP (Stiner et al. 2012) The lack of evidence of marine shellfish or fish species used for subsistence or ornaments at the cave indicates that the site was strictly utilized as an upland residential site. This suggests little to no mobility of coastal resources to the area. Klissoura Cave ranged between 30 and 15 km from the coastline throughout the Late Pleistocene to Holocene. Sea-levels at the time of abandonment of the site (8,500 cal BP) had yet to reach modern levels (Lambeck et al. 2014). Paleolithic

16

peoples within the region were not transporting marine resources to the site, indicating possible low ranges of mobility or a complete shift in subsistence strategies from one site to another, as nearby Franchthi Cave displays a very different pattern. 2.3.2 Franchthi Cave Franchthi Cave is a limestone cave currently situated 11 m from the Mediterranean Sea in southern Greece. The site is roughly 50 km south of the Klissoura Gorge sites (see Figure 5). The cave contains one of the most intensely occupied Late Pleistocene-Early Holocene sites (39,000-8,000 cal BP) in southern Europe (Stiner et al. 2012). Early occupation patterns fluctuate between minor utilization, punctuated by periods of abandonment. The oldest stratigraphic unit at Franchthi is a volcanic tephra stratum dating to 39,000 cal BP (Stiner et al. 2012). Lithic materials identified within this stratum are sparse, consisting of non-diagnostic material. An increase in utilization of the site is identified between 36,650 and 35,300 cal BP (Starkovich et al. 2017) representing short occupations of small groups (Perlès 1999; Stiner et al. 2012). Human activity at the site fluctuates from 35,300 to 15,000 cal BP, with long occupational hiatuses lasting thousands of years, particularly between 20,000 to 15,000 cal BP. Lithic material recovered during this period consists of single and double-backed points and bladelets along with end scrapers (Stiner et al. 2012). During this period, the environment surrounding the cave was generally steppe vegetation associated with a cool dry climate (Stiner et al. 2012). Occupation at Franchthi Cave intensified from 15,000 to 8,000 cal BP as the cave developed into a significant, centralized residential site. This period is marked with an expansion of hearths, middens, and cemetery features. The lithic assemblage shifts from an industry reliant on single and double-backed points, to a reliance on microlithic technology (Perlès 1999; Stiner et al. 2012). Exotic lithic material, including obsidian from the island of Melos, appear after

17

Figure 6. Submerged Coastal Plain Profile Offshore of Franchthi Cave. 11,000 cal BP, indicating the Mesolithic peoples at Franchthi had developed a maritime industry (Stiner et al. 2012). Subsistence and occupational patterns at Franchthi Cave fluctuate with the changing environment which directly correlate to changes in sea-levels. The earliest subsistence patterns during the sporadic habitation period (39,000 to 15,000 cal BP) mimic those of Klissoura cave. Red deer and European wild ass were especially prevalent, with sporadic evidence of aurochs, wild pig, and wild goat (Stiner et al. 2012). During this period, the coastline would have been roughly 7 km from the site (see Figure 6). After the LGM, subsistence patterns dramatically shift from exclusively terrestrial to a mixed of marine and terrestrial species. The utilization of land snail increased by 368% during a thousand-year period spanning between 15,000 and

18

14,000 cal BP. This period also included the first evidence of marine shellfish processing at the site. Marine shellfish are in the highest abundance during the initial use of littoral resources (Stiner et al. 2012). From 13,300 to 12,700 cal BP, there is a reduction in small terrestrial game, supplemented by an increase of small nearshore fish remains of gilthead (Sparus aurata). A shift in lithic technology occurred as a marked change in marine resources took place. Small nearshore species are replaced by larger pelagic species such as tunny (Thunnus thynnys) and barracuda (Sphyraena sphyraena). Evidence of both exotic lithic material from Melos, and pelagic fish species strongly indicate that hunter-gather groups at Franchthi Cave adapted to maritime based industry by 11,000 cal BP. The sudden emergence of resources that required boats for procurement may not represent the sudden development of maritime technology. The appearance of these materials coincides with MWP 1B, during which sea-levels increased by 20 m in a 500 year period (Fairbanks 1989; Joy Chapt. 6; Lambeck et al. 2014). Sea-level rise may have submerged older maritime sites that were closer to the shoreline and more convenient for storage of boats prior to the MWP. After the MWP, the shoreline was within 3 km of Franchthi cave, making the location more attractive for maritime culture. Unlike the Klissoura site, Franchthi Cave continued to be utilized throughout the Holocene. Stiner et al. (2012) interpret the change in subsistence economy and the intensified occupation of the Franchthi site as a reaction to the local availability of food resources due to increased pressures on food supplies. Simply put, as the available menu changed locally at the site, so did the local diets. I propose an additional interpretation involving shift in demographics coupled with stressors on food resources. The highest occupation intensity at Klissoura site falls between 39,000 and 17,000 cal BP, while the occupation at Franchthi during this period is relatively low. The lower occupation

19

patterns by upland hunter-gather groups at Franchthi Cave imply that the site was a relatively unimportant location from 39,000-15,000 cal BP. Franchthi Cave’s location may have been situated in a “middle ground” between two cultural groups during times of expanded landscapes due to lower sea-levels. Culturally-divided landscapes in northern Florida have also been proposed by Thulman (2011). As the coastal plain and littoral landscape were submerged, littoral hunter-gather groups would progressively be pushing inland by the encroaching sea. Coastal landscapes would have experienced an accelerated rate of submergence during MWP 1A, lasting from roughly from 15,000 to 13,500 cal BP (see Chapt. 6; Lambeck et al. 2014). This correlates to the rise in occupation intensity at Franchthi Cave (Stiner et al. 2012). The intensification of inhabitants, shift in lithic technology, and evolution in resource economy may reflect a shift in demographics at Franchthi cave. The cave may also have become more attractive as a habitation site as the proximity to the sea increased, making it a prime location for peoples adapted to a coastal economy. The expansion of the site to a residential hub may have been two-fold. As rising sea-levels forced groups inland toward the Franchthi site, a clear reduction in upland food resources (Starkovich et al. 2017) may have enticed upland groups at Klissoura Cave to migrate south toward the resource rich coast. This may have led to a decrease in utilization of Klissoura Cave after the LGM, to the complete abandonment of Klissoura Cave by 8,500 cal BP, and the increase population at Franchthi Cave. Changes in lithic technologies and subsistence patterns at both Klissoura and Franchthi Caves provide insight into ranges in mobility and adaptations to coastal environments. Klissoura Cave was strictly an upland, terrestrial-species-reliant site. Throughout its occupation, the distance to the sea ranged between 30 to 15 km. The site lacks any evidence of marine ornaments or resources, indicating its occupants likely did not make forays to the coastline or trade with

20

peoples who did (Starkovich et al. 2017; Perlès 2016; Stiner et al. 2012). The evidence at Franchthi Cave demonstrates a transition from a terrestrial to a marine economy as sea-levels increased. Periods of sporadic usage by upland hunter-gathers were punctuated by a sudden shift in both subsistence patterns and lithic technologies. The shift occurs distinctly at 15,000 cal BP, when sea-levels begin to rapidly increase. Sea-levels were approximately 100 m lower at this time, with the shoreline located roughly 6 km from the Franchthi site (Lambeck et al. 2014). This indicated that the potential range for coastally-adapted hunter-gather groups in southern Greece should fall between 10 and 6 km from the coastline. 2.4 Coastally Adapted South America There are several coastally adapted sites on the west coast of South America, especially on the coast of Peru (see Figure 7). Coastally adapted sites in Peru extend back to the dawn of

Figure 7. Peruvian Coastal Sites

21

the peopling of the Americas (Dillehay et al. 2017). The elevated number of coastal sites is due to Richardson’s Law, which states that coastal sites with great antiquity should be located in regions with steep coastal relief (Richardson 1998). This section reviews several Paleoindian coastal sites identified on South America’s steep western coast. 2.4.1 Huaca Prieta Huaca Prieta is an open-air site located on the Pacific coast in northern Peru. Evidence for human occupation of the site has been identified in stratigraphically-secure contexts dating to the Late Pleistocene (15,000 cal BP). Occupations at the site represent short periods of habitation likely characterized as camp sites from 15,000 to 8,000 cal BP. Faunal remains identified within cultural features indicate a heavy reliance on coastal resources including marine mammals, fish, and shark (Dillehay et al. 2017). An extensive lithic tool and faunal assemblage has been recovered from the site spanning the breath of the Paleoindian period (15,000-10,000 cal BP), allowing for a comprehensive analysis of changes in resource economy as the proximity to the sea increased. The Huaca Prieta site is located on the Chicama River alluvial fan situated on the tip of a geological uplift called the Sangamon Terrace. The terrace sits just above the present-day Pacific shoreline and consists of boulder to cobble sized oligomictic conglomerates with a sandy matrix and a calcite cement (Dillehay et al. 2017). The terrace is roughly 2 km long and 1 km wide at its northern end, where the site is located. The surface of the terrace is 13.9 m above the modern sea level and would have been a notable landmark along the coastal plain. Huaca Prieta is under a large Holocene-aged mound complex measuring approximately 30 m high, 65 m wide, and 165 m long (Dillehay et al. 2017). The mound has protected the Late Pleistocene site beneath it from centuries of human mound construction activities and fair weather oceanic disturbances.

22

The site is approximately 100 m from the modern coastline but during initial occupation (15,000 cal BP), sea-levels were 100 m lower than modern levels. The coastline would have been located roughly 30 km from the site (Dillehay et al. 2017; Lambeck et al. 2014). By the end of the Paleoindian period (10,000 cal BP), sea-levels had risen to just over 30 m below modern sealevels, placing the encroaching coastline between 8 and 12 km from the site. The varying topography on the continental shelf would have created environments that consisted of both littoral and wetland setting including vernal pools, brackish inlets, and barrier island back waters (Dillehay et al. 2017) (see Figure 8). Faunal and lithic remains indicate that resources were gathered from both distant shoreline and forests within the mountains and transported tens of kilometers to terrace. Huaca Prieta is located within the shortest coastal plain separating the ocean from the mountains, which are connected by the Chicama River valley (Dillehay et al. 2017). The central location of the site within the river valley expresses the importance of both mountain and the sea environments, and the range of mobility for early coastal hunter-gatherers. Excavation at Huaca Prieta resulted in a plethora of faunal material and over 1 million lithic artifacts recovered at the site (Bird et al. 1985; Dillehay et al. 2017). The preservation at the site was excellent due to the dry desert climate. The stratigraphic units consist of fine to coarse sands that have undergone varying levels of pedogenesis with inclusions of charcoal flecks, ash, shell, bone, rush stems, lithic tools, and other cultural materials (Dillehay et al. 2017). Extensive faunal remains were recovered from the occupations spanning 15,000 to 10,000 cal BP, including 281 specimens of marine fauna. Throughout the earliest layers at the site, evidence suggests a heavy reliance on nearshore and vernal pool wetland species. Shallow,

23

Figure 8. Submerged Coastal Plain Profile Offshore of Huaca Prieta. warm water requiem sharks (Carcharhinus sp.) were extensively hunted in the back-water pools. Sharks and rays represent the highest utilized littoral species at the site (27.55% of taxa at the site). Sea lions (15.85%) and marine birds (13.96%) were also hunted on the beaches. Marine bony fish (12.83%) such as drum (Sciaena deliciosa) were also collected from the back-waters. Shellfish were not as extensively collected as marine mammals and shark taxa. Remains of limpets (9.43%), marine gastropods (7.55%), and marine bivalves (8.30%) were identified at the site, yet they were only observed in the earliest stratum. Remains of terrestrial species were scarce at Huaca Prieta, with only three specimens of white tail deer (O. virginianus) identified at the site (Dillehay et al. 2017). Lack of evidence for fish hooks or harpoons indicate that the marine species were captured through netting or trapping. Woven rush fabric was recovered in most units at the site that may have been used in the construction of netting or traps (Dillehay et al. 2017). 24

Floral remains recovered at the site consist of a variety of upland species including chile pepper (Capsicum spp.), squash (Cucurbita sp.), bean (Phaseolus sp.), and avocado (Persea sp.). The presence of these species either indicates an extensive trade network with upland peoples or that people were very mobile, moving up and down the river basin. A substantial lithic assemblage was recovered from Huaca Prieta, including over one million expedient tools and debitage spanning the roughly 15,000 years of continued habitation. Seventy-one lithic artifacts were recovered in the strata that span the greatest changes in sealevels (15,000 to 10,000 cal BP). All lithic material was produced from local sources of rhyolite, basalt, andesite, and quartzite cobbles that were eroded from the Sangamon Terrace conglomerate. Two exotic flakes of silicate skarn, sourced from the mountain region to the east were also identified (Dillehay et al. 2017). Unequivocal lithic materials include 38 flakes, 5 spalls, 2 split cobbles, a core fragment, a chopper on a cobble, and two denticulates on flakes produced by direct percussion. A perforated pebble was also recovered at the site despite the lack of intricate tools such as drills or points. All lithic tools identified throughout the Paleoindian period at the site consisted of expedient unifacially-flaked tools. There is no evidence of bifacial flaking from the earliest assemblage (15,000 to 13,500 cal yr BP). Tools are comprised of large sharp primary flakes with little retouching and evidence of use wear, including striations and polishing (Dillehay et al. 2017). The assemblage recovered from the 13,500 to 11,500 cal BP strata contained smaller unifacial tools with a variety of cutting, wedging, scraping, and pounding tools (Dillehay et al. 2017). The youngest assemblage (11,500 to 10,000 cal BP) contained smaller unifacial flakes that were utilized as scrapers. The majority of tools (90%) showed macroscopic indications of use wear.

25

The lithic technology at the site changed little over thousands of years and encroaching sea-levels. Tools consisted of expedient, unifacial flakes associated with ephemeral occupation. There is a distinct lack of bifacial projectile points or complex tools such as Fishtail and Paijan types (13,000 to 10,000 cal B.P.) despite both being identified in the foothills east of Huaca Prieta (Dillehay and Kaulicke 2011). The extensive reliance on marine resources and local lithic material, coupled with a departure from complex lithic technologies, may reflect separate cultural groups between the coastal plains and the interior mountains. Interregional trade is supported by evidence of upland resources recovered at Huaca Prieta (Dillehay et al. 2017). The littoral cultural patterns at Huaca Prieta appear to differ from the Old-World sites. The proximity to the coastline appears to have had little effect on the utilization of marine resources. A pattern of increased utilization of marine resources does occur from the Late Pleistocene into the Early Holocene, when coastlines were within 8 to 12 km. However, the most substantial procurement of shark dates to the earliest strata when the coastline was over 30 km from the site (Dillehay et al. 2017). Regardless of the ocean’s proximity, Paleoindian peoples on the Peruvian coastal plains were transporting marine resources inland to a significant landmark. Despite the ephemeral nature of the occupations, coastal peoples were traveling extended distances from the coastline to process marine resources at the site. The relatively young age of the site restricts the long-term analysis of cultural adaptations during periods of encroaching sealevels as compared to Old-World sites, but the occupants appear to have been coastally adapted since their first use of the site. The mobility range of coastally-adapted Late Pleistocene/Early Holocene peoples in northern Peru extends to upward of 30 km (Dillehay et al. 2017; Lambeck et al. 2014).

26

2.4.2 Quebrada Jaguay Quebrada Jaguay site is an open-air site located on the southern coast of Peru. Occupation of the coastal site dates to 12,948 ±241 cal BP, when sea-levels were 70 m lower than modern levels (Lambeck et al. 2014; Sandweiss et al. 1998). Due to the steep terrain and tectonic uplift in the region, the site has always remained relatively close to the shoreline. Several postholes arranged in a 5 m circular pattern indicate that a semi-subterranean residential structure was erected at the site dating to the Terminal Pleistocene (Sandweiss et al. 1998). The presence of the structure suggested the site may have been a longer-term residential site. Nearly 100% of all faunal remains identified at the site were marine resources. Evidence of subsistence procurement technologies include the use of exotic lithic material, a lack of formal tools such as projectile points, and the possible use of fishing nets (Sandweiss et al. 1998).

Figure 9. Southern Peruvian Coastal Sites

27

The Quebrada Jaguay site (QJ 280) is located on an alluvial terrace adjacent to an ephemeral stream within Jaguay Canyon. The terrace is one of several alluvial fan terraces in the region created by long-term tectonic uplift and sea-level fluctuation. The site is currently located 2 km from the modern-day shoreline, approximately 40 m above sea-level. Before the LGM, the site would have been located approximately 7 to 8 km from the coast (Sandweiss et al. 1998) (see figure 10). Substantial evidence has been identified at the site for a strict reliance on marine resources, including fish, crustaceans, and marine mollusks dating to the Terminal Pleistocene.

Figure 10. Submerged Coastal Plain Profile Offshore of Quebrada Jaguay. Over 96% of the number of identified species (NISP) recovered from the site were fish, followed by 3.2% of mammals. Mammals consist of mostly rodents that may simply have died at the site

28

(Sandweiss et al. 1998). The subsistence strategy focused on the procurement of small species of nearshore fish such as drum (Sciaenidae). The modal standard length of the fish was 172 mm, which was calculated from the recovery of 534 otoliths (Sandweiss et al. 1998). Knotted cordage recovered at the site and the standard size of the fish remains indicates the fish were caught using nets. The mollusk remains identified at the site were also monospecific. The wedge clam (Mesodesma donacium) made up over 99% of the mollusk species recovered. The people using Quebrada Jaguay had highly-specialized marine subsistence strategies, which may have been in place prior to the first occupation. The lithic technology at Quebrada Jaguay mimics that of the previously-discussed coastal sites in this review. The lithic material recovered at the site consisted of flake and some unfinished, bifacially worked tools. No projectile points of any stage of manufacturing were identified at the site. The raw lithic material consisted of petrified wood and obsidian. The obsidian flakes (n=30) were subjected to Instrumental Neutron Activation Analysis to determine the source of the material. The obsidian was mined from the Alca region which can be reached by following the Quebrada Jaguay stream valley to its head at the Cotahuasi River, a distance of 130 km (Sandweiss et al. 1998). Quebrada Jaguay has always been a coastal location from the beginning of human occupation at the site. During the low stand of -130 m, the site was only 14 km from the shoreline. The furthest distance from the coast during occupation was 7 to 8 km at roughly 13,000 cal BP (see Figure 10). The lithic and subsistence patterns fit the emerging model of heavy reliance on marine resources coupled with a lithic assemblage that lacked projectile points and the location of the site within 12 km of the coastline.

