STORAGE , DECISION - MAKING, AND RISK MANAGEMENT IN NON - SEDENTARY SOCIETIES By Kathryn M. Frederick A DISSERTATION Submitted to Michigan State University i n partial fulfillment of the requirements f or the degree of Anthropology Doctor of Philosophy 201 9 ABSTRACT STORAGE, DECISION - MAKING, AND RISK MANAGEMENT IN NON - SEDENDARY SOCIETIES By Kathryn M. Frederick Food storage , or the act of extending the shelf life of foodstuffs, often formed an important part of the adaptations of certai n small - scale hunter - gatherer and low - level horticulturalist societies. Research on stora ge in small - scale societies has, until recently, narrowly focused on determining the form an d scale that food stor age took, and its relatedness to increasing social co mplexity. This dissertation, instead, looks at the purposeful decision - making behind the use of food storage as a risk management strategy in non - sedentary soc ieties. For storage to be a risk - averse mana gement strategy, the technology required for successf ul storage must be reliable. Knowledge of proper storage pit construction and food preservation techniques is required before the risk and uncertainty of food storage can be mitigated. Due to the increas ed front - end energy output, the risk associated with the processing and storage of the foodstuffs must be minimal, as compared to the seasonal risk of food scarcity and potential energy/caloric gain. Subterranean storage pits appear in the archaeological landscape of the northern Great Lakes after ca. AD 1 000 and tribal communities continued to use them through the historic period . During the late Late W oodland period (ca. AD 1000 - 1600 ), subterranean food storage containers were systematically used by tr ibal communities with a spatially and seasonally rest ricted fisher - forager - horticulturalist subsistence system to create a stable food supply . The act of food storage, actualized through the technology of subterranean storage pits, allowed groups to increa se their communal capacity for surv ival, success, and regeneration; these capacities were stressed by the increasingly restrictive setting of the Late Prehistoric sociopolitical landscape. Combining experimental archaeology, ethnographic and ethnohistoric data, along with archaeological data on food storage , this research examines the technology and behavioral patterns for use of subterranean food storage utilized by hunter - gatherers. This research aimed to understand the use of food storage through the pe rspective of the technology of the physical container and the decision - making behind this risk - management strategy. A series of experiments were performed to understand the technology and technical know - how required to create a risk - averse storage feature. Data was then collected on food storage practices, b oth past and present, across the globe. Factors, such as movement strategy, climate, environment, economy, and to a lesser extent socio - political triggers, were considered. These factors were then compar ed and analyzed for patterns in decision - making. Wi th a baseline for understanding the technology of storage and an evident pattern in storage use by hunter - gatherers, my research considered whether similar variables were at play in the Late Woodland period, and what other factors drove the decisions to st ore. The collected data exhibited two prominent patterns for storage use, reliant and redundant. When these patterns were applied to the case study of northern lower Michigan a model for storage practices and their effect on the larger settlement and sub si stence practices was created. The proliferation in the use of food storage during the late Late Woodland indicates a socioeconomic shift that made the previously risk - prone act of storage, risk - averse. I argue that the northern lower Michigan Late Woodla nd people incorporated redundant food storage practices into their existing risk management strategies as a response to increased population and reduced territory. Copyright by KATHRYN M. FREDERICK 201 9 v ACKNOWLEDGEMENTS They say it takes a village to raise a child M y son, Elliott, has confirmed this sentiment. I f I can compare a dissertation to the raising of a child, a process of putting blood , swe a t and tears into the production of a worthy document, I would argue it takes a small city to raise a dissertation. The goal, as a parent, is to shape children into good human beings as they grow . Like a child, a dissertation needs to be shaped into something good, worthy; however , unlike children, a dissertat ion will not grow on its own - it needs constant nurturing. A dissertation takes a small city to not only help shape its outcome, but to ensure its growth. As my dissertation is at its end, I need to th ank the small city that has raised my dissertation . The multitude of people involved in this research is probably enumerable, but I shall do my best to thank those who have aided in the development of my baby, aka my dissertation. earching, and the re doing it. On the academic side, I could not have done this without my remarkable Dissertation Committee . I must start with my advisor and chair, Dr. William Lovis . Since day one of my graduate career at MSU, Bill has steadfastly encouraged my research, but never gave me an easy out. He slowly, maybe painfully, guided me to my research question, allowing me to learn the process of scientific research, fostering a long - term understanding of getting at a problem, rat her than just handing me a project. My de epest gratitude is in the fact the Bill does not stand for , and will not allow his students to conduct, shoddy science . Because of this, I have put forth a product in which I have complete confidence. I also owe a special thank you to Dr. Meghan Howey . Meghan was the PI on my first field school in 2006. Her passion for archaeology and willingness to engage in meaningful vi conversations about her research with curious undergrads solidified my ambitions to become an a rchaeologist. My career has been heavily she has invited research , and collaborative publications. Meghan was the first to enc ourage my ambitions and provided a contin ued support system/motivational speaker when I faltered. In complete honesty, without Meghan this dissertation would have never happened. For all of this, and more, I am truly grateful. The other members of my comm ittee include Dr. Lynne Goldstein, Dr. Jo and Dr. Randy Schaetzl. I sincerely appreciate their candid feedback and constructive criticism. This dissertation is what it is because they each took the time to explain to me how it could be better; or how I should step back and reassess the problem. Being on a dissertation committee is a task that requires a lot of time but comes with little reward and for this I am deeply appreciative of this. My research would not have been possible without early supporters including Bob Vande who was the first to excitedly suggest experime ntal cache pits, The Burt Lake Band of Ottawa Indians for telling me your stories an d giving me insight into traditional practices, Leslie Bourquin from the Department of Food Science and Nutrition for guiding my food safety studies , and Dean Anderson and all of the SHPO staff for providing excavation permits and all of the cache pit data for the State of Michigan. Appreciation also needs to be given to my colleagues Susan Kooiman, Sean Dunham, Mike Hambacher, Frank Raslich, and Nicole Raslich for not only acting as a sounding board, or providing words of encouragement, but also never hes itating to share their data , or buy me a vii beer. A big thank you to Annie and Sara for completing the mind - numbing task of editing my prose and formatting my slop. On the pe rsonal side , the side that motivated the maturation and completion of this dissertat ion , my family is to thank. Without th eir stability and patience this dissertation would have remained stagnant. I am grateful to my parents for never discouraging my far - fetched dream of becoming an archaeologist. Thank you to my sisters, Meg and Annie, f or always keeping me grounded and providing an air - gasping, belly laugh when necessary. Also, thank you to my gra ndfather, whose thirst for knowledge, evidence d in his mile - high stack of books next to his night stand, instilled in me at an early age a desi re to ask why. Though he is too young to remember my hours of maternal neglect while holed away writing, I am gr ateful to my son, Elliott, for, hopefully, never holding it against me and for helping me perfect the skill of multi - tasking. Thank you for sho wing me the meaning of unconditional love. My deepest gratitude goes to Evan, m y husband, my high school and for ever sweetheart, my best friend, and the most selfless person I know. The spouse of an archaeologist leads an interesting, and difficult life. Not only are archaeologists gone for weeks or months at a time, but we ask weird things of our spouses, like soak ing birch bark in our bathtub, or spending hours collecting acorns, or even buying seven food dehydrators so I can dry 50lbs of blueberries. To find someone who responds well to this lifestyle is probably difficult, but I got damned lucky. Evan has not onl y supported all of my endeavors, he has done so without a single word of doubt, or a grain of guilt. In all honesty, I have missed many a propo sed deadline in my PhD career and a five - year plan turned into seven; Evan never asked , , can I make it Once we became parents our life required more juggling, but even the viii countless hours and late nights I h ad to spend in my office writing , Evan never held a single second over my head. That is true selflessness. So, thank you, to my Evan, for being by my side as we juggle life . ix TABLE OF CONTENTS LIST OF TABLES ................................ ................................ ................................ ................................ .. x i i LIST OF FIGURES ................................ ................................ ................................ ............................... x ii i 1.0 INTRODUCTION ................................ ................................ ................................ ............................. 1 1.1 Direction of Research ................................ ................................ ................................ ................... 3 1.2 Theoretical Orientation: Decision - Making and Optimal Foraging Theory ................................ .. 6 1.3 Methodology and Dissertation Organization ................................ ................................ ................ 9 2.0 ENVIRONMENTAL AND CULTURAL BACKGROUND ................................ ......................... 13 2.1 Environmental Background ................................ ................................ ................................ .......... 13 2.1.1 Glaciers and Land Formations ................................ ................................ ............................ 13 2.1.2 Southern Lower Michigan versus Northern Lower Michigan ................................ ............ 15 2.1.3 Discussion ................................ ................................ ................................ ........................... 19 2.2 Cultural Background ................................ ................................ ................................ .................... 19 2.2.1 The Biotic Zones Model ................................ ................................ ................................ ..... 29 2.2.2 Inland Shore Fishery ................................ ................................ ................................ ........... 31 2.2.3 Ecosystem and Land Use Model ................................ ................................ ......................... 32 2.2.4 Inland Foragers Model ................................ ................................ ................................ ........ 36 2.3 Late Woodland Models and Foo d Storage ................................ ................................ ................... 3 8 2.4 Summary ................................ ................................ ................................ ................................ ...... 3 9 3.0 RISK MANAGEMENT STRATEGIES FOR NON - SEDENTARY SOCIETIES .................... 4 1 3.1 Understanding Risk ................................ ................................ ................................ ...................... 4 1 3.1.1 Defining Risk ................................ ................................ ................................ ...................... 4 3 3.2 Risk Management Strategies ................................ ................................ ................................ ........ 4 6 3.2.1 Risk Management a nd Exchange ................................ ................................ ........................ 4 6 3.2.2 Risk Management and Diversification ................................ ................................ ................ 4 7 3.2.3 Risk Management and Movement ................................ ................................ ...................... 4 9 3.3 Risk Management and Storage ................................ ................................ ................................ ..... 5 1 3.3.1 Types of Food Storage ................................ ................................ ................................ ........ 5 1 3.3.2 Food Storage Temporal and Scalar Variability ................................ ................................ ... 5 3 3.3.3 Food Storage and Move ment ................................ ................................ .............................. 5 6 3.4 Shortcomings of Food Storage T erminology ................................ ................................ ............... 5 7 3.5 Measuring Risk and Uncertainty ................................ ................................ ................................ .. 5 9 3.5.1 OFT and Front - Back Loading Model ................................ ................................ ................. 6 1 3.5.2 Diet Breadth, Front - Back Loading and Food Storage ................................ ........................ 6 2 3.5.3 Critiques of OFT ................................ ................................ ................................ ................. 6 2 3.6 Summary ................................ ................................ ................................ ................................ ...... 6 3 4.0 MICHIG AN STATE UNIVERSITY SUBTERRANEAN STORAGE RE SEARCH EXPERIMENT (MSU SStoRE) ................................ ................................ ................................ ....... 6 4 4.1 Experimental Archaeology ................................ ................................ ................................ ........... 6 5 4.2 Previous Food Storage Experiments ................................ ................................ ............................ 6 7 4.3 Experimental Subterranean Food Storage ................................ ................................ .................... 6 9 4.3.1 Round One ................................ ................................ ................................ .......................... 70 4.3.2 Round Two ................................ ................................ ................................ ......................... 7 4 x 4.3 .3 Round Three ................................ ................................ ................................ ........................ 76 4.3.4 Discussion ................................ ................................ ................................ ........................... 8 1 4.4 Acorn Processing Experiment ................................ ................................ ................................ ...... 8 1 4.4.1 Acorns in the diet ................................ ................................ ................................ ................ 8 2 4.4.2 MSU SStoRE Part two - Acorn Processing Pit ................................ ................................ .... 8 3 4.4.3 Conclusions ................................ ................................ ................................ ......................... 8 6 4.5 Discussion: The Work of Storing Food ................................ ................................ ........................ 8 7 4.6 MSU SStoRE Conclusions ................................ ................................ ................................ .......... 9 1 5.0 HUNTER - GATHERER ST ORAGE PRACTICES ................................ ................................ ....... 9 3 5.1 EnvCalc Data Set ................................ ................................ ................................ .......................... 9 3 5.2 Using the Human Area Relations Files ................................ ................................ ........................ 9 4 5.3 The Dataset ................................ ................................ ................................ ................................ ... 9 6 5.4 The Variables ................................ ................................ ................................ ............................... 9 9 5.5 Statistical Analysis ................................ ................................ ................................ ....................... 100 5.5.1 The Results ................................ ................................ ................................ ......................... 100 5.6 Variation of Food Storage ................................ ................................ ................................ ............ 10 5 5.6.1 Gi lyak ................................ ................................ ................................ ................................ . 10 5 5.6.2 Tubatulabal ................................ ................................ ................................ ........................ 10 6 5.6.3 Quinault ................................ ................................ ................................ .............................. 10 7 5.6.4 Pomo ................................ ................................ ................................ ................................ .. 10 8 5.6.5 Copper Inuit ................................ ................................ ................................ ....................... 10 9 5.6.6 Ingalik ................................ ................................ ................................ ................................ 10 9 5.6.7 Ojibwa - Round Lake ................................ ................................ ................................ .......... 1 10 5.6.8 Discussion ................................ ................................ ................................ ........................... 1 11 5.7 Understanding Hunter - gatherer Movement ................................ ................................ .................. 11 4 5.8 Conclusion ................................ ................................ ................................ ................................ .... 11 7 6.0 MODELIN G HUNTER - GATHERER FOOD STORAGE ................................ ........................... 11 9 6.1 Food Storage in Michigan ................................ ................................ ................................ ............ 11 9 6.1.1 Landscape Distribution of Storage Features ................................ ................................ ....... 12 4 6.1.2 Movement and Food Storage ................................ ................................ .............................. 12 7 6.1.3 Evidence for Foodstuffs Stored ................................ ................................ ......................... 12 8 6.2 Late Woodland Settlement and Subsistence Chan ge s ................................ ................................ .. 12 9 6.2.1 Fitting Food Storage into Michigan Models ................................ ................................ ....... 1 30 6.3 Modeling Late Woodland Seasonal Gap Storage in Northern Lower Michigan .......................... 13 2 6.3.1 Late Woodland Setting ................................ ................................ ................................ ....... 13 2 6.3.2 Seasonal Gap Storage Model ................................ ................................ .............................. 13 5 6.3.3 Delayed Returns and Back - loaded F oo dstuff ................................ ................................ ..... 13 6 6.3.4 Mass Capture and Selectionist Models ................................ ................................ ............... 13 8 6.3.5 The Role of Women in Resource Selection ................................ ................................ ........ 13 9 6.3.6 Storage Features as Emplaced Facilities ................................ ................................ ............. 1 40 6.4 Summary ................................ ................................ ................................ ................................ ...... 1 4 1 7.0 CONCLUSION AND FUTURE RESEARCH ................................ ................................ .............. 14 3 7.1 Research Results ................................ ................................ ................................ .......................... 14 7 7.2 Summary of Research ................................ ................................ ................................ ................... 1 50 7.3 Future Research ................................ ................................ ................................ ........................... 15 3 APPENDICES ................................ ................................ ................................ ................................ ......... 1 5 6 xi A PPENDIX A : The Green Site ................................ ................................ ................................ .......... 1 5 7 A PPENDIX B : The Ferrell Ridge Site ................................ ................................ ............................... 1 6 1 A PPENDIX C : Description of Excavated Storage Features from the Late Woodland Period in ................................ ................................ ................................ ............. 1 7 1 REFERENCES ................................ ................................ ................................ ................................ ........ 1 7 4 xii LIST OF TABLES Table 2.1: Inferences and Expectations regarding current models and food storage ................................ 38 Table 4.1: Water Activity Analysis, Round 3. ................................ ................................ .......................... 7 8 Table 4.2: Mycotoxin Analysis for Round 3 Foodstuffs.. ................................ ................................ ......... 80 Table 5.1: Hunter - ga therer groups that can be found in both eHRAF and EnvCalc ................................ 9 7 Table 5.2: K - means clusters organized into their respective groups. ................................ ....................... 10 4 Table 5.3: Ethnographic groups and their respective movement ................................ .............................. 11 5 Table 6.1: Recorded Storage Feature (cache pit) sites i n Michigan ................................ .......................... 1 20 xiii LIST OF FIGURES Figure 2.1: L eft , Map of Michigan depicting the Mesic - Frigid soils boundary . Right, Map of Michigan depicting the boundary for the Spodosol - Alfisol Provinces ................................ ... 18 Figure 2. 