29

2.4.3 Quebrada Tacahuay Quebrada Tacahuay is a coastal open-air site situated 400 m from the Pacific Ocean in southern Peru. The briefly occupied site appears to have been a sea-bird butchering site which contained a hearth, lithic material, and burned bird and fish bone. The site was quickly abandoned due to debris flows that currently cover the site. Quebrada Tacahuay is located on an alluvial fan just south of the Ilo district in southern Peru (see Figure 11). The site was discovered during a roadway and waterline construction project. Quebrada Tacahuay was covered in 7 m of clastic debris flow from the nearby

Figure 11. Submerged Coastal Plain Profile Offshore of Quebrada Tacahuay. ephemeral stream. The site was exposed in a road cut where bird bones and a 50 x 8 cm hearth were observed protruding from the cut. Currently, the site is 400 m from the shoreline, yet when

30

the site was first occupied in 12,700 cal BP the ocean would have been 3 km away (Keefer et al. 1998) (see Figure 11). Like Quebrada Jaguay, the location has always been in a coastal setting since human occupation. Quebrada Tacahuay appears to be an ephemeral campsite where sea birds (n=16) were butchered. The species included: guanay cormorant (Phalacrocorax bougainvillii), Neotropical cormorant (Phalacrocorax brasilianus), booby (Sula spp.), cormorant (Phalacrocorax spp.), and one immature pelican (Pelecanus sp.) (Keefer et al. 1998). Many of the remains had butchering marks consistent to cutting breast meat portions away from the carcass (Keefer et al. 1998). The remains of small schooling fish including: anchoveta (Engraulis ringens), anchovy (Anchoa spp.), and indeterminate bony fish (Osteichthyes uid) were also recovered at the site. Both the size of the fish and the species’ tendency to school indicate that they were caught by using nets. Many of the faunal remains were burnt and/or recovered from within the hearth (Keefer et al. 1998). Lithic material was recovered in proximity to the faunal remains. The chalcedony lithic material consisted of 17 flakes and two unifacially-constructed tools. The debitage consists of eight smaller flakes that were produced during tool manufacture and nine utilized flakes. The tools were unifacial scrapers that appear to be common to coastal sites. No projectile points or complex knives were recovered at the site (Keefer et al. 1998). 2.4.4 Ring Site The Ring site is an open-air, ring-shaped shell midden site located roughly 40 km north of Quebrada Tacahuay (see Figure 12). The midden consists of a large variety of marine fauna including shellfish, fish, marine mammals and birds. The Ring site is unique in the that it is located on a marine terrace and not an alluvial fan. The modern coastline is currently 750 m from

31

the site, yet when the site was originally occupied in 11,832±239 cal BP the coast was 5 km away (Sandweiss et al. 1989). Like many of the other sites in the region, the Ring site has always been occupied by coastally-adapted people. The most prominent feature at the Ring site is the Holocene aged ring-shaped shell midden. The midden is 26 m in diameter with an indentation in the center. The invertebrate remains include: bivalves, gastropods, mollusks, and a few crustacean species such as barnacles and crabs. A few fragments of land snails (Scutalus spp.) and echinoderm (sea urchin) were also

Figure 12. Submerged Coastal Plain Profile Offshore of Ring Site. identified at the site. Wedge clam (Mesodesma donacium) were the most utilized species of invertebrate at the site (Sandweiss et al. 1989). Although, there appears to be a relatively even distribution between invertebrates that originated in sandy environments as there are for rocky

32

environments. The extensive use of shellfish did not become part of the marine subsistence strategy until later in the site’s occupation. The base of the shell ring dates to 5806±75 cal BP when sea-level began to stabilize (Lambeck et al. 2014; Sandweiss et al. 1989) (see Figure 12). Sea birds were also an important resource for the earliest peoples inhabiting the site. Many of the species include grebe (Podiceps major) from coastal marshes, shearwaters (Procellariidae), petral (Pelecanoides garnoti), pelicans (Pelecanus spp.), boobies (Sula spp.), and cormorants (Phalacrocorax spp.) from (Sandweiss et al. 1989). Cormorants were extensively used in the marine subsistence strategy and represent the second most abundant vertebrate taxon at the Ring Site, next to fish (Sandweiss et al. 1989). Fish remains were the most abundant taxon identified in the earliest strata at the site despite the use of only ½” and ¼” screen during excavation. Twenty species of fish were identified, with two species of drums (Sciaena deliciosa and S. gilberti) being specifically targeted. The two drum species represent 33.5% of the MNI at the site (Sandweiss et al. 1989). Many of the fish remains show signs of being burnt, while others appear to be used for nondietary purposes, such as for necklaces or other decoration (Sandweiss et al. 1989). Aquatic mammals such as river otter (Lutra feline) and sea lion (Pinnepedia) were also identified within the midden. There was a distinct absence of terrestrial faunal remains identified at the Ring site. Only small amounts of new world mice (Cricetidae) were identified. The lithic assemblage recovered at the Ring site is similar to the surrounding coastal sites in the region. Lithic materials (n=124) were concentrated in the levels below the shell ring where many of the vertebrate fauna were recovered. The lithics include several pieces of chipping debris, utilized flakes, and unifacial scrapers. No lithic projectile points were recovered at the Ring site (Sandweiss et al. 1989). Seven bone and shell tools were recovered at the site,

33

including: a compound fishhook, a barbed bone harpoon, and several culturally-modified shells. The reliance on drum and the presence of compound fishing hooks indicate that the fish were caught using hook and line (Sandweiss et al. 1989). 2.4.5 Monte Verde Monte Verdein Chile, dating between 14,220 and 13,980 cal BP (~12,310 and 12,290 radiocarbon years before present [rcybp])., is most widely known for containing unequivocal evidence for the earliest human habitation in the Americas. The site also contains the earliest

Figure 13. Monte Verde in Southern Chile.

evidence in the Americas for the use of seaweed, (Dillehay, Ramírez, et al. 2008). The site contains hearths with the remains of nine species of marine algae (Dillehay et al. 2008).

34

Evidence of coastal resources, including four species of seaweed (Durvillaea, Porphyra, Mazzaella, and Sarcothalia) were recovered within the hearths at the site (Dillehay et al. 2008). Although the site contains marine resources, it lacks the marine faunal resources identified in coastally-adapted sites (Marean 2014). Dillehay et al. (2008) state that the coastline “was situated ~90 km to the west and ~15 km to the south” of the site (Dillehay et al. 2008:784). Dillehey et al. used curves generated by Lambeck et al. (2002) and Siddall et al. (2003) to determine the depth in sea-level during the time of occupation. Lambeck et al. (2002) and Siddall et al. (2003) project sea-levels at 14,500 cal BP to be -100 m deep, yet Dillahey et al. models sea-levels at -60 m deep (Dillehay et al. 2008:SOM). The error may have been a confusion in radiocarbon versus calendar dates. Both Lambeck et al. and Siddall et al. present their curves in calendar years, not radiocarbon years. The depths at 12,500 cal BP are -60 m for both sea-level curves; the date in radiocarbon years of the seaweed is approximately 12,310 rcybp. It does not take a stretch of the imagination to see how the error may have occurred. The actual distances from the coasts at the time of initial occupation was 60 km to the west and 20 km to the eastern inlet. The reevaluated distances place the site outside the expected range for coastally-adapted sites. In summary, South America contains the oldest evidence for the utilization of marine resources in the Americas (Dillehay et al. 2017). The steep topography of the western coast prevented coastal sites from becoming inundated since the LGM. The negligible changes in the proximity to the sea prohibits an evaluation of the changes in cultural material as the distances increase or decrease. The sites appear to have been coastally-adapted from their initial occupation. Undiscovered sites inundated offshore may contain earlier examples of coastal adaptation that were in place before the sites reviewed in this section were occupied. The sites

35

indicate the same pattern noted in Old-World sites. There was an extensive reliance on expedient flaked tools and a near-exclusive reliance on marine resources. Coastal ranges tended to fall within 10 to 12 km from the coastline, yet faunal remains at Huaca Prieta extend the range up to 30 km (Dillehay et al. 2017). 2.4 Coastally Adapted North America There are only a few Terminal Pleistocene coastal sites in North America, all on the West Coast. The steep topography of the western coastline has kept some sites high and dry and secure from becoming inundated. This section will discuss the Paleoindian coastal sites on the Channel Islands and how they are the exception to the coastally-adapted model and should be considered the remains of a maritime-adapted culture. 2.5.1 Channel Islands The Channel Islands lay 42 km from the California coast. They have always been separated from the mainland, yet during the LGM they were 65% larger, and their distance from the California coast has increased by upwards of 10 km due to submergence (Erlandson et al. 2011) (see Figure 14). The group of Channel Islands were connected in one large island landmass, Santarosae, during the LGM sea-level low stand (Rick et al. 2005). Maritime travel would have been required to access the island. This fact skews our model as the Paleoindians that inhabited the islands were already maritime-adapted from the earliest occupations. The differences in cultural material, in comparison to coastally-adapted sites, may be a result of the technological adaptation to maritime way of life. The Channel Islands contain some of the most extensive maritime adapted sites in the Americas, with over 40 (n=44) documented sites spanning 12,000 to 7,000 cal BP (Rick et al. 2005). The oldest human occupation on the Channel Islands is at Arlington Springs site on Santa

36

Figure 14. Channel Islands Sites Off the Coast of California. Rosa Island, where human remains have been recovered dating between 13,000 to 11,500 cal BP (Johnson et al. 2002). The earliest inhabitants of the island relied extensively on marine resources (Cassidy et al. 2004; Erlandson et al. 2011; Erlandson and Jew 2009; Erlandson et al. 1998, 2007; Rick et al. 2005; Rick, Kennett, et al. 2005). The evidence for shellfish reliance is overwhelming in the earliest layers in the Channel Island sites. Sites such as Daisy Cave and CA-SRI-6 show that the subsistence strategy of shellfishing was supplemented by fishing or bird hunting, respectively (Rick et al. 2005). CA-SRI-6 is located at the bottom of Arlington Canyon, one of the largest drainage systems on Santa Rosa Island. The site would have had a reliable source of freshwater yearround during the LGM (Erlandson et al. 1998). CA-SRI-6 has been radiocarbon dated to approximately 9,300 cal BP (Erlandson et al. 2011). The midden at CA-SRI-6 is roughly 15 m long and 3.5 to 10 m thick. The shellfish remains comprise of 17 taxa of shellfish including 37

abalones, mussels, limpet, barnacles, and small gastropods. California mussel (Mytilus califtrnianus) and black abalone (Haliotis cracherodii) are the most prevalent shellfish species in the earliest occupation levels (Erlandson et al. 1998). A few remains of California sheephead (Semicossyphus pulcher), rockfish (Sebastes sp.), and smaller fish such as anchovies and sardines (Clupeiformes) were recovered at the site. The remains of snow geese (Anser caerulescens) and barn owl (Tyto alba) were also recovered at the site. The reliance on shellfish outweighed other marine resources at the site during the Terminal Pleistocene (Erlandson et al. 2011). As the environment shifts from Terminal Pleistocene to the Early Holocene, many of the midden sites on the Channel Islands show an increase in the use of birds to equal exploitation of shellfish (Rick and Kennett et al. 2005). Daisy Cave contains some of the earliest evidence of fishing economy in the Americas (Rick et al. 2001). The earliest layers at Daisy Cave contain only small amounts of rockfish recovered from a stratum dating to approximately 11,500 cal BP (Rick et al. 2001). Shellfish, specifically black turban (Tegula funebralis), surpass all faunal remains recovered in the early levels. As habitation at the site continued into the early Holocene, the subsistence strategy shifted to a reliance on fish. Researchers recovered more than 27,000 fish bones from 252 individuals and 18 different taxa. Over 90% of the fish remains recovered belong to nearshore species such as surfperch (Amphistichus rhodoterus), rockfish (Sebastes sp.), cabezon (Scorpaenichthys marmoratus), and sheephead (Semicossyphus pulcher) (Rick et al. 2001). Several fish gorges (n=30) were recovered in the Early Holocene levels, along with roughly 2,000 pieces of grass woven cordage. The faunal remains indicate that a combination of line fishing and dip netting were utilized by the inhabitants of Daisy Cave (Rick et al. 2001).

38

The lithic technology on the Channel Islands is unique in comparison to the other coastal sites reviewed in this chapter. The earliest inhabitants appear to employ expedient tool production, utilizing large flake tools, but contrary to the emerging lithic technology model, these expedient tools were also associated with carefully-manufactured bifacial tools including Channel Island Barbed points, stone crescents, and fish gorges. The lithic assemblages were predominantly constructed from locally-sourced chert, but also included an exotic obsidian artifact sourced to outcrops 300 km to the northeast (Erlandson et al. 2011). The presence of stemmed points and crescents indicate that inhabitants of the Channel Islands may have been associated with the Western Stemmed cultural group more generally associated with the Great Basin (Erlandson et al. 2011). The faunal remains recovered at the roughly 40 sites in the Channel Islands certainly reflect a subsistence strategy that revolves around marine resources. More than 97% of the faunal bone remains identified in Channel Islands assemblages are of marine origin (Cassidy et al. 2004) However, the sites differ from other coastal sites reviewed in this chapter. Paleoindians brought with them a maritime technology and an advanced lithic tool kit that differed from coastally-adapted sites in the Americas. The difference in lithic technology may reflect the difference between coastally-adapted and maritime-adapted strategies, and/or a difference in cultural groups. The ranges to the coastline during the LGM never extended further than 10 km, making it difficult to assess the differences in subsistence strategy versus coastline distances. However, the sites were all located in the direct proximity to freshwater (Rick et al. 2001), giving us a proxy for potential site locations. Shifts in environmental conditions from the Terminal Pleistocene may have played a larger role in the shifting subsistence strategies on the island than changing coastlines (Erlandson et al. 1998). The sites in the Channel Islands

39

represent the closest (both in distance and temporally) examples for modeling coastally-adapted patterns for the Southeastern United States, although until we identity a single Eastern Paleoindian coastal site, it is difficult to assess how similar these sites may be. 2.6 Discussion The goal of this chapter is to define what it means to be coastally adapted, how this adaptation is reflected in material culture and subsistence strategies, and the range in mobility of coastally adapted groups. During the review of coastal sites from around the globe, spanning the breadth of modern human behavior, a definitive pattern begins to emerge. Groups that become coastally adapted make a concerted shift from a terrestrial subsistence economy to a strictly marine diet. The lithic tool kit also shifts with the changes in diet. Carefully-crafted unifacial and bifacial projectile points and knifes are replaced by expedient utilized flake tools crafted from local material. The dramatic shift in the lithic tool kit may represent different cultural groups versus a single group moving back and forth from upland to coastal sites, recreating their tool kits as they move from site to site. If the latter scenario was the case, one would expect evidence from upland tool kits, such as knives or projectile points, and, possibly, terrestrial faunal remains to be recovered at coastal sites. However, this is overwhelmingly not the situation. Trade between upland and coastal hunter-gathers is certainly taking place, yet on a relatively small scale. This is indicated by evidence of a small amount of exotic upland lithics in coastal sites and marine resources in upland sites. The one exception to the lithic technology pattern is seen at the Channel Island sites. The complex tool kit associated with the Channel Island peoples maybe a product of maritime cultures versus coastal. Examinations both at Franchthi Cave and the Channel Islands illustrate how the introduction of boats can vastly change subsistence economy and mobility range reflected in the material culture.

40

Mobility ranges for coastally-adapted groups are generally confined between 10 to 12 km from the coastline, yet in some cases they can extend to 30 km inland. The extension in range appears to center around a specific land form in arid environments such as Boodie Cave in Northern Australia and Huaca Prieta in Peru. Boodie Cave may have served as a retreat from the heat of Northern Australia’s summers, whereas Huaca Prieta served as a high spot on the coastal plain. Pleistocene sites such as Pinnacle Point and Franchthi caves have a dramatic reduction in marine resources and habitation frequency as sea levels falland the coastline recedes further than 15 km from the site. Although Monte Verde appeared to fall within this range to the coastal inlet to the south of the site, and the site contained marine resources, it lacked the material culture to indicate that peoples inhabiting Monte Verde were coastally adapted. When the correct sea-level was used to determine the distance of Monte Verde to the coast for the habitation period, the site fell outside of the 10 to 12 km coastal range (see Table 1). Creating a predictive model for submerged coastal Paleoindian sites in the Gulf of Mexico should include the following: proximity to the coast within 10 to 12 km, local availability of freshwater such as a spring or stream, and local availability of a lithic source. Sites should also be within close proximity to paleo-river channels and/or alluvial fans. They should contain an extensive shellfish or nearshore fish midden with a lack of projectile points and an elevated amount of utilized flake tools. Faunal remains should have high concentrations of a single marine resource, with smaller amounts of birds and sea mammals. Submerged coastal Paleoindian sites in the Gulf of Mexico may contain high concentrations of eastern oyster and may resemble Late Archaic middens (Russo 1988). Submerged sites covered in Chapter 2 may serve as a more modern/contemporary analog to Paleoindian coastal sites further offshore.

41

Table 1. Sites Discussed in Chapter 2 with Range, Lithic Technology, and Subsistence Site

Pinnacle Point

1-10 km

Expedient Lithic Tool Technology X

Boodie Cave

3-24 km

X

X

Klissoura Cave

12-30km

2-8 km

X

X

Fish, wedge clam

Huaca Prieta

.1-30 km

X

X

Daisy Cave

7-.5km

Franchthi Cave

.4-7km

Shark, sea lion, fish, sea birds Shellfish, nearshore fish Shellfish, tuna, barracuda Seaweed

Quebrada Jaguay

Monte Verde

Range to the Coastline

Strictly Marine Subsistence Economy X

X X

X

20-60 km

CA-SRI-6

.5-7 km

Ring Site

.75-5 km

X X

42

X

Faunal Type

Shell fish, Whale, Sea Lion Fish, dolphin, 40 species of mollusk Deer, ass, pig, ibex, aurochs

Shellfish, nearshore fish Shellfish, sea birds, nearshore fish

CHAPTER 3 REVIEW OF SUBMERGED GULF OF MEXICO ARCHAEOLOGICAL SITES K.O. Emery and R. Edwards (1966) first suggested the potential for archaeological sites on the Atlantic Continental Shelf. They called into question the difference between data collected from coastal midden sites and midden sites located inland. The inland dates were substantially older than coastal midden dates (Emery and Edwards 1966). Emery (1966) postulated that older prehistoric coastal sites could have been inundated by rising sea-levels since the end of the LGM. This notion set the stage for archaeologists throughout the Americas to begin surveying the continental shelf for submerged prehistoric sites. In the mid-1980s, a team of archaeologists including Richard Anuskiewicz, James Dunbar, Joe Latvis, Michael Faught, and others began searching Florida’s western continental shelf in the Big Bend region (Cook-Hale 2016). The crew was successful in locating several sites in two field seasons. Throughout the 1990s, work continued as the team identified upwards of 40 submerged archaeological sites, while probing deeper and further offshore. Between 2008 and 2009, Jim Adovasio and Andrew Hemmings conducted two field seasons surveying Florida’s Middle Grounds for Paleoindian sites. The Middle Grounds are approximately 80 to 100 miles off of Florida’s panhandle. Their team surveyed a large area, adding nearly 100 miles of geophysical data to the submerged paleo-Suwannee River channel (NOAA 2009). Although they collected substantial data, they did not locate any archaeological sites or paleo-coastlines (NOAA 2009). One of the major contributing factors to their failure may have been the lack of understanding of coastal geological processes, and the lack of a refined sea-level curve. The Middle Grounds are a geological feature that was subaerially formed

43

roughly 9,000 cal BP. The Middle Grounds are remnants of a Holocene coral-reef buildup that was created subsequent to the sea-level transgression (Brooks 1962; Brooks and Doyle 1991;

Figure 15. Overview of Apalachee Bay Submerged Sites, Florida (Faught 1996).

Reich et al. 2013). Only a handful of submerged prehistoric Gulf of Mexico archaeological sites have been thoroughly-investigated in the last three decades. To understand the taphonomic processes that occur to archaeological sites during and after transgression, a geoarchaeological assessment must be conducted and the results compared between sites. In this chapter, I present a summary of published submerged sites in Apalachee Bay including: Ray Hole Springs (8TA171), Fitch Site

44

(8JE739), Econfina Channel (8TA139), and J&J Hunt (8JE740) and a geoarchaeological interpretation of the stratigraphy. 3.1 Ray Hole Springs (8TA171) Ray Hole Springs (8TA171) is a submerged site surrounding a freshwater spring located about 32 km offshore in Apalachee Bay, Florida (see Figure 15). The spring is 12 mbsl, 7.6 m in diameter and 5 m deep. The north side of the sinkhole grades into the spring at a gradual southeast direction. The southeast side is a vertical wall of limestone (Anuskiewicz and Dunbar 1994). Sport divers discovered the site and reported archaeological material surrounding the sinkhole to the Bureau of Archaeological Research (BAR) (see Figure 16). In 1986, archaeologists with the Bureau of Archaeological Research, in conjunction with Minerals Management Services, conducted the first archaeological survey of the spring. Preliminary testing, including two sets of cores and test units, focused on the area within the spring. Excavation within the springs terminated in a thick layer of recently deposited marine sediment (Anuskiewicz and Dunbar 1994). Excavations outside the spring identified a layer of carbonate sand overlying limestone bedrock. Excavations were then conducted in a 15 centimeter (cm) wide crevice that was identified near the rim of the spring. Archaeologists recovered multiple pieces of lithic material (chert) that appeared to be culturally modified. The material was recovered at 15 to 20 cm below the ground surface (cmbs) (Anuskiewicz 1988). Anuskiewicz (1988) refers to these artifacts as pseudo-artifacts due to the inability to positively identify the lithic material as being culturally modified. The material was recovered within the marine layer and unequivocally in secondary context. A layer of eastern oyster was encountered

45

Figure 16. Profile of Ray Hole Springs. Developed from (Anuskiewicz 1988)

at 75 cmbs (Anuskiewicz and Dunbar 1994). Two sample of the oyster (Crassostrea virginica). were radiocarbon dated to 8,243±74 and 9,162±128 cal BP. Excavations continued to 100 cmbs where pieces of live oak (Quercus virginiana) were identified in contact with the limestone bedrock. The dates indicated a 1,000-year range between the freshwater environment in which

46

the live oak was deposited and the transition to a brackish water habitat of the oyster (Anuskiewicz and Dunbar 1994). The initial results of the examination of Ray Hole Springs sparked archaeological surveys in 1989, 1990, and 1992. Attempts to extract a sediment core from the site in 1990 and 1992 were hampered by mechanical malfunctions and bad weather (Anuskiewicz and Dunbar 1994.) Dive teams searched the rim of the sinkhole for additional deep crevasses comparable to the one excavated during the initial survey. The team also excavated smaller holes around the rim and encountered two sediment layers within the excavation pits (Anuskiewicz and Dunbar 1994.) The first was a disturbed marine carbonate sandy layer that extended on average to 10 cmbs. Many of the pseudo-artifacts were recovered from this layer. The second layer consisted of unsorted rubble deposits that typically continued to approximately 25 to 35 cm deep. This layer also appears to be an unweathered deposit of a bioclastic shell detritus and pebble to cobble-size rock mixed with sand, silt, and clay (Anuskiewicz and Dunbar 1994). The team was successful in recovering two positively-identified secondary reduction flakes, one in 1990 and the second in 1992. Both flakes were recovered from the second sediment layer, and showed multiple conchoidal pattern flake scars, and one flake had a bulb of percussion (Anuskiewicz and Dunbar 1994). The second unit excavated in 1992 was terminated prior to reaching a sterile level or encountering bedrock due to time constraints. The field season had ended just as they were coming down on the brackish layer. The team never returned to finish the unit. In 2009, a National Oceanographic and Atmospheric Administration (NOAA) archaeological survey, led by Andrew Hemmings and James Adovasio, attempted to relocate the site. Unfortunately, the team was unsuccessful in their efforts.

47

3.2 Econfina Channel Site (8TA139)

The Econfina Channel Site (8TA139) was identified by Dunbar and Faught during the 1986 Apalachee Bay Survey (Faught 1989). The site is located 4.8 km (3 mi) from the mouth of the Econfina River in water depths that range from 6 to 15 ft. (Faught and Donoghue 1997) (see Figure 17). The site is located on the southwestern side of the PaleoEconfina River channel. After a revisit to the site, Cook Hale (2018) described the site layout. A prominent shell midden is located on the northern end of the site oriented east to west. There are several chert and dolomite outcrops that surround the midden with widespread lithic debitage. The PaleoEconfina channel is positioned south-

Figure 17. Overview of discussed Submerged Sites in Apalachee Bay, Florida (Faught 1996). southwest along the quarry zone. The 1986 survey detected a freshwater spring feature 50 m west of the midden and was confirmed as still flowing during the 2015 survey (see Figure 18) (Faught 1989; Cook-Hale 2018).

48

During the 1988 field season, four test units were excavated to determine the stratigraphy of the site. Initial results indicated that sediment deposits were truncated at the site. Test Unit D was the deepest unit, which was terminated at the dolomite layer at 165 cmbs (Faught 1989). The first layer consisted of a marine sediment of shell and limestone fragments. This marine sedimentary unit lacked any structure and extended to a depth of 15 cmbs. Most of the artifacts recovered were from this layer (Faught 1989).

Figure 18. Site Map of Econfina Channel (Cook-Hale 2016). The 1988 field survey excavations and surface collections yielded 542 pieces of lithic material. Of the 542, 1% (n=5) of the artifacts were bifacially worked, with one identified as a

49

Marion or Putnam Archaic period projectile point (Faught 1989). Of the total flakes collected, 44% (n=239) of them had varying degrees of cortex remaining on their surface. The smaller, tertiary flakes 21% (n=114) showed evidence of bifacial lithic scars. There were three color classifications for the chert debitage: white (23%), grey (55%), and black (22%) (Faught 1989). The high percentage of cortex present on the lithic debitage signifies that Econfina Channel Site was used both as a quarry and bivalve processing area (Faught 1989). Surveys were conducted in 2014 and 2015 by Jessica Cook-Hale of the University of Georgia focusing on the shell midden (Cook Hale et al. 2018). Three additional test units were excavated throughout the site, and the dimensions of the midden were obtained. The midden measured 9 m along a north to south axis and 6 m along an east to west axis (Cook-Hale 2018) (see Figure 18). The species of shell observed were eastern oyster, scallop (Pectinidae), and freshwater gastropods (Cook-Hale 2018). Test Unit 1 was placed on the midden. Test Unit 2 was excavated in an area with rocky outcrops north of Test Unit 1. Test Unit 3 was placed northwest of the shell midden (Cook-Hale 2018). One flake was recovered from Test Unit 1, while several were discovered during the excavations of Test Units 2 and 3 (Cook-Hale 2018). No diagnostic lithic material was discovered during the 2014 and 2015 field seasons. The abundance of oyster, scallop, and freshwater gastropods suggests a juxtaposition to the coast, making these sites more likely to date no older than 5,000 cal BP (Cook-Hale 2018). Yet, the interpretation of the site has been hampered due to the repeated disturbances to the stratigraphy during severe weather conditions.

50

3.3 Fitch Site (8JE739) The Fitch Site (8JE739) is located 10 km southwest of the mouth of the Aucilla River in approximately 5.2 m of water (Faught 1990) (see Figure 17). The site is located in an area scattered with karst depressions and hills. The topography at the site consists of an alternating

Figure 19. Profile of the Fitch Site (Faught 1990).

51

linear island of limestone bedrock with shell hash marine deposits in between. The center of the site rests on top of a slight rise consisting of rounded limestone bedrock. Surface collection conducted in 1989 consisted of 200 m long transects across the site. Three 1x1 m excavation units were excavated in 20 cm arbitrary levels using an induction dredge. Test Unit 1 was located near Loci 1, and Test Units 2 and 3 were located near Locus 2 (see Figure 19). 3.3.1 Stratigraphy The stratigraphic profile was developed from the sediment observed in Test Unit 3. The surface deposit of sediment consisted of fine-grained sands and biogenic organics including remnant marine shell fragments from deceased shellfish species (Faught 1990). The marine surface sediment extended from 10 to 15 cmbs. Several artifacts were recovered in this stratum. The layer was referred to as “marine hash” which likely represents a conflated surface which was disturbed during storm surge. The marine hash layer transitioned into Sedimentary Unit 2 at an abrupt, wavy interface. Unit 2 consisted of compact dark grey, fine silty/sand. This layer contained small amount of shell inclusion of an unidentified species. A limited number of lithic artifacts were recovered in this level (Faught 1990). This layer may have been deposited under brackish conditions shortly after transgression. Unit 2 transitioned into Unit 3 at an abrupt interface. Unit 3 extended from 60 to 80 cm below Unit 2 and included large pieces of shell and wood (3 to 7 cm in diameter). Shells were deposited in a matrix of fine grained sand. Lithic material was recovered within this unit (some possessing bulbs of percussion) and the density increased with depth. Faunal remains of unidentified species of turtle were recovered in Unit 3 (Faught 1990). Unit 3 represents freshwater sediments deposited before the transgression.

52

Unit 4 is described as a “natural pavement” located under Unit 3 (Faught 1990:12). This pavement consisted of rounded cobble to boulder sized pieces of lightly colored dolomite. This dolomite layer has been encountered in the J&J Hunt and Econfina Channel sites. The process under which this layer was created is unclear (see Chapter 5). Under the cobble and boulders of Unit 4 is Unit 5 which consists of a light grey layer of silty/clay with gastropod shells. Lithic shatter and the disarticulated remains of extinct fauna including: giant sloth (Eremotherium sp.), manatee (Sirenia), horse (Equus sp.), and camel (Camelops hesternus) were identified within this level (Faught 1996). The lithic shatter was not positively identified as being culturally-modified. The silty/clay was described as friable (Faught 1996), indicating that the sediment may have been subaerially-exposed and have undergone pedogenesis. During surface collection and excavation of three 1x1 m units, 402 artifacts were recovered. The average weight of the artifacts was 101.21grams (g) with several pieces weighing over 200 g. Many of the artifacts recovered were found either on bedrock outcrops or within the first 10 to 15 cmbs of marine sediments (Faught 1990). There were two separate loci of artifact concentrations. The highest concentrations of artifacts were located in the southern end of the scatter, tapering off in size and number as it progressed to the north. Lithic material in the first stratum was likely conflated and disturbed by the transgression and subsequent storms throughout the Holocene. Cultural material recovered in Units 2 and 3 was minimally disturbed. Fitch Site lacks a paleochannel like those found at Econfina and J&J Hunt; it may be associated with the PaleoPinhook River (Faught 1990). Manatee and parrot fish remains indicate the site was connected to a river system (see Table 2). Of the 402 artifacts collected, 6% (n=23)

53

Table 2. Faunal Remains at the Fitch Site (Faught 1990). Class Mammalia Racoon

Class Reptillia Procyon lotor

Extinct sea cow

Dugingidea sp.

Giant land tortoise Box turtle

Pleistocene horse Short limbed llama Class Osteichthyes Bowfin Parrot fish

Equus sp.

Soft shell turtle

Paleolama mirfila

Snapping turtle Red eared turtle

Amia calva Diodon sp.

Mud turtle Alligator

Geochelone crassiscuttata Terrapene carolina Apalone ferox Macroceclemys temminki Chrysenys scripta Kinosternon sp. Alligator Mississippiensis

exhibited unequivocal evidence of being culturally modified. Lithic material that lacked either bulbs of percussion or flake scars 68% (n=273) could not be ruled out as non-cultural. Cortex was observed on the surface of 22% (n=88) of the lithic material recovered (Faught 1990). Surface staining associated with many of the lithic materials recovered from the region’s tannic rivers was observed (Faught 1990). When the chert was fractured, the unstained interior appeared pink with white spots. Chert sources of this type have been identified near Hillsboro County (Jon Endonino, personal communication July 2016). Two “discoidal cores” and several large flakes were recovered, and some displayed signs of heat treatment (Faught 1990). Discoidal cores have been associated with Paleoindian lithic reduction techniques, yet heat treatment is regarded as an Archaic period technology (Faught 1990). Other flakes showed evidence of blade-like flake reduction techniques (Faught 1990). The bedrock surrounding the site possesses high amounts of medium to fine grained chert imbedded within the limestone. Fitch may have been a quarry site, though many of the artifacts recovered had low amounts of

54

cortex on their surfaces. Due to the lack of diagnostic artifacts, occupation dates for the Fitch site are estimated by sea-level data to between 12,000 and 7,000 cal BP (Faught 1990). 3.4 J&J Hunt Site (8JE740) The J&J Hunt Site (8JE740) was first identified by Michael Faught and James Dunbar in 1989 during the Aucilla River Prehistory Project (ARPP) survey of the submerged PaleoAucilla in Apalachee Bay (Faught 1997). J&J Hunt site is located 6.5 km offshore from the mouth of the modern Aucilla River (see Figure 17). The site and associated loci include a submerged channel segment approximately 1.25 km in length and five large sinkholes (Faught and Donoghue 1997). The main site consists of a 10,000-square meter section surrounding a large sinkhole that was once part of the PaleoAucilla River system. Accompanying the site are three loci referred to as Areas A, B, C and one Locus, L. Areas A, B, and C are in proximity to J&J Hunt in both location and chronology yet are not considered to be one continuous occupation event (Arbuthnot 2002) (see Figure 20). Over six years, the ARPP excavated 33 test pits, recovered 1,740 artifacts (Faught 2004), extracted several vibracore samples (including seven from Locus L) (Faught and Donoghue 1997), and mapped 238-square meters of the site. (Arbuthnot 2002). 3.4.1 J&J Hunt, Locus L, and Area A, B, C Locus L is one of the largest features of any observed from the mouth of the Aucilla River to the J&J Hunt site (13 km). The locus is situated just southwest of J&J Hunt proper, (Faught and Donoghue 1997). Locus L represents an area of high activity during the Early and Middle Archaic. A projectile point of Middle Archaic age and a broken deer antler were located near the oak tree stump at Locus L. The presence of the artifacts suggests that humans were present just before the site was transgressed (Faught and Donoghue 1997:443).

55

Area A is located north-northeast of the sinkhole in Locus L. The artifacts recovered from this area included Paleoindian/Early Archaic period Kirk and Bolen projectile points, Hendricks, Edgefield, and Turtleback uniface scrapers. The artifacts in this area are more dispersed than the other areas and represent secondary lithic reduction and tool resharpening (Arbuthnot 2002). Several unidentifiable terrestrial mammal long bones were also present at Area A (Faught and Donoghue 1997). The dispersal of the lithic material may be a result of disturbances during transgression. The artifacts recovered in Area A were stained black at a

Figure 20. Site Map of J&J Hunt (Arbuthnot 2002).

56

higher rate than the other surrounding areas. This staining can still be observed for eroded artifacts laying on the floor of the modern Aucilla River. Area B is located 115 to 150 m north northwest of Locus L and represents an older Paleoindian period occupation based on the artifact assemblage (Faught 2004). A Suwannee preform, thumbnail scrapers, and tertiary flakes associated with bifacial reduction were recovered in this location. The flakes were discovered in a localized, high concentration (Faught 2004). Artifacts recovered in Area C included a Paleoindian period fluted biface at the margins of the channel (Faught and Donoghue 1997). In 1992, a test unit was excavated in Area C, fully exposing a stump discovered in the previous field season. A sample of the outer ten rings of the oak stump was subjected to radiocarbon testing and resulted in a date of 8070±101 cal BP (Faught 2004) (All dates have been recalibrated using Oxcal 4.3) Eastern oyster shells were also recovered from around the stump. The oyster shells were dispersed in a linear manner 20 to 30 m east and west and suggest the presence of an extinct oyster bioherm (Faught and Donoghue 1997). Radiocarbon dating was conducted on the eastern oyster shell and resulted in a date of 6848±98 cal BP (Faught 2004). 3.4.2 Stratigraphy The stratigraphic series and pollen analysis for the site was determined by vibracore sample 91-3. The first sedimentary unit was a mixed marine layer of coarse sand and shell hash that was deposited post transgression. This unit extended to a depth of 28 cmbs. This sandy layer would have been periodically disturbed during major storms impacting Apalachee Bay. Dating for this unit would range between the date of inundation, 6848±98 cal BP to present (Arbuthnot 2002). The second unit was separated into two sections, the upper transition zone and lower

57

brackish water zones. These two units represent a brackish water deposition, which consisted of a black richly organic layer with preserved wood. The upper layer (35 to 40 cmbs) was high in cypress (Taxodium) pollen (77%) followed by pine (Pinus) 9% and traces of gum-tree (Nyssa biflora). Gum-tree is primarily considered a species associated with a swamp environment. The absence or presence of gum-tree pollen within the samples was used as an indicator of the amount of surface water on the surrounding landscape (Arbuthnot 2002). The lower brackish layer consisted of a dark gray silty clay with pollen counts for pine (36%) are comparable to that of cypress (30%) (Arbuthnot 2002). Foraminifera and ostracodes (Cyprideis americana) identified as brackish species as well as eastern oyster were found throughout the brackish unit (Faught and Donoghue 1997). A radiocarbon assay for the brackish unit determined

Table 3. Faught and Donoghue (1997) Radiocarbon Samples from J&J Hunt. Samples Were Calibrated Using OxCal 4.3. Wood Samples Were Calibrated Using Incal13 and Oyster Using Marine13. Sample Wood in marine sand deposit Oyster recovered near stump Brackish water wood from marl Fresh water wood in sandy clay Wood from stump

Radiocarbon Date 6,100±60

Calendar Year Range 99.7% 7,241-6,745

Median Date

Water Depth

6,979±97

-3.7 m

6,375±75

7,154-6585

6,848±98

-4.5 m

6,785±80

7,920-7,436

7,636±70

-5.5 m

7,130±75

8,178-7,711

7,955±80

-6.4 m

7,240±100

8,379-7,786

8,070±101

-4.5 m

from a sample of wood (Lab # AA-8895) of unknown species from the middle of the layer. The wood returned a date of 7636±70 cal BP (Faught and Donoghue 1997) (See table 3 for dates). This layer represents a period of encroaching sea-levels. During this period, global sea-levels

58

would have been 4 m below modern sea-levels (Lambeck et al. 2014). The site would have resembled modern-day marshes at the mouth of the Aucilla River (Faught and Donoghue 1997). Underlying the brackish layer was a fine-sandy grey clay sediment unit that extended from 48 to 80 cmbs. Pollen samples were extracted from the upper and lower sections of this unit. Upper level pollen results had a mix of pine (40%) and cypress (31%). Gum tree pollen was absent from this sample. The high levels of pollen from grasses and trace amount of hickory (Carya sp.) indicated a transition from a dryer, cooler climate to a wetter, warmer climate. It also may denote the beginning of marsh conditions (Arbuthnot 2002). Samples from lower in the unit were fine-sandy grey clay with pollen levels similar to the upper sample: pine 36%, cypress 30%. An increase in grass pollen was not observed in this sample. Radiocarbon samples gathered from above and below this level places the deposition between 8070±101 and 7955±80 cal BP (Faught and Donoghue 1997). This sediment unit was likely deposited when freshwater flowed through the PaleoAucilla. The final layer was a fine-sandy grey clay analogous to the sediment in the previous layer. The pollen counts differed dramatically with pine levels at 68% followed by cypress at 12% (Arbuthnot 2002). The core was terminated due to a rock impasse caused by a dolomite deposit. A radiocarbon date of 7955±80 cal BP was returned for a sample collected in the clay matrix surrounding the rocks. At the Fitch Site, Faught and Dunbar were able to penetrate the dolomite deposit and recovered Pleistocene era faunal material within this layer (Arbuthnot 2002). 3.4.3 Site Occupation J&J Hunt site had two occupational periods based on diagnostic artifacts and radiocarbon dates: Paleoindian and Bolen/Early Archaic (Arbuthnot 2002). Of the 1700 artifacts recovered, 25 of them were diagnostic projectile points or tools. Fourteen of the diagnostic points were

59

associated with the Paleoindian period, including a fluted point base and a Suwannee point preform. Although there is a lack of unequivocal dates for Suwannee projectile points, they are believed to be Middle Paleoindian period (Smith et al. in press; Thulman 2012). During this period, the Gulf of Mexico would have been 200 km (124 mi) from J&J Hunt (Chapter 6). This section of PaleoAucilla would have resembled the discontinuous segments of the modern river near Half Mile Rise or the Little River. The landscape surrounding the site during the Paleoindian period would have been thick forests of oak and hickory trees with vast amounts of grasses and herbs (Arbuthnot 2002). The site would have been an ideal area for Pleistocene megafauna to thrive. A lower water table during this period would have confined fresh water sources to sinkholes, such as J&J Hunt, which would have made the site an important watering hole (Thulman 2009). Six of the diagnostic artifacts were associated with the Paleo/Archaic transition period including Kirk and Bolen projectile points andHendricks, Edgefield, and Turtleback uniface scrapers (Arbuthnot 2002). During this period, sea-levels had increased to -35 m bmsl, with the coastline located 124 km (77 mi) from the site. Pollen counts indicate that the area was dotted with cypress and pine forests. By 6,000 cal BP, ocean levels would have been within 3 m of modern levels. Brackish conditions would have intruded into the PaleoAucilla up to the J&J Hunt site. The coastline would have been roughly 3 km (2 mi) from the site. Evidence suggests brackish shellfish species (eastern oysters and scallop) were harvested at the site during this period (Faught and Donoghue 1997). The remaining five artifacts recovered were classified as intermediate and could not be definitively dated (Arbuthnot 2002).

60

3.4.4 Bchron Analysis of J&J Hunt Radiocarbon Samples The suite of radiocarbon samples were subjected to the sample chronological Bayesian statistical model utilized on the reevaluated sea-level curve (see Chapter 6). The six dates were

Figure 21. Bchron 95% Chronology of Radiocarbon Samples from J&J Hunt (Faught and Donoghue 1997). entered into the program to identify outliers and to build a chronology of deposition. The mean P-values for the sample was .29 out of .50. The date reversals for the eastern oyster and the oak stump samples, differing by more than a thousand years may be the cause of the small mean Pvalue for the model. Sample AA-6714, the oak stump, was the only sample considered an outlier by the Bchron program, due to its growth well before the transgression event. However, the model shows that the sediments are time transgressive (see figure 21)

61

3.5 Discussion Throughout this summary, there have been several similarities between sites. Although these sites have been transgressed, these sites do not fit the subsistence or lithic assemblages associated with coastal sites. Instead, the sites appear to have been inhabited when coastlines were hundreds of kilometers from the sites, as will be presented in chapter 6. (see Sea Level Figure 33) Second, a majority of artifacts were recovered in the first 15 cm of what has been described as “marine hash.” This marine hash appears to have little structure, and diagnostic artifacts from separate time periods are located in the same levels. There are several processes that may have occurred. The marine hash may have been deposited during high-energy storm surge and all artifacts associated with this level are in a secondary context. Another hypothesis is that the artifacts have remained roughly in place and the sediment in which they were originally deposited has been eroded away. In either case, artifacts that are discovered within this layer must be considered to be in secondary context. A future study could monitor varying lithic materials of different weights and sizes as they are transported across the sea-floor. The third similarity with all the offshore sites in this summary is the dolomite layer. As mentioned above, dolomite can form under varying processes. Geologists state that dolomite can form in as little as a few years (Coudray and Montaggioni 1986). If dolomite can be accurately dated using U-series, then the dolomite layer could be used as proxy data for sea-level rise in the Gulf of Mexico. The final similarity between the sites is the lack of excavations reaching sediment from the Paleoindian time period. Paleoindian artifacts have been recovered at several sites, yet unit excavations are terminated at the dolomite layer, which appears younger than the Paleoindian period. At the Fitch site. where the dolomite layer has been penetrated. a plethora of Pleistocene

62

faunal remains were discovered. To understand how the Paleoindian aged diagnostic artifacts were deposited in a Late Archaic sediment, an in-depth analysis of the sediments and taphonomics of the site must be completed. This analysis will determine what layers are intact and what artifacts are in primary context (see Chapter 4).