2 : The biotic zones with their correlated historic Native American cultural groups , as designated by the Biotic Zone Model ................................ ................................ ...................... 30 Figure 3.1: Hypothetical Z - Score Model ................................ ................................ ................................ .. 4 4 Figure 3.2: Z - Score Model with an increased R ................................ ................................ ....................... 4 5 Figure 4.1 : L eft , Unexcavated cache pit, visible as a sl ight surface depression. R ight , Cross - section of an excavated cache pit ................................ ................................ ................................ ........ 7 1 Figure 4.2 : Cross - section profile of storage feature excavated by Meghan Howey; 2007 . ...................... 7 1 Figure 4.3 : MSU SStoRE Round 1 pit lined with birch bark and insulated with straw. ........................... 7 2 Fig ure 4.4 : Blueberry and trout storage containers after Round 1 ................................ ............................ 7 3 Figure 4.5 : Round 2 b lueberries after reopening the pits ................................ ................................ .......... 7 5 Figure 4.6 : Round 2 acorns after reopening ................................ ................................ .............................. 7 6 Figure 4.7 : Round 2 data logger results for Cache Pit 1 ................................ ................................ ........... 7 6 Figure 4.8 : Left , Roun d 3 Acorns, Hominy, and Flint C orn after reopening. Right , B lueberries recovered after Round 3. ................................ ................................ ................................ ......... 7 9 Figure 4. 9 : Green Site Acorn Processing Pit ................................ ................................ ............................ 8 2 Figure 4.1 0 : Round 2 Acorn Charring ................................ ................................ ................................ ...... 8 7 Figure 4.1 1 : Round 2 remnants of charred acorns after burning. ................................ ............................. 8 7 F igure 5.1: PCA output plotted two dimensionally with PC1 and PC2 ................................ .................... 10 2 F igure 5.2: PCA plot with K - means cluster analysis ................................ ................................ ................ 10 3 Figure 6.1 : Estimated hunting territory for a hunting - gathering group inhabiting northern lower Michigan ................................ ................................ ................................ ................................ 13 4 Appendix Figure 1: U SGS Indian River Quadrangle. ................................ ................................ .............. 1 64 1 1.0 INTRODUCTION Food storage , or the act of extending the shelf life of foodstuffs, often formed an important part of the adaptations of certain small - scale hunter - gatherer and low - level horticulturalist societies. Res earch on stora ge in small - scale societies has, until recently, narrowly focused on either establishing the conditions under which food storage was utilized, i.e. evaluat ing what combination of seasonal/annual resource predict ability, abundance, and reliabi lity led to fo od storage among hunter - gatherers; or, determining the form an d scale that food storage took, i.e. small - scale, portable storage or larger - scale, container or other fixed storage, and its relatedness to increasi ng social complexity. Although these lines of inquiry have been , and remain , critically important, much current research has not focused sufficiently on the role of the physical storage container and the associated larger system of decision - making and ris k management within which such physical facilities are embedded. An expanded and altered focus results in some alternative questions, such as: What were the technological risks associated with the technology? How was the risk in technology negotiated? What other risk management systems were utilized by hunter - gatherers in conjunction with food storage? For storage to be an applicable risk management strategy, the technology required for successful storage must be reliable. Knowledge of proper storage pit co nstruction and food preservatio n techniques is required before the risk and uncertainty of food storage can be mitigated. Due to the increased front - end energy output, the risk associated with the processing and storage of the foodstuffs must be minimal, a s compared to the seasonal risk of food scarcity and potential energy/caloric gain. The northern Great Lakes region provides a particularly robust location within which to explore the impact of food storage technology because during certain time periods, cache pits 2 (subterranean food st orage containers) are the most common archaeological feature in the region (Howey 2006) . Subterranean storage pits appear in the archaeological landscape after ca. AD 1000 and tribal communities continued to use them through the historic period (Dunham 2 000). During the late Late W oodland (ca. AD 1000 - 1600 ), subterranean food storage containers were systematically us ed by tribal communities with a spatially and seasonally restricted fisher - forager - horticulturalist subsistence system to create a stable fo od supply ( Dunham 2000; Holman and Krist 2001; Howey and Parker 2008; Howey and Frederick 2016). The act of food sto rage, actualized through the technology of subterranean storage pits, allowed groups to increase their communal capacity for surv ival, succe ss, and regeneration; these capacities were stressed by the increasingly restrictive setting of the Late Prehistoric sociopolitical landscape. Combining experimental archaeology, ethnographic and ethnohistoric data, along with archaeological data on food storage, this dissertation examines the technology of subterranean food storage utilized by hunter - gatherers and mix ed horticulturalists. This research aimed to understand the use of food storage through the perspective of the technology of the physical co ntainer and the decision - making behind this risk - management strategy. Additionally, data was collected on food stora ge practices, both past and present, across the globe. Factors, such as movement strategy, climate, environment, economy, and to a lesser ex tent socio - political triggers, were considered. These factors were then compared and analyzed for patterns in decisi on - making. Further, this research and case study is set into the Late Woodland landscape of northern lower Michigan. With a baseline for u nderstanding why storage is selected in other, similar settings, my research considered whether similar variables we re at play in the Late Woodland period, and what other factors may have driven the decisions to store. This research then aimed 3 to understan d the underlying factors that affected the decision - making behind the use of food storage. Ultimately, in my case st udy, I wanted to understand how subterranean storage technology was utilized during this period. Was it narrowly used as a risk management s trategy; a cushion for times of need? Or, was it a regularized mainstay technology used throughout the year? If it w as a risk management strategy, why in northern Michigan was it selected over other risk management strategies such as mobility, diversificat ion or exchange? Did the primacy of food storage change over time? 1.1 Direction of Research To focus my researc h efforts, I laid out three primary questions, along with expectations of research results and how each question was evaluated. The overarch ing goal of this dissertation was to offer insight into the question of: Why is there a proliferation of subterranea n food storage in Michigan after AD 1000? Understanding the decision - making behind hunter - storage requi res an initial understanding of the technology of food storage. I maintain that the technology of subterranean food storage is best comprehended by, first, successfully recreating the act of food storage. My early research (see Chapter Four) focused on the technology of the food storage feature, its reliability in preserving food, and its inherent risk. Several iteratio ns of this actualistic experimental archaeology research have helped answer some initial questions concerning the technology of food storage , but they have also expectably generated additional questions regarding the decision - making behind the use of food storage as a risk - management strategy. 4 Research Q uestion #1: How much risk is involved in the act of food storage? Expectations: Comparat ive analysis of global food storage practices, along with data from experimental research (see Chapter Four), will be used to evaluate risk. I expect to find that when food storage is used as a primary risk buffering strategy the stor age pits are risk aver se. It is not until the technology of subterranean food storage is reliable, or presents minimal risk, that food storage is utilized. I also expect to find that in hunter - gatherer communities where food storage is known to be less rel iable, it is not used as a primary risk management strategy. Research thus far has indicated that after the technology of food storage had been perfected, it was a reliable risk management strategy for the Late Woodland people of northern lower Michigan. I expect that subterran ean food storage was equally reliable in other regions. I will draw on the data collected from Michigan State University Subterranean Storage Research Experiment ( MSU SStoRE ) research to evaluate risk, namely whether storage in a temp erate climate is risk averse or risk prone. This experimental data will also be used to fill informational holes in the ethnographic/ethnohistoric record. The data collected from the Ethnographic/Ethnohistoric record will be evaluated for mention of risk w hen utilizing food sto rage. A Z - score model , which measures the level of risk versus reward, will be used to determine if and when food storage is risk averse. Research Q uestion #2: What was the range of variation of hunter - gatherer and mixed horticultural food storage tec hnologies as ethnographically documented around the world? And how does this variation determine whether food storage is a primary risk management strategy? 5 Expectations: I expect to find that hunter - food storage pr actices follow similar patterns based on differing causal or primary decision variables, i.e. groups in similar climates will utilize food storage in a simila r manner. a, along with the Electronic Human Area Relations Files (eHRAF) to determine which ethnographically documented hunter - gatherer groups utilized food storage. O nce the results are narrowed, I will do an in - depth data collection on the ethnographic groups to understand the specifics of their storage behaviors; including foodstuffs stored, storage technology, front - loaded vs. backloaded processing, correlation of s torage to other risk management strategies, and individual vs. group behavior. These variables wil l then be statistically evaluated to determine reationships between and across variables. If patterns in the variables exist, then these patterns will be comp ared to the storage behavior taking place in the Late Woodland of northern lower Michigan. Resear ch Q uestion #3: Why was subterranean food storage selected as the primary risk management strategy during s L ate Woodland period ? Expectations: I exp ect to find that by the late Late Woodland period, the technology and methodology for subterranean food storage had been enhanced, leading to reliable food preservation, and therefore was selected over other risk management strategies. I anticipate that the pattern of subterranean food storage in Michigan will manifest as a more encompassing behaviora l pattern of food storage around the world. That is, the decision - making behind the food storage in Michigan, is more than likely, similar decision - making in 6 other ethnographically documented cases. It will be argued that there are a certain set of factors that come into play in the Late Woodland period of Michigan that propelled hunter - gatherers into a new risk - management strategy, a strateg y that revolved around food storage. By understanding the use of food storage around the world and the decisions behi nd its use, this research will examine the possible factors that led to successful food storage in Michigan. 1.2 Theoretical Orientation : Decision - Making and Optimal Foraging Theory , tuates around a hunter - gatherer/mixed econo my groups , this essential factor is often food. A population cannot exceed what it can feed during the leanest times of any given year . Smith (1972) explains that when food resources become scarce, population pressure s increase , requiring changes in subsis tence strategy, leading to decision - making. Decision - making is weighing the choice - decis ions, or accumulated knowledge and experience, and deciding on a positive outcome (Mithen 1990). Hunter - gatherers rely on meta - decisions ga thered through information exchange, stored information, and environmental cues to determine what strategy will allev iate their population pressures (Mithen 1990). Increased logistical hunting, highly mobile residential hunting, de - aggregation, expanded fo od sources, and food storage are all risk management strategy options for hunter - gatherer decision - making (Smith 1972 ). Mithen (1990) argues that, at its core, decision - making is based on adaptation, or successful achievement of intermediate goals that lea d to adaptation. Successful decision - making for hunter - gatherers would result in more efficient hunting/gathering beh avior (M ithen 1990:5). Although Law posits that food is the limiting factor for sustained population growth, Boserup (1993) assert 7 in turn, an increased foo ds supply (Boserup 1993:12). Boserup (1993) posits that because increased food production accomplished through decreasing the fallow (time between plantings), results in reduced output and increased labor input , it is a poor adaptive decision, unless the l arger population required s , if extended to hunter - gatherers, implies that decision - making leading to c hanges in subsistence strategies is an adaptive response to food scarcity. I argue that due to the increased labor in put, food storage is only an adaptive response to food shortage when the reliability of the technology outweighs the risk of failure. The decisions behind the specific resources collected is another important consideration in the study of food storage. De cision theory attempts to model rational behavior by lities and human decision s are based on weighted options that result in non - random outcomes; generally human decisions are goal oriented (Brouwer 2011). Decisions a re weighted depending on the factors at hand; whether it is resource extraction, fight or flight reactions, or front - loading labor to create food storage, decisions are dependent on the activity to be undertaken, but are weighed based on goals (Brouwer 201 1; Jochim 1976). The decision - based model operates under the assumption that the hunter - gatherers of the Late Woodlan d period of Northern Michigan made rational decisions about economic and settlement pursuits, which were generally organized well in advan ce . Since decision - making is non - random and based on weighted options it has the potential to be predicted. There is one more important caveat to make regarding decision - making theory. The idea oncept that can never truly predict all human decisions 8 motivations, circ umstances, and information can, (Brouwer 2011:187). However, it is a useful tool when mapping decision making. - maker, as this greatly affects the weighted options (Hansson 1994). The state of the decision - knowledge falls into three categories: certainty, risk, and uncertainty (Hansson 1994). Jochim among hunter - gatherers, the would seem to operate, since the exact probabili ties of the consequences of various economic choices are not known, but at best are estimated from previous experie nc (Jochim 1976:5). Since food storage is a risk management activity created for times of uncertainty, creati ng a model around such decision - making potentially creates an uncertain outcome. To understand strategies behind r esource selection, I will use principles derived from evolutionary ecology and Optimal Foraging Theory (OFT). OFT posits that human behavior is rational, and that decision - making will be based on the most energy efficient food selection (Kelly 1995). Using a Front - Back loadi ng model (Tushingham and Betting er 2013) to determine whether a food requires more energy to forage and collect or to proc ess also explains food selection of hunter - gatherers. Like Bettinger et al. (1997), Tushingham and Bettinger (2013) find that because acorns are a back - loaded food resource (more energy to process) , they are often chosen over other resources, (salmon in th is case) even though acorns provide less total nutrition. In the case of risk - management, acorns were chosen for th (Tushingham and Bettinger 2013). OFT models usually predict that hunter - gatherers will always choose la rge animals ove r small animals in order to counteract en ergy output (Kelly 1995 ; Madsen and Schmitt 1998; Ugan 2005 ). The energy output must equal or be greater than the 9 energy input. However, when groups have the ability to mass collect low - level resource options, the energy balance is altered (Ugan 2005; Madsen and Schmitt 1998). If a technology is created that allow s for mass capture of an otherwise low - level resource, then in some cases the low - level resource is more cost effective than a high - level res ource (Madsen and Schmitt 1998). Ugan (2005) finds that mass collecting is not always chosen over high - level resour ces because the back - loading costs sometimes outweigh the mass capture. It is more time efficient to process one large animal as opposed to h undreds of small animals (Ugan 2005). Further, the selectionist model will be applied to explain how an initially l ow - ranked resource can acquire a higher ranking when the technology (storage) behind its intended purpose gains fitness (Church and Nass 2002 ); if the technology of food storage becomes reliable, then the food stored acquires a higher ranking. The OFT and selectionist model will both aid in interpreting the decisions behind the continued utilization of storage technology. 1.3 Methodology and Dissertation Organization The methodology for this research is multi - pronged and includes, ethnographic/ethnohisto ric data collection and modeling, replicative archaeology, and comparative analysis of storage practices globally. Each methodological prong will converge for the final analysis of food storage in northern lower Michigan during the Late Woodland period. Th e ethnographic/ethnohistoric research has been utilized in two ways: to inform on the replication of the storage features, and additionally, ethnographic/ethnohistoric data was collected during the comparative analysis of storage practices, along with data from the Human Relations Area Files (HRAF). The replicative archaeology data will aid in evaluating the risk of food storage practices globa lly. The data from the replicative archaeology and the data from the 10 comparative analysis are then merged to interp ret the behavior behind the utilization of food storage during the Late Woodland period of northern lower Michigan. Chapter Two provides a b glaciers and the ensuing peopling of Michigan up unt il European contact in the mid - Seventeenth Century. This background will provide insight into the decision - making in settlement and subsisten ce patterns and shed - light on why the patterning of the Late Woodland of northern lower Michigan becomes an outlier from preceding patterns. Chapter Two also lays out the current explanatory models for the shift in settlement and subsistence patterns and e xamines which model correlates best with the utilization of food storage. Chapter Three addresses the concepts of risk management and storage. Though storage can be utilized in other capacities, this dissertation focuses on its role as a risk management s trategy; or a strategy utilized to mitigate uncertainty. Storage is only one mechanism used to manage risk. Others include; movement, diversification, and exchange; these are briefly discussed. Definitions of storage, along with all of its manifestations, and storage terminology are explained in this chapter along with an overview of general storage research. A critiqu e of the current storage definitions and research is laid out and a more focused terminology is offered. The Michigan State University Subte rranean Storage Research Experiment (MSU SStoRE), a data set based on replicative archaeology, was the initial prong of this research. Chapter Four explains MSU SStoRE in detail and how the results aided in the interpreta tion of food storage risk. This ser ies of experiments aimed to recreate the behavior and technology present in the subterranean storage features excavated in northern lower Michigan (Howey and Frederick 2016). The first prong of the research was to underst and the technology and mechanics, a nd in turn the inherent risk, of subterranean food storage through a series of 11 experiments, i.e. actualistic experimental archaeology. The data collected from MSU SStoRE were necessary to interpret the storage behavior oc curring in northern lower Michigan in the Late Woodland period. MSU SStoRE data have given a more robust understanding of the technology of storage, and the inherent risk when the technology is not reliable. Ethnographic and ethnohistoric data are necess ary to plug the information gaps le ft by archaeology. Chapter Five provides background on the two ethnographic data sets utilized to perform a cross - cultural comparative analysis of food storage. Ethnographic and ethnohistoric research provides decision - ma king context for behaviors found in the archaeological record. In the Human Relations Area Files (HRAF). A coding system was created to collect data on variables such as, location, facility type, foodstuffs stored, the societal structure of the communities using the storage, the role of the storage as risk - management, and the decision - making behind these actions (i.e. is this a primary risk management strategy). O nce this data was narrowed down, th e newly coded data was input into the EnvCalc Program (Binford and Johnson 2014). The results of potential settlement and subsistence patterns in relation to food storage are discussed in detail. The data collected was an alyzed to create a decision - making model that weighed the variables from the collected data to determine when subterranean food storage became the most practical option. In later chapters this was applied to the use of storage in the Late Woodland of north ern lower Michigan. Chapter Six is a culmination of the research where all the data prongs come together to analyze the utilization of food storage in northern lower Michigan during the Late Woodland period. The current models for the Late Woodland are an alyzed against the food storage pat terns 12 for best fit and the model laid out in Chapter Five is tested against the food storage pattern in Michigan. Finally, Chapter Seven reports the final conclusions. I explain why further analysis should be conducted to support the conclusions I reache d. Additionally, I offer suggestions for future research and detail how this research can be applied globally. 