63

CHAPTER 4 COASTAL GEOLOGY AND TAPHONOMIC PROCESSESS AT TRANSGRESSED SITES Taphonomic processes at any archaeological site, either terrestrial or submerged, play a vital role in site interpretations. Yet, taphonomic processes at transgressed sites on Florida’s Gulf of Mexico continental shelf have yet to be fully explored (Duggins et al. 2018; Faught 1992; Marks 2006). Understanding the geological processes that take place during a transgression event is a vital aspect in interpreting a site. Although many different factors may control the level of site disturbance including topography, sediment types, and MWPs, a fundamental understanding of how these events present themselves in the stratigraphic profile is paramount. This chapter reviews the geological processes of the Holocene transgression event at J&J Hunt site because it is the most intensely surveyed offshore site in the region with an excellent geoarchaeological record. 4.1 Idealized Stratigraphic Description of Transgression Event The transition from glacial to interglacial involves a massive amount of meltwater being reintroduced into the ocean. The additional meltwater increases global sea-levels, which changes the sedimentary deposition regime for areas that are transgressed by the sea. These changes in depositional environment can be identified within lithostratigraphic profiles. Freshwater pond and low energy fluvial deposits are associated with gray and greenish-gray smectic clay with sand laminations in higher energy zones and from wash-in deposits. Sedimentary structures in the higher flow regime may consist of tabular and trough cross-bedding, while pond deposits may be massively deposited clay muds with inclusions of freshwater fauna such as turtle (Chrysemys picta), and extinct Pleistocene terrestrial species such as sloth (Eremotherium sp.), horse (Equus sp.), and camel (Camelops hesternus). As sea-levels begin to transgress the area,

64

tidal zones that are protected from high energy wave oscillations will form muddy tidal flats deposits. Tidal flats consist of dark gray muddy sandy silt deposits with very fine sand laminations, including flaser bedding occurring during episodes of high energy events. Other sedimentary structures may include ripple and oscillation bedding and may include plant and bivalve remains. In protected areas such as lagoons or estuaries, fines may settle out and form silty clay facies or, in lower clastic areas, even carbonates. These facies will have an abundance of oyster and foraminifera and ostracodes (Cyprideis americana) identified as brackish species with evidence of vertical bioturbation. Nearshore deposits will consist of grey to olive grey marine hash consisting of sands with occasional silty laminations with organic inclusion deposited in the low energy coastal environment of Apalachee Bay (Cattaneo and Steel 2003). 4.2 Stratigraphic Description of J&J Hunt Site J&J Hunt site is located 6.5 km offshore from the mouth of the modern Aucilla River (see Figure 17). The site includes a submerged channel segment approximately 1.25 km in length, including five large sinkholes (Faught and Donahue 1997). The site consists of a 10,000-square meter section surrounding a large sinkhole that was once part of the PaleoAucilla River system. The area of the site where the sediment samples were extracted were freshwater subaqueous sediments prior to sea-level transgression. The paleoenvironmental reconstruction of the site indicates that the area was a shallowly filled freshwater karst feature (Arbuthnot 2002). Coastal sedimentary deposits are distinct to the environment in which they are aggregated. The amounts of energy and the parameters for chemical reactions dictate the deposition of specific clastic grain sizes, sedimentary structures, and carbonate formation. The sedimentary structures and facies expected in the research area should contain low energy pond and low clastic fluvial

65

deposits and thin overbank facies on the periphery of the karst feature. These deposits should be overlaid by retrogradational transgression tidal flats and nearshore facies (Eros et al. 2012). 4.3 Core Descriptions Core 01-02B was driven 140 cm into the sediment (see Figure 22). The first 4 cm consisted of a rounded fine to medium sand, pale yellow (2.5Y 8/2) in color. Below the sand is a “marine hash” that extends from 4 to 33cm. The sediment is very dark grayish brown (2.5Y 3/2) in color and consists of silty fine sand with well-rounded quartz grains and remains of fractured eastern oyster. The following layer transitions at an abrupt interface and consists of a brackish layer that encompasses the sample from 33 to 50cm. The sediment is comprised of a gray (2.5Y5/1) fine sand with trace silt and organic material (wood fragments). The fragments of wood were radiocarbon dated with a resulting date of 7636±70 cal BP (Faught and Donahue 1997). A high energy event that may have been associated with the transgression and/or a large storm was identified by flaser lamination from 52 to 64cm (see Figure 22). This layer contains rounded fine sands with trace silt and clay muds. This layer, when mixed, is gray (2.5Y 6/1) in color and transitions at an abrupt interface. Below the flaser lamination is a layer that is associated with a low energy freshwater environment. This layer extents from 64 to 94cm and consists of sandy clay loam that is dark gray (2.5Y 4/1) in color and contained organic material. This layer was also Figure 22. Profile from Core 01-02B from J&J Hunt Site Apalachee Bay, Florida.

66

radiocarbon dated, and its resulting date was 7955±80 cal BP (Faught and Donahue 1997). This layer terminated at an abrupt interface with dolomite layer with the sandy clay loam filling the spaces between the dolomite cobbles. The dolomite continued throughout the core to termination at 140cm. Chapter 5 further investigates the dolomite layer and possibility for the layer to be dated using U-series dating, with the potential to be used as a sea-level proxy data source.

67

CHAPTER 5 DOLOMITE FORMATION AND POTENTIAL AS A SEA-LEVEL PROXY Since the 1960s, sea-level transgression models for the end of the last glacial maximum have relied on radiocarbon dating of proxy data. The accuracy of radiocarbon dating has improved greatly, yet computing radiocarbon reservoir corrections fall short of addressing local changes in marine carbon environments. Fluctuations in atmospheric CO2 levels and in air to sea carbon exchange can alter carbon intake in local marine environments. Researchers have begun to use uranium series dating of coral to avoid the inaccuracies associated with radiocarbon dating. Corals that have been employed as sea-level indicators, such as elkhorn coral, grow in southern end of the Gulf of Mexico. Elkhorn coral has a difficult time surviving in the northern Gulf due to the clastic input from fluvial systems. Modern dolomite may have the potential to replace coral as sea-level proxy data in the northern sections of the Gulf of Mexico. During a review of submerged archaeological surveys conducted on Florida’s continental shelf, a common occurrence has been reported: a layer of dolomite has been identified during excavations at varying stratigraphic depths. Modern dolomite formations have been reported from the Persian Gulf, South Australia, the Bahamas, and the Florida Keys and can precipitate during sea-level rise within seawater subtidal environments. Modern dolomites can range in age from a few years to about 4,000 cal BP. Dolomite can be dated using uranium series methods and potentially be used as sea-level rise proxy data. Uranium series dating of the dolomite layers has never been attempted at any of Florida’s off-shore sites and has the potential to be excellent proxy for oceanic transgression.

68

5.1. Dolomite The French naturalist Deodat de Dolomieu first described dolomite in 1791. Dolomites are calcium carbonate sedimentary rocks with a mineral content of Calcium, Magnesium, and Carbon Trioxide (Ca,Mg,CO3) (Al-Awadi et al. 2009). Dolomite appears as white, gray, brownish white, or reddish white with an average density of 2.84 g/cm3. Diaphaneity is generally transparent to translucent and is brittle to very brittle producing small, conchoidal fragments when fractured. Its habit is blocky with rhombohedral shaped coarse crystals and has a hardness of 3.5 to 4 (Steinfink and Sans 1959). They can be found disseminated throughout the geologic record, ranging from Precambrian to Holocene (Bahniuk et al. 2015). Dolomites are associated with limestone and are also commonly associated with evaporates. Coarsely crystalline dolomites are secondary rocks formed by diagenetic replacement of older limestones (Braithwaite 1991). Fine-crystalline dolomites lack such textural evidence of replacement and cannot be proven to have originated by diagenetic alteration of limestones (Lumsden and Lloyd 1997). The “dolomite problem” has been coined due to the abundance of Middle Triassic aged dolomite in sedimentary rocks, in contrast to the rarity of modern dolomite (Blendinger et al. 2015). Modern dolomite formations have been reported from the Persian Gulf, Russia, South Australia, the Bahamas, and the Florida Keys (Boggs 2006). Dolomite can be dated using the classical radiocarbon, U-Th, and amino-acid racemization methods (Rasbury and Cole 2009). Radiocarbon estimated ages of these dolomites range from a few years old to about 4,000 years old (Boggs 2006). Discovery of relatively young dolomite was considered as evidence that it can be precipitated geologically as a primary deposit (Rodriguez-Blanco et al. 2015)

69

5.2 Models for Dolomite Formation at Fitch and J&J Hunt Sites Dolomite can materialize in two ways: precipitation and replacement dolomitization. Replacement dolomitization is the process of replacing the calcite in existing limestone and transforming it into dolomite. This process is important in the formation of ancient dolomites. This research will focus on the process of precipitation of younger dolomites. There are three models for the modern formation of dolomite: (1) The Hypersaline Model, (2) The Mixed-Water Model, and (3) The Seawater Subtidal Model (Al-Awadi et al. 2009). 5.2.1 Hypersaline Model The Hypersaline, or Reflux, Model requires seawater to become restricted in a shallow lagoon or sabkhas. As the sea-water evaporates, it is converted to a hypersaline brine that is much denser and descends through the lagoon floor. The hypersaline water seeps through porous sediments and, as it does, magnesium from the brine reacts in a neogenesis process that replaces calcium within facies that contain aragonite and/or calcite. The concentrated hypersaline water increases the Magnesium/Calcite (Mg/Ca) ratios. The Mg/Ca ratio in modern seawater is about 5:1. As the ratios are increased in the hypersaline water to 10:1, carbonates within the sediments are effectively converted to dolomite (Rodriguez-Blanco et al., 2015). Hypersaline dolomites are associated with other evaporate minerals, such as anhydrite, gypsum, and halite, which incorporates much of the Ca minerals, increasing the Mg ratios further (Al-Awadi et al. 2009). In areas where the hypersaline model is responsible for dolomitization, concentration of both dolomite and evaporates decrease with diminishing proximity to the lagoon/sabkhas environments (Al-Awadi et al. 2009). Sulfate-reducing bacteria play a vital role in the diagenesis of primary dolomite in lagoons. Sulfates have been shown to inhibit dolomite production. The metabolic process of

70

lagoon bacteria reduces the levels of sulfate in the hypersaline water, allowing dolomite to precipitating in anoxic, organic-carbon-rich marine sediments (Bahniuk et al. 2015). Most Holocene dolomites were formed under the hypersaline process (Richter 2016). This process is called Seepage Refluxion (Adams and Rhodes 1960). 5.2.2 Mixing-Water Model Mixing of fresh and saline water in subsurface zones of coastal areas can also create conditions for dolomitization. This model is referred to as the Dorag Model (Badiozamani 1973). Dolomitization arises in the brackish water zones as seawater supplies the Mg ions, and dissolution of CaCO3 occurs as the two waters mix. The mixing of freshwater with saline water causes under-saturation of calcite, while increasing dolomite saturation. Calcite is replaced by dolomite in saline waters that have been diluted by freshwater by 5 to 30% (Badiozamani 1973). The chemical equation for this reaction is expressed as 2CaCO3+ Mg2+=>CaMg(CO3)2+ Ca2+ (Al-Awadi et al. 2009). Ratios of Mg/Ca ions in waters conducive for dolomitization range from normal seawater values of about 5:1 to as low as 1:1 (Land, 1973; Rodriguez-Blanco et al. 2015). These dolomites can be distinguished from hypersaline dolomites by the absence of evaporates. This process can also take place at spring heads that exchange saline and fresh water during changes in tides. As tides rise, saline water is forced into the aquifer system, creating a subterranean brackish zone. As low tide approaches, the saline water is displaced by the fresh water as the pressure of the denser saline water is reduced (Al-Awadi et al. 2009). 5.2.3 Seawater Subtidal Model The Seawater Subtidal dolomitization concept works by constantly flushing seawater through the sediment during the changing of tides. Seawater is forced upward and downward through Holocene carbonate mud during the rise and fall of spring tides (Folk and Land 1975).

71

As the saltwater passes through the sediment, it is consistently supplied with renewed sources of Mg ions. As the Mg ions replace the Ca ions, they are removed from the sediment. The accumulation of the Mg ions within the sediment causes dolomite to form (Xun and Fairchild 1987). This type of dolomitization has been reported to occur in Sugarloaf Key, Florida (Boggs 2006). Dolomitization has been suggested to occur during sea-level rise as ocean levels rise and cause a seawater subtidal environment to move landward (Tucker 1993). 5.2.4 Additional Models Additional models have been derived for environments that are also conducive to dolomite diagenesis, yet these environments are unrelated to this study area. Briefly, these models include the Burial Diagenesis Model, where buried diagenesis of dolomites can precipitate directly as cement or as replacements. The chemical process takes place as hot Marich hydrothermal waters are forced through porous limestone, replacing Ca ions with Ma ions (Wierzbicki et al. 2006). In Hydrothermal Dolomitization, water temperature is increased, creating a catalyst for dolomitization limestones along fault lines (Allan and Wiggins 1993). 5.3 Dolomite Formation at Fitch and J&J Hunt Sites The dolomite formation at J&J Hunt site appears to date to the Holocene. Yet, modern formation of dolomite is extremely rare (Al-Awadi et al. 2009; Arvidson and Mackenzie 1999; Braithwaite 1991; Folk and Land 1975; Lu 2008). The stratigraphic location of the dolomite layer is firmly in sandy clay. Evidence for high energy redeposition of the dolomite cobbles is absent. The sandy clay loam not only surrounds the dolomite layer but is also interspersed between the cobbles (see core Figure 22). The low energy sediments do not indicate an environment with high enough energy to have deposited limestone cobbles that have undergone subsequent dolomitization. This indicates that the carbonate source for dolomitization was

72

deposited congruently with the Holocene sandy clay loam. Examining the foraminifera and the faunal remains from Unit 5 at the Fitch site may give insight into the condition in which the dolomite formed. The remains of camel and horse do not limit the time frame of when the dolomite formed. American horses and camel are believed to have survived well into the Holocene (Meltzer 2015). The presence of sea cow, parrot fish, and bowfin however, does give an indication on environment before the dolomite formed. The bowfin is a freshwater species that lives in the muddy bottoms of slow-moving rivers (Ross 2001). The parrot fish serves as an excellent indicator of an environmental shift. Parrot fish are a marine species that feed on algae that grow on coral. They can also be found in brackish lagoons when they are young (Sime 2005). The Fitch site was most likely part of a marine estuary system prior to the dolomite formation, based on the faunal remains. As the sea-levels began to rise, the site became a brackish estuary (Arbuthnot 2002), creating an environment conducive to dolomite formation (Badiozamani 1973). The topography of the karst features at both Fitch and J&J Hunt sites may have created a protected lagoon during which hypersaline conditions may have existed in karst features. If a shallow marine lagoon overtook both locations, it would have created conditions similar to Wilson’s facies 7 and/or 8 (shelf lagoon or lagoon and tidal flats with limited circulation). Tidal flats with limited circulation conditions are conducive to dolomite formation (Wilson 1975). The samples lacked any evidence of evaportes such as gypsum of halite. This indicates that the dolomite was not formed in a sabkhas like conditions. The absence of evaportes supports the both the Mixing-Water and Seawater Subtidal Model. The presence of marine, brackish, and freshwater species at Fitch site further support the Mixing-Water Model. Determining the age of dolomite formation is troublesome when using faunal remains and superposition alone. The two most recent periods in which sea-levels were high enough to

73

create conditions where carbonates could form are the most recent Holocene interglacial and an interglacial that took place 135,000-105,000 years ago (Lambeck et al. 2002). The preceeding period where sea-levels were at the correct position to form carbonates at the two sites was 200,000 years ago. Lack of evidence of an unconformity and the Holocene sediment that directly over and interlays the dolomite layer does not support the formation of the dolomite as far back as 200,000 years ago. Direct dating methods must be applied to the dolomite samples to obtain a definitive time for its formation. 5.4 Thin Section Sample Descriptions Two dolomite samples were chosen from the core 01-02B from 100 to 110 cm section. The two samples were then cut in half using a table mounted rock saw (see Figures 23 and 24). Hydrochloric acid was applied to the cut surface with a negative result for effervescent reaction. A small amount of the sample was powdered, and the acid was reapplied, resulting in a minor

Figure 23. Dolomite Sample 01 from J&J Hunt Site.

Figure 24. Dolomite 02 from J&J Hunt Site.

Figure 23. Dolomite Sample 01 from J&J Hunt Site

Figure 24. Dolomite Sample 02 from J&J Hunt Site

74

Figure 25. Dolomite Sample 01 Under 10x and 20x Power Cross Nichols. effervescent reaction associated with dolomite (Al-Awadi et al. 2009). The two samples differed in both color and density and inclusions. Sample 01 was 10YR 7/2, light grey, in color and was denser with no faunal inclusions (see Figure 23). Sample 02 was 10YR 9/1, white, in color with a weaker structure and extensive inclusions of gastropods and bivalve remains (see Figure 24). The two samples were then sent to Wagner Petrographic and impregnated with clear epoxy and cut into thin sections. Half the samples were stained using calcite staining. The samples were then examined under 40x magnification for crystalline structure.

75

Figure 26. Dolomite Sample 02 Under 2.5x Power Plain Polar

Sample 01 consisted of blocky, rhombohedral shaped coarse crystals. The matrix in sample 01 consists of drusy calcite cement with organics and pyrite inclusions (see Figure 25). Allochems were completely dissolved with well-developed moldic porosity. Sample 02 also had drusy calcite matrix with abundant mollusk and foraminifera including the remains of bivalves, gastropods, evolute coils, milioids, uniserial, and peliods (see Figure 26). Several foraminifera were identified as Brizalina Translucens, Ammonia Parkinsoniana tepida, and Amphistegina gibbosa, which are found primarily in reef or estuary environments. Sample 02 resembles bindstone. The isopachous cement and the foraminifera indicate that the samples underwent diagenesis in the meteoric phreatic zone, in either a reef or estuary environmental setting.

76

5.5 Overview of U/Th Dating Carbonates such as corals or speleothems can be accurately and precisely dated using Useries disequilibrium methods. The uranium thorium disequilibrium dating technique is constructed based on radioactive decay of radionuclides within decay chains. The date of a sample is calculated by determining the degree of equilibrium that has been restored between parent material uranium-234 and daughter material thorium-230. The half-lives of uranium and thorium are 245,620±260 and 75,584±110 years, respectively (Cheng et al. 2013). Materials that are formed in a closed system in which parent-daughter disequilibrium is maintained, such as marine and terrestrial carbonates, can be precisely dated by U-series disequilibrium methods (Scholz and Hoffmann, 2008). Marine based carbonate organism such as corals, foraminifera, and bivalves and inorganically-precipitated sediments, speleothems, peat deposits, and dolomites can also be dated utilizing uranium series methods (Polyak et al. 2016; Rasbury and Cole 2009; Scholz and Hoffmann 2008). Coral species, such as elkhorn coral, have proven to be outstanding sources of sea-level proxy data (Toscano and Macintyre 2003). Elkhorn coral live at shallow depths of up to 5 m. If water levels become deeper, and the coral cannot keep pace with rising sea-levels, the coral will perish (National Marine Fisheries Service 2014). Uranium series dating is preferred over radiocarbon dating due to the unknown variability in prehistoric CO2 levels and the necessity of independent calibration (Edwards and Peltier 1995; Fairbanks 1990; Pirazzoli 1996). Unfortunately, many species of shallow water coral cannot survive in the waters of the northern Gulf of Mexico due to the increased particle content of the Northern Loop currents (National Marine Fisheries Service 2014). Dolomite that has been identified within secure stratigraphic context at all offshore submerged sites may be a viable alternative to elkhorn coral.

77

5.6 Dolomite Dating Methods U-series dating was conducted on two dolomite samples using a multi-collector multi collector-inductively coupled plasma mass spectrometer (ICP-MS) located at the University of Miami’s Ocean and Atmospheric Geology laboratory. The (ICP-MS) uses an inductively coupled plasma source, which removes electrons from the atoms creating positively charged ions. The ions are then focused into a beam by accelerating them utilizing a series of electrostatically charged plates. The beam then passes through an energy filter and a magnetic field where the ions are separated by their mass to charge ratio. The isotope ratios are then calculated by converting the beams to voltage and passing them through a collector (Anon 2017). Once the uranium and thorium ions are separated and counted, a date is obtained using the known decay rates of 245,620±260 for uranium and 75,584±110 years for thorium (Cheng et al. 2013). Two samples of dolomite collected from core 01-02B within the 115-120 cm subsample were processed for U/Th dating. Sample 01 was a piece of coral extracted from a cobble of dolomite. Sample 02 was a small piece of bivalve extracted from the center of a dolomite cobble. The samples were then washed in deionized water and pulverized using a mortar and pestle. The samples were then weighed using a microbalance. Sample 01 weighed 912.1 mg and sample 02 weighed 27 mg. The samples were then poured into 50-ml polypropylene centrifuge tubes and 5 milliliters (ml) of 6 molar (M) Nitric acid (HNO3) was slowly added to the samples. The tubes were then placed in a sonicator for 5 min until the samples had fully dissolved. Then .5 g of 229

Th-233U-236U spike mixture was added to the samples. The "spike mixture" artificially

enriches the isotopic tracer to determine concentrations and parent-daughter ratios of the uranium and thorium isotopes by dilution (Pourmand et al. 2014). The sample was then mixed in a centrifuge at 6000 RPM for 30 min. The samples were then analyzed using the ICP-MS.

78

5.7 Results The isotope ratios for the samples were established by the ICP-MS, yet the dates for the samples were inconclusive. The ideal ratios for U-series dating is 10:1 uranium to thorium. The high clay content surrounding the dolomite layer and the open nature of the uranium thorium system within the samples produced ratios for sample 01 that were 1:1.5 and 1:1 for sample 02. U-series dating was conducted on sample 02 due to the favorable uranium to thorium ratio. Unfortunately, the levels of thorium were too high for the test, and a date for the samples could not be produced (Ali Pourmand personal communication). Further testing of dolomite samples with less clay inclusions is required to obtain a viable date.

79

CHAPTER 6 THE TROUBLE WITH THE CURVE 6.1 Review of Balsillie and Donoghue (2004) Balsillie and Donoghue's (2004) sea-level curve has been the standard model for oceanic transgression in the Gulf of Mexico for over a decade despite differing from other Gulf of Mexico and global curves (Bard et al. 1990; Fairbanks 1989; Lambeck et al. 2014; Toscano and Macintyre 2003). This chapter addresses issues with sampling, dating, and analyzing sea-level proxy data and introduces new data and methodologies to enhance the Gulf of Mexico sea-level transgression model. This chapter presents a new Gulf of Mexico sea-level curve utilizing both linear regression and Bayesian modeling. Balsillie and Donoghue’s pivotal reconstruction of oceanic transgression over the Gulf of Mexico continental shelf has been a critical dataset utilized by researchers from a wide range of disciplines (Osterman et al. 2009; Paine et al. 2012; Shennan et al. 2015). Without their contribution, the Gulf of Mexico submerged landscape and predictability models would lack the localized data required to create models. Balsillie and Donoghue’s curve consisted of data points collected from 23 researchers over four decades. They “assumed that investigators involved in radiocarbon dating work have responsibly reported their findings” (Balsillie and Donoghue 2004:3), yet the standards in radiocarbon methodologies and reporting have drastically changed over four decades. Sea-level transgression models for the end of the last glacial maximum have relied on radiocarbon dating of proxy data collected from both secure and unsecure stratigraphic contexts. Using Pirazolli (1996) as a guide, their goal was to comprehensively engage the compilation of irregular sealevel datasets for the Gulf of Mexico to produce an accurate sea-level curve. A comprehensive

80

sea-level curve for the Gulf had been previously lacking due in part to the contradictory datasets by several researchers (Curray 1960; Fairbridge 1961,1974). Balsillie and Donoghue recognized the shortcomings associated with the data, including sampling methods, radiocarbon calibration reservoirs, and the type of sea-level indicator sampled. They devised three data editing procedures (one geological, the other two statistical) to reduce the number of equivocal sea-level proxy data samples. The geological editing involved simply selecting samples from locations that were considered tectonically stable. The statistical methods involved using a 75-meter acceptance envelope in which any data points that fell outside were considered outliers. The second utilized a 7-point floating average that averaged the first seven data points with every individual point within the data collection in the hope to reduce the noise within the curve. Balsillie and Donoghue elected to use 353 data points from areas of the Gulf of Mexico, while considering 134 data points from eustatic curves (Balsillie and Donoghue 2004). The radiocarbon dates were corrected using CALIB (Rev 4.4.2), correcting terrestrial and freshwater samples using IntCal98 and marine samples with Marine98. Twelve data points were identified as outliers and removed from the curve, resulting in 341 sea-level indicators to construct their curve. Balsillie and Donoghue used the Siddall et al. (2003) eustatic sea-level curve as a benchmark to compare their results. The Siddall et al. (2003) model was also a statistical curve derived from dates taken from δ18O analyses of foraminifera from Red Sea sediment. When the two models are compared in an overlay, ocean levels differ by 20 m, and seldom do they correspond or intersect. When the two models are compared to the more recent (Lambeck et al. 2014) curve, the depths differ between the two models by 25 m.

81

Figure 27. Fairbanks (1990) Barbados Overlay Balsillie and Donoghue (2004).

Figure 28. Bard (1990) Barbados Overlay Balsillie and Donoghue (2004).

82

Figure 29. Lambeck (2014) Eustatic Overlay Balsillie and Donoghue (2004).

Figure 30. Toscano and Macintyre (2003) Caribbean Overlay Balsillie and Donoghue (2004).

83

When the curve was compared to more local data, Toscano and Macintyre (2003) differs by 15 m, Bard and Fairbanks (1990) Uranium Thorium (TH/U) curve differed by 20 to 25 m, respectively. (see Figures 27-30) So, what is the trouble with the curve? 6.2 Radiocarbon Calibrations Issues Willard Libby discovered radiocarbon dating in 1949 and won the Nobel Prize in Chemistry in 1960 for his discoveries. Radiocarbon dating works by using the known decay rate of Carbon 14 isotopes. Cosmic radiation creates Carbon 14 isotopes high in Earth’s atmosphere by replacing one of nitrogen's protons converting it to carbon. The Carbon 14 is dispersed throughout the planet in the form of carbon dioxide. The carbon dioxide is taken in by oceanic and terrestrial plant life and redeposited throughout the global food chain. Thus, radiocarbon dating cannot be conducted on a non-living, non-organic sample because inorganic objects do not uptake Carbon 14. When an organism dies, it stops photosynthesizing or consuming plants and animals, ceasing the uptake and replenishment of Carbon 14. As Carbon 14 ceases to be replenished, the carbon begins to decay by emitting negatively charged electrons. During this process, Carbon 14 is converted back to nitrogen. Carbon 14 decays at a known rate with a halflife of 5730 years (Mook and Van de Plassche 1986:14). The amount of Carbon 14 remaining in a sample can be calculated and a date extrapolated for when Carbon 14 began to decay (Talma and Vogel 1993). Organic samples can be radiocarbon dated providing that enough organic material and Carbon 14 remains. Choosing a proper sample type in the field can determine how precise the resulting dates will be. Bones that lack collagen and plant material that have become fossilized cannot be radiocarbon dated with our current technology. Charcoal or large pieces of wood could be affected by the “old wood paradox”. The old wood paradox pertains to old pieces of wood

84

that are introduced to a location from trees that may have died hundreds to thousands of years earlier. For instance, wood that may have potentially been preserved for thousands to tens of thousands of years in a submerged context may be redeposited in a much younger stratigraphic unit. The resulting dates on wood samples maybe are hundreds to tens of thousands of years older than other samples within a stratigraphic unit. Extracting organic objects with a short life span, such as seeds, leaves, or twigs may avoid the old wood paradox. Some organic samples such as peats can also be contaminated with younger or older sources of carbon. Identifying the species of the sample is also important. Different species of plants use 3-Carbon acid (C3) or 4Carbon acid (C4) photosynthesis pathways, which uptake carbon at different rates. C4 plants intake Carbon 14 at a higher rate than C3 plants. For example, if one dated a piece of corn (C4) that died the same year as an oak tree (C3), without correction, the corn would appear to date hundreds of years older due to the excess Carbon 14 within the sample. The reservoir effect is an issue that concerns most sea-level proxy data that has been radiocarbon dated. The radiocarbon dating method is based on the circulation of Carbon 14 in the atmosphere. Some organisms acquire Carbon 14 from marine and freshwater sources that may significantly differ from atmospheric levels. For instance, apples snails that live along Florida’s karst river systems will assimilate "dead" or old carbon from chemically-weathered limestone. A modern snail shell can return a date hundreds, or even thousands, of years older than when they were living (Ludwig and Renne 2000). Radiocarbon dating marine organisms can present a similar issue. Shellfish incorporate carbon not from the atmosphere, but from the water. Oceans are a huge carbon reservoir, storing more "old" carbon than the atmosphere. Radiocarbon dating relies on the assumption that, in the past, marine and atmospheric environments were similar to modern conditions. Yet, levels of

85

carbon in the atmosphere and oceans fluctuated over time. Radiocarbon calibration curves have been developed for both atmospheric (IntCal) and marine (Marine) samples. Radiocarbon calibration has improved greatly over the decades, yet the controls for computing radiocarbon reservoir corrections fall short of addressing local changes in marine carbon environments over the last 50,000 years. Changes in atmospheric carbon dioxide levels and fluctuations in air to sea carbon exchange can vastly alter carbon intake in regional and local marine environments (Wagner et al., 2009). Marine organisms can date older by upwards of 400 years, making it crucial to calibrate the samples with the proper curve. Radiocarbon dating standard deviation or “error” is not distributed as a normal curve. Dates within the ranges of error have different probabilities values. Using the median of the error is an inaccurate portrayal of the date (Parnell et al. 2008). Median points are commonly used in the construction of sea-level curves if their range to 95% of the error is also represented (Bard et al. 1990; Lambeck et al. 2014; Pirazzoli 1996). Recent Bayesian chronological modeling now include the entire probability distribution within the models (Bronk Ramsey 2008; Parnell et al. 2008). Testing the hypothesis that the Balsillie and Donoghue curve simply needed to be recalibrated to IntCal13 and Marine13 was straightforward. Corrections to the radiocarbon calibration models has made marked advances in adjusting for carbon reservoir issues, making radiocarbon dating more accurate. Calibrating the data samples from IntCal98 and Marine98 to IntCal13 and Marine13 required identifying the type of material originally sampled, and then using the appropriate radiocarbon reservoir calculation. In most cases, the material that was sampled was listed in both Balsillie and Donoghue (2004) and the original research publication. In other cases, the material dated was neither reported in Balsillie and Donoghue (2004) or in the

86

original publication, as in Fairbridge (1961,1974). These data were excluded from the reevaluation due to the inability to assign the appropriate calibration curve to the samples. The remaining data were then calibrated via OxCal (Rev 4.3), utilizing the Marine13 calibration for marine samples and IntCal13 for terrestrial samples (Reimer et al. 2013). The data then were scatter plotted in Excel, and a 7-point floating average was applied using the same methodologies employed by Balsillie and Donoghue (2004). However, the resulting curve still included the differences in depth and time, as previously stated as compared to other global curves.

Figure 31. Gulf of Mexico Sea-level Curve Developed from Balsillie and Donoghue (2004). If the issues with the curve did not involve the calibrations, then it was possible that the sample data was the issue. Testing this involved taking a closer look at the data that were originally utilized in the construction of the 2004 curve. All publications employed by Balsillie

87

and Donoghue were scrutinized for sampling, dating, and reporting methodologies to ensure that the data being used would pass the modern sampling standards. 6.3 Sampling Issues with Balsillie and Donoghue (2004) 14

Balsillie and Donoghue state that, “While there is error associated with the C age dating methodology, the bulk of error is undoubtedly associated with the indicator material chosen to represent sea-level elevation” (Balsillie and Donoghue 2004:ix). Balsillie and Donoghue understood the nature of the level of error associated with sea-level proxy data, yet they fell short of devising methods to deal with these issues. The editing methods that they did employ were successful in identifying only 12 spurious data points. During a literature review of the original research, it was clear that many of the original authors had doubts about the validity of their data (Davies 1980; Kuehn 1980; Schnable and Goodell 1968; Schroeder et al. 1995; Shier 1969). Researchers such as Schnable and Goodell 1968 state that their peat samples could be contaminated with younger material. They further state that the oyster samples they collected could have been redeposited (Schnable and Goodell 1968). Similarly, Davies (1980) is concerned with younger mangrove roots contaminating the samples of peat that were collected. Kuehn (1980) excluded dates from her research due to an unconformity, yet those excluded data points were used in the Balsillie and Donoghue curve (Kuehn 1980). Researchers such as Schroeder et al. (1995) collected weathered and encrusted oyster samples from unsecure stratigraphic contexts on the surface. Shier’s (1969) samples did not report the radiocarbon error and a standard deviation of 150 was arbitrarily assigned to the samples, while other dates had to be “eyeballed” from core descriptions. Regardless of the stratigraphic security, some sea-level data are inherently complex and complicated sample sources, such as eastern oyster (Crassostrea virginica) (see Chapter 6.5).

88

Despite using three forms of editing in the construction of their 2004 Gulf of Mexico curve, Balsillie and Donoghue’s methods failed to identify spurious data that a literature review would have revealed. Creating a sea-level curve that is accurate to ±5 m requires a stringent method of editing the dataset for outliers. New methods for both sea-level reconstruction and dataset editing are presented in the following section. 6.4 New Curve Methods The reevaluated Gulf of Mexico sea-level curve collected and assessed over 650 samples (n=654) from 32 separate publications spanning five decades of marine biological and sea-level research (see Apendix B). All data were calibrated via OxCal (Rev 4.3), utilizing the Marine13 calibration for marine samples, and IntCal13 for terrestrial samples (Reimer et al. 2013). The median date was utilized as the single data point on the curve (A. Bayliss personal communication Feb. 2017). Error bars were assigned to each sample representing 2 sigma standard deviations for median dates with 95% probability distribution and vertical bars for ranges in depth. The error bars allow researchers to fully evaluate the data being utilized including the possible ranges in time and depth in association with the linear curve (Pirazzoli 1996). All calibrated data were entered into an Excel spreadsheet and placed into descending order according to date. The samples were then evaluated for spurious data and outliers based on a scoring system developed by Pirazzoli (1996). Samples that received a 5 were considered excellent, followed by 4 that may suffer from inaccuracies from reservoir calibrations. Samples that were rated 3 may be stratigraphically secure but may experience contamination from younger or older carbon. A rating of 2 was assigned to samples that may have undergone translocation, redeposition, or were poor environmental indicators. Samples that may have a combination of the aforementioned conditions, and/or were underreported by modern standards

89

(Bayliss 2015), excluding vital information for reevaluation, were assigned a rating of 1 (Graf 2009). Fairbridge (1974) neglected to publish both the material sampled or the standard deviation associated with the radiocarbon dates (n=51). Due to the inability to accurately calibrate the sample dates, Fairbridge (1974) was excluded from this research. In locations where the landmasses are tectonically stable, such as Florida’s 8.7 meters of uplift over 1.6 million years (Willett 2006), the range of vertical movement was considered negligible and not factored into sample depth. In areas that are considered relatively tectonically stable, such as Barbados’ 34 centimeters (cm) per 1000 yr uplift, the difference in vertical displacement was factored into the sample’s reported depths (n=206) (Fairbanks 1989). 6.5 Rating System Coral species, such as elkhorn coral (Acropora palmata), have proven to be outstanding sources of sea-level proxy data (Toscano and Macintyre 2003). Elkhorn coral samples that have been utilized in this research include not only horizontal error bars, but also vertical error bars associated with their survival depth of 5 m. Researchers working with sea-level curves have begun to use uranium series dating of coral to avoid the inaccuracies associated with radiocarbon dating (Abdul et al. 2016; Bard et al. 1990; Brock et al. 2008; Fairbanks 1990; Toscano and Macintyre 2003; Toscano 2016). Uranium series dating should be preferred over radiocarbon dating due to the unknown variability in prehistoric CO2 levels and the necessity of independent calibration (Edwards and Peltier 1995; Fairbanks 1990; Pirazzoli 1996). Using coral for sea-level proxy data has its drawbacks. Redeposition of corals can introduce inaccuracies into sea-level curves. Translocated corals can be washed up or down slope and redeposited at different levels relative to their growth locations. Core samples must be thoroughly inspected to avoid redeposited corals. Sampling method for researchers working with

90

corals were scrutinized for evidence of redeposition before being included within this research. Coral samples that were recovered in secure stratigraphic context and had undergone uranium series dating were valued higher than samples dated by radiocarbon. Corals that were dated using uranium series were given a rating of 5 and those dated with radiocarbon were given a rating of 4. In the special case of Fairbanks (1990), identical samples were dated using radiocarbon and Th/U230 (n=20) resulting in two different dates. Both reported dates were utilized in the curve with the notion that once a moving average was applied, the trendline would split the average of the two dates. Samples of wood or peats were given a mid-range rating due to the “old wood problem” (Olsen et al. 2013; Talma and Vogel 1993) and contamination and compaction, respectively. Contamination of younger materials and compaction can also occur during the formation of peat and mangroves (Mook and Van de Plassche 1986; Turetsky et al. 2004). As mentioned above, Davies (1980), who was utilized in the Balsillie and Donoghue curve, expresses the concern that the peat and mangrove samples collected were contaminated with younger material. Peat samples received a score of 3 and were scrutinized when discovered in an outlier position. Oyster has been utilized by researchers as a proxy for prehistoric brackish conditions and an indicator for coastline proximity despite its tolerance for vast shifts in salinity and depth. Oyster can grow anywhere from miles up rivers to benthic depths greater than 30 meters (98 feet) (Barnes et al. 2007). The limits for species survival are ultimately determined by salinity and the availability of substrate to attach to. Adult oyster can tolerate salinities ranging from 0– 42.5 parts per thousand (ppt), with normal distribution occurring between 5 and 40 ppt (Lorio and Petrone 1994). The optimum salinity for growth and reproduction is between 10 and 28 ppt

91

(Barnes et al. 2007). Larvae will not metamorphose into spat when salinity is less than 6 ppt (Wilson et al. 2005), and adult oyster can endure indefinitely in salinities up to 35 ppt (Buroker 1983). Under modern conditions in Florida, oyster have been identified living 14 km inland and 7 km offshore (Florida Fish and Wildlife Conservation Commission 2017). This presents an issue when determining where the coastline was relative to the recovery location of the oyster being dated. The Gulf of Mexico was a much different body of water during the terminal Pleistocene and influxes of meltwater into the Gulf significantly changed both temperature and salinity (Flower and Kennett 1995). Flower and Kennett's (1995) study of planktonic foraminiferal assemblages from Orca Basin cores (EN32-PC4 and EN32-PC6) gives a detailed look into what types of changes took place in the Gulf of Mexico as it pertains to temperature and salinity throughout the last deglaciation. Flower and Kennett's (1995) cores were taken 290 km south of the Mississippi delta near the edge of the continental shelf. Globigerinoides ruber is a species of plankton that can survive in lower salinity sea-water more successfully than other planktonic species. The plankton can tolerate salinity levels as low as 22 ppt, where many other forms of plankton cannot survive drastic drops in salinity below 36 ppt. (Bijma et al. 1990). At 16,000 cal BP, warm-water foraminifera began to become more prevalent as cold-water plankton species were replaced in response to early deglaciation (Flower and Kennett 1995). G. ruber populations exploded during melt-water pulse 1A (MWP 1A), reaching their maximum proliferation from 15,500 to 12,500 cal BP, with its peak population at the maximum meltwater discharge at 13,500 cal BP. G. ruber were ubiquitous in the Gulf of Mexico throughout the low-salinity conditions during the meltwater influx from 16,000 to 12,800 cal BP. MWP 1A was suddenly interrupted by an episode of higher salinity and cooler sea-surface temperatures associated with the Younger Dryas

92

at 12,500 and continued until 11,200 cal BP. The cooling and increase in salinity is marked by a reduction in G. ruber populations and the reappearance of cold-water, higher salinity foraminifera. G. ruber increased again with MWP 1B from 11,200 to 10,700 cal BP, where afterwards, the plankton populations stabilized to modern Holocene levels (Flower and Kennett 1995). G. ruber made up on average about 35% of the planktonic foraminiferal assemblages during the last glacial period. At the onset of deglaciation, maximum population densities increased to 70%, followed by a rapid decrease to 30% after the meltwater influxes. The populations then increase slightly into the late Holocene to modern populations. This indicates that the salinity in the Gulf of Mexico was much lower during the terminal Pleistocene/Holocene transition, which may have afforded eastern oyster free rein over the continental shelf.