13 2.0 E NVIRONMENTAL AND CULTURAL BACKGROUND In an effort to contextualize the preceding research, this chapter will provide a brief backgrou nd on the environment, including geology, climate, vegetation, and soils as well as odland period, ending with the European entrance into the region. While entire vo lumes can, and have been , written on thes e topics, it is my intention to only focus on the environmental and cultural events that impacted the socio - political climates of the Late Woodland period and directly pertains to this research. Hunter - gatherers mak e decisions about their settlement and subsistence strategies based on their surrounding environment. Daily reevaluations let them know if current strategies are effective, or further alterations are required. Understanding the environment in which the dec isions were being made leads to a better understanding of the changes that were implemented to create a successful strategy. 2.1 Environmental Background 2.1.1 Glaciers and Land Formations Prior to human occupation, Michigan was shaped by deglaciation. The modern terrain of northern lower Michigan is attributable to a complex glacial history. The Laurentide ice sheet, which covered most of northern lower Michigan , began its retreat around 13,000 years ago. This initial retreat created drumlins and esker s, forms of subglacial landforms, recessional morainal features, drainage systems, and the basins for the Great Lakes (Dunham 2014; Schaetzl et al. 2000). Subseque nt oscillating readvance and retreat episodes further sculpted the landscape, resulting in g lacial landforms including moraines, kames, kettle lakes, and glacial outwash 14 plains. The final ice advance, the Greatlakean (formerly known as Valderan) advance, b egan 11,800 years ago, and covered parts of northern lower Michigan. The Greatlakean episode had retreated by 11,200 years ago, but is responsible for glacial deposits and landforms such as the drumlin fields of northern lower Michigan (Schaetzl et al. 200 0). As the ice sheets continued to retreat north, a series of postglacial lakes, with chang ing ( Larsen 1999; Egan 1993). The main phase of Lake Algonquin began 11,200 years ago. At its maximum, Lake Alg onquin (184 m) filled the Lake Michigan basin, the Lake Huron basin, covered portions of the northern tip of the Lower Peninsula, a majority of the Upper Peninsula, and the Lake Superior basin (Larsen 1999; Schaetzl et al. 2000). Continued retreat of the i ce margins resulted in partial draining of Lake Algonquin around 10,000 years ago, leading t o a reduced water level, known as the Stanley low (48 m) (Larsen 1999; Schaetzl et al. 2000; Egan 1993). As surrounding lake outlets began to rise, the water level in the Great Lakes began to rise. By 8,000 years ago , the Nipissing lake stage (181 m) took hold and continued to rise to the Algoma lake stage, the maximum stage, 4,000 - 3,000 years ago. This series of lake levels are evident in the terraces that parallel the shores of Lake Michigan and Lake Huron. Modern Great Lakes levels (177 m) became steady around 3,000 years ago (Larsen 1999; Egan 1993). The story of the glacial history of Michigan is an important one; not only did the oscillating glaciers create t he landforms in Michigan, the melting glacial water filled the Great Lakes. The Great Lakes ultimately provided an integral food source for early inhabitants of Michigan, in the form of fish, and the sheer size of the lakes create a unique climate and weat her pattern for Michigan (Howey 2006; Schaetzl 2005). The moisture created by the lakes lea ds to - 15 in the winter months ( Keene 1993 akes warm slowly in the n creates an ameliorating affect which prolongs the growing season (or frost - free days) along the coastal zones of Topograph y combined with climatic differences caused by lake - effect weather has created two distinct ecosystems across the Lower Peninsula; southern lower Michigan and northern lower Michigan. To give further background on the environment of Michigan, it is necessa ry to separate these two main ecosystems. 2.1.2 Southern Lower Michigan versus Northern Lo wer Michigan There are several differing names for the two environmental zones of Michigan, Mesic - Frigid Boundary (soil temperature), Spodosol - Alfisol Provinces (Soil type), Tension Zone (vegetation), topography, climate (specifically snow - pack), depending on the topic being studied, but the line of transition is similar for all of these environmental variables. When considering the environment, it is important to understand the ecological interactions. Climate, vegetation, and soil have symbiotic relations hips; process es of one affect the other (Kapp 2000). To understand ecosystem. The topography of southern lower Michigan consists of a rather homogenous series of rolling hi lls and flat lake plains (Howey 2006). Lake Huron and Lake Erie make up the eastern boundary, and Lake Michigan the Western. The Saginaw Bay is the eastern border that creates the distinctive mitten shape of Michigan. Northern lower Michigan i s bounded by the conjoined Lake Michigan basin to the west and Lake Huron basin to the east, forming the Michigan - Huron basin. These two lakes meet at the Mackinac Straits, the area that separates the Lower Peninsula 16 from the Upper Peninsula. The topograph y of northern lower Michigan can be divided between the low coastal zones and the interior High Plains (Howey 2006). The High Plains are distinguished by their end moraines and glacial outwash, which creates a prominent plateau in the interior of northern lower Michiga n (Howey 2006; Albert 1995). The Maritime Tropical airmasses that form over the Gulf of Mexico and the ameliorating ad to warmer weather and a longer g rowing season compared to northern lower Michigan. This longer growing season, 120 - 140 days, potentially allow ed for productive precontact agriculture in this region. The juxtaposed topography of northern lower Michigan leads to climatic differences. The coastal zone s of northern Michigan are affected by the ameliorating effects of lake - effect weather. This weather phenomenon, created by the air masses moving over the Great Lakes, leads to the air warming more slowly in the Spring, and cooling more slowly in the Fall (Albert 1995). The prolonged cooling in the Fall allows for a longer growing season along the coastal zones. The interior high plains do not receive the ameliorating effects of the Great Lakes, and instead have a harsh climate, intense winters with a deep snow pack and single - digit air temperatures due to the higher altitudes and northerly latitude. A north - south boundary is also evident in the flora ; and Wagner 19 81). The warmer climate of southern lower Michigan has a vegetation profile that includes deciduous forests, oak savannas, beech - sugar maple forests, elm - ash forests, oak - hickory forests, and tall grass savannas (Schaetzl and Isard 1991; Alber t 1995; Howey 2006). North of this floristic boundary 17 consists of mixed coniferous - deciduous (hemlock and birch) forests and stands of red and white pines (Schaetzl and Isard 1991; Albert 1995). Soils are another integral environmental variable that shoul d be discusse d. Not surprisingly, the soils of Michigan have a similar dichotomous pattern to the climate and vegetation patterns. The soils in Michigan can be divided north from south through the Mesic - Frigid (M - F) Boundary (Schaetzl et l. 2005). The M - F Boundary was (Schaetzl et al. 2005:2036); mesic soils were more suitable for corn agriculture, whereas frigid soils were only suitable for hardier agric ulture. Southern Lower Michigan is distinguished by its Mesic soils or soils whose mean average temperature is 8° - 15°C . Northern lower Michigan is in the Frigid soil boundary, which has average soil temperatures between 0 ° - 8 ° C. Figure 2.1 s hows the amel iorating effect of the Great Lakes on soil temperature in the coastal zones. Grand , - climate that allows for productive fruit agriculture (Schaetzl and Isard 1991). 18 Figure 2.1: Left, Map of Michigan depicting the Mesic - Frigid soils boundary (adapted from Schaetzl et al. 2005). R ight , Map of Michigan depicting the boundary for the Spodosol - Alfisol Provinces ( A dapted from Schaetzl and Isard 1991) . The boundary for the Spodosol - Alfisol Provinces has a similar trajectory across Michigan as the M - F Boundary (see figure 2. 1 ) (Schaetzl and Isard1991). Across southern lower Michigan the soils mostly consist of Alfisols, soils with a clay - enriched B horizon (Schaetzl and Isard 1991). The clay enriched B horizon also makes Alfisols more productive for agriculture. The soils of northern lower Michigan are primarily Spodosols; sandy, acidic soils that are characterized by the subsurface accumulation of Al and Fe oxides (Schaetzl http://geo.msu.edu/extra/geogmich/spodosols.html ). Development of Spodosols in northern lower Michigan is aided by the early on - set snow pack that insulates the soil, preventing deep freezing. Lack of frozen soil allows for more water infiltration during the spring snow melt, and water infiltration is the primary mechanism for podzolization, or the creation of Spodosols (Schaetzl and Isard 1991). 19 2.1.3 Discussion Contextualizing the physica l environment of Michigan allows for a better understanding of the decision - making of hunter - gatherers. There is a significant environmental change as one travels north through Michigan, and this would have influenced economic strategies. Based on climate and soil charact eristics, the majority of northern lower Michigan would have been unsuitable for reliable agriculture, except in the immediate coastal zones. It is also important to point out that the Spodosols that are prevalent in northern lower Michiga n are an uncommo n soil order (there are 11 major soil orders) (Sch ae tzl http://geo.msu.edu/extra/geogmich/spodosols.html ). Spodosol soils, in a Frigid soil temperature regime, in a climate wi th a deep, insul ating snow pack, are all variables that allow for efficient subterranean storage. This is a topic that will be further explored in Chapter Four. 2.2 Cultural Background Since the Paleo - Indian period, Native American communities living in Michigan have c reated socio - economic systems that negotiate the hostile environment of northern lower Michigan. Understanding the decisions behind settlement and subsistence strategies of these Native American communities at different periods throughout history is integr al in comprehending their risk management strategies during the Late Woodland period. The initial peopling of the Great Lakes region began at the end of the Pleistocene epoch, between 12,000 - 10,000 B.P. (Shott and Wrigh t 1999). The Paleoindian period (12,0 00 - 8,000 BC) has regional variation both chronologically and culturally across North America. In Michigan, a majority of the Paleoindian sites occur in the southern half of the Lower Peninsula, with only a handful of si tes occurring in northern lower Michi gan (Holman, Ho lman and Cleland 20 1998). Gainey, the earliest phase of Paleoindian occupation in Michigan, marked by the fluted Gainey points, dates to 10,900 B.P. Th e Gainey phase is followed by the Parkhill phase, chara cterized by Barnes points, 10,700 B.P . and then Crowfield phase, with Crowfield points, 10,500 B.P. (Holman, Holman and Cleland 1998). While these phases occur in southern lower, the Holcombe phase, characterized by minimally fluted Holcombe points and dat ed to 10,300 B.P., has been found in northern lower Michigan. The northern most Paleoindian sites to be recorded in Michigan include the Samels Field site (20GT2) near Traverse City (Dekin 1966; Cleland and Ruggles 1996), and the Kalkaska Bi - face locale 30 kilometers to the east (Lovis and Bo le 2002). The density of Paleoindian sites in southern lower Michigan, compared to the scarcity of sites in northern lower Michigan, indicates an initial peopling of Michigan from south to north (Holman, Holman and Cle land 1998). Before 12,000 B.P., north ern lower Michigan was a landscape still evolving from deglaciation; it was cold, tundra - like, consisting of newly emergent acidic soils, and few plant species. Lake Algonquin covered swaths of land and resulted in a se ries of archipelagos (Lovis 1991). Th is region turned to tall grass tundra (11,800 B.P), then to open spruce forests (11,300 B.P.). By 11,300 B.P., southern lower Michigan consisted of a more diverse habitat of deciduous forest (Holman, Holman and Cleland 1998). Though habitable, northern low er Michigan was less appealing to the Pleistocene fauna due to the insufficient soil nutrients. The deficient soils produced vegetation lacking required minerals and nitrates, making the habitat less than appealing to t he fauna of the Pleistocene (Holman, Holman and Cleland 1998). Since Paleoindian hunter - gatherers followed their food, if the food neglected northern lower Michigan, then so too, did the hunters. 21 It was long assumed that the Paleoindian economy was focuse d around the hunting of big game. Mor e recently, broader hunter - gatherer models have been used to interpret the new data. Many agree (Surovell 2000; Kelly and Todd 1988; Anderson and Gillam 2000) that Paleoindians were highly mobile, using a combination of residential and logistical mobility to adapt to the changing environment, leading to a rapid colonization of the Americas. Because the to aid in their hunting and gatherin g. A more generalized subsistence strategy was used. E astern North America would have been a fairly seasonally stable en vironment, leading to a species - rich landscape ( Meltzer and Smith 1986) . In turn, a species - r ich la ndscape is more amenable to a general ized subsistence strategy ; however, due to the lack of local plant knowledge, fauna would have been relied on heavily over flora ( Kelly and Todd 1998). Lithic analysis also indicates a flexible use technology. Raw mater ial was transported hundreds of miles (Kelly and Todd 1988; Meltzer and Smith1986) and used to create a toolkit that maximized potential functions, while minimizing number of tools required. Unlike Paleoindian sites to the north, in Ontario, which were sit uated along the coasts of Lake Algonq uin, sites in Michigan are located on sand ridges which consisted of varied habitats. This strategy allowed for Michigan Paleoindians to exploit various habitats, increasing their hunting success (Holman, Holman, and Cl eland 1998). This flexible, yet funct ionally non - specific toolkit indicates a generalized subsistence strategy. The Archaic period, divided into the Early Archaic (10,000 - 8,000 B.P.), Middle Archaic (8,000 - 5,000 B.P.), and Late Archaic (5,000 - 2,800 B.P.) i s often viewed as a period of continu al and dynamic re adaptation; it was an initial adaptation to the Holocene, and then an adaptation to the hypsithermal (Anderson and Hanson 2009; Koldehoff and Walthall 2009). In northern lower 22 Michigan, environmental reconstructions have been developed to better understand the landscape (Lovis 1991). There is almost no evidence for occupation of this region during the lower lake levels from 11,000 - 5,500 B.P. (Lovis 1991). It is not until the Nipissing sta ge (5,500 - 3,200 B.P.), or the Late Archaic period, wh en evidence occurs for occupation in northern lower Michigan. The Screaming Loon site along modern - day Burt Lake in the Inland Waterway dates to 3637 - 3591 B.P. (Lovis 1991). This site, situated on an Alg oma stage terrace, is surrounded by poorly drained ma rsh and wetlands, generally a productive habitat for multiple fauna. This site is also important because it is during a time of marked environmental transition within this region. The drainage of Inland Waterway shifted from Little Traverse Bay in the west , to Lake Huron in the east (Lovis 1991). M ajor environmental changes, such as changing lake levels, likely required a change in subsistence strategies, which in turn, led to changes in mobility and soci al organization. By the end of the Archaic period, th e rapid environmental changes that had been occurring since deglaciation had stabilized, but the Great Lakes were 60 m below modern levels (Shott 1999). By the Late Archaic, modern vegetation communities existed across Michigan, but Great Lake levels excee ded modern shores and fluctuated often (Robertson et al 1999). Though adaptations in subsistence strategies are the most evident characteristic to trace archaeologically, it is important to understand t he true complexity of the cultural systems and unders tand their interconnectedness (Anderson and Hanson 1988; Emerson and McElrath 2009). Adaptation to changing environments is most often accomplished by tweaking or refining existing strategies, not adopti ng completely new strategies (Styles et al. 1983). In the Great Lakes region, strategic decision making around subsistence strategies is also evident in the Saginaw Basin (Lovis et al. 2005). Middle Archaic groups used a logistical mobility strategy to 23 tak e advantage of the resources in the uplands and lowla nds, resulting in an efficient resource extraction strategy (Lovis 1999; Lovis et al. 2005). Late Archaic populations adapted to this dynamic environment by freely moving across multiple resource zones, creating extensive kinship networks, participating i n long - distance exchange of exotic material, and engaging in cooperative buffering (Robertson et al. 1999; Howey 2012). Though full - on plant domestication in Eastern North America is not evident until t he Terminal Archaic/Early Woodland threshold, there i s evidence for plant cultivation as early as the Middle Archaic ( Simon and Parker 2006 ; Simon 2009). Cultivation implies a level of decision - making around activities that boost plant productivity (Simon and Parker 2006; Simon 2009). While domesticated plan ts have been cultivated, not all cultivated plants are domesticated (Simon 2009). Simon (2009) explains that because domestication is a process, not an event, the behavior behind the process of domestica tion began in the Middle Archaic. Middle Archaic site s, including Notcha and Diane in the American Bottom, have evidence for use of sumpweed, chenopod, and ragweed, all plants in the Eastern Agricultural Complex (Simon and Parker 2006). These plants, and o thers in the EAC (squash, little barley, erect knotwe ed, goosefoot, marshelder, sunflower, and maygrass) all thrive in disturbed habitats (Simon and Parker 2006; Smith and Yarnell 2009). Other sites in the American Bottom including McLean, Go - Kart, Marge, Floyd, and Meyer all follow the pattern of cultivatin g plants in the soon - to - be EAC. These plants were added to existing subsistence strategy, and through cultivation, became domesticates (Smith and Yarnell 2009; Simon 2009). The earliest evidence of domes tication comes from the Philips Spring site in Missou ri where a domesticated squash seed was dated to 5025 B.P. Other early domesticates include marshelder (4400 B.P. Napoleon H o llow, I llinois ) 24 and sunflower (4840 B.P. Hayes Site, T ennessee ) (Smith and Yar nell 2009; Simon 2009). Three EAC species (chenopod, squash, and sumpweed) show cultivation by the end of the Middle Archaic (4000 BC), and two more (sunflower and ragweed) are evident by the end of the Terminal Late Arch aic. The Terminal Late Archaic/ Earl y Woodland transition has a seed record that indicate s low - level food production economies in several areas of the Eastern Woodland (Simon 2009). The Early Woodland period (800 BC AD 1) is marked by the emergence of ceramics; however, the socio - economic strategies of the Archaic continued mostly unchanged , but with a smattering of ceramics. There is very little evidence for any formal occupation o f northern lower Michigan during the Early Woodland period. This is an indication of the continuation of the environmentally volatile landscape of this region. It was only adequate for short - term logistical hunting that left only ephemeral traces of hunter - gatherer presence. Significant cultural developments become a defining factor of the Middle Woodland period (200 BC AD 400). Southern lower Michigan saw the a doption of the Middle Woodland Hopewell culture, regional integration activities that include long - distance exchange of exotic material and an elaborate mortuary ritual visible through the creation of large - scale mound centers (Charles and Buikstra 2002; C hivis 2016). In Michigan, the local cultural expressions of Hopewell are known as the Norton ( west Michigan) Tradition and the Saginaw (east Michigan) Traditions; these are known as Havana - Hopewell ( Kingsley et al. 1999 ; Chivis 2016). The influence of the Hopewell interaction sphere lasted from 200 B.C. to A.D. 400 (Seeman and Branch 2006). Complex ity of the Hopewell interaction sphere is evidenced by the expansive interaction sphere, construction of elaborate mounds and earthworks, and overarching mortuary ceremonialism that included items such as mica cutouts, obsidian blades, and exotic pipestone 25 pipes (Yerkes 2002; Chivis 2016). Defining traits of Hopewell were not evident at every Hopewell site, rather the local expressions of Hopewell cherry - picked ele ments to adopt that fit into their preexisting cultural values (Chivis 2016). Cowan (2006) ana lyzed lithic technology and determined that the abundance of bifaces at Hopewell sites indicated a highly mobile settlement system. Stable isotope analysis done o n Hopewell burials indicated that while weedy cultigens (sumpweed, goosefoot and chenopods) we re consumed, they were only a supplement to a diet based on fish, game, and wild plants ( Yerkes 2005 ; Chivis 2016). There is also a lack of evidence for agricultu re in terms of deep storage pits, agricultural tools, and skeletal markers like tooth wear (Ye rkes 2005). Based on the hunter - gatherer economy, the Hopewell mound centers would have been used as gathering places for these widespread communities. The mound centers would have acted as a place to schedule aggregation where dispersed groups come togeth er to maintain social ties (Yerkes 2002). Northern lower Michigan never adopted Hopewell. Instead , the onset of the Woodland is marked by the widespread adoption of ceramics (Dunham 2014). Ceramic traditions in northern Michigan are primarily characterized as Laurel (Middle or Initial Woodland) , which spread from northern lower Michigan up across the Upper Peninsula (Brose and Hambacher 1999; Lovis 1991). Settleme nt and subsistence patterns in northern lower Michigan suggest high levels of residential mobility with seasonal fishing, collecting, and hunting, and an increasing emphasis on exploitation of aquatic resources (Brose and Hambacher 1999; Cleland 1982). Con tinued fluctuations in Great Lake levels brought fluctuating cl imatic shifts; cool wet periods followed high lake levels, while warmer, dryer weather corresponded to low lake levels. 26 Until recently, maize was considered the trigger that catapulted Eastern North America into an agricultural, fully se dentary lifestyle (Parker 1996 ; maize from the south was quickly adopted by Middle and Late Woodland populations and became the dominant crop that led to increased sedentism, co mplex organization, and social hierarchy. In Michigan, this dra matic shift occurred around AD 1000, which saw an increase in village settlements, restricted territories, and increased complexity of social organization (Fitting 1975). It has since been esta blished that maize was introduced , through down - the - line exchan ge from Illinois, well before the dramatic shift and was slowly and increasingly adopted into the existing subsistence pattern throughout hundreds of years (Raviele 2010) . At the Schultz site i n the Saginaw River drainage of Michigan, evidence for maize us e was dated to 2120 BP, and in Ontario maize is evidenced in the diet by AD 500 (Hart and Lovis 2013 ; Raviele 201 0 ). Hart and Lovis (2013) argue that there is no direct link between the introd uction of maize and settled village life. Instead, the intensif ication of maize agriculture revolved around regional decisions occurring at differing times, based on long held traditions of plant cultivation and domesticated landscapes (Hart and Lovis 2013 ). According to Hart (2014), hunter - gather ers domesticated thei r landscape in that they understood how to take advantage of species and resources in their landscape for survival. Maize would have easily been folded into the existing domesticated landscape; the resulting increased human attention modified maize enough, creating a more efficient, reliable, and productive crop (Hart and Lovis 2013). This pattern of maize cultivation being folded into the domesticated landscape is evident throughout Michigan, though clustered around the southern half of the lower peninsul a (Parker 1996). The climate for southern Michigan is adequate for maize agriculture, though th e amount 27 of maize found at sites points to a mixed cultivation strategy (Parker 1996). In northern Michigan, the coastal micro - climates provide enough frost - free days for maize cultivation, but it that there is a large lag between the introduction of maize and any significant cultural changes occurring throughout Michigan, he a rgues that once maize takes hold, it becomes a catalyst for social change. Others (Parker 1996 ; Raviele 2010 ) argue that while maize was used throughout Michigan, it was not the main trigg er for the cultural changes that were occurring in the late Lat e Woodland period . The beginning of the Late Woodland in Michigan (AD 600) is not marked by any significant changes in settlement or subsistence strategies , only a passage of time. The souther n areas that were influenced by the Hopewell Interaction Sphere ceased to exhibit Hopewell traits, while northern Michigan continues the diffuse subsistence strategies practiced since the Archaic (Howey 2012; Brashler and Holman 1985). Unlike the rest of E astern North America, Michigan , and particula rly northern Michigan and the upper Great Lakes, was almost completely unaffected by the Mississippian Period influence and its resulting highly complex social organization. While some Mississippian style cerami cs have been found at Late Woodland sites, su ch as Moccasin Bluff in southwestern Michigan, there is no evidence of overwhelming cultural influence throughout Michigan . The early Late Woodland in Michigan took a similar form throughout much of Michigan. A diffuse subsistence strategy was employed tha t used rounds of residential mobility to take advantage of seasonal ly abundant resources (Howey 2012; Holman and Lovis 2008 ; Holman and Brashler 1999; Martin 1999 ). During t he Mackina c Phase (AD 800 - 1000), in no rthern Michigan, these mobility rounds consis ted of warm season habitation along the coasts of Lake Michigan and Lake Huron, near streams situated for the spring spawning (Smith 2004). 28 Warm season habitation sites were larger, aggregated sites, organized w ith extended family units. During the fall, g roups would move to the Straits of Mackina c in order to take advantage of the fish spawning in the shallows (Holman and Lovis 2008; Howey 2012). Winter sites were located in the interior of Michigan, near the in land lakes, which provided shelter and a wide range of habitats for winter hunting. Similar residential mobility patterns occurred in southern Michigan. During the Mackina c Phase, the dispersed settlement layout during the warm season indicates that famili es were independent and group membership was flexible (Holman and Lovis 2008). It i s not until around AD 1000 substantial social and economic changes occur across Michigan. Visibility of hunter - gatherer occupation of northern lower Michigan increase d drama tically; there are more sites, and the sites were more densely occupied (Lovis 1991). Beginning in the Boise Blanc phase (AD 1000 - 1200) and culminating during the Juntunen phase (AD 1200 - 1400) of northern Michigan (McPherron 1967) , communities show an incr ease in population size, a shift towards defi ned social boundaries, a change in subsistence focus from a broad subsistence strategy to one that relied on intense exploitation of seasonally dense resources , and a dramatic increase in subterranean food stora ge (Howey 2012; Holman and Lovis; Milner and Evidence suggests that at least in the Inland Waterway subregion, unlike the Middle Woodland period, with its generalized site locations, Late Woodland site selection was determined by specific extraction needs (Lovis 1991). For example, Sawdust Pile (20CN22), Portage (20EM22) and Columbus Beach (20CN14) sites are all Laurel Middle Woodland sites, and all exhibit similar variables for site selection based on a generalized subsistence st rategy (Lovis 1991). Late Woodland occupation of the same region includes sites like the Johnson, 29 MacAlpin (20CN9), and Ponshewa ing Point (20EM18 ) sites , which all have drastically different location variables; these variables indicate a narrowly focused s eries of activities at specific sites (Lovis 1991). This variation in site characteristics is indictive of a change in settlemen t and subsistence strategies (Lovis 1991). Another site variable that is being selected for during the Late Woodland is a lands cape that permits subterranean storage. A glaring difference between Middle and Late Woodland sites is the prevalence of subterr anean storage features at specific Late Woodland sites. By the Late Woodland period , the landscape of Michigan had evolved in su ch a way as to create an ideal region for storage. As the water levels dropped, sand dunes began to form, and sandy beaches and ridges began to stabilize. It is at the bases and crests of Lake Nipissing and Algoma shorelines and sand ridges where storage f eatures are most common and obvious. Hunter - gatherers of the Late Woodland period understood the advantages of the sandy soils a nd saw an opportunity to utilize the evolved landscape. Several models lay out settlement and subsistence patterns that account for these Late Woodland cultural shifts. These models can be grouped into three broad subsistence and settlement pattern categor ies : The Inland Shore Fishery model (Cleland 1982; B. A. Smith 2004; S. Martin 1985); Inland Foragers and Exchange model ( Milner 1991; 2006), and the Ecosystem and Landuse model (Lovis et al. 2001; Holman and Lovis 2008; Dunham 2014; Rav iele 2010). 2.2.1 The Biotic Zones Model The Biotic/Historic model, first synthesized by Fitting and Cleland (1969) (see also Yarnell 1964; Cleland 1966) sets up three main biotic zones in Michigan; the Carolinian, the Canadian, and the Transitional or E dge Area. Each zone is known for its specific climate and 30 vegetation, and the Late Woodland inhabitants of these zones adapted s pecifically to environmental variables ( Fitting and Cleland 1969). Communities within each zone utilized both the coastal and in terior regions, but they stayed within their environmental region. These subsistence adaptation models were based on historic ac counts of groups living in these areas during European contact (ca. 1650). Figure 2. 2 : The biotic zones with their correlated historic Native American cultural groups, as designated by the Biotic Zone Model (adapted from Fitting and Cleland 1969; Howey 2012:39) The Carolinian Biotic Province is found in southern lower Michigan, not including deciduous forests (Fitting and Cleland 1969). The historic Potawatomi settl ement pattern corresponded with the Carolinian Biotic Province . This pattern consisted of large, permanent summer villages that relied primarily on agriculture, and large residential winter hunting camps (Fitting and Cleland 1969). The Canadian Biotic Prov ince , which occupies northern Michigan, is juxtaposed to the Carolinian Biotic Province in both environment and settlement patterns. This province has a cool 31 climate, with a short growing season, and coniferous forests (Fitting and Cleland 1969). Using the Chippewa settlement pattern, it posits that summers a nd falls were spent at large, aggregated villages along the northern coasts of the Great Lakes. These large aggregation sites relied on the spring and fall fish spawn as a primary resource (Cleland 1982 ). The Juntunen site on Bois Blanc Island in the Strai ts of Mackinac is the type site for the large summer aggregation villages camps specific to resource extrac tion sites (i.e. hunting, maple - sugar). The Transitio nal Zone, the third biotic province the Canadian and Carolinian Biotic Provinces overlap; it is an ecotonal or transitional area. This zone consists of a diverse array of plants and animals, and it has a growing sea son sufficient for agriculture, but not always reliably . The Carolinian - - pe rmanent coastal villages utilized a mixed economy of f isher - forager - horticulturalists. During the winter months, logistical hunting parties returned to the main village with their hunted resources. Though this early model for subsistence strategies in Mic higan has lived past its prime, and is no longer seen as a viable model, it sets the stage for the later models. 2.2.2 Inland Shore Fishery The Inland Shore Fishery model is a technoeconomic model that posits an intensification of fishing, with the aid of deep water gill nets, around the Great Lakes shorelines (Cleland 1982). Gill net technology allowed for an intensified harvesting of fish th at spawn in deep waters in the fall, specifically Lake Whitefish and Lake Trout. This intensification of fishing required an aggregation of people, coupled with a labor force that could aid in the procurement and processing of the fish (Cleland 1982). In t urn, aggregation led to larger settlements, increased 32 residential duration, and intensified inter/intra - group coo peration. Finally, the ability to harvest and process an abundance of fish and create a surplus resulted in a storable product that could feed a larger number of people throughout the winter, leading to an increase in population (Cleland 1982). In this mod el, the interior is only used during the winter, when groups dispersed due to scarce resources (Cleland 1982). It is the combination of the int roduction of the gill net technology, increased social cooperation, and utilization of storage that led to the se ttlement and subsistence changes in the Late Woodland (Cleland 1982; Smith 2004). Others (Martin 1989; Smith 2004) have suggested that though g ill nets became prevalent in the Late Woodland and created popular fall fishing locales, groups continued to util ize well - established spring fisheries located in the interior. 2.2.3 Ecosystem and Land Use Model The second set of models that interprets the settlement and subsistence changes in the Late Woodland period can be considered the Ecosystem and Land Use m odels (Lovis et al. 2001 ; Holman and Lovis 2008; Dunham 2014). These models focus on a suite of factors, including environmental changes that l ed to an adaptive strategy requiring a flexible system of resource selection. Similar to the Biotic model, the e cosystem and land use models consider the effect of different ecosystems on the settlement and subsistence strategies and how the people of the Mackinac and Juntunen phases of the Late Woodland period capitalized economic return within the differing ecosys tems (Lovis et al. 2001; Holman and Lovis 2008; Dunham 2014). Rather than confining these groups to one specific biotic zone, this model explai ns how the risk management strategies of mobility and buffering were utilized across ecosystems (Holman and Lovis 2008). 33 The earliest phase of the Late Woodland in northern Michigan is the Mackinac Phase, 800 - 1000 CE (McPherron 1967). Lovis and Holman (2 008) posit that during the Mackinac phase, the Chippewa pattern of residential mobility fits well with the archae ological record. Their model suggests that, like the Chippewa, the Mackinac phase communities were (Holman and Lovis 2008: 290). The majority of sites found for th e Mackinac phase are located in ecosystems similar to those utilized by the historic Chippewa. However, there are also sites along the northern coast of Lake Michigan; these coastal settings were utilized in the warm months and were part of their seasonal mobility rounds (Holman and Lovis 2008). When excavated, the coastal sites of Mackinac phase groups indicate repetitive occupations by entire f amilies, evidenced by a range of residential activities rather than task - specific sets. These coastal sites also periodically contain ceramics from more southern groups, indicating a territorial buffer, one that can be stretched in times of need (Holman an d Kingsley 1996; Holman and Lovis 2008). The Juntunen site on Bois Blanc Island in the Straits of Mackinac is a n example of the Mackinac phase fall site (McPherron 1967; Holman and Lovis 2008). During the fall, Whitefish and Lake Trout spawn, making the Juntunen site an ideal location, showing dense occupation. These models argue that at sites like Juntunen, there would have been an aggregation of local bands. This aggregation would have served several purposes, including information exchange and a reinfo rcement of social networks (Holman and Lovis 2008). During the winter months, Mackinac phase communities would have headed inland and camped near inland lakes. These inland lakes provided a sheltered location and micro - environments where fishing and hunt ing would have been productive (Holman and Lovis 2008). 34 The ceramic assemblages indicate these interior sites wou ld have been limited to Mackinac phase groups. During maple sugar season, these communities would move to their sugar bush and collect maple sa p. The Mackinac phase groups utilized a system of cooperative buffering to mitigate seasonal risk (Holman and Lo vis 2008). Cooperative buffering allowed for use of neighboring territories in exchange for the same privilege. This is most evident at sites a long the Lake Michigan coast where ceramics and non - local chert sources overlap (Holman and Lovis 2008). Enviro nmental changes, in the form of fluctuating lake levels around AD 900 (Lovis et al. 2012 ; Lovis, Monaghan et al. 2012 ) resulted in less predict able near coastal resources, i.e. shallow water fishing (Dunham 2014). This may have led to the increased utiliza tion of deep water gill netting. An unstable coastal environment resulted in additional risk management strategies including resource diversifi cation in the form of acorn and wild rice ( U.P. specific) and storage (Dunham 2014). Evidence from the Juntunen phase of the Late Woodland, 1200 - 1650 CE, corresponds to all fishing locales, long - involved winter residential mobility (Holman and Lovis 2008: 295). Similar to the Mackinac phase sites, the people Juntunen phase utilized both the coa stal fishing sites and interior lakes (Holman and Lovis 2008). Unlike the Ottawa, the Juntunen groups selected co astal sites for their productive fishing, not for maize agriculture. Evidence for the consumption of large game animals like deer and caribou a t the coastal sites indicate a system of logistic hunting, whereby hunters brought the food back to the aggregate d villages (Holman and Lovis 2008). 35 Subsistence e vidence and site locations suggest an intensification of resources from the interior (Dunham 2014). The increased use of logistic movement indicates a targeted exploitation of resources like acorns and wil d rice (Dunham 2014). During the Juntunen phas e, there is evidence for an in tensification of fall fishing, through 1988). The Juntunen phase, like the Mackinac phase o f the Juntunen site, exhibits a level of band aggregation. Additionally, buria l mounds and ossuaries on the island allude to a demarcation of argues the mortuary rituals that surround the burials acted as a mechanism to reinforce social ide ntity (Dunham 2014; Holman and Lovis 2008). Like the adaptive pattern of the Ottawa, the Juntunen communities mitigated seasonal risk through reciprocal exchange (Holman and Lo vis 2008). Interaction with neighboring groups is evidenced by ceramic styles that begin to share attributes with neighboring groups. The exchange exhibited at the earthwork complexes is seen as part of this risk mitigation strategy. Most likely, exchange occurred at social boundaries, or areas that were not territorialized in order to facilitate a friendly encounter (Holman and Lovis 2008). The Juntunen phase groups had a flexible, yet critically timed and organized mobility system that allowed them to m ove inland for the winter, when necessary (Holman and Lovis 2008). Winter site s were occupied by extended families, and based on ceramic evidence, only consisted of Juntunen peoples. Like the Mackinac phase people, Juntunen people used the natural shelter of the inland waterway that had resource - rich micro - climates (Holman and Lovis 2008). 36 According to the Ecosystem and Land Use model, by the late Late Woodland (Juntunen Phase) the change in the subsistence and settlement pattern indicates a flexible syst em of movement and resource selection and extraction. Though still reliant on deep - water fall fishing, a series of risk management strategies were implemented to counteract the potential pitfalls of relying on a single resource. Resource diversification, s torage, and social integration were all utilized to mitigate risk. 2.2.4 Inl and Foragers Model lower Michigan. This model argues that the coastal lake - effect zones allo w for a hospitable growing season of 120 fros t - free days. Late Woodland hunter - gatherers took advantage of this agricultural zone and not only productively grew maize along the coasts, but intensified production enough that maize became a primary subsisten intensified coastal maize production th at led to the dynamic late Late Woodland period. Increased territorialism, due to demarcation of productive maize areas, began to cut off communities from their traditional sum mer coastal fishing villages. This, in turn, created a dichotomy between the c inland areas of Michigan are not suitable for agriculture, the inland foragers had to continue their socio - econo Howey 2006). Though maize agriculture is possible along the coastal regions, the fluctuating climate makes agriculture a risky endeavor. W ithout the ability to travel inland to f orage, the coastal 03). Both the inland foragers and 37 coastal farmers began to intensify their respective resource zones, but another risk management strategy was utilized: long - ea 2003). According to the Inland Foragers model, a series of ritual monuments in the form of Howey 2006; Howey 2012). Ritualized burial mounds were created in areas of hi gh resource potential. These mounds symbolized an ownership of the resources, and acted as an intratribal gathering place, and a mechanism for tribal integration (Howey 2012). Burial mounds are found in resource rich areas in both the interior (inland lake argues that these intratribal burial mo unds aided in solidifying group identities, and in turn, help ed Intertribal monuments, in the form of earthwork comp lexes, were created to facilitate inter - tribal interaction and exchange throug h ritualized activities (Howey 2012). Yearly intertribal aggregation at ritualized monuments acted as a risk management strategy. Resource pooling and exchange helped to mitigate the risk inherent in the increasingly circumscribed territories. Unlike the intratribal burial mounds, the intertribal mound complexes were removed from productive resource zones. Additional features that are common across the monument complexes include: enclosure shape and pairing, water feature, and numerous subterranean storage pits (Howey 2012). The Inland Forager model posits that the settlement and subsistence changes that occur in the Late Woodland period are the result of a symbiotic relationship between the coastal horticulturalists and the inland foragers that relied on exchange and interregional social rituals 38 exploitation in their respective territori 2.3 Late Woodland Models and Food Storage It is not the intention of this paper to determine which Late Woodland Model is correct, mainly because this research exposes only one variable and multiple variables would require evaluation to take on such a task; however, as part of this researc h, the role of food storage in each of these models will be evaluated for best fit. Which, if any, of these models best accounts for the proliferation of food storage in the late Lake Woodl and (AD 1000 - 1650)? Table 2.1: Inferences and Expectations regardi ng current models and food storage 39 2. 4 Summary This chapter has shown the dynamic environmental events that - glacial landscape. The landscape of northern lowe r Michigan, though slow to develop (both ecologically and culturally), was an obstacle course of fluctuating lakes levels, harsh winters, and at times scarce resources, not to mentio n the socioeconomic challenges of increasingly restricted territories. Eac h of these challenges required settlement and subsistence strategies that could compensate for difficulties and ensure survival. Throughout the culture history of Michigan, shifts in strategies are evident through the changes in settlement and subsistence patterns. The presence of storage features in the Late Woodland period is one such example of a subsistence shift; this shift entailed an intensification of resources which allowed f or storage. It is this shift to focus on storage that is the crux of this dissertation; what can this shift in strategy tell us about the Late Woodland period in northern lower Michigan ? Several models have been put forth to explain the settlement and sub sistence changes that took place during the Late Woodland period. The prol iferation of storage during this period is an important marker for subsistence change and as such, requires a thoughtful understanding of how and why it is used. Using the variable o f storage, and data collected on the global use of storage by hunter - gathe rers, these models will later be evaluated using Optimal Foraging Theory (OFT) to determine best fit. The following chapter will discuss how these broad changes are best evaluated at a local level, using a selection - based model. Selection - based explanati ons focus on the individual actors making decisions based on their local environment. Chapter three will also provide an in - depth discussion on risk, how it can be evaluated using Op timal Foraging Theory models, consideration of hunter - gatherer risk manage ment strategies, and the utilization of storage to 40 mitigate risk. This discussion will be followed by a detailed evaluation of food storage as a risk management strategy by hunter - ga therers globally, framed by an OFT, the Diet Breadth Model. 