Table 4. Eastern Oyster Samples from Balsillie and Donohue Dataset.

The Balsillie and Donohue sea-level curve data contained 36 samples of oyster, with roughly half of those samples originating when salinity in the Gulf was extremely low. Given the

93

proper substrate to which they could attach, oysters could have populated regions hundreds of kilometers from the coastline. The Schroeder et. al. (1995) samples may represent this phenomenon. Examining these data indicate that sea-levels increased only 9 m (-40 to -31 m) between 18,000- 10,200 cal BP (Balsillie and Donoghue 2004). During the same period, Lambeck et al.'s (2014) eustatic sea-level increased by 75 m, Siddall et al. (2003) by 72 m, and Balsillie and Donoghue (2004) by 87 m. Shepard (1960) also contains samples of oyster that statistically date to the same age, yet they differ in water depth by over 20 m (see Table 4). Oyster data should be considered a poor coastline indicator and sea-level proxy data due to fluctuating salinity in Gulf of Mexico, as well as the vast ranges of habitats. Samples of shellfish, including eastern oyster, received a rating of 2 due to their range in habitat, likelihood of redeposition, and uncertainty involved in radiocarbon calibration. Table 5. Sample Rating System. Rating

Sample Type

Dating Method

5

Elkhorn Coral

U/Th

4

Elkhorn Coral

Radiocarbon

3

Wood, Peats, Mangroves

Radiocarbon

2

Shellfish, Foraminifera

Radiocarbon

1

Various Material

Batch Radiocarbon

Dates that were a result of bulk or batch testing were also given a lower score than uranium series and radiocarbon accelerated mass spectrometry (AMS) samples. When radiocarbon dating was first established, dates were obtained from counting the beta emissions with an ionization detector. Since the 1980s, accelerator mass spectrometry (AMS) has been 94

used to count the remaining Carbon 14 isotopes, which has vastly improved accuracy and the size of the sample required for testing. Batch testing increases the chances that the sample has been contaminated with younger or older material, yielding an inaccurate date. It is no longer necessary to conduct bulk or batch sampling to collect the required amount of organic material to conduct radiocarbon dating. For this reason, all batch samples from Balsillie and Donoghue (n=42) were gave a rating of 1 for consideration in the new curve. To determine which data points were outliers, samples that varied in depth by more than 10 m from the surrounding eight samples, four lower and four higher, were removed. The rating system assisted in determining which data may have been outliers due to sample methodologies, versus a local sea-level transgression phenomenon. Of the 603 samples evaluated, 425 were utilized in this research (71% of the collected data [see appendix]). While editing reevaluated data, samples of oyster (n= 41) were considered for the new sea-level curve, and 76% were deemed outliers. Balsillie and Donoghue’s original research consisted of a statistical envelope of 75 m in which data were considered viable. Editing the data in this manner allowed spurious samples to remain in the data collection, vastly affecting the trendline within the 75 m envelope. In addition, the difference of 75 m of sea-level rise in a low gradient continental shelf environment, such as Florida’s western shelf, which ranges from 0.2 to 4.0 m of slope per km (Davis 2017), renders the curve ineffective for locating paleo-shorelines. This research strived to tighten the outlier envelope to ±5 m accuracy by effectively assessing the data for errors associated with sampling methods. Sea-level curves have been constructed by gathering proxy data and determining the sample’s depth and date. The samples were scatter plotted on an X,Y graph and a trendline was applied to represent the transgression event (Pirazzoli 1996). Sea-level curves have been created

95

Figure 32. Joy Gulf of Mexico Sea-level Curve with Environmental Indicators. using this method over several decades (Bard et al. 1990; Balsillie and Donoghue 2004; Fairbanks 1989; Lambeck et al. 2014; Pirazzoli 1996). Plotting data points using this method

96

disregards the environmental indicator the samples represent. Samples that represent terrestrial environments that predate the transgression event have the same power to move the trendline as samples that represent post-transgressive environments. Plotting the data without considering the environmental indication will influence the trendline with equal value regardless of the environment in which the sample was deposited. In other words, terrestrial wood and marine coral samples may have the same date, but drastically differ in depths, pulling the trendline in opposite directions for the same time period. This may have been the cause of the “noise” reported by Balsillie and Donoghue (2004). The reevaluated curve retained the environmental depositional data by plotting the samples in groups according to environment. Using this method eliminated the need to place a 7point floating average on the majority of the trendlines. This method also allows researchers utilizing the curve to properly evaluate the data and identify when the LGM transgression took place. Coral samples were plotted separately and represent the earliest period at which the ocean transgressed. Brackish samples represent the closest actual time and depth at which sea-levels transgressed, followed by freshwater peat and terrestrial samples which represent the upper limits of oceanic levels. 6.6 New Curve Results The goal of this research was to create a Gulf of Mexico sea-level curve that is accurate enough to predict the location of the coastline throughout the last 22,000 years. Utilizing the new method of editing outliers and separating environmental sample types, the reevaluated curve achieves this goal of ±5 m accuracy. Samples from the first 8,000 years (22,000 to 14,000 cal BP) of the curve are coral samples that represent the lowest possible location of sea-levels. At 22,000 cal BP, sea-levels are approximately -125 m below modern sea-levels (bmsl). From

97

22,000 to 17,000 cal BP ocean levels begin to steadily rise to -110 m bmsl. From 16,800 to 14,700 cal BP there is a gap in sea-level data. This gap precedes MWP 1A, making the exact

Figure 33. Coastlines and Area Submerged by Millennia (with contributions from CookHale). timing of the start of the pulse unclear. By 14,700 cal BP, water levels have risen to -94 m bmsl and continue to rapidly rise to 13,800 cal BP to -70 m bmsl. The transgression rate slows after MWP 1A and the onset of the Younger Dryas, yet levels continue to rise from 13,800 to 11,200 cal BP to -60 m bmsl. MWP 1B can be distinctly identified at 11,200 where sea-levels rise to -40 m bmsl in a 400-year period. Sea-levels slow to a steady increase from 10,800 cal BP through 8,400 cal BP to -18 m bmsl. At 8,200 cal BP, MWP 1C surges sea-level rise to -8 m bmsl until

98

by 7,000 cal BP. MWP 1C is the last of the melt water pulses, as sea-levels progressively rise to modern levels at 2,500 cal BP (see Figure 32). 6.7 High Stands and Low Stands The evidence for Holocene Gulf of Mexico high stands has been a postulated by several researchers (Morton et al. 2000; Otvos 2004). The reevaluated curve does indicate a high stand between 6,300 to 4,700 and 2,000 to 1,200 cal BP, yet this may be a result of issues within the sampling. The new method of separate environmental samples illustrates that all the samples that indicate high stands are shellfish or foraminifera from the western Gulf. The lack of high stand indicators, such as marine peats or corals, could indicate that the shellfish have been redeposited, giving the facade of a high stand. Proving the presence or absence of a Gulf of Mexico high stand is beyond the scope of this research. There appears to be a low stand centered at 4,000 cal BP which is indicated by both the coral and the shellfish datasets. The shellfish can be problematic for the aforementioned reasons. Four of the seven more trustworthy elkhorn coral samples collected during this period were living well below the modern 5 m water mark. The archaeological record my also reflect the 4,000 cal BP regression event. Several coastal sites in Florida were abandoned during the same period of the regression (Russo 1988). This abandonment may have been a demographic shift moving closer to marine resources on the newly re-exposed coastal landscape (Sassaman et al. 2016).

99

6.8 Bayesian Analysis of Gulf of Mexico Transgression Event The construction of sea-level curves relies on plotting samples of varying environmental settings to indicate whether sea-levels were present at a given location in the past. Although the samples do not represent the transgression event, they do represent an environmental indicator. Using Bayesian statistics, core samples can be bracketed to determine the depth and age associated with an event such as the transgression (Parnell et al. 2008). This method considered samples at several known depths and their corresponding ages. Between these samples are depths with unknown dates which may include the event being researched. The statistical model can constrain the number of probable dates between two known depths and ages, reducing the amount of uncertainty in the range of dates between the two knowns (Parnell et al. 2008). The model also addresses uncertainties involved in sea-level modeling and has the ability to identity and reject outliers. 6.8.1 Bchron: Age-Depth Modeling Including the uncertainties of sea-level proxy data is crucial for the development of sealevel curves (Parnell and Gehrels 2015; Pirazzoli 1996). Uncertainties include: ranges in habits for samples, tectonic instability, and age uncertainties of the sea-level proxies. These uncertainties are amplified when proxy data have undergone antiquated sampling and dating methods. Parnell and Gehrels (2015) have developed a statistical model (Bchron) to estimate the uncertainty in the chronology to predict rates of sea-level rise. Bchron is a non-parametric chronology model designed to estimate unknown age/depths within sediment cores using a Compound Poisson-Gamma model (Haslett and Parnell 2008). Bchron uses a modified Markov chain Monte Carlo algorithm which converges known data

100

points to make probability estimations for sections of cores without know dates. The algorithm has also been modified to identify outliers. Bchron functions by converting the Law of Superposition into a mathematical principle know as monotonicity. Monotonicity is a function of order that only increases or decreases (Royden and Fitzpatrick 2010). The Parnell and Gehrels (2015) chronology model relies on the depths and dates obtained from sediment cores collected from salt marshes, in combination with an estimation in sedimentation rates. The aggradating nature of low energy salt marsh sedimentation during increases in sea-level conform to a monotonic function. Utilizing the Bchron model, dated strata within the core can be used to determine the probability distribution of the estimated age of undated strata (Parnell and Gehrels 2015). The monotonic restriction allows the probability distribution of date ranges to be further confined by eliminating probability ranges that pre- or post-date samples lower or higher in the sediment core. Using this method also increases the ability to identify outlier dates within the series (Parnell and Gehrels 2015). The Balsillie and Donoghue and the reevaluated sea-level curves both utilize the mean point of the dates’ probability distribution to plot the time position within the graph. This method is problematic because the probability distribution of radiocarbon dates is not normally distributed. Utilizing chronology models, such as Bchron, factors in the actual probability distribution of the calibrated radiocarbon dates into the model creating a more precise range for the chronology (Telford et al. 2004). Creating a precise chronology for radiocarbon dates within the dataset is the first step in addressing the uncertainties within the sea-level curve. Using a standard linear regression model, or in the case of the Balsillie and Donoghue curve, a 7-point floating average, disregards the

101

menagerie of uncertainties associated with errors in sampling, date calibration, or the nature of sea-level rise. Creating a single, overly precise, trendline representing the sea-level curve can create misleading results. Approaching sea-level curves by accounting for errors in variables creates a more realistic range for the rate in sea-level rise. This method will reduce the accuracy of a trendline, yet increase the precision by accounting for uncertainties (Parnell and Gehrels 2015). New editing methods utilized in this reevaluated sea-level curve has reduced the level of “noise” in the Balsillie and Donoghue curve, revealing a monotonic pattern for a majority of sealevel rise in the Gulf of Mexico. The statistical principles used by Parnell and Gehrels (2015) for core sample should directly translate to the dataset utilized in this new sea-level curve. To reduce the complication in uncertainties with the dataset, coral proxies (n=290), averaging one sample every 40 cm, were used to test the viability of the model on a regional scale. As discussed in Section 2.4, elkhorn coral have a constrained, predictable habitat range and can be dated using more reliable dating methods such as U-series dating. The result of the Bchron model reflect those of the reevaluated sea-level curve, through the Terminal Pleistocene into the Early Holocene (see Figure 35). The higher variability in sea-levels, including a possible low stand at 4,000 cal BP, were not detected in the resulting model due to the non-monotonic pattern of the data. 6.8.2 Methods Bayesian modeling creates a probability distribution parameter of unknown data based on known data points. The probability distribution is referred to as the “posterior distribution”. Bayesian statistical chronology modeling considers the sedimentation rate, known dates, and the identification of outliers to create posterior distribution for the unknown ages of the sediment at

102

all depths (Parnell and Gehrels 2015). Bchron also includes the uncertainties associated with radiocarbon ages. Bchron utilizes Bayes’ theorem through the relationship “posterior is proportional to likelihood times prior” or posterior ∝ likelihood X prior. Mathematically this

relationship is represented with parameters written as θ and data as x. P(θ|x) ∝ p(θ| x) p(θ)

where p() is probability density, and the vertical bar C means “given” (Parnell and Gehrels 2015). Here x is adjusted to combine the radiocarbon age of a sample with standard deviation σ and associated calendar age 𝜃𝜃. This calibration curve is written as r(𝜃𝜃) which produces a

radiocarbon age from the calendar age (Parnell and Gehrels 2015). The likelihood is expressed as: x|θ~N(r(θ), σ2) When utilizing datasets with multiple dates, xi, i = number of radiocarbon ages, (x1,, x2, x3…n), with the standard deviation σi and calendar age 𝜃𝜃I (Parnell and Gehrels 2015). The

likelihood now becomes a product of normal densities:

xi ~ N(r(𝜃𝜃1i),), σ2),p(x1,, x2, x3…xn| 𝜃𝜃1, 𝜃𝜃2, 𝜃𝜃3… 𝜃𝜃n)=∏p(xi| 𝜃𝜃i) Next, the depths must be represented as continuous positive values d. Depths (d) are represented with the associated radiocarbon and calendar ages, x(di) and 𝜃𝜃 (d). The distribution represents the monotonic nature of the aggrading sediments where the depth d is a sum of positive random variables (Parnell and Gehrels 2015): N(k)+1

(d)- (d+k)=∑ gi I=1

103

k is a positive value, N(k) is the number of depositional units in the depth range (d, d + k), and gi is a positive random variable. The equation expresses that the age difference between depths d, and d + k is the sum of a number of aggrading events each of varying thicknesses. In Bchron, gi can vary corresponding to a Gamma distribution, whereas in OxCal it is a fixed value (Bronk Ramsey 2008). The rate of the Poisson distribution and the additional parameters within the gamma distribution are deemed supplementary parameters and estimated as portion of the posterior distribution (Parnell and Gehrels 2015). The full prior distribution model for Bchron is expressed as:

p(𝜃𝜃(d), ϕ|x(d1),…,x(dn))

n

n-1

i=1

i=1

∝∏p(x(di)|𝜃𝜃(di))×∏p(𝜃𝜃(di)-𝜃𝜃(di+1)|ϕ)×p(ϕ)

Here, ϕ are the constraints governing the deposition rate (the Gamma and the Poisson rate) required for the prior distribution. The chronological models are then derived from Markov chain Monte Carlo algorithm that samples the probability distribution. Estimates are then made of the calendar ages at varying depths where the ages were previously unknown. The algorithm guesses the values for the calendar ages for the deposition, then it proposes a new set of values for calendar years for the same depositional event. The new values are calculated to establish the likelihood that they match the prior values. If they do, then the values are accepted. If they do not match the likelihood, then they are accepted with the probability given the ratio of likelihood multiplied by the prior for the older values (Parnell and Gehrels 2015). New values that fall outside the acceptance region are considered outliers, and the values are rejected (Andrieu et al. 2003). Iterations of this process can be run for upwards of a million times to ensure convergence. The results of the chronology model are then plotted with a 95% confidence interval for the depths and ages.

104

Parnell and Gehrels model collected dates from a salt marsh sediment core to extrapolate sedimentation rates and to verify the monotonic pattern in deposition. Using the known dates and the sedimentation rates, Bchron can calculate “chronologies” that include predictions for ages on undated sections of the core while detecting outliers within the dataset (Haslett and Parnell 2008). For the purposes of the model, sediment accumulation rates were replaced with the increase in water levels. The coral dataset was utilized in the Bchron chronology model for several reasons. First, the thicknesses of sediments between dates must be used in the creation of the Bchron chronology model. To create a chronology from the Gulf of Mexico sea-level data, the thickness of a sea-level rise must be used in the calculations. Determining the increase in water levels using peats, wood, or shellfish would be problematic. The constrained habitat range of elkhorn coral of 5 m is more ideal for estimating thickness of the increase in water levels. Second, many of the samples were U-series dated and do not require calibration. This simplifies the Bchron likelihood algorithm by eliminating the need for the program to calibrate the samples (Haslett and Parnell 2008). The unedited coral dataset (n=290) was utilized for the chronology model. Arbitrary identification numbers were assigned to the sample as required by the Bchron program. Both uncalibrated radiocarbon and U-series dates and their associated errors were entered into an Excel spreadsheet. A thickness of 4 m was assigned each sample, subtracting 1 m from the elkhorn coral habitat range due to regional tide fluctuations. Samples that were radiocarbon dated were assigned a Marine13 reservoir calibration while U-series dates were entered as “normal” and did not receive calibration. The file was then saved as a CSV. file. The statistical program R was used to run Bchron. The following code was used to load the coral CSV. file and to run the Bchron program. (see Appendix A for Bchron R code)

105

6.9 Results The results of the Bchron chronology curve utilizing the unedited coral dataset was remarkably close to the reevaluated sea-level curve (see Figure 35). The convergence test for the model had a mean rate P-value of .49 out of .50. Bchron also has an output function that presents the quantiles of predicted ages by depth (see Table 6). Areas where the two models differed may have been a result of lowstands that may have broken the monotonic pattern required for Bchron to work properly. There appears to be a low stand indicated by both coral and shellfish datasets from 4,000 to 3,500 cal BP. Other areas such as 14,700 to 16,700 cal BP lack data points altogether, yet the Bchron model was able to fill the missing data to within a 5 m depth range. The Bchron model also validated the new editing methods utilized in the reevaluated curve for identifying outliers. The results of the chronology model show that a sea-level curve utilizing linear regression can be a viable method, provided that the dataset has been edited for outliers, yet it surpasses a linear regression model when large areas of data are missing. The new reevaluated sea-level curve developed during this research is capable of locating paleo-coastline to within ±5 m of water depth. The new editing methods, both using the rating system and Bchron statistical model, increases the identification of spurious data points within the dataset by 93% versus the methods used by Balsillie and Donoghue (2004). Utilizing statistical models and linear regression, this research improved paleo-coastline probability models by 75%.

106

Table 6. Quantiles of Predicted Ages by Depth Depth in Meters 0

2.5%

10%

50%

90%

99.7%

2,946.9

3,399.8

4,324

4,511.1

4,593

10

6,679.8

6,747.9

6,899

7,036

7,166

20

7,851

7,923

8,064

8,183

8,314

30

9,266.9

9,315

9,422

9,545

9,630

40

10,428

10,465

10,522

10,574

10,640

50

11,037

11,079

11,149

11,237

11,336

60

11,631

11,667.9

11,760

11,852

11,970

70

12,924

12,958.9

13,043

13,102

13,151

80

13,789.9

13,826.9

13,887

13,928

13,974

90

14,116

14,157

14,257

14,373

14,528

100

15,027

15,173

15,486

15,807

16,193

110

16,827

16,975.9

17,272

17,567

17,838

120

18,722

18,805

18,948

19,068

19,253

130

19,355

19,460

19,693

19,970

20,274

107

Figure 34. Bchron 95% Chronology for 290 Unedited Coral Samples

108

Figure 35. Joy Gulf of Mexico Sea-level Curve with Bchron Overlay.