41 3.0 R ISK M ANAGEMENT S TRATEGIES FOR N ON - SEDENTARY S OCIETIES 3.1 Understanding Risk Environmental variability, whether for hunter - gatherers or mixed low - level horticultural economic groups, p oses a selective pressure on behavior, often resulting in long - term social unpredictable, the survival of a group is, at any given moment, hinged on just one or two limi ting 1989). Limiting factors are situational and can quickly change; a group could be desperate for water during a drought, then faced with a looming forest fire. Food is a basic limitin g factor, needed on a regular basis, but often scarce due to variability i 1990). Response m echanisms embedded in cultur al behaviors are suited to mitigate many of shortages will definitely occur, hunter - gatherers create strategies to counteract the potential for shortages (Colson 1979; Halstead a culturally ingrained or embedded across many different dimensions. Because the se cultural constructions become so embedded in the daily lives of hunter - gatherers they can be used to inform on any nu mber of anthropological questions, from economic structures to changes in settlement patterns (Edwards 2017). For example, Frink (2007) explains that at the Qavinaq site in Alaska, a change in location of storage from inside the home to a communal village area, after European contact, indicated a cultural shift where women were cut off from their role as food managers. Strategies that cop e with limiting factors and lessen the risk of food scarcity are known as 42 predictability, scale, and severity of potential environmental variables that lead to resour ce shortages are important factors to consider when studying risk management strategies (Edwards 2017). According to Hal - gatherer groups to anticipate the patterned variation in resource availa bility. For example, in the Great Lakes region, a temperate climate, winter is a predictable season on scarcity. Inhabit ants of this region will know that winter brings low temperatures and snow, which results in scarce plant and animal resources. Of cours e, as the severity and length of winter fluctuates from year to year, there is a generalized pattern of predictability. The patterned predictability of lean winters can be mitigated with risk management strategies. Though seasonal variation is predictable , other weather and environmental events, like blizzards or droughts, are less predictable and may require additional mi tigation techniques. Since droughts may not occur every year, mechanisms are embedded to account for long - term cycles. For example, Cols on (1979) found that the effects of long - term cycles are stored in the memories of group elders or formed into oral trad itions. Because not all events that effect resources are predictable, groups utilize multiple strategies. Groups that inhabit extremely environmentally volatile climates are more likely to have buffering mechanisms that account for the long - term environmen tal cycles (Colson 1979). Buffering mechanisms are applied through a hierarchical system ranging from low - level to high - level mechanisms 1989). Low - level mechanisms are the most efficient and reliable; whereas high - level ing mechanism is best used to mitigate risk, in a given context, is based on meta - decisions (Mithen 1990) that revolve a round the spatial - level 43 buffering mechanisms are risk - averse, while high - level buffering mechanisms are risk - prone with high rewards. The potential outcome of risk - averse ac tions makes high - level buffering 3.1.1 Defining Risk Risk refers to the potential for loss (Marston 2010). In the case of hunter - gatherers, the main risk bei ng mitigated is the amount of food that can be counted on throughout the year. Mitigation techniques are then put into place to reduce t his risk, though it is the uncertainty that can never be accounted for. Fitzhugh (2001) posits that the use of the term the probability of an outcome (Fitzhugh 2001; Marston 2010). Risk is not being avoided, rather it is being gambled. Risk it self is actually a buffering mechanism because decisions behind the selection of the buffering mechanism are weighed on a continuum from risk - averse to risk - prone (Fitzhugh 2001; Cashdan 1990). Return rates in the diet can be predicted based on probability ; levels can be compared against the required diet to determine the rate of insufficient returns (Edwards 2017; Fitzhugh 2001). Using a Z - score model, developed by ecologists, the risk of failure of specific diets can be determined (Stephens and Charnov 1982; Winterhalder and Goland 1997 ; Edwards 201 7). Z - score models allow for measuring of risk by graphing multiple risk management strategies and the resulting levels of success (Stephens and Charnov 1982; Edwards 2017). Models like Z - score, can demonstrate how hunter - gatherers decide on a risk managem ent strategy that 44 maximizes return rates by weighing the risk of the specific strat egies and providing the best probability of survival (Winterhalder and Goland 1997). The Z - score model, depicted in Figure 3.1 exhibits a normal distribution of harvest rat es; the x - axis measures Net Harvest Rate and the y - axis measures Frequency. The dot ted gray line represents R, the required return rate for survival, and Z is the probability of falling below the minimum required return rate (Winterhalder and Goland 1997). Strategy A (red line) and Strategy B (blue line) represent the different strategie s used to buffer risk. Because Strategy A has a lower mean rate, it has a lower maximum potential compared to Strategy B (Edwards 2017). Strategy B represents a high - risk an d high reward scenario because it has a high variance; Strategy A, on the other han d, has a predictable return rate (Edwards 2017). Figure 3.1: Hypothetical Z - Score Model . A and B represent two different hypothetical strategies and R represents the minimum required harvesting rate to feed the population. ( Adapted from Edwa rds 2017. ) 45 The scenario in Figure 3.1 is risk averse because both strategies will meet R. Though A is less risky than B, the benefi ts of the higher risk B would most likely be chosen over the safe bet of Strategy A (Edwards 2017). If R is increased, like in Figure 3.2 and neither Strategy A nor B have a guarantee of providing the minimum required amount of food, then risk is gambled ( Cashdan 1990). Ethnographic data (Cashdan 1990) indicates that in these situations, the high - risk option is often selected. Figure 3.2: Z - Score Model with an increased R ( Adapted from Edwards 2017 ) . 46 3.2 Risk Management Strategies There are four main categories of buffering mechanisms used to counteract yearly variability Colson 1979; Cashdan 1990). According to a Z - score model, every strategy has inherent potential and inherent risk. Under normal conditions, i.e. non - extreme conditions l ike seasonal variation, low - risk (or risk averse) strategies will be selected over risk - prone strategies (Edwards 2017; Cashdan 1990). The four types of risk manage ment strategies, mentioned above, are all utilized to reduce variance, or at least accommoda te for variance. Understanding the risk management strategies that were utilized by hunter - gatherers to mitigate risk can allow for inferences about the decision - ma king behind specific contexts and techniques. 3.2.1 Risk Management and Exchange Exchange is one mechanism for risk management. In its broadest sense, exchange is trading one object in the expectation of receiving another object. Exchange as a risk management (1981) embeds exchange in the mechanism of storage and explains that it can be considered social storage; food is exchanged for the moral obligation to someday reciprocate (see also Spielmann 1986). In some instances, groups exchange meaningful to kens that can be cashed in during times depends on environmental factors, such as local distribution of resources, area affected by food shortage, and physical geograph y that co uld prevent exchange; and cultural variables such as takes, it can be considered a low or high - level mechanism. Exchange can be divided into two main cate gories, b 47 redundant and complementary). Buffering exchange occurs when short - term shortages are alleviated through access to food in neighboring territories or given by neighboring group s; while mutualism is the more traditional concept of exchange where complementary subsistence environment, with its unique mix of resource abundance, reliability and predi ctability, these two types of exchange are not mutually exclusive and could be used separately in a risk management system. Winterhalder (1986) considered the effectiveness of exchange in reducing risk and created a model through which it could be tested. goal revolve around minimizing risk, instead of maximizing energy acquisition and found that risk is reduced when meat is shared with fellow hunters, rather than just caching the surplus for later (Win terhalder 1986). Not all hunter - gatherer sharing is a form of risk management exchange. In the case of the Meriam, meat sharing is a social mechanism used for prestige, and not used out of the expectation of reciprocity (Bird and Bird 1997). When exchange is not ti ed to a social obligation like reciprocity it is not part of a risk management system. 3.2.2 Risk Management and D iversification as risk management. Diversific ation all ows hunter - gatherers to expand their food choices, or to fall back on less palatable famine foods. Because seasonal variation is predictable (generally, not specifically), hunter - gatherer and mixed economy groups accounted for yearly resource shor tages by incorporating the risk management strategy of diversification into their yearly cycle Marston 2011 ). Holman (1984) discusses how maple sap was used as a yearly famine food. Though the constant consumption of maple sugar and syrup wre aked havo c on the digestive system, it was one of the few resources that was reliable in the early spring (Holman 48 1984). Scars found on pine trees across Northwest North America indicate a consistent utilization of famine foods (Prince 2001). The cambium, a tree ba rk layer, of pine trees was harvested on a yearly basis to supplement carbohydrates during the early spring. While the cambium was used consistently by prehistoric populations in British Columbia, groups without regular access to salmon relied mor e heavily on the consumption of cambium during the lean months of spring (Prince 2001). Hunter - gatherer groups who had access to salmon stored the surplus catch for the lean months and were less reliant on cambium; however, it was still consistent in their diets (P rince 2001). Nagoaka (2002 ) uses a prey choice model to determine hunter - gatherer decision - making during times of resource depression. Studying hunter - gatherers in New Zealand, she finds that when the population of the moas, a large game bird, dec lines the foraging efficiency of the hunter - gatherers declines due to their limited choice of lower ranked taxa (Nagoaka 2002). Smaller taxa that were previously ignored, due to their low return ranking, were added to the diet as a risk management strategy . Diversi fication is evident in the archaeological record of prehistoric farmers when wild plants are evident in the paleoethnobotanical record (Marston 2011). During poor yield years, crops were supplemented with wild plants in order to buffer the potenti al risk o mixed economy groups, seasonal scarcity is predictable and planned for, resulting in a consistent reliance on resource diversification. The Pawnee supplemented their crops with bison hunts; while the Hu ron relie d on fishing and storage of famine foods (i.e. acorns, blueberries, and variation, the Huron diversified by increased exploitation of wild game; while the Pa wnee leng 49 3.2.3 Risk Management and Mo vement Mobility is a hunter - gatherer strategy used to situate oneself closer to, or within an appropriate distance of, a food resource ( Kelly 1992 ). B inford (1980) described a mobility continuum, with foragers with high residential mobility at one end, and collectors with high logistical mobility on the other. In foraging systems people move to resource locations, and in collecto r systems resources are moved to people. Collector systems are more common when resources compete in either timing of availability, or spatial location. Although mobility is viewed as a strategy for survival, it is important to separate the concept of mobi lity from the idea of mo vement . The term movement accounts for the human agency at play in specific mobility strategies and is a more accurate term when discussing the decision - making of hunter - gatherers and mixed horticulturalists (Oetelaar and Meyer 200 6; Oetelaar and Oetelaar 2007 ). After human mobility is used within what has commonly been termed a landscape, that landscape becomes a cultural landscape; human knowledge and experience is then mapped onto the cultural landscape, and purposeful movement o ccurs within the landsca pe (Oetelaar and Meyer 2006). Seasonal movement strategies are not just based on the ecology of the landscape, but rather, on the combination of ecology, landforms, knowledge, and history (Oetelaar and Oetelaar 2007). An important a spect of collectors is t hat logistical mobility is used to monitor environmental indicators that signal resources (Kelly 1983; Whallon 2006). The flexibility of movement allows risk management in the form of social stress reduction (Savishinsky 1981). Reso urce monitoring is often embedded in the logistical mobility and allows groups to observe resources in order to harvest them at their peak (Kelly 1983). The necessity of hunter - gatherer movement can be explained beyond just the need of situating the group near exploitable resourc es; movement also allows for a flow of information and creates and maintains social networks (Whallon 2006). 50 Informational mobility refers to movements in which the gathering of information is primary (Whallon 2006). Ritual and cere monial duties can be see n as a means to ensure that groups travel to and aggregate at specific times of the year for information gathering and solidifying social networks (Whallon 2006). Using a pre - contact Michigan case study, Holman and Kingsley (1996) p osit that the aggregatio n of large villages in the summer months allowed for information exchange and cooperative buffering strategies. In seasons of scarcity, groups had to know they had a back - up location to utilize and the confidence to know that locati on would not already be populated (Holman and Kingsley 1996). Aggregation months along the coasts of Lake Michigan solidified social networks and allowed for information exchange so a buffering strategy for the new year entailed all pertinent information; i.e. it may be a poor ye ar for fishing (Holman and Kingsley 1996; Whallon 2006). In northern lower Michigan, hunter - gatherer groups used logistical mobility during the winter, but aggregated in the early spring as maple sugaring camps (Holman 1984). Movem ent is used as a long - te rm adaptive strategy and as a seasonal risk management strategy (Thompson and Turck 2009; Holman 1984). Numerous variables affect these movement decisions. Variables including seasonality of resources, dispersal of resources, mobili ty of resources (i.e. se asonal migration), knowledge of the landscape, and proximity to neighboring groups. All these variables need to be negotiated when movement is used as a risk management strategy. Food storage, covered in the next section, is the fo urth strategy to mitigat e risk and the focus of this dissertation. The most affective risk management systems utilize multiple strategies. Diversification is often used with movement and exchange is complemented with storage. 51 Therefore, to understand the r ole of food storage as a risk management strategy, it is necessary to understand the subsidiary role of other mitigation strategies. 3.3 Risk Management and Storage 3.3.1 Types of F ood S torage Storage is the final mechanism used to buffer economic risk. In its simplest defini tion storage defers consumption and caloric use and intake, and thereby levels - out seasonal variation It almost always requires smal ler or larger surpluses and must often recognize that there is a trade - off in the f orm of reduced nutritional values. Storage can first be delineated between direct, any strategy that prolongs the shelf - life of foodstuffs, and indirect, or any process that 1981). One type of indirect storag e is e cological stor age, the interruption of the natural cycle of energy from plant to animal by the concentration of nutrients within that cycle (Ingold 1983:555). For exam ple, feeding scraps to animals tha t will later be eaten would be ecological storage . The food waste , that under other circumsta nces would have been discarded, is deposite d in another usable resource. Soffer (1989) has a similar storage category she calls s torage in the self; an example being gaining weight before the lean season. Social storage, a type of indirect storage, is the acc umulation of resources with an end result of rights for distribution (Ingold 1983:559 ). With thi s ty pe of storage, the objects being stored become a symbol of wealth or prestige such that the ownership of the resources becomes important. Social storage is evident in extensive excha nge systems, but is contingent upon resources exchanged and distance and e fficiency of transport between communities 981:172). Social storage is mos t evident in hunter - gatherer groups as the risk management strategy of 52 Long - term food cycles, as opposed to yearly cycles, are also accounted for through buffering mechanisms i n the form of storage and transmission of knowledge (Colson 1979). Practical storage , a direct storage strategy, is the category that is most a ssociated with hunter - gatherer populations. This type of storage involves a series of planned stora ge facilities that create food supplies at planned stops on mobile rounds (Ingold 1983). Mo re broadly speaking, it is the stra tegic planning of the facilities, the accumulation of foo dstuffs, and the processing of foodstuffs for storage (Ingold 1983:558). Practical sto r age enables group movement by guaranteeing the food availability at sev eral locations (Ingold 1983). Practical s into material, portable, and permanent storage (S offer 1989). Material storage is simply the accumulation and processing of physical resources. Material storage allows for mobility if the foodstuff is portable; drying, smoking, and salting lead to portability (Soffer 1989). P ermanent sto rage requires the construction of immovable structures, like storehouses and pits for the st orage of foodstuffs and therefore lim its mobility (So ffer 1989). Deciding which type of storage to utilize is heavily dependent on types of foodstuff available, climate, and existin g technology - existent south of 28 degrees latitude; therefore, hunter - gatherer groups in those areas rely on indirect storage strategies or other risk management options. It is eviden t that storage, as a risk management strategy, is a broad concept; storage can refer to the setting aside of foo d to the building up of obligatory favors and can even overlap with the concept of exchange. Therefore, it is integral to continue to narrow dow n the specific type of food storage that is being discussed. 53 3.3.2 Food Storage Temporal and Scalar Variabilit y Food storage, its sociopolitical ramifications, and its role in a hunter - gatherer economy can be understood when temporal and scalar variabili ty are considered. Large - scale storage is often seen as a trigger for sedentism and socioeconomic inequalities ( Testart 1982; Kuijt 2009; see also Ingold 1983; Soffer 1989). Large - scale food storage is storage beyond the stockpile necessary for the lean se ason, it is food surplus that can be converted into lasting goods through exchange (Cunningham 2011; Testart 198 2; Kuijt 2009; Soffer 1989). While at its core, food storage is a risk management strategy used to mitigate times of scarcity, the results of i nstitutionalized food storage show direct correlation with socioeconomic hierarchy (Rowley - Conwy and Zvelebil 19 89 ; see also Testart 1982; Kuijt 2009; Ingold 1983). Surplus is any food above and beyond the quantity needed to make it through a standard year , though this quantity is socially defined (Kuijt 2009; Berrier 2011). Surplus is not visible in the archaeologi cal record, only remnants of behavior related to surplus are evident (Berrier 2011; Rowley - Conwy and Zvelebil 1989). Berrier (2011) traces surpl us at Moundville by considering storage facilities and the use of surplus labor (i.e. feasting and monumental co nstruction). He finds that large ceramic vessels indicate people of all echelons were participating in creating food surplus. Kuijt (2009) consi ders the evidence for surplus at several Neolithic sites and finds the evidence lacking because feasting does no t equate to surplus. Rowley - Conwy and Zvelebil (1989) argue that because food storage plans for shortages during the extremes, a surplus is left in the average years. Surplus is eventually manipulated into symbolic capital through social storage ( Wesson 19 99 ). Subterranean food storage surplus will never provide symbolic capital because by its very essence it is concealed and unknown (DeBoer 1988) . Surplus must be visible or else it holds no value. By exchanging surplus food into social storage (see Ingold 1983) by 54 redistributing food (i.e. through feasting) one gains symbolic capital, or prestige (Wesson 1999; Rowley - Conwy and Zvelebil 1989). Addi tionally, as food storage entails a tethering to a specific resource area, construction of storage facilities, a nd reliance on delayed returns, food storage limits mobility, and encourages increased sedentism (Rowley - Conwy and Zvelebil 1989; Wesson 1999; S offer 1989). Soffer (1989) explains that increased sedentism, or reduced residential mobility, takes away the Wesson (1999) considers the relationship between food storage, prestige goods, symbolic capital and the collapse of Mississippian chiefly power. He finds that when individual families stockpile thei r own surplus, evident through storage pits within dwellings, the structure of the chiefs begin to wane because others are gaining symbolic capita l through the accumulation of prestige goods and threatening the power structure (Wesson 1999). Cunningham (20 11) argues that because small - scale food storage is not considered a gateway to complexity, like large - scale storage, small - scale storage has been all but overlooked. Testart (1982) sets up a dichotomy of hunter - gatherers who store and those that do not. I n the analysis of hunter - gatherers that store food, Testart only considers hunter - gatherer groups that utilize large - scale storage, and explains t hat small - scale storage does not lead to socioeconomic inequalities (Testart 1982). Understanding food storage variability beyond large - scale is integral to the further understanding of the role of food storage in the social and economic aspects of hunter - gatherer lifeways (Cunningham 2011; Kuijt 2009; Ward 1985). Small - scale food storage does not produce surplus . However, that does not mean it is not intensive food storage, or that it requires less work/labor input than large - scale storage (Cunningham 201 1; Soffer 1989). Small - scale food storage can take different forms, from storage of small amounts of several d ifferent types of food, to several small - scale storages 55 strategically placed across the landscape (Cunningham 2011). Cunningham argues that a comb ination of small - scale storage strategies equates to large - scale storage in terms of amount of food stored. Wh ile it is generally believed that food storage limits mobility because storage facilities tether groups to one area, Cunningham (2011) and Soffer (1989) both argue that portable food storage (i.e. pemmican) accommodates hunter - gatherer mobility. Certain pr eservation techniques like drying, salting, or smoking increase food portability, allowing groups to maintain mobility; whereas freezing or cachin g limits mobility (Cunningham 2011; Testart 1983; Soffer 1989). Testart (1982) posits that because preservatio n techniques like drying and salting require more energy expenditure and time, these methods will rarely lead to large - scale storage because of th e front - loading requirements. The scale of food storage is less important than the shelf - life of food storage (Soffer 1989). Large - scale food storage is long - term because grains are generally relied on to accumulate surplus (Testart 1982). Long - term is def ined as the storage of food beyond three months, therefore small - scale food storage can easily be long - term. S offer (1989) explains that the longer the shelf - life of the food, the longer the use - value through consumption or social manipulation. Long - term f ood storage has more impact on sociopolitical relations (Soffer 1989; Cunningham 2011; Testart 1982). The comb ination of s cale and shelf - life of food storage can be an analytical tool for understanding the role of food storage in a society. Large - scale foo d storage by hunter - gatherer groups indicates a primacy of storage as a risk management strategy due to the or ganization and planning required. Additionally, manipulation of shelf - life, rather than physical storage, alludes to movement being a primary risk mitigation, and storage being subsidiary. Understanding the decisions revolving around storage, such as how t o store, and how much to store can inform on the role of food storage. 