109

CHAPTER 7 CONCLUSION Nearly 40% of the modern global population currently live within 100 km of the coastline (Neumann et al. 2015). Lower than modern sea-levels spanning 130,000 years extended global coastal plains onto the continental shelves, increasing habitable land by 13% (168 million km2) during maximum regression (Lambeck et al. 2002). The use of the coastal environment for subsistence ranges far into human antiquity. The coastal plain environments have played a role in both the physical and cultural development of modern humans (Marean 2010), yet archaeological research on the submerged landscapes is still in its infancy. The global archaeological record is missing vital information concerning the cultural development and migrations of human populations in coastal environments during the last glacial period (130,00021,000 cal BP). Answering vehemently debated questions such as ‘Who were the first Americans?’ and ‘How did they arrive on the continent?’ are further complicated by the lack of research on such an important landscape. This thesis is designed to create research models for the exploration and identification of submerged Paleoindian coastal sites in the Gulf of Mexico to begin filling in the gaps within the archaeological record. Predictive modeling of submerged coastal Paleoindian and Early Archaic (17,000-7,000 cal BP) sites on Florida’s western continental shelf requires precise locations of paleo-coastlines, a thorough understanding of coastal geological processes, the ability to identify fluvial and karst features in sub-bottom profile data, and knowledge of local cultural history (Faught 2018). In Chapter 2, a review of global coastally and non-coastally adapted sites was conducted to better identify submerged coastal sites. The analysis of lithic material and subsistence strategies indicate that many coastally adapted sites contain lower amounts of lithic

110

material, fewer complex lithic tools, and extensive shellfish remains. The coastally adapted groups frequently ranged between 10-12 km from the coastline but could travel upwards of 30 km in harsh environments. This model was designed based on just a handful of unequivocal sites from around the globe. This model could be enhanced if a more comprehensive review was conducted of available coastal Pleistocene sites. Chapter 3 reviews the known submerged sites in Apalachee Bay in Florida. The sites contain significant amounts of cultural material spanning the Paleoindian through Woodland periods. The cultural material at the majority of the sites represent upland habitation patterns during periods when coastline would have been 10s to 100s of km from the region. The Econfina Channel site is the youngest of the reviewed sites and represents a coastal site. The shallow depth of water at the site has led to repeated disturbances during inclement weather conditions. The disturbances have conflated the cultural material, making the site’s interpretations a difficult task. Coastal geological processes discussed in Chapter 4 give researchers insight into the taphonomic processes of submerged sites and the extent of disturbances involved in the transgression. Many of the offshore sites had high levels of disturbance in the first 15 cm of sediment. Many of the artifacts recovered at the sites were in this layer. However, artifacts were recovered in lower undisturbed layers and may represent artifacts in primary context. More excavations at sites such as J&J Hunt and Fitch should be conducted primarily to further understand the depositional chronology at the sites, but to also further understand the taphonomy of the sites. Chapter 5 explores the formation of a geological pavement of dolomite identified at several submerged sites in Apalachee Bay. Using U-series dating [though unsuccessful in this

111

research] may prove to be an invaluable chronological maker in future research into submerged sites in the northern Gulf. Quaternary-aged dolomite formation is very rare, and only a handful of locations where it forms have been identified. The dolomite in Apalachee Bay presents a rare opportunity for geologists to solve the modern dolomite formation problem. Further research should be conducted in dating the formation of the dolomite by either directly dating samples sheltered from the contamination of surrounding clay or by establishing a relative date by dating the freshwater layer under the dolomite. Chapter 6 presents a new reevaluated sea-level curve capable of locating paleo-coastline to within ±5 m. Utilizing the editing methods outlined in Chapter 6 decreases the chances of including spurious samples into the dataset by 93% versus the methods utilized by Balsillie and Donoghue (2004). By using both linear regression and statistical models, the reevaluated curve improved paleo-coastline probability models by 75%. Using statistical age/depth modeling is an outstanding compliment to constructing sea-levels curves and is invaluable when data points are missing from large time swaths in the dataset. Despite the assistance for age/depth modeling, more elkhorn coral samples should be collected from 17,000-15,000 cal BP (110-100mbsl) shoreline to fill in the gap in the data. The 4,000 cal BP regression is also a phenomenon that should be further investigated and may prove vital to understanding demographic changes in coastal regions in Florida during this time period. Evidence of this regression was recently noted during excavation at Manasota Key site off the coast of Venice Beach. Divers noted a shift in the sediment from brackish sediments to marine, then back to brackish sediments, before permanently shifting to marine sediments (Ryan Duggins personal communication July 2017). Extracting cores from submerged offshore

112

sinkholes in the future would assist in understanding both the timing and extent of this regression. The crux of this thesis is to create a comprehensive dataset designed for researchers to locate, identify, and investigate submerged coastal Paleoindian sites. Using these models, researchers can now hone in on offshore locations where coastal sites may be identified. Focusing on locations where coastal regions were quickly transgressed during melt water pulses contain the highest probability of intact sediments and minimally disturbed sites. These regions fall within a 60 km swath that ranges from 100 to 40 m in depth. It is imperative to collect high resolution geophysical data, including multibeam and subbottom profile data, within these regions to identify potential archaeological targets. Statistically, fifty percent of Paleoindian sites in Florida have been discovered within 1.2 km of a river or stream (Duggins 2012). The sub-bottom data should be analyzed to identify submerged paleochannels and river convergences near paleo coastal areas. Remotely operated vehicles can explore these potential targets. Where cultural material is identified, divers can recover the artifacts at these depths which are well within technical diving limits. Until global submerged coastal plains are investigated for Pleistocene and Early Holocene archaeological sites, the story of human history will be missing a vital part of the narrative. Littoral landscapes have played such a major role in the Holocene archaeological record since sea-levels stabilized. Archaeological surveys on the submerged landscape will begin to answer questions about human interaction with the coastlines during the largest sea-level increase in modern human history. These answers lay waiting on distant shorelines, currently drowned on the continental shelf.

113

REFERENCES CITED

Abdul, N. A., R. A. Mortlock, J. D. Wright, and R. G. Fairbanks 2016 Younger Dryas sea level and meltwater pulse 1B Recorded in Barbados reef crest coral Acropora palmata: Younger Dryas Sea Levels and MWP 1B Barbados Corals. Paleoceanography 31(2):330–344. DOI:10.1002/2015PA002847. Al-Awadi, Mishari, William Clark, William Ray Moore, Michael Herron, Tuanfeng Zhang, Weishu Zhao, Neil Hurley, Djisan Kho, Bernard Montaron, and Fadhil Sadooni 2009 Dolomite: Perspectives on a Perplexing Mineral. Oilfield Review 12(3). Andrieu, Christophe, Arnaud Doucet, Nando de Freitas, and Michael I. Jordan 2003 An Introduction to MCMC for Machine Learning. Machine Learning 50:5–43. Anon 2017 Multicollector-Inductively Coupled Plasma Mass Spectrometer (MC-ICPMS). Techniques https://serc.carleton.edu/research_education/geochemsheets/techniques/MCICPMS.html, accessed March 25, 2018. Anuskiewicz, R.J. 1988 Preliminary Archaeological Investigations at Ray Hole Spring in the Eastern Gulf of Mexico. Florida Anthropologist 41; 181–185. Anuskiewicz, R.J., P.R. Gerrell, J.S Dunbar, and J.F. Donoghue 1994 Evidence of Prehistoric Man on the North American Continental Shelf from a Sinkhole Located 32 Km. Offshore in 12 m Seawater. 59th Annual Meeting of the Society for American Archaeology, Anaheim, CA. Arbuthnot, Michael 2002 A Paleolandscape Reconstruction of the J&J Hunt Site (8JE740) and Its Surrounding Area. Unpublished Master’s, Thesis, Department of Anthropology Florida State University, Tallahassee, Florida. Arvidson, Rolf S., and Fred T. Mackenzie 1999 The Dolomite Problem; Control of Precipitation Kinetics by Temperature and Saturation State. American Journal of Science 299(4):257–288. Balsillie, James H, and Joseph F Donoghue 2004 High Resolution Sea-level History for the Gulf of Mexico Since the Last Glacial Maximum. Florida Geological Survey, Tallahassee. Bard, Edouard, Bruno Hamelin, Maurice Arnold, and Daniele Buigues 1990 (Super 230) Th/(Super 234) U and (Super 14) C Ages Obtained by Mass Spectrometry on Corals from Mururoa Atoll, French Polynesia. Radiocarbon 192

33(2):173. Barnes, T. K., A. K. Volety, K. Chartier, F. J. Mazzotti, and L. Pearlstine 2007 A Habitat Suitability Index Model for the Eastern Oyster (Crassostrea virginica), a Tool for Restoration of the Caloosahatchee Estuary, Florida. Journal of Shellfish Research 26(4):949–959. Bayliss, Alex 2015 Quality in Bayesian Chronological Models in Archaeology. World Archaeology 47(4):677–700. DOI:10.1080/00438243.2015.1067640. Bijma, J., W. W. Faber, and C. Hemleben 1990 Temperature and Salinity Limits for Growth and Survival of some Planktonic Foraminifers in Laboratory Cultures. The Journal of Foraminiferal Research 20(2):95– 116. DOI:10.2113/gsjfr.20.2.95. Binford, Lewis R. 1967 Smudge Pits and Hide Smoking: The Use of Analogy in Archaeological Reasoning. American Antiquity 32(01):1–12. DOI:10.2307/278774. 1980 Willow Smoke and Dogs’ Tails: Hunter-Gatherer Settlement Systems and Archaeological Site Formation. American Antiquity 45(01):4–20. Bird, J.B., J. Hyslop, and M.D. Skinner 1985 J. B. Bird, J. Hyslop, M. D. Skinner, Preceramic Excavations at Huaca Prieta, Chicama Valley, Peru. Anthropological Papers of the American Museum of Natural History. Bochenski, Z.M., and T. Tomek 2010 The Birds of Klisoura Cave 1: A Window into the Upper Palaeolithic Greece. Eurasian Prehistory 7(2):91–106. Braithwaite, C. J. R. 1991 Dolomites, A Review of Origins, Geometry and Textures. Transactions of the Royal Society of Edinburgh: Earth Sciences 82(2):99–112. Brock, J. C., M. Palaseanu-Lovejoy, C. W. Wright, and A. Nayegandhi 2008 Patch-reef Morphology as a Proxy for Holocene Sea-level Variability, Northern Florida Keys, USA. Coral Reefs 27(3):555–568. DOI:10.1007/s00338-008-0370-y. Bronk Ramsey, Christopher 2008 Deposition Models for Chronological Records. Quaternary Science Reviews 27(1–2):42–60. Brooks, Gregg R, and Larry J Doyle 1991 Geologic Development and Depositional History of the Florida Middle Ground: A midshelf, Temperate-zone Reef System in the Northeastern Gulf of Mexico. Society

193

for Sedimentary Geology 46:189–203. Brooks, H.K 1962 Observations on the Florida Middle Ground. Geological Society of America Special Publication (68):65–66. Buroker, N. E. 1983 Population Genetics of the American Oyster Crassostrea Virginica along the Atlantic Coast and the Gulf of Mexico. Marine Biology 75(1): 99–112. Carver, R.E. 1971

Procedures in Sedimentary Petrology. Wiley Interscience, New York, NY.

Cassidy, Jim, Mark Raab, and Nina Kononenko 2004 Bones, Boats and Biface Bias: The Early Holocene Mariners of Eel Point, San Clemente Island, California. American Antiquity 69:109–130. Cattaneo, Antonio, and Ron J. Steel 2003 Transgressive deposits: a review of their variability. Earth-Science Reviews 62(3– 4):187–228. DOI:10.1016/S0012-8252(02)00134-4. Cheng, Hai, R. Lawrence Edwards, Chuan-Chou Shen, Victor J. Polyak, Yemane Asmerom, Jon Woodhead, John Hellstrom, Yongjin Wang, Xinggong Kong, Christoph Spötl, Xianfeng Wang, and E. Calvin Alexander 2013 Improvements in 230Th Dating, 230Th and 234U Half-life Values, and U–Th Isotopic Measurements by Multi-collector Inductively Coupled Plasma Mass Spectrometry. Earth and Planetary Science Letters 371–372:82–91. Cook-Hale, Jessica 2016 Springs, stone, and shell: recent excavation at the Econfina Channel Site, a submerged Archaic site, Apalachee Bay, Florida, U.S.A. 81st Annual Meeting of the Society for American Archaeology, Orlando, Florida. 2016 Cook Hale, Jessica, Nathan L. Hale, and Ervan Garrison 2018 What is Past is Prologue: Excavations at the Econfina Channel Site, Apalachee Bay, Florida, USA. Southeastern Archaeology. February 1. Compton, John S. 2011 Pleistocene Sea-level Fluctuations and Human Evolution on the Southern Coastal Plain of South Africa. Quaternary Science Reviews 30(5–6):506–527. Coudray, J., & Montaggioni, L. F. 1986 The Diagenetic Products of Marine Carbonates as Sea-level Indicators. In O. Van de Plassche (Ed.), Sea level research: Manual book for the collection and evaluation of Data (pp. 311–360). Norwich: Geo Books.

194

Davies, T.D. 1980 Peat Formation in Florida Bay and its Significance in Interpreting the Recent Vegetation History of the Bay Area. Unpublished Ph.D. Dissertation, Penn. State University, College Station. Davis, Richard A. 2017 Sediments of the Gulf of Mexico. In Habitats and Biota of the Gulf of Mexico: Before the Deepwater Horizon Oil Spill: Volume 1: Water Quality, Sediments, Sediment Contaminants, Oil and Gas Seeps, Coastal Habitats, Offshore Plankton and Benthos, and Shellfish, edited by C. Herb Ward, pp. 165–215. Springer New York, New York, NY. Dillehay, Tom D., Steve Goodbred, Mario Pino, Víctor F. Vásquez Sánchez, Teresa Rosales Tham, James Adovasio, Michael B. Collins, Patricia J. Netherly, Christine A. Hastorf, Katherine L. Chiou, Dolores Piperno, Isabel Rey, and Nancy Velchoff 2017 Simple Technologies and Diverse Food Strategies of the Late Pleistocene and Early Holocene at Huaca Prieta, Coastal Peru. Science Advances 3(5) Dillehay, Tom D., and Peter Kaulicke (editors) 2011 From Foraging to Farming in the Andes: New Perspectives on Food Production and Social Organization. Cambridge University Press, New York. Dillehay, Tom D., C. Ramírez, M. Pino, M. B. Collins, J. Rossen, and J. D. Pino-Navarro 2008 Monte Verde: Seaweed, Food, Medicine, and the Peopling of South America. Science 320(5877):784–786. DOI:10.1126/science.1156533. Dillehay, Tom D., Carlos Ramirez, Mario Pino, Michael B. Collins, Jack Rossen, and J. D. PinoNavarro 2008 Monte Verde: Seaweed, Food, Medicine, and the Peopling of South America. Science 320(5877):784–786. Duggins, R. M 2012 Florida's Paleoindian and Early Archaic: A GIS Approach to Modeling Submerged Landscapes and Site Distribution on the Continental Shelf Department of Anthropology. Dissertation. Florida State University, Tallahassee, FL. Duggins, Ryan M., Franklin Price, Melissa Price, Ivor Mollema, Neil Puckett, and Paulette S. McFadden 2018 Manasota Key Offshore: A Prehistoric Cemetery in the Gulf of Mexico. Presented at the Landscapes, Entrepôts, and Global Currents, Society for Historical Archaeology January 7, New Orleans, La. Dunbar, James S, and Faught, Michael K. 1999 Field Notes Taken During 7/12/1999 Econfina Channel Site 8TE139. Manuscript on file, Florida Department of State, Tallahassee, Florida. Dunbar, James S. and. Vojnovski, Pamela K

195

2007 Early Floridians and Late Megamammals: Some Technological and Dietary Evidence from Four North Florida Sites. In Foragers of the Terminal Pleistocene in North America, edited by R. B. Walker and B. N. Driskell, pp. 167-203. University of Nebraska Press, Lincoln. Edwards, R. Lawrence, and W. R. Peltier 1995 Paleotopography of Glacial-Age Ice Sheets. Science 267(5197): 536–538. Erlandson, Jon M., Michael H. Graham, Bruce J. Bourque, Debra Corbett, James A. Estes, and Robert S. Steneck 2007 The Kelp Highway Hypothesis: Marine Ecology, the Coastal Migration Theory, and the Peopling of the Americas. The Journal of Island and Coastal Archaeology 2(2):161–174. DOI:10.1080/15564890701628612. Erlandson, Jon M., and Nicholas Jew 2009 An Early Maritime Biface Technology at Daisy Cave, San Miguel Island, California: Reflections on Sample Size, Site Function, and Other Issues. North American Archaeologist 30(2):145–165. DOI:10.2190/NA.30.2.c. Erlandson, Jon M., Torben C. Rick, Todd J. Braje, Molly Casperson, Brendan Culleton, Brian Fulfrost, Tracy Garcia, Daniel A. Guthrie, Nicholas Jew, Douglas J. Kennett, Madonna L. Moss, Leslie Reeder, Craig Skinner, Jack Watts, and Lauren Willis 2011 Paleoindian Seafaring, Maritime Technologies, and Coastal Foraging on California’s Channel Islands. Science 331(6021):1181. DOI:10.1126/science.1201477. Erlandson, Jon, Torben Rick, Ren L. Vellanoweth, and Douglas J. Kennett 1998 Maritime Subsistence at a 9300 Year Old Shell Midden on Santa Rosa Island, California. Journal of Field Archaeology 26(3):255-265. Eros, J.M., I.P. Montañez, D.A. Osleger, V.I. Davydov, T.I. Nemyrovska, V.I. Poletaev, and M.V. Zhykalyak 2012 Sequence Stratigraphy and Onlap History of the Donets Basin, Ukraine: Insight into Carboniferous Icehouse Dynamics. Palaeogeography, Palaeoclimatology, Palaeoecology 313–314:1–25. DOI:10.1016/j.palaeo.2011.08.019. Fairbanks, R.G. 1989 A 17,000-year Glacio-Eustatic Sea-level Record; Influence of Glacial Melting Rates on the Younger Dryas Event and Deep-Ocean Circulation. Nature 342(6250):637–642. 1990 The Age and Origin of the Younger Dryas Climate Event in Greenland Ice cores. Paleoceanography 5:937–948. Emery, K.O., and Edwards, R.L. 1966 Archaeological Potential of the Atlantic Continental Shelf. American Antiquity 31, 733–737. Faught, Michael K.

196

1989 Preliminary Report on the Excavation of the Econfina Channel Site: An Inundated Archaic Lithic Quarry in the Apalachee Bay, Florida. Manuscript on file. Florida Bureau of Archaeological Research, Tallahassee, FL. 1990 Report of the 1989 Offshore Field Session. Manuscript on file. Florida Bureau of Archaeological Research, Tallahassee, FL. 1992 Report of the 1991 Offshore Field Session: Discovery of Paleo-River Channel Sediments and Paleoindian Diagnostics on the Continental Shelf. Manuscript on file. Florida Bureau of Archaeological Research, Tallahassee, FL. 1996 Clovis Origins and Underwater Prehistoric Archaeology in Northwestern Florida. Ph.D. Dissertation, Department of Anthropology, University of Arizona, Tucson. 1997 Field notes taken during an 4/8/1997 Fitch Site 8JE739. Manuscript on file, Florida Department of State, Tallahassee, Florida. 2002 Submerged Paleoindian and Archaic Sites of the Big Bend, Florida. Journal of Field Archaeology 29:273-290. 2003 Geophysical Remote Sensing and Underwater Cultural Resource Management of Submerged Prehistoric Sites in Apalachee Bay: A Deep-Water Example, Site Predictive models, and Site Discoveries. In Proceedings: Twenty Second Annual Gulf of Mexico Information Transfer Meetings, 2002, edited by Melanie McKay pp.115-129. Minerals Management Service, New Orleans. 2004a The Underwater Archaeology of Paleolandscapes, Apalachee Bay, Florida. American Antiquity 69(2):275–289. DOI:10.2307/4128420. 2004b Submerged Paleoindian and Archaic Sites of the Big Bend, Florida. Journal of Field Archaeology 29(3–4):273–290. DOI:10.1179/jfa.2004.29.3-4.273. 2018 State-of-the-art: Reconstructing paleo landscapes for maritime CRM projects presented at the 51st Annual Conference On Historical and Underwater Archaeology, New Orleans, La. Faught, Michael K., and D. J. Donahue 1997 Marine Inundated Archaeological Sites and Paleofluvial Systems: Examples from a Karst-Controlled Continental Shelf Setting in Apalachee Bay, Northeastern Gulf of Mexico. Geoarchaeology Vol. 12(5):417–458. Faure, Hugues, Robert C Walter, and Douglas R Grant 2002 The Coastal Oasis: Ice Age Springs on Emerged Continental Shelves. Global and Planetary Change 33(1–2):47–56. DOI:10.1016/S0921-8181(02)00060-7. Fisher, Erich C., Miryam Bar-Matthews, Antonieta Jerardino, and Curtis W. Marean 2010 Middle and Late Pleistocene Paleoscape Modeling Along the Southern Coast of