56 3.3.3 Food Storage and Mo vement Many researchers addressing these issues (Testart 1983; Rowley - Conwy and Zevelibil 1989; Brenton 1988) argue that storage is incompatible with mobility, especially as a risk buffering mechanism. Because storage tethers groups to a location, it limits hunter - options for movement (Testar t 1983; Rowley - Conwy and Zvelebil 1989). While food storage does restrict the mobility of hunter - gatherers, a strategic use of storage allows for an efficient mobility system. Morgan (2012) divides hunter - gatherer storage into two types, caching and centra l place storage. The Mono of California u sed a system of central place storage, or residential storage, for their immediate risk relief, in conjunction with caching that accommodated for long - term shortage and risk (Morgan 2012). The Mono were hunter - gathe rers but had semi - sedentary winter hamlet s. Caching away from the hamlets expedited acorn collecting by minimizing energy costs, while allowing for flexible mobility that kept them in range of the caches (Morgan 2012). Hunter - gatherers in Michigan utilized caches and central place storage in a di fferent system than the Mono (Holman and Krist 2001). Storage was structured to accommodate patterns of seasonal movement by creating central place storages at sites before they left, so food was available upon the seasonal return (Holman and Krist 2001). For example, fish and berries were stored at late summer camps, to be consumed on return in late spring/early summer; while acorns were stored at fall locations and consumed in early spring during maple sugar season (Holman and Krist 2001). Additionally, c aches were set up along hunting routes, at key locations where they would determine their next winter move. Holman and Krist (2001) found that storage pits were often located in environmental areas that held overlap ping seasonal resources. This seasonal ov erlap gave hunter - gatherers options for resources and mobility upon their return (Holman and Krist 2001). 57 Early hunter - gatherers on the plains used cold lava tube caves to cache their surplus bison meat (Henrikson 2003). Lava tube caves are natural freeze rs that would enable the hunter - gatherers to cache their meat at minimal energy costs (since it would be frozen and not require processing) and return to it when necessary (Henrikson 2003). Cannon and Yang (2006) an alyzed aDNA to ascertain that pink salmon was the fish of choice for groups on the Northwest coast because the high fat content made for reliable storage. Though groups living at Namu in British Columbia were semi - sedentary, they relied on salmon storage t o get them through the winter hunting mon ths (Cannon and Yang 2006). Storage is most efficient with a mobility system that has planned seasonal rounds (Holman and Krist 2001; Smith 2003; Brenton 1988). Collection and storage of pinon nuts by hunter - gathere rs in Wyoming was such a reliable resourc e base that permanent structures were built at high - resource production areas (Smith 2003). There was no evidence that the nuts were brought back to camp; the nuts were cached away from residential camps and utilize d during mobility rounds (Smith 2003). Each type of storage (social, practical, ecological) is connected. Constraints in one type of storage will lead to constraints in another, and vice versa is also true (Brenton 1988). When physical storage is effecti ve enough to minimize food spoilage, and therefore risk, it leads to a more stable social network, which in turn leads to more risk management options (Brenton 1988). 3.4 Short comings of Food Storage Terminology This overview of risk - management strategi es and how food storage has been researched results in a rather confusing list of terms and definitions, specifically when referring to storage in its many dimensions. One primary goal of this research is to simplif y terms revolving around 58 storage and appl y more succinct and clear terminology. Though these aforementioned definitions have merit in their own right, they tend to muddle the focus of this research and stray from the main topic of food storage as risk mana gement. One glaring issue with the curre nt suite of storage terminology is that it is directionally biased; it views food storage as a path towards complexity, i.e.it is causal and explanatory, rather than an important social function in its own right. Th is function of historical changes in anth ropological theory as applied to the evolution of human systems over long time spans. There is no denying that the scale of food storage can and does play a role in social complexity, but for this research the only pertinent scale is the amount which lasts through the lean season. Pigeon - holing food storage as a cog in social complexity finds similar issues when defining temporal aspects of food storage. Evaluating the extension of shelf - life through the lens of soc ial capital return, i.e the longer the sh elf - life the longer the use - life, overlooks the initial goal of survival. The stored food needs to last long enough to make it through the targeted lean season. For the purpose of this paper, food storage is from h ere forward defined as any level of stora ge that accommodates the seasonal flux in availability of food resources. Scale and shelf - life of food storage are culturally defined and revolve around survival through the lean season. This will be referred to as Seasonal Gap Storage (SGS), which is risk management food storage where scale and shelf - life are measured only by survival. By decluttering the concept of food storage, this research homes in on the day - to - day decisions behind food storage as a risk manag ement strategy. How does SGS operate with in the settlement and subsistence system? How is labor divided to accommodate planned SGS? What other risk management strategies complement SGS? 59 3.5 Measuring Risk and Uncertainty Winterhalder and Goland (1997) u sed the Diet Breadth Model to understand the series of risk management decisions that led to plant domestication. The diet breadth model, adapted from an ecological Optimal Foraging Theory (OFT), gives a ranking to all hunter - gatherer food resources, and t heorizes that hunter - gatherers will selec t/hunt the resource with the highest ranking. While the goal is to harvest the highest - ranking resource, variables must be weighed in the process of hunting and gathering. For example, if the hunter sets out to kill a deer, but along the way comes across a rabbit, variables of pursuit and handling costs must be weighed against the rank of the resource. Based on the ranking of resources, the model is set up to hypothesize the decisions hunter - gatherers make for their food selection ( Winterhalder and Goland 1 997 ). Resources with a high rank will always be pursued if encountered, and resources with a rank so low it puts them out of the average diet, will never be pursued. An important aspect of this model is that if abundance of the highest ranked resources dec lines, and in turn the foraging efficiency of said resource declines, then the diet will expand to include a lower ranked item (Winterhalder and Goland 1997). This shift in diet can potentially be traced archaeo logically (see Ugan 2005 in previous section) . Another shift that can be traced archaeologically is the proliferation of a previously low - ranked resource. The diet breadth model posits that any resource that becomes more efficient to harvest or process wil l rise in the resource rankings (Winterhalder and Goland 1997). Winterhalder and Goland (1997) argue that the best method to understanding the decision - making of hunter - gatherers is from a selectio n - based , rather than a functionalist viewpoint. Unlike fu nctionalist models that use adaptive response s as explanations for change, selection - based models begin with selection and attempt to infer the behavior that led to the 60 causal effects (Winterhalder and Goland 1997). Hunter - gatherers probably did not forese e an end result of economic stability when th ey transitioned to agriculture, instead they made case - by - case selection decisions (Winterhalder and Goland 1997). Another factor of the selection - based model is the focus on non - normative environmental variabl es. Risk management strategies are put in pla ce to account for predictable shortages; it is the uncertainty, or unpredictable events that cause dramatic change (Winterhalder and Goland 1997). Finally, selection - based models reject prime mover arguments (i .e. climate change) for major structural shif variable to the actual conditions and options experienced by an individual organism, the more likely it will be of causal importance in evoluti 1997 :126). Though large - scale and broad environmental changes often occur, it is the immediate effects at the local level that create the change (Winterhalder and Goland 1997; Hart 1993 ; Edwards 2017). For example, to argue that an increase in population cause d settlement and subsistence changes overlooks the more powerful local - level interpretation that the risk management measures taken last year were not enough to stave off starvation, therefore a change in strate gy should be implemented. A specific example from Michigan is explained by Dunham (2014) when he argues that the Medieval Climatic Optimum after AD 900 resulted in fluctuating lake levels and in turn, less predictability for near coastal fishing. This unpr edictability in coastal fish led to a series of decision - making events that resulted in a transition to a deep - water fishery (Dunham 2014). 61 3.5.1 O FT and Front - Back Loading Model To apply the diet breadth model to risk management strategies, specifical ly food storage, an additional framework is r equired. The Front - Back Loading Model can be used to explain how certain resources increase their rankings in a diet breadth model (Tushingham and Bettinger 2013). The front - back loading model considers the rank of a resource based on its storability. The model divides resources into front - loaded, those that are time - consuming to harvest/collect and store but expedient to prepare for eating, versus back - loaded resources, those that are easy to collect and store, but are costly to process for consumption (Tu shingham and Bettinger 2013). This model can help explain why certain resources are selected over others. Tushingham and Bettinger (2013) argue that in the case of hunter - gatherers that rely on storage, back - loa ded resources are often selected over front - l oaded resources because there is less up - front risk if the to their storability, even though they are costly to proce ss. Groups continued a mobility strategy that of wasted labor costs, than an intensification of salmon fishing which would have required increased group coope ration to compensate for the front - loaded pro cessing time (Tushingham and Bettinger 2013). Along similar lines, the Sel e ctionist Model can be applied to explain how an initially low - ranked resource can acquire a high ranking when the technology behind it s intended purpose gains fitness (Church and Nass 2002). In this case, if the technology of food storage becomes reliable and efficient, then storable resources acquire a higher ranking. When the storability of a resource is factored into its ranking and t - back loading qualities ar e 62 considered, then a more accurate scenario of hunter - gatherer decision - making and resource selection can be created. 3. 5 .2 Diet Breadth, Front - Back Loading and Food Storage In the next chapter a series of models will be laid out that explain the settle ment and subsistence changes in Michigan during the Late Woodland period. Later, in chapter 6, these models will be analyzed using the diet breadth, front - back loading, and sele ctionist models to determine which model, if any, best explains the prevalence of food storage during the Late Woodland period. 3.5.3 Critiques of OFT Several archaeologists have offered their critique of OFT and its applications (Smith et al. 1983; Gremillion 2002; Zeder 2012). The most common criticism of OFT models is that they tend to be over simplistic and reductionist in terms of human behavior ( Sm ith 1983 ; Gremillion 2002 ). Though, it can be argued that as a tool, OFT aims to simplify the real world by separating informative data from clutter to create a broadly applicable s cenario. Many pitfalls of OFT occur when these models are used in isolatio n; OFT should be used in conjunction with other methodological techniques (Smith 1983). Along the same lines, OFT is of ingredients create predictable human behavior. Another critique of OFT is that it hinges on Darwinian assumptions of human behavior, that the goal of decision - defends OFT and its use of Darwinian adapta many of the results of nat Oversimplification and incorrect application of OFT models can cause many issues, and more than likely invalidates the model. OFT models, like the one used by Winterhalder and 63 Goland (1997) were developed to be broadly applica ble. Models, like diet breadth, are often tailored to specific regions in order to account for individual taxa, technology, and social organization (Edwards 2017). 3.6 Summary T he goal of this chapter was to explain the systems of risk management and the complexity of storage in primarily hunter - gatherer systems. Risk management strategies rarely operate in a bubble and are most effective when coupled with multiple strategies an d utilized in a systematic program of hierarchical selection. Low - level risk management strategies are the most efficient, but only for predicted shortages; whereas high - level strategies are created to buffer extreme shortages, but they require more effort and maintenance. High - level strategies are generally deeply embedded practices and are more likely to cause cultural change (Halstead and Storage is a complex concept because it is often studied as an instigator of change; a mechanism that can create power dynamics. By breaking down the blunt terminology of food storage and redefining its basic premise, group survival, the focus of research can be sharpened. Reframing the terminology from causal to decision - based, Seasonal Gap Storage consi d ers the role of storage specific to hunter - gatherers and their need to mitigate risk. Finally, this chapter provided an initial framework with which to study the decisions that lead to food storage. The diet breadth model can provide a structure to under s tand decision - making, in regards to the selection of food storage as a risk management strategy. Further, the front - back loading and selectionist models can add input as to why certain resources acquire a higher - ranking when the technology of storage is u t ilized. 64 4.0 M ICHIGAN S TATE U NIVERSITY S UBTERRANEAN S TORAGE R ESEARCH E XPERIMENT (MSU SStoRE) A primary question in this research was, Are subterranean food storage features risk averse, or risk prone? For this to be determined, measurement of risk had t o be ascertained. Michigan State University Subterranean Storage Research Experiment (MSU SStoRE) was created to evaluate and measure the risk in food storage. The variables which are brought into play when the decision to use food storage as a risk manage m ent strategy are better comprehended if one understands the specific technology and associated processes behind food storage. The work and planning required to store food are factors in the decision - making process. For example, how much work was required t o construct and fill a storage feature; and could this work be incorporated into an existing collecting/hunting/horticulture pattern? To fully understand the technology and know - how required for successful food storage, it became necessary to actualistica l ly replicate the food storage process. To this end a series of archaeological experiments were conducted between 2010 and 2014 to determine the efficiency of the processing and storage of wild and domestic foodstuffs. The purpose of these archaeological e x periments was to determine the potential reliability, efficiency, and capacity of subterranean storage features. Additional forms of information explored, included timing and monitoring of storage, preservation techniques for wild foodstuffs, and nuanced s torage techniques were also tested. A second set of experiments were conducted to understand the technology and behavior behind the preservation of acorns for eventual storage. With a better comprehension of the processes behind food storage, one can more completely understand the decisions made in selecting a risk management strategy. 65 4.1 Experimental Archaeology Traditional excavation techniques can only answer so many questions of interest to the archaeologist. An excavated site only sheds light on on e point in time , and human behavior rarely expresses itself statically. At a certain point, it often becomes useful to attempt to recreate past human behaviors by attempting to re - produce that behavior in the present. Experimental archaeology is a method u sed to understand human behavior in which a hypothesis is tested , and one variable is controlled (Ingersoll et. al. 1977). This method has been debated in the field of archaeology because of its alternative approach to data collection (Binford 1983). Exp er imental archaeology should, of course, be used critically and can yield flawed results if insufficiently considered. But over the years, its usefulness and ability to explain human behaviors has made it an integral part of archaeology. Ingersoll et al. (1 977) identified four categories of experimental archaeology: those dea li collection techniques, but for the purpose of this paper only the replication and ethnographic types will be discussed. An excavated site only sheds light o n one point in time and human behavior rarely expresses itself statically. Eventually it becomes useful to attempt to recreate past human be haviors. Experimental archaeology method s attempt to understand human behavior by testing a hypothesis through the r e - creation of the behavior (Ingersoll et. al. 1977). The most prevalent type of experimental archaeology involves replicated studies. This method employs an imitative experiment that seeks to reproduce a past behavior. Ingersoll et. al. 66 stress that an unde rs an accurate reconstruction of the data (1977:xiii). Because the goal of replicated studies is to create a precise scenario, experiments are typically limited to issues of subsis te nce and technology. In archaeology, this translates into an infinite number of experimental possibilities. va 1977:xv). Ethnoarchaeology provides insight into behaviors that can be seen through butchering practices (1977). By witnessing the act t hat creates the cut marks on the bones, which are material correlates discovered by the ethno - archaeologist, a direct link can be made between behavior and artifacts (Yellen 1977). Regardless of the category, all archaeological experiments share somethi ng in common; they all seem to address problems in the archaeological record based on incomplete survival and 1973:14). Coles numerates rules that should be consi de red and applied to all experimental archaeology projects. First, all the materials used should be accurate for the time period. Second, the methods used should also be consistent with presumed original technologies when recreating acts. Third, the use of m odern technology should be avoided. Fourth, the experiment should be replicable so future research can build on the results. Fifth, the answer should be unknown. Sixth, the experiment can never demonstrate absolutely; p roof should never be claimed, only su ggested conclusions. Finally, the experiment should ask the right questions, ones that can be appropriately answered (Coles 1973:15 - 18). These basic standards of experimental archaeology provide guidelines for the conduct of responsible research and data c ollection. Experimental 67 Binford (1968) argued that if a correct methodological strate gy could be applied, the totality of a given cultural system was potentially discernable from the archaeological record inherent in the nature of the archae ol are advanced through methodology , then an understanding of human behaviors of the past can be more compre he nsive. This first - hand knowledge of food storage aided in the cross - cultural analysis by informing on the decision - making behind food storage as a risk management strategy. For example, timing of food storage can now be viewed from a technological stan dp oint, in addition to a sociocultural standpoint. 4.2 Previous Food Storage Experiments The replication of traditional subterranean storage pits requires the collection of archaeological, ethnographic, and ethnohistoric data. Previously collected archaeo logical data from northern lower Michigan (Howey and Frederick 2016; Howey and Parker 2008 ) were used to inform on the size and shape of the pit, and on the foodstuffs stored during the Late Woodland period. Additionally, the few previous experimental stud ies conducted on the use of subterranean food storage guided the replicative experiments. Reynold (1974) tested the subterranean storability of grain in lowland Britain, Arzigian et al. (2007) and Grooms (1996) both experimented with corn, the former in Wi sconsin and the latter in Ohio; and Cunningham (2005) 68 studied the storage of hazelnuts in Northern Ireland. Reynolds found that subterranean storage of grain was most likel y not for planting because although the grain did germinate, the germination rate po st storage was greatly reduced. Experiments at Fort Ancient were successful in storing cor n with high post storage germination rates (Grooms 1996); while experiments of Oneota storage pits showed more success in the germination of full ears of corn, at 94% , over shelled corn at 5 - 10% (Arzigian et al. 2007). Reynolds (1974) explains that in orde r to prevent germination during storage a proper anaerobic environment must be present. This process occurs only if the storage container is sealed, otherwise the lac k of carbon dioxide buildup will not allow for anaerobic atmosphere (Reynolds 1974). Reyno lds (1974) argues that if the storage pit is opened and closed repeatedly (or even once) the carbon dioxide will not build up, causing grain loss through germination. reopened periodically , resulting in partial grain germination (Grooms 1996). The consideration of pit lining to prevent mold and pest infestation is another integral aspect of storage construction. In the storage of hazelnuts, the storage pits that utilized baskets as liners had the grea test level of success (Cunningham 2005). When modeling storage pits after (1996) lined storage pits with Big Bluestem ( Androp ogon gerardi) grass which absorbed much of t he moisture . Arzigian et al. (2007) found that the crops stored in the pits without a grass lining developed several toxic mold s . A basket - lined pit resulted in almost total recovery of the stored grain, but the lining could not be reused due to microflora buildup (Reynolds 1974). An unlined grain sto rage pit resulted in the formation of a grain skin along the wall of the pit ; thus some grain loss occurred ; however the larger the pit the less percentage of grain l oss due to grain skin. Therefore, a large unlined grain storage pit would have crop loss, but in a patterned, predictable manner (Reynold 1974). 