197

South Africa. Quaternary Science Reviews 29(11–12):1382–1398. Florida Fish and Wildlife Conservation Commission 2017 Oyster Beds in Florida | Florida Fish and Wildlife Conservation Commission. http://geodata.myfwc.com/datasets/a78160f5acaf4439b49f9fbef4c100ac_5?geometry=84.04%2C30.043%2C-83.98%2C30.056&selectedAttribute=OYSTER, accessed August 31, 2017. Flower, Benjamin P., and J Kennett 1995 12. Biotic Responses to Temperature and Salinity Changes During Last Deglaciation, Gulf of Mexico | Effects of Past Global Change on Life | The National Academies Press. In Efficts of the Past Global Change on Life. The National Academies Press, Washington, DC. Folk, R.L., and L.S. Land 1975 Mg/Ca Ratio and Salinity: Two Controls Over Crystalization of Dolomite. American Association of Petroleum Geologists Bulletin 59:60–68. Graf, Kelly 2009 “The Good, the Bad, and the Ugly”: Evaluating the Radiocarbon Chronology of the Middle and Late Upper Paleolithic in the Enisei River Valley, South-Central Siberia. Journal of Archaeological Science 36(3):694–707. Halligan, J. J., M. R. Waters, A. Perrotti, I. J. Owens, J. M. Feinberg, M. D. Bourne, B. Fenerty, B. Winsborough, D. Carlson, D. C. Fisher, T. W. Stafford, and J. S. Dunbar 2016 Pre-Clovis Occupation 14,550 years ago at the Page-Ladson Site, Florida, and the Peopling of the Americas. Science Advances 2(5): e1600375 Halligan, Jessi Jean 2012 Geoarchaeological investigations into Paleoindian adaptations on the Aucilla River, Northwest Florida. Ph.D. Dissertation, Department of Anthropology, Texas A&M University. College Station, Texas. Haslett, John, and Andrew Parnell 2008 A Simple Monotone Process with Application to Radiocarbon-dated Depth Chronologies. Journal of the Royal Statistical Society: Series C (Applied Statistics) 57(4):399–418. DOI:10.1111/j.1467-9876.2008.00623.x. Henshilwood, Christopher S., Curtis W. Marean, Philip Chase, Iain Davidson, Clive Gamble, Trenton W. Holliday, Richard G. Klein, Sally Mc Brearty, João Zilhão, Christopher S. Henshilwood, and others 2003 The Origin of Modern Human Behavior: Critique of the Models and Their Test Implications 1. Current Anthropology 44(5):627–651. Hill, Kim, Michael Barton, and A. Magdalena Hurtado 009 The emergence of human uniqueness: Characters underlying behavioral

198

modernity. Evolutionary Anthropology: Issues, News, and Reviews 18(5):187–200. Hine, Albert C, and Stanley Locker 2011 Florida Gulf of Mexico Continental Shelf. Gulf of Mexico Origin, Waters, and Biota: Volume 3, Geology 3:101. Jerardino, Antonieta, and Curtis W. Marean 2010 Shellfish Gathering, Marine Paleoecology and Modern Human Behavior: Perspectives from Cave PP13B, Pinnacle Point, South Africa. Journal of Human Evolution 59(3–4):412–424. DOI:10.1016/j.jhevol.2010.07.003. Johnson, John R., Thomas W. Stafford Jr, Henry O. Ajie, and Don P. Morris 2002 Arlington Springs Revisited. In Proceedings of the fifth California Islands Symposium, 23. pp. 541–545. Santa Barbara, CA: Santa Barbara Museum of Natural History. Karkanas, Panagiotis, and Paul Goldberg 2010 Site formation processes at Pinnacle Point Cave 13B (Mossel Bay, Western Cape Province, South Africa): Resolving Stratigraphic and Depositional Complexities with Micromorphology. Journal of Human Evolution 59(3–4):256–273. Keefer, David K., Susan D. deFrance, Michael E. Moseley, James B. Richardson, Dennis R. Satterlee, and Amy Day-Lewis 1998 Early Maritime Economy and El Niño Events at Quebrada Tacahuay, Peru. Science 281(5384):1833. DOI:10.1126/science.281.5384.1833. Kuehn, D.W. 1980 Offshore Transgressive Peat Deposits of Southwest Florida: Evidence for a Late Holocene Rise of Sea-level. Unpublished M.S. Thesis, Pennsylvania University, Philadelphia, PA. Lambeck, K., H. Rouby, A. Purcell, Y. Sun, and M. Sambridge 2014 Sea level and Global Ice Volumes from the Last Glacial Maximum to the Holocene. Proceedings of the National Academy of Sciences 111(43):15296–15303. Lambeck, Kurt, Tezer M. Esat, and Emma-Kate Potter 2002 Links Between Climate and Sea-levels for the Past Three Million Years. Nature 419(6903):199. Lorio, J. W., and C. Petrone 1994 The Cultivation of American Oysters. Crassostrea virginica. Southern Regional Aquaculture Center Volume 432 Lu, Feng H. 2008 Pristine or Altered: Low-Mg Calcite Shells Survived from Massive Dolomitization? A Case Study from Miocene Carbonates. Geo-Marine Letters 28(5–

199

6):339–349. Ludwig, Kenneth R., and Paul R. Renne 2000 Geochronology on the Paleoanthropological Time Scale. Evolutionary Anthropology: Issues, News, and Reviews 9(2):101–110. Marean, Curtis W. 2010 Pinnacle Point Cave 13B (Western Cape Province, South Africa) in Context: The Cape Floral Kingdom, Shellfish, and Modern Human Origins. Journal of Human Evolution 59(3–4):425–443. DOI:10.1016/j.jhevol.2010.07.011. 2014 The Origins and Significance of Coastal Resource Use in Africa and Western Eurasia. The Role of Freshwater and Marine Resources in the Evolution of the Human Diet, Brain and Behavior 77:17–40. DOI:10.1016/j.jhevol.2014.02.025. Marean, Curtis W., Miryam Bar-Matthews, Jocelyn Bernatchez, Erich Fisher, Paul Goldberg, Andy I. R. Herries, Zenobia Jacobs, Antonieta Jerardino, Panagiotis Karkanas, Tom Minichillo, Peter J. Nilssen, Erin Thompson, Ian Watts, and Hope M. Williams 2007 Early Human Use of Marine Resources and Pigment in South Africa During the Middle Pleistocene. Nature 449(7164):905–908. DOI:10.1038/nature06204. Marean, Curtis W., Miryam Bar-Matthews, Erich Fisher, Paul Goldberg, Andy Herries, Panagiotis Karkanas, Peter J. Nilssen, and Erin Thompson 2010 The Stratigraphy of the Middle Stone Age Sediments at Pinnacle Point Cave 13B (Mossel Bay, Western Cape Province, South Africa). Journal of Human Evolution 59(3–4):234–255. DOI:10.1016/j.jhevol.2010.07.007. Marks, Brian S. 2006 Site Formation Processes and Activity Areas at Ontolo (8JE1577): A Submerged Prehistoric Human Occupation Site in Apalachee Bay, Florida. Unpublished Dissertation, Department of Anthropology. Florida State University, Tallahassee, Fl. Meltzer, David J. 2015 Pleistocene Overkill and North American Mammalian Extinctions. Annual Review of Anthropology 44(1):33–53. DOI:10.1146/annurev-anthro-102214-013854. Mook, W. G., and O. Van de Plassche 1986 Radiocarbon dating. Sea-Level Research, pp. 525–560. Springer. National Marine Fisheries Service 2014 Recovery Plan for Elkhorn (Acropora palmata) and Staghorn (A. cervicornis) Corals. Prepared by the Acropora Recovery Team for the National Marine Fisheries Service, Silver Spring, Maryland. National Oceanic and Atmospheric Administration 2009 Exploring the Submerged New World http://oceanexplorer.noaa.gov/

200

explorations/09newworld/.html accessed 11/04/2016 Olsen, Jesper, Jan Heinemeier, Karen Margrethe Hornstrup, Pia Bennike, and Henrik Thrane 2013 ‘Old wood’ Effect in Radiocarbon Dating of Prehistoric Cremated Bones? Journal of Archaeological Science 40(1):30–34. Osterman, Lisa E., David C. Twichell, and Richard Z. Poore 2009 Holocene Evolution of Apalachicola Bay, Florida. Geo-Marine Letters 29(6):395–404. Paine, Jeffrey G., Sojan Mathew, and Tiffany Caudle 2012 Historical Shoreline Change Through 2007, Texas Gulf Coast: Rates, Contributing Causes, and Holocene Context. GCAGS Journal, v. 1 (2012), p. 13–26. Parnell, A.C., J. Haslett, J.R.M. Allen, C.E. Buck, and B. Huntley 2008 A Flexible Approach to Assessing Synchroneity of Past Events Using Bayesian Reconstructions of Sedimentation History. Quaternary Science Reviews 27(19– 20):1872–1885. DOI:10.1016/j.quascirev.2008.07.009. Parnell, Andrew C., and W. Roland Gehrels 2015 Using Chronological Models in Late Holocene Sea-level Reconstructions from Saltmarsh Sediments. In Handbook of Sea-Level Research, pp. 500–513. John Wiley & Sons, Ltd. Perlès, C 1999 Long-term Perspectives on the Occupation of the Franchthi Cave: Continuity and Discontinuity. British School at Athens Studies no. 3. In Proceedings of the ICOPAG Conference, pp. 311–318. Ioannina, Greece. Perlès, Catherine 2016 Food and Ornaments: Diachronic Changes in the Exploitation of Littoral Resources at Franchthi Cave (Argolid, Greece) During the Upper Palaeolithic and the Mesolithic (39,000–7000 cal BC). Quaternary International 407:45–58. Pirazzoli, Paolo Antonio 1996 Sea-level Changes, the Last 20,000 years. John Wiley, New York, NY. Polyak, Victor, Paula Provencio, University of New Mexico, and Yemane Asmerom 2016 U-Pb Dating of Speleogenetic Dolomite: A New Sulfuric Acid Speleogenesis Chronometer. International Journal of Speleology 45(2):103–109. Pourmand, Ali, Tissot François L. H., Arienzo Monica, and Arash Sharifi 2014 Introducing a Comprehensive Data Reduction and Uncertainty Propagation Algorithm for U‐Th Geochronometry with Extraction Chromatography and Isotope Dilution MC‐ICP‐MS. Geostandards and Geoanalytical Research 38(2):129–148.

201

Rasbury, E. Troy, and Jennifer M. Cole 2009 Directly Dating Geologic Events: U-Pb Dating of Carbonates. Reviews of Geophysics 47(3). DOI:10.1029/2007RG000246. Reich, Christopher D., Richard Z. Poore, and Todd D. Hickey 2013 The Role of Vermetid Gastropods in the Development of the Florida Middle Ground, Northeast Gulf of Mexico. Journal of Coastal Research:46–57. 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 Hogg, 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, and Johannes van der Plicht 2013 IntCal13 and Marine13 Radiocarbon Age Calibration Curves 0–50,000 Years cal BP. Radiocarbon 55(04):1869–1887. DOI:10.2458/azu_js_rc.55.16947. Richardson, J.B. 1998 Looking in the Right Places: Pre-5000 B.P. Maritime Adaptations in Peru and the Changing Environment. Revista de Arqueolog´ıa Americana 15:35–36. Rick, Torben C., Jon M. Erlandson, and René L. Vellanoweth 2001 Paleocoastal Marine Fishing on the Pacific Coast of the Americas: Perspectives from Daisy Cave, California. American Antiquity 66(4):595–613. Rick, Torben C., Douglas J. Kennett, and Jon M. Erlandson 2005 Preliminary Report on the Archaeology and Paleoecology of the Abalone Rocks Estuary, Santa Rosa Island, California. In Proceedings of the Sixth California Islands Symposium, pp. 55–63. National Park Service Technical Publication CHIS-05-01, Institute for Wildlife Studies Arcata California. Rick, Torben, Jon Erlandson, René L. Vellanoweth, and Todd Braje 2005 From Pleistocene Mariners to Complex Hunter-Gatherers: The Archaeology of the California Channel Islands. Journal of World Prehistory 19:169–228. Ross, S.T. 2001

Amia calva Linnaeus, Bowfin. Inland fishes of Mississippi:92–95.

Royden, H.L, and P.M. Fitzpatrick 2010 Real Analysis (Fourth Edition) Pearson Education, Inc., Publishing as Prentice Hall. Russo, Michael 1988 Coastal Adaptations in Eastern Florida: Models and Methods. Archaeology of Eastern North America 16:159–178.

202

Sandweiss, Daniel H. 2008 Early Fishing Societies in Western South America. In The Handbook of South American Archaeology. Helaine Silverman and William H. Isbell, eds. Pp. 145–156. New York, NY: Springer New York. Sandweiss, Daniel H., Heather McInnis, Richard L. Burger, Asunción Cano, Bernardino Ojeda, Rolando Paredes, Marıá del Carmen Sandweiss, and Michael D. Glascock 1998 Quebrada Jaguay: Early South American Maritime Adaptations. Science 281(5384):1830. DOI:10.1126/science.281.5384.1830. Sandweiss, Daniel H., J.B. Richardson, Elizabeth Reitz, Jeffery Hsu, and Robert Feldman 1989 Early Maritime Adaptations in the Andes: Preliminary Studies at the Ring Site, Peru. In Ecology, Settlement, and History in the Osmore Drainage, Peru., edited by D.S. Rice, C. Standish, and P.R. Scarr, pp. 35–84. British Archaeological Reports International Series 545. Oxford, UK. Sassaman, Kenneth E., Neill J. Wallis, Paulette S. McFadden, Ginessa J. Mahar, Jessica A. Jenkins, Mark C. Donop, Micah P. Monés, Andrea Palmiotto, Anthony Boucher, Joshua M. Goodwin, and Cristina I. Oliveira 2016 Keeping Pace With Rising Sea: The First 6 Years of the Lower Suwannee Archaeological Survey, Gulf Coastal Florida. The Journal of Island and Coastal Archaeology 12(2):173–199. Schnable, J.E., and H.G. Goodell 1968 Pleistocene-Recent Stratigraphy, Evolution, and Development of the Apalachicola Coast, Florida. Florida Geological Society of America Special Paper 112:72. Scholz, Denis, and Dirk Hoffmann 2008 230Th/U-Dating of Fossil Corals and Speleothems. Quaternary Science Journal 57(1–2):52–76. Schroeder, William W., Albert W. Shultz, and Orrin H. Pilkey 1995 Late Quaternary Oyster Shells and Sea-level History, Inner Shelf, Northeast Gulf of Mexico. Journal of Coastal Research:664–674. Shepard, F.P. 1960 Rise of Sea-Level along Northwest Gulf of Mexico: Recent Sediments, Northwest Gulf of Mexico. American Association of Petroleum Geologists: 338-344. Shier, D.E. 1969 Vermetid Reefs and Coastal Development in the Ten Thousand Islands, Southwest Florida. Geological Society of America Bulletin 80:485–508. Siddall, M., E. J. Rohling, A. Almogi-Labin, Ch. Hemleben, D. Meischner, I. Schmelzer, and D. A. Smeed 2003 Sea-level Fluctuations During the Last Glacial Cycle. Nature 423(6942):853–858.

203

Sime, Patti 2005 St. Lucie Estuary and Indian River Lagoon Conceptual Ecological Model. The Society of Wetland Scientists 25(4):898–907. Smith, Marian W. 1956 The Cultural Development of the Northwest Coast. Southwestern Journal of Anthropology 12(3):272–294. DOI:10.1086/soutjanth.12.3.3629085. Smith, Morgan, Shawn Joy, Michael R. Waters, Jessi Jean Halligan, Michael K. Faught, Angelina Perrotti, Brendan Fenerty, Ryan M. Duggins, and Adam Burke In Press Liquid Landscapes- The Contributions of a New Wave of Submerged Prehistoric Archaeology to the Paleoindian and Early Archaic Record of the Southeastern United States. In The Paleoindian and Early Archaic Southeast: The Next 20 Years. Snow, Dean R. 1972 Rising Sea Level and Prehistoric Cultural Ecology in Northern New England American Antiquity 37(2):211–221. DOI:10.2307/278207. Starkovich, Britt M., Natalie D. Munro, and Mary C. Stiner 2017 Terminal Pleistocene Subsistence Strategies and Aquatic Resource Use in Southern Greece. Quaternary International. DOI:10.1016/j.quaint.2017.11.015. Stiner, Mary C., Natalie D. Munro, and Britt M. Starkovich 2012 Material Input Rates and Dietary Breadth During the Upper Paleolithic Through Mesolithic at Franchthi and Klissoura 1 Caves (Peloponnese, Greece). Quaternary International 275:30–42. DOI:10.1016/j.quaint.2011.08.006. Talma, A. S., and J. C. Vogel 1993 A Simplified Approach to Calibrating 14C Dates. Radiocarbon 35(02):317–322. Telford, R.J., E. Heegaard, and H.J.B. Birks 2004 The Intercept is a Poor Estimate of a Calibrated Radiocarbon Age. The Holocene 14:296–298. Thulman, David K. 2009 Freshwater Availability as the Constraining Factor in the Middle Paleoindian Occupation of North-Central Florida. Geoarchaeology 24(3):243–276. 2011 Discriminating Paleoindian Point Types from Florida Using Landmark Geometric Morphometrics. Journal of Archaeological Science 39:1599–1607. 2012 Paleoindian Occupation Along the St. Johns River, Florida. The Florida Anthropologist 65(1–2):77–83.

204

Toscano, M. A., and I. G. Macintyre 2003 Corrected Western Atlantic Sea-level Curve for the Last 11,000 years Based on Calibrated 14C Dates from Acropora palmata Framework and Intertidal Mangrove Peat. Coral Reefs 22(3):257–270. DOI:10.1007/s00338-003-0315-4. Toscano, Marguerite A. 2016 Revised Paleoenvironmental Analysis of the Holocene Portion of the Barbados Sea-level Record: Cobbler’s Reef Revisited. Coral Reefs 35(2):641–653. Turetsky, Merritt R., Sturt W. Manning, and R. Kelman Wieder 2004 Dating Recent Peat Deposits. Wetlands 24(2):324–356. Veth, Peter, K Aplin, L Wallis, T Manne, T Pulsford, E White, and A Chappel 2007 The Archaeology of Montebello Islands, North-West Australia: Late Quaternary Foragers on an Arid Coastline. Archaeopress, Oxford, UK. Veth, Peter, Ingrid Ward, Tiina Manne, Sean Ulm, Kane Ditchfield, Joe Dortch, Fiona Hook, Fiona Petchey, Alan Hogg, Daniele Questiaux, Martina Demuro, Lee Arnold, Nigel Spooner, Vladimir Levchenko, Jane Skippington, Chae Byrne, Mark Basgall, David Zeanah, David Belton, Petra Helmholz, Szilvia Bajkan, Richard Bailey, Christa Placzek, and Peter Kendrick 2017 Early Human Occupation of a Maritime Desert, Barrow Island, North-West Australia. Quaternary Science Reviews 168:19–29. Waelbroeck, Claire, Laurent Labeyrie, E. Michel, J. Cl Duplessy, J. F. McManus, Kurt Lambeck, E. Balbon, and M. Labracherie 2002 Sea-level and Deep Water Temperature Changes Derived from Benthic Foraminifera Isotopic Records. Quaternary Science Reviews 21(1):295–305. Waters, Michael R., Thomas W. Stafford, Brian Kooyman, and L. V. Hills 2015 Late Pleistocene Horse and Camel Hunting at the Southern Margin of the Ice-free Corridor: Reassessing the Age of Wally’s Beach, Canada. Proceedings of the National Academy of Sciences 112(14):4263–4267. DOI:10.1073/pnas.1420650112. Willett, Michael 2006 Effect of Dissolution of the Florida Carbonate Platform on Isostatic Uplift and Relative Sea-Level Change.PhD Disertation. Florida State University, Tallahassee, Fl. Williams, Alan N., Sean Ulm, Tom Sapienza, Stephen Lewis, and Chris S.M. Turney 2018 Sea-level Change and Demography During the Last Glacial Termination and Early Holocene Across the Australian Continent. Quaternary Science Reviews 182:144–154. Wilson, J.L 1975

Carbonate Facies in Geologic History. Geological Magazine 113(6):584–585.

205

BIOGRAPHICAL SKETCH Shawn Joy, B.A., has over a decade of experience in historic, maritime, prehistoric, and submerged prehistoric archaeological fieldwork and research. Mr. Joy joined SEARCH in 2016 as an archaeologist with the Tallahassee office. Shawn received his Bachelor’s degree in Anthropology from the University of Massachusetts, Amherst and his Master’s degree from Florida State University. Mr. Joy’s research focuses on geoarchaeology and submerged Paleoindian sites. Shawn has been involved in numerous phase three submerged prehistoric excavations including Page-Ladson, Guest Mammoth, and Sloth Hole. He has conducted surveys for federal, state, and private archaeological projects throughout the Mid-West and Eastern United States. He has worked extensively with the US Army Corps of Engineers, Department of Transportation, US Forest Service, and National Park Service. Mr. Joy has conducted desktop analyses and submerged paleolandscape reconstructions for both utility and research projects in the Gulf of Mexico and Atlantic Coast. He has served as archaeological liaison while participating in several NAGPRA repatriations. He has worked for the Massachusetts Board of Underwater Archaeological Resources where he assisted in developing the reporting guidelines for maritime archaeological reports and advanced an extensive New England shipwreck database. During Mr. Joy’s time in New England, he was also named the official on-call archaeologist for the Springfield Armory, the oldest armory in the country. Mr. Joy has operated on several major pipeline and utilities contracts as field director, coordinating multiple companies throughout all phases of archaeological investigation. As an underwater archaeologist Mr. Joy has logged hundreds of hours of scientific diving and specializes in underwater photography. His photographs have been featured on CNN, Smithsonian Magazine, and Popular Archaeology.

206