69 Though aspects of other storage experiments could be used to inform in the creation of MSU SStoRE, these previ ous experiments focused primarily on grains and maize, all domesticated, crop foodstuffs. that held only wild, non - crop foodstuffs, making MSU SStoRE the first experiment of its kind. Genera lly, the storage of domesticated crops is considered a success if the crop seeds do not ge rminate; while the measure of success in the storage of wild, non - crop foodstuffs is edibility. 4.3 Experimental Subterranean Food Storage The Michigan State Unive rsity Subterranean Storage Research Experiment (MSU SStoRE) aimed to replicate the storage behavior found in underground archaeological storage features. While specifically directed at the late Late Woodland within the Inland Waterw ay of northern lower Mic higan, the experiments and results have broader applicability across similar contexts both in North America and other parts of the world. In order to understand how such facilities operate within a hunter - gatherer/mixed horticultura list community, and more specifically the Late Woodland setting of the Inland Waterway, it is necessary to understand the mechanics of the physical creation and use of the cache pits. For this series of three experiments I utilized archaeological and eth nographic/ethnohistoric data in order to replicate a subterranean food storage pit that preserved wild foodstuffs. Three seasons of experimentation were conducted with revisions cumulatively implemented every season based on compiled successes and failur es culminating in a su ccessful food storage episode. An unintended outcome of the experiment was realization of the labor and seasonal forethought and timing required in the construction of a food storage pit. 70 MSU SStoRE was conducted in three separate experiments, each buildi ng off the previous failures and successes. 4.3.1 Round One The first round of MSU SStoRE, conducted in the Fall (September) of 2010 through Spring (April) of 2011, considered the numerous variables in the development of a cache p it; including soil type, natural soil drainage class, foodstuff, pit lining, season, and containers. Additionally, ethnographic and ethnohistoric data were collected to ensure an accurate representation of traditional construction methods. Excavation of s everal prehistoric cache pits in the Inland Waterway by Howey and colleagues, and the resulting botanical and faunal analysis of the soil samples revealed that the cache pits there would have normally been filled with berries, acorns, and fish (Howey and P arker 2008; Howey and Fr ederick 2016). Excavations of the cache pits indicated that they were on average 1.5 meters in diameter and 1.5 meters in depth. Ethnographic and archaeological research revealed that pits would have been lined with bark, insulated with dry grass, and cove red with logs and a layer of soil (Howey and Parker 2008; Howey and Frederick 2016; Densmore 1929; Wilson 1917). Because of the foodstuff recovered and as further ethnographic research confirmed, the cache pits would have been fille d and sealed after the a corn harvest in late September and reopened in late March, or early April. Large ceramic sherds were found in two of the excavated cache pits (Howey and Parker 2008; Howey and Frederick 2016), leading to the inference that at least some of the foodstuffs w ould have been stored in ceramic vessels. 71 Figure 4.1 : Left , Unexcavated cache pit, visible as a slight surface depression. Right , Cross - section of an excavated cache pit (Howey et al. 2016). Figure 4.2 : Cross - section pro file of storage feature excavated by Meghan Howey; 2007. 72 Round 1 of the MSU SStoRE experiment entailed the construction of four subterranean pits within the Inland Waterway, located in sandy, well - drained soils. Each pit was 2 meters in diameter and 1.5 meters deep. Two of the pits were each filled with 3 kg of dried blueberries (a substitute for wild raspberries), and 21.4 k of parched acorns, and the other two pits were filled with 49 kg of smoked salmon (a substitute for lo cal whitefish or lake trout). Locally harvested birch bark was used to line the inside of each pit, while dried straw (substitute for dried grasses) was used for insulation (Densmore 1974). Additionally, two data loggers (iButton DS1923L), placed in each p it, recorded temperature and humidity every 244 minutes. Each pit was covered with logs from downed trees and covered with leaves. The food was cached September 24 th of 2010 and the pits were re - opened on April 2 nd , 2011 (Frederick 2011). Figure 4.3 : MSU SStoRE Round 1 p it lined with birch bark and insulated with straw. 73 Results: The Round 1 experiment was unsuccessful, although the data collected were useful to build upon. Upon opening the pits, it was quickly discovered that all of the f ood had been consumed by animals . The two pits that contained the smoked salmon had been repurposed as coyote dens and lacked any stored food. The only useful data from Round 1 were the temperature and humidity recorded on the data collectors. These data i ndicated that the pits remained at a steady 34°F (1°C) - not including the pits that recorded the body temperature of the coyotes. According to the USDA ( https://www.fsis.usda.gov ), rapid bacteria growth begins at 40°F (4.4°C), therefore the experimental p its maintained a temperature tha t prohibited bacteria growth. The main factor to overcome for round two, beyond the preservation of edible food, was keeping the animals out. Figure 4.4 : Blueberry and trout storage containers after Round 1 (note both containers lacking foodstuffs). 74 4.3.2 Round Two Preparation: Learning from Round 1, several modifications were made for Round 2 (Winter 2012 - 2013) of MSU SStoRE. The first modification cam e with the realization that monitoring was most likely required for successful fo od storage. Therefore, the location of the experiment was moved locally, to Lansing, so that periodic monitoring by the investigator could be implemented. The next modificat ion was the timing of the caching and opening of the pits. Further ethnographic/e thnohistoric research indicated that a caching timeframe of early November correlated with the fall spawning of lake sturgeon, a known time of aggregation for the hunter - gathe rers of the Late Woodland period (Smith 2001). Not only do fish make for an abund ant resource to store at the end of the warm season, but the work required to construct subterranean storage pits would have been offset by the abundance of labor. It was infe rred that if fish were being stored, then early November was the logical caching season. A final modification was in the pit construction. The final capping of the pit in Round 1 did not create an anaerobic environment, and also resulted in pest disturba nce. Further ethnographic research indicated that the pits would have been mounde d with a thick cap of soil in order to isolate the smell of any foods within, and in turn to keep the animals out. Thus, a soil cap was used for the pits in Round 2. All of these changes were implemented for the 2012 - 2013 Round 2 of MSU SStoRE. Four pits were dug, measuring 1.5 meters in dimeter and 1.5 meters in depth. The soil was a Spinks loamy sand (USDA Soil Maps), which is a well - drained soil type. Each pit was filled with 2.6 kg of parched acorns, 0.5 kg of dried blueberries, and 1.4k g of corn meal . Instead of covering the pits with logs, for this round they were capped with 30 cm of sediment. Each pit was equipped with a data logger that recorded temperature and humi dity. Additionally, the pits 75 were visited weekly to inspect for animal activity. Th e pits were loaded and sealed on November 15, 2012, and re - opened on April 12 th and 14 th , 2013. Results: Results from Round 2 were vastly improved; food remained in the pits. The weekly monitoring did not reveal evidence of animal activity in or around th e pits. All cached food was accounted for. However, there was an issue of moisture build up in the pits. T he data loggers recorded that the humidity levels in the pits topped out - at 90% humidity - after the first three weeks. Resulting in moldy blueberrie s and corn meal. The best preservation came from the acorns, which displayed minimal mold, no signs of ins ect infestation, and no sprouting. Round 2 was considered a successful caching season; however, the question was raised as to whether the food was act ually edible, or simply that it was visually edible. Figure 4.5 : Round 2 blueberries aft er reopening the pits. Notice the mold on the upper left corner of the berries (note, pocket knife for scale). (Photo by W.A. Lovis) 76 Figure 4.6 : Round 2 acorns after reopening. There is no evidence for mold. (Photo by W.A. Lovis) Figure 4.7 : Round 2 data logger results for Cache Pit 1. The x - axis represents time and the y - axis is temperature and percent humidity y within the first couple of weeks. (note, the solid lines represent the upper and lower alarms set on the data loggers). 4.3.3 Round Three Preparati on: Round 3 was able to build from both Rounds 1 and 2 and correct mistakes in the technology of food storage in cache pits. The main issue to overcome was the humidity in the 77 pits which had resulted in moldy growth. Continued research in the ethnographic /ethn ohistoric record alluded to a possible miscalculation on our part. The operat ing assumption was that the cache pits would have been opened near the beginning of spring, but more than likely the opening of the pits would have corresponded with the mapl e sug aring season (Holman 1984). The running of the maple sap occurs in mid - March in Michigan, almost one month prior to the date when the pits were re - opened in Round Two. Generally, the ground is still cold or nearly frozen during this time of year, ther efore there would have been less water infiltration from April rains and snowmelt. Utilization of the stored foodstuffs during maple sugaring season would also have provided carbohydrate and protein diet to offset the maple sugar. Round three of MSU SStoR E aga in built upon the earlier experiments and culminated in a very successful caching season. For this round, only two pits were dug, each measuring 1.5 meters in diameter and 1.5 meters deep. The pits were dug in the same location as Round 2 so the soil was S pinks loamy sand, a well - drained soil. Each pit was filled with 6.8 kg of dried blueberries, 6.8 kg of charred acorns, 113 g of dried hominy, and .5k g of dried flint corn kernels. The pits were filled on November 15th, 2013 and re - opened March 31st, 2014. Results: Upon opening the pits, the stored food was visually edible. In order to confirm their edibility samples of each of the stored foods were collected and analyzed further, using modern standards for evaluation of food safety. Samples of foodst uffs from both pits were collected and sent to the Michigan Department of Agriculture and Rural Development (MDARD) for further analysis. Microbiological tests were conducted by MDARD and analyzed for aerobics, yeasts, molds, general coliforms, and E. coli . The results indicated that all the stored foodstuffs were edible. Additionally, I performed tests on each of the samples to analyze the 78 the vapor pressure is 80 perc ent of that of pure water. The water activity increases with temperature. The moisture condition of a product can be measured as the equilibrium relative https://www.fda.gov/ICECI/Inspections/InspectionGuides/InspectionTechnicalGuides/ucm07291 6.htm ). In foods with a low pH, the water activity should be be low .86aw, or risk the growth of bacteria, yeasts and molds (personal correspondence with Dr. Leslie Bourquin, MSU Food Science and Human Nutrition). As figure 11 indicates, the blueberries and flint corn tested below .86aw, but the hominy and acorns were right at the cusp of bacteria growth and above, respectively. Foodstuff Tested Water Activity (aw) Blueberries 1 - pit #2 0.668aw Blueberries 2 - pit #2 0.576aw Blueberries 3 - pit #3 0.588aw Blueberries 4 - pit #3 0.646aw Acorns 1 - Pit #2 0 .92aw Acorns 2 - Pit #2 0.945aw Acorns 3 - pit #3 0.942aw Acorns 4 - pit #3 0.946aw Corn 1 - pit #2 0.754aw Corn 3 - pit #3 0.726aw Table 4.1 : Water Activity Analysis, Round 3. 0.86 aw is stable based on food safety standards. 79 Foodstuff Tested Water Activity (aw) Hominy 1 - pit #2 0.893aw Hominy 2 - pit #2 0.878aw Hominy 3 - pit #3 0.894aw Hominy 4 - pit #3 0.881aw Figure 4. 8 : Left , Round 3 Acorns, Hominy, and Flint C orn after reopening. Right , B lueberries recovered after Round 3 also (Photo by W. A. Lovis) . Because bacteria, yeasts, and molds begin to grow at .86aw and above and some of the foodstuff exceeded this value, samples of the acorns and flint corn were then tested for mycotoxins. Mycotoxins are any fungal by - product that directly leads to health hazards (Kenst 2002). There are several types of mycotoxins potentially lead ing to a wide range of health hazards from cancer to organ disorders, some more acutely toxic than ot hers (Kenst 2001). The samples were analyzed for two types of Zearalenone, Vomitoxin, T - 2 Toxin, DAS, TxSpec Table 4.1 : ( ) 80 Myco and four types of Aflatox ins. A ll the foo dstuffs tested below the threshold for toxicity, except the acorns from Pit B which tested positive f or a detectable level of Zearalenone, at 3.5ppm. This mycotoxin is the bi - product of the Fusarium fungus species which thrives off moisture. Zearalenone ca n lead to fertility issues, al though this most often occurs in swine (Kuiper - Goodman et al. 1987). Ho wever, toxicity studies done on zearalenone indicate that humans are fairly resistant to its effects, and would not be sensitive to its toxicity at this le vel (Kuiper - Goodman et al. 1987). Table 4.2 : Mycotoxin Analysis for Round 3 Foodstuffs. The red circle indicates that only the acorns tested positive for a Mycotoxin, Zearalenone. Based on modern food science standards , the foodstuff cached in the subt erranean pits for five months met the curren t acceptable parameters for food safety. MSU SStoRE therefore created a successful cache pit that would have yielded healthy, edible food and offset seasonal resource scarcity and provided us with the necessary c alories during the leanest time of the year. After three rounds of experimentation, MSU SStoRE determined that one method to create a successful and reliable cache pit, one that preserves food throughout a winter season, in 81 northern lower Michigan require d: a) a soil matrix of well drained sand, b) the construction of cache pits that are approximately 1.5 meters in diameter and 1.5 meters in depth (though this may be variable), c) foodstuffs with a water activity level below .86 aw, or in this case dried b lueberries and parched acorns, d) dried birc h bark to line the inside of the pit, e) dried grass or straw to insulate the foodstuffs in the pit, f) and a soil capping that is mounded at a minimum of 30cm. Additionally, the food should be cached around mid - November and opened in mid - March before the winter thaw. 4.3.4 Discussion The MSU SStoRE experiments created a diagnostic baseline against which to further test the efficiency and reliability of subterranean storage pits. It is clear that subterranean ca ching is a highly refined technolo gy requiring detailed knowledge about construction, resource processi ng, packaging, and most i mportantly timing . Now tha t the parameters are known for creating a basic, successful , and reliable cache pit, further experimen ts can be performed that examine the effects of other variables such as specific food proc essing techni ques, vessel types ceramic ver sus bark containers or baskets lining, and degree of accessibility during the winter. Comprehensively understanding the risks and uncertainties involved in creating, filling, and relying on cache pits for survival will lea d to a better understanding of the socioeconomic organization required for such a resource system. 4.4 Acorn Processing Experiment The excavation of a feature at the Green Site in Perry, Michigan revealed a subterranean pit with carbonized acorns (see Ap pendix 1 for Green Site report). After further excavation and analysis of recovered material, it was hypothesized that this feature, which had a 14 C age of 82 1240 ± 70 B.P. (cal AD687 - 901), was an acorn processing pit. Using the field and lab data, a replicati ve archaeological experiment was created to confirm our functional hypothesis. Figure 4. 9 : Green Site Acorn Processing Pit. (Photo by W.A. Lovis) 4. 4.1 Acorns in the diet Multiple ethnographic and ethnohistoric (Densmore 1974; Hilger 1992; Smith 1923) accounts detail the utilization of acorns in the Great Lakes region. Smith (1923) details how the ) explains the timing of acorn collection. Acorns were an abundant resource that were easily gathered and processed (Dunham 2009). Studies have shown that in terms of nutritional characteristics, acorns are on par with wild rice an d maize (Speth and Speilm ann 1983; Malainey et al. 2001). As a carbohydrate replete in fat, acorns would have been a nutritional compliment to a diet that was 2009). In addition, the storabil ity of acorns, in either the whole form or as flour, made it an appealing resource. 83 Archaeologically there is evidence for time depth in the utilization of acorns in the Upper Great Lakes region. The Butternut Lake site (47FR137) in Wisconsin has evidence for the processing of acorns that dates to the Late Archaic (Bruhy et al. 1999). A feature at the Butternut Lake site was determined to be an acorn parching pit, with remnants of charred acorn shell. Unlike the Green site parching pit, the parching pit at the Butternut Lake site is located within a dwelling and contained only fragments of charred acorn shell. 4.4.2 MSU SStoRE Part two - Acorn Processing Pit Preparation: It was determined, by measuring the carbonized whole acorns, that the acorns were predominately from white oak trees (see Appendix 1). Oak trees in Michigan fall under two broad categories, red oaks and white oaks (sometimes referred to as the black oaks) (Dunham 2009). It is unclear whether white oak acorns would h ave been purposefully selected, but there are a number of advantages in collecting white oak acorns over red oak acorns. Firstly, white oak acorns contain less tanic acids than red oaks, which me ans less leaching is required before consuming white acorns ( Dunham 2009). It has also been argued that the acorns of some species of white oak, like bur oak, do not even require leaching to be edible (Densmore 1974). Another advantage of white oak acorn s pecies is that they mature in one year and fall to the ground , whereas red oak acorns require two years to mature. Therefore, in general, white oaks will produce more acorns than red oaks (Dunham 2009). A final reasoning for selection of white oaks over red oaks can be found in collection and processing time. In an experiment that considered the calories gained after collection and shelling of the acorns, Dunham (2009) found though red oak acorns were mor e efficient to collect (2001/hour vs. 1520/hour for bur oak and 1612 for white oak), white oaks (specifically bur ) were much more efficient to shell (5 hours vs. 16 hours for red oak), leading to a caloric 84 benefit of 3533 calories/hour of collection for bu r oak, and 1011 calories/hour of collection for red oak (Dunham 2009). These lines of evidence suggest that white oak acorns have processing and caloric advantages over red oak acorns and were likely purposefully selected and collected. Therefore, the fir st step in the experiment was to collect 25lbs/11.3 kg of white oak acorns. In late October 2016 we coordinated w ith Dr. Frank Telewksi, MSU plant biologist, to ensure the collection of only white oak acorns. Dr. Telewski monitored the acorn collection pro cess and acted as a quality control, so a non - contaminated sample could be insured. After collection, the acorns were laid out to air dry and prevent mold, since there was a lapsed time of one week between collection and preservation. Additionally, the ac orns were monitored for insect activity, and any infested acorns were discarded. An estimated 10% of the collecte d acorns were discarded due to insect activity. Soil samples collected from the Green site determined the soil to be sand and silty sand (Ran dall Schaetzl, personal correspondence 2016), consequently, the location selected for the experiment contained si milar sediment conditions. The Green site feature suggested that after the pit was dug, a fire was built within the pit, then the acorns were placed on the hot coal bed, covered with a layer of sand, and left to smolder. Thus, in the experiment, a charri ng pit was dug to a depth of 50 cm and a diameter of 1 meter. A fire was ignited within the pit and allowed to burn for 1 hour, creating a red - hot coal bed. The acorns were then dumped directly onto the coal bed, topped with a piece of birch bark, and seal ed with a layer of sand to a depth of 7 cm. The birch bark was used as a buffer between the acorns and the sand. The acorns were left to smolde r and dry for two days. 85 Results: Upon opening the charring pit, it was discovered that although the acorns were dried, they were not charred. Therefore, the experiment did not directly replicate the observations recorded on the Green site feature. However , our results indicated that most likely the carbonized acorns recovered from the Green site were an unintentiona l outcome since they represent processing spoilage. Because the goal was to replicate the archaeological findings, a second round of experiment ation was conducted in order to produce charred acorns. Round 2: A second round of experimentation was required t o achieve the desired outcome of carbonized acorns. An additional 25 lbs/11.3 kg of acorns were collected for Round 2 of the experiment. Collec tion and experimentation occurred on the same day, and it was assumed that immediate exposure to fire would dispe l any insect activity. Again, a hole was dug to a depth of 50 cm and 1 meter in diameter. This time the location was in sandier soil. For the second round, a fire was built in the pit and allowed to burn for an hour, but fuel (wood) was continually added to ensure flames and not let it burn to coals and embers. After an hour, the acorns were loaded into the fire. After the acorns were loaded int o the fire the flames ceased, but the acorns continued to sizzle and pop for almost 30 minutes after exposure. Unlike the replication in Round 1, for Round 2 a layer of sand was not placed over the acorns, the mass of the acorns was enough to douse the f lames and stop the active fire. Once it was evident that the fire was extinguished, and the acorns were no longer showing signs of burning (smoking, popping) the pit was covered with a sheet of plywood and left to cool. Because it would have been difficult to extract the acorns from a hot fire, the acorns were cooled for two days (in accord with Round 1) and then rem oved. 86 Results: Upon extracting the acorns, it was evident that the results of Round 2 more closely resembled our archaeological observations. T he bottom layer of acorns, the layer that directly doused the flames, were heavily charred, and a handful were fu lly carbonized. The upper layers were not charred but were parched from the escaping heat of the fire. 4.4.3 Conclusions Data obtained from the second round of the acorn charring experiment most closely resembled data collected from the Green site acorn feature. Applying these result s to the acorn feature, I believe that the acorns were poured from baskets or bags onto an active fire, not a be d of embers. The completely carbonized acorns were inedible, and therefore left behind. These unwanted acorns are what remained in the feature at the Green site. The acorn feature at the Green site alludes to a time depth as early as cal AD783 for the pr ocess of parching acorns for preservation and storage. According to our experiments, in the process of acorn parc hing, acorns are placed onto a fire with open flames and parched by heat. Most of the acorns become parched, a percentage of the acorns become charred, and some become carbonized and inedible. This method for acorn preservation prolonged the shelf - life of the acorns by inhibiting mold and insect damage as well as germination, and therefore was an effective method of processing before storage. Bas ed on the AMS age of the feature, it is apparent that the technology for acorn preservation processing was known and practiced well before the popular use of subterranean storage in the Late Woodland of Michigan. The decision to more intensively utilize su bterranean food storage three hundred years later was, therefore, a low risk decision. 87 Figure 4.1 0 : Ro und 2 Acorn Charring. Notice the flames visible during parching. Figure 4.1 1 : Round 2 remnants of charred acorns after burning. (Photo by K. Frederick) 4.5 Discussion : The Work of Storing Food An integral revelation during the course of these experiments was the planning and labor required for successful food storage. These experiments indicated that food storage is not a last minute, off - the - cuff decision, it is precisely planned and coordinated throughout the year. 88 For birch bark to b e a sus tainable resource it must be harvested when the birch tree is engorged with sap. This period of engorgement enables the bark to be harvested without damaging the cambium. The cambium is responsible for tree growth, and in turn, growth of the tree ba rk; if the cambium is damaged the tree will die (Bergman 2004). During this engorgement the birch bark is at peak harvesting time; the bark will easily pop off (quite literally makes a eason, the bark is difficult to remove, and the removal will more than likely kill the tree. The peak season for birch sap in strawberries are in bloom - Keshic k, Elder from Sault St. Marie Tribe). Birch bark had many uses besides lining storage pits (see Densmore 1979), therefore an adequate amount had to be collected in the Spring to meet all needs for the remainder of the year. The late spring and early summer is also the beginning of wild berry season. Wild strawberries ripen first, followed by blackberries, then raspberries and end with blueberries in the late summer and early fall (Densmore 1979). Each of these berries had to be collected en masse , th en proc essed in a manner that extended their shelf - life. More than likely, the early berries (strawberries and blackberries) would have had an early caching episode, rather than waiting until fall, or were consumed upon harvest. This early caching would ha ve requ ired labor and planning to maximize the front - loading required for berry storage. The fall would have been the busiest caching season. Nut masts begin in September and continue through October. Several nut varieties would have been available, inclu ding ac orn, hazelnut, beechnut, and chestnut. As the acorn charring pit and resulting experiment indicated, acorns (and other nuts) would have required parching before being cached. 89 The final harvesting episode during the caching year is the fall fish spa wn. Dur ing the fall fish migrate to the shallow waters of the Great Lakes to spawn (Smith 2004; Cleland 1982). This allowed for an abundant resource to be harvested in surplus, aided by the gill net (Cleland 1982). Catching the fish may have been easier du ring th e spawn, but there was still the same amount of work required to process them for storage. If the weather permitted, the fish could have been frozen for storage, but if it was too warm then they required the labor - intensive process of smoking. Fish, the fi nal resource before winter, would have capped off the season of storage. A handful of studies (Bettinger et al. 1997; Morgan 2012; Howey and Frederick 2016) have been done to estimate the labor cost of filling a food storage container. Though impe rfect, these estimations can provide a general idea of the planning required to sustain a community throughout the winter, and the risk in the labor cost if storage fails. Howey and Frederick (2016) calculated the average size of the subterranean storage feature s excavated around Douglas Lake in northern lower Michigan to be 207 cm wide (diameter at surface) and 143 cm deep (note: this is a similar size estimate reported by Densmore). Using the formula for a truncated cone , the average volume for pit capacity was 2.50 . This volume was then converted to liters and quartered, to account for pit lining, insulation, and use of containers, leading to a mean of 652 liters of estimated storage capacity (Howey and Fre derick 2016 ). Though evidence indicates that the storage features in northern lower Michigan would have been filled with differing foodstuffs, from raspberries to fish, storage of acorns is currently the most quantifiable. Previous research (Bettinger et al. 1997) determined the amount of unhulled acorns storable per liter is 1.45 kg. Using this figure, it is estimated that the storage features around Douglas Lake could hold up to 906.25 kg of acorns. 90 Well established figures (Bettinger et al. 1997; M orga n 2012; personal data of MSU SStoRE acorn collection) have offered a baseline for acorn collection rate of 9 kg per hour. Plugging this collection rate into the average estimated storage capacity of these features results in a figure of 100.7 hours. It wou ld take one person, working 6 - hour days, 16.5 days to fill one subterranean storage feature (Howey and Frederick 2016). Taking these calculations a step further, an estimation of caloric return can be established. Dunham (2009) found that, on average, Red Oak acorns provided 3.14 calories per gram and White Oak acorns provided 2.38 calories per gram. Therefore, a liter of Red Oak acorns would provide 4553 cal, and 3407.5 cal per liter of White Oak acorns. Using the figure of 906.25 kg of acorns per sto rage pit, it is estimated that a single pit could provide between 215,628 - 284,484 cal. Kelly (1995) calculated that the average hunter - gatherer required 2,440 cal/day; a single storage pit, then, could sustain an individual somewhere between 88.4 116.6 day s, if they were living off acorns alone. In sum, 16.5 days of acorn collection could provide around 100 days of sustenance for an individual, or 4 days of work could provide a month (25 days) of sustenance for a family of four. This, of course does not acc ount for the processing required before caching, or the processing of the tannins before consuming the acorns. It is evident that all aspects of food storage required an immense amount of labor, in addition to the labor required for daily activities. Prov isioning a cache of food to survive the winter months mandated an organized and properly timed decision - making strategy that began as early as spring and continued until early winter. This level of work and organization would not have been utilized unl ess food storage was the primary risk management strategy. 91 4.6 MSU SStoRE Conclusions This series of experiments has created a foundation for a further understanding of food storage and preservation in northern lower Michigan. Archaeological data has sho wn that subterranean food storage was en vogue towards the end of the Late Woodland, and a greater comprehension of the mechanics behind said food storage can further aid in assessing the causation for the change in subsistence practices. MSU SStoRE h as s hown that food storage of wild foodstuffs, like acorns and berries, was a reliable, and if done correctly, low risk endeavor. Additionally, the examination of an acorn processing technique that predated the popular use of subterranean food storage indi cate s an efficient preservation method that could have been seamlessly folded into the practice of food storage, making the act of food storage even more reliable. It is evident that available foodstuffs, like acorns, would have been a nutritious and sto rabl e resource. Not only were acorns an easily collected resource, the majority of the required processing (i.e. leaching) could have been backloaded, making them an ideal resource to store. Parching acorns before storage would have dried the nut meat, pre vent ed bug infestation, and ceased germination. Additionally, acorns would have been a nutritional compliment to a diet that relied heavily on fish, and the storage of fish. Finally, MSU SStoRE allowed for an evaluation and measurement of the possible ri sk i nherent in food storage. This series of experiments showed that subterranean food storage is a reliable, effective, and risk averse management strategy. Once the processing and timing were perfected, MSU SStoRE verified the risk management potential fo r fo od storage. Chapter Five will begin to explore the utilization of storage by hunter - gatherers around 92 Area Files, Chapter Five considers the variables that revol ve a round hunter - gatherers and their use of storage. Do patterns of storage use emerge as variables are analyzed and compared? Using common variables, can a model for storage as a primary risk management strategy be developed? 93 5.0 HUNTER - GATHERER STORAGE PRACTICES To have a more comprehensive understanding of the decision - making behind the utilization of food storage by hunter - gatherer societies it is advantageous to compare food storage on a global level. In order to accomplish thi s, I used two existing d ata sets, Binford and - Gatherer Frames of Reference (EnvCalc) and the Human Area Relations Files (HRAF). Combining these data sets allowed for an analytical comparison of hunter - gatherer deci sion - making. This chapter will provide background on the data sets and explain the methodology of the data collection and the subsequent analysis of the data. Numerous archaeological studies from the nineteenth century up through t he present (see Stopp 20 ethnohistoric data to archaeological context in order to creat e a more complete and co mprehensive picture of past behaviors. 5.1 EnvCalc Data Set culminating work Constructing Frames of Reference: An Analytical Method for Archaeological Theo ry Building Using Hunter - Gatherer and Environmental Data Sets ( 2001), in which he records over 200 variables on 339 hunter - gatherer groups from around the world. The data in - gatherers and the potential patte rned responses to environmental variables (Binford 2001). The documented 94 hunter - gatherer variability in his volume is then organized into a frame of reference that can then be compared to the archaeological record (Binford 2001). To make the data even m ore accessible, the EnvCalc Program (Binford and Johnson 2014) was created. This program not only provides the original data set, it also allows for further analysis of the input variables. EnvCalc uses data from 339 ethnographicall y documented hunter - gath erer groups, 1429 weather stations. Each record in a basic input file represents a location for which information on latitude, longitude, elevation, distance to the nearest coast, soil type, vegetation type, mean monthly temperature and rainfall are record ed. The output file includes a number of variables which measure properties of the environment including : heat indices, water balance, available plant productivity and the structure of the plant community. These calculated variables are then used to projec t both ungulate biomass and a variety of variables which indicate projected behavior of hunter - gatherers, like those ethnographically documented, in the organiz ation in similar environmental conditions. data nor the EnvCalc program dire ctly address the use of food storage by hunter - gatherer groups. However, the vast amount of data present in EnvCalc provides a jumping off point from which to compare environmental and organizational variables that can be c ompared to identify and elucidate patterns of behavior. Since EnvCalc does not record the practice of storage, a secondary data set, HRAF, was required as a tandem vehicle for the ethnographic analysis. 5.2 Using the Human Area Relations Files The Human Area Relations Files (HRAF) are a collection of information about almost 400 cultural groups around the world. This collection consists mostly of primary ethnographic 95 source materials and facilitates cross - cultural research. As of 2008, Electroni c HRAF (eHRAF) had a collection of 165 cul tures available online ( http://hraf.yale.edu.proxy2.cl.msu.edu/cross - cultural - research/basic - guide - to - cross - cultural - research/ ). Data collection for this research pulled exclusively from the eHRAF resources. Every paragraph of every document in eHRAF has been indexed, allowing for expedient searches of over 1,000 subjects; the subjects are concept based and not keyword based. Once a subject or culture is searched, the results can be narrowed by subsistence type and sample. There are eight categories of sub sistence type, ranging from hunter - gatherers to commercial economy. Additionally, the results can be organized by sample type; Ethnographic Atlas (EA), Probability Sample Files, Standard Cross Cultural Sample (SCCS ), and Simple Random Sample. It is necessa ry to create a search with constant variables to ensure consistence results. It is important to note that there have been multiple studies (Naroll 1964; Naroll 1965 ; Shaefer 1969 ; Naroll 1970 and Ember 2007 ) conducted on the best methodology for utilizing cross - cultural comparisons and HRAF. Ember (2007) emphasis a consideration of the time period. A proper cross - cultural comparison cannot be conducted if the researcher is comparing different time periods (Em P roblem (Nar oll 1965), or th e issue with also requires a planned approach to cross - cultural studies. The researcher must determine whether are the result of cultural diffusion, ra ther than culturally func tional (Naroll 1965). Schaefer (1969) argues that (Schaefer 1969:299) is only the first step in creating a cross - cultural comparison and that a li nked pair method (Naroll 1964) is required to ov 96 study by checking the similarity analysis of functional, rather than di ffused behaviors. My methodology for creating a cross cultural comparison using eHRAF consisted of hunter - gath s defined as 56% or more dependence on hunter, - defined as 86% or more dependence on hunting, gathering and fishing, exclude d too many cultural groups. Finally, I selected the Ethnographic Atlas (EA) sample. The EA sample contains 1,264 societies and was created to be an exhaustive list of world cultures. The advantage of EA samples is they contain information focused on a spec ified time and place. This allowed for a search control to ensure a comparison across similar time periods. 5.3 The Dataset A dataset was created that used information from eHRAF to add a variable of storage to the EnvCalc database. Because eHRAF allow s for a topic search across cultures, a narrow search was created that pinpointed hunter - ethnographic groups, 77 are found in eHRAF. Of those 77, 62 (see Table 5.1 for complete list) mention the practi ce of food storage. All 62 groups were compared in the statistical analysis. However, because the level of analysis required for research entailed an in - depth look into specific storage practices, six groups from the 62 groups were selected for direct comp arison. These groups were selected for direct comparison primarily because they had the most documentation related to food storage. 97 It is noteworthy to mention that for a group to be considered for the storage variable, the ethnographic literature had to specifically mention the forethought of setting food aside for the lean season. Cultural groups that had mention of food preservation, but not specific to a stockpiling of food, in the sense of Seasonal Gap Storage, were omitted. Table 5.1 : Hunter - gath erer groups that can be found in both eHRAF and EnvCalc (Binford and Johnson 2014). Group Name Location Storage (no/yes) SHOMPEN NICOBAR I S. No SEMANG MALAYSIA No VEDDAH SRI - LANKA Yes AINU_HOKKAIDO JAPAN Yes GILYAK RUSSIA Yes NGANASAN R USSIA Yes SHIRIANA VENEZUELA No YARURO - PUME VENEZUELA Yes SIRIONO BOLIVIA No BOTOCUDO BRAZIL Yes ONA ARGENTINA No AKA CONGO No BAYAKA CONGO No DOROBO - (OKIEK ) KENYA No TIWI AUSTRALIA - NT No EASTERN - POMO CALIFORNIA Yes CLEAR - L AKE - POMO CALIFORNIA Yes POMO - SOUTHERN CALIFORNIA Yes YUKI - PROPER CALIFORNIA Yes POMO - NORTHERN CALIFORNIA Yes ATSUGEWI CALIFORNIA Yes YUROK CALIFORNIA Yes KLAMATH OREGON Yes YAVAPAI - ARIZONA Yes UTE - TIMANOGAS UTAH Yes UTE - WIMONANT CI UTAH Yes UINTAH - UTE UTAH Yes COMANCHE TEXAS Yes CROW WYOMING No 98 Group Name Location Storage (no/yes) PLAINS - OJIBWA NORTH DAKOTA Yes PIEGAN ALBERTA Yes BLACKFOOT ALBERTA No ASSINIBOINE SASKATCHEWAN No PLAINS - CREE SASKATCHEWAN Yes BLOOD ALBERTA Yes NOOTKA BRIT. - COLUMBIA Yes CHINOOK OREGON Yes QUINAULT WASHINGTON Yes MAKAH WASHINGTON Yes BELLA - COOLA BRIT. - COLUMBIA Yes CHUGASH ALASKA Yes ALEUT ALASKA Yes OJIBWA - KITCHIBUAN MICHIGAN Yes MICMAC NEW - BRUN SWICK Yes RAINY - R. - OJIBWA ONTARIO Yes NO. - SAULTEAUX ONTARIO Yes PEKANGEKUM - OJIBWA ONTARIO Yes ROUND - LK. - OJIBWA ONTARIO Yes NIPIGON - OJIBWA ONTARIO Yes WEAGAMON - OJIBWA ONTARIO Yes MONTAGNAIS QUEBEC Yes KASKA BRIT. - COLUMBIA Yes TAHLT AN BRIT. - COLUMBIA Yes INGALIK ALASKA Yes HOLIKACHUK ALASKA Yes NASKAPI QUEBEC Yes KOBUK - INUIT ALASKA Yes KOTZEBUE - SOUND - INUIT ALASKA Yes LABRADOR - INUIT NEWFOUNDLAND Yes GREAT - WHALE - INUIT QUEBEC Yes CARIBOU - INUIT NW - TERRITORIES Yes N0ATAK - INUIT ALASKA Yes NUNAMIUT - INUIT ALASKA Yes MACKENZIE - INUIT NW - TERRITORIES Yes SIVOKAMIUT - INUIT ALASKA Yes POINT - HOPE - INUIT ALASKA Yes COPPER - INUIT NW - TERRITORIES Yes UTKUHIKHALINGMIUT NW - TERRITORIES Yes AIVILINGMIUT - INUIT NW - TERRITORIES Yes INGULIK - INUIT NW - TERRITORIES Yes Table 5.1: ( ) 99 5.4 The Variables EnvCalc program utilizes dozens of cultural and environmental variables so the first step was to extrapolate which variables related to the practice of food storage. By testing the relationship of multiple variables, it was determined that 12 variables (pe rcentage of hunting, percentage of gathering, pe rcentage of fishing, mean size of smallest residential group, mean size of largest residential seasonal camps, mean size of periodic regional camps; area in 100km 2 occupied by the group, population density, n umber of residential moves per year, mean total mileage moved throughout the year, Effective temperature, and total yearly rainfall) would be best used to inform on the practice of food storage. These 12 variab les were selected because they best represent ed both cultural and environmental aspects of groups with storage. The first three variables (hunting, gathering, and fishing) express the percentage of hunting, gathering, and fishing undertaken by each group. This variable informs on economic strategy an d environment. The next 3 variables (GRP1, GRP2, and GRP3) directly relate to movement strategies and group size. The area occupied by the group and population density (AREA and DENSITY) variables inform on the possible population pressure and available re sources. The variables number of residential moves per year (NOMOV) and mean total mileage moved throughout the year (DISMOV) can be used to understand movement strategies. The Effective Temperature (ET) variabl e is one created by Binford (2001) which Group Name Location Storage (no/yes) BAFFIN - ISLAND - INUIT NW - TERRITORIES Yes NETSILIK - INUIT NW - TERRITORIES Yes TAREUMIUT - INUIT ALASKA Yes Table 5.1: ( ) 100 meas ures the amount of solar energy available, based on the mean temperatures of the warmest 2001). Total yearly rainfall (CRR) is the final variable and, along with ET can inform on the local environment. 5.5 Statistical Analysis Statistical analysis was necessary to see if these 12 variables share a relationship and if so, does that relationship inform on the practice of foo d storage. In order to do this, two multivari ate analyses were performed, Principle Component Analysis (PCA) and K - Means analysis. Principle Component Analysis measures the distance between variables and creates principle components (Baxter 2015). These pr inciple components explain the variance in th e variables. Performed in an R program (R Core Team 2017), the K - Means analysis measures the distance between variables and pinpoints the number of clusters present among those included in the analysis. The el bow method was used to determine the optimal number of clusters to analyze with K - means clustering. This method runs K - means clustering on the dataset for a range of values ve. The ere is a large jump in the sum of the squared errors, is the best value for k (Baxter 2015) . It was determined that for this dataset, five was the optimal number of clusters. 5.5.1 The Results Though the PCA measures 12 dimensions, based on the 12 varia bles, its visual output, a bivariate plot, exhibits the variance of the first two dimensions. Figure 5.1 is the PCA plot with 101 Principle Component One as the x - axis and principle component two as the y - axis. PCA1 explains 60.53% of the variance and PCA2 exp lains 31.12% of the variance, meaning that the first two PCAs account for 91.75% of the variance. Simply put, an overwhelming majority of the variables have a statistical relationship suggesting they are part of the same population. The red lines (vector s) in Figure 5.1 approximate the correlations between variables (Baxter 2015). It is the angles between the vectors that deter mine relationship. The data indicates that CRR (total rainfall) and AREA are poorly correlated, but the remaining variables are st rongly correlated. Figure 5.2 shows the PCA output once the K - means cluster analysis was performed. The different colors repr esent the five cluster groups, each group is related based on the measurement of their variables. A cluster analysis, like the one performed, shows correlation of groups that are otherwise difficult to interpret. 102 Figure 5.1: PCA output plotted two dime nsionally with PC1 and PC2. Red vector lines indicate relationships. 103 F igure 5.2: PCA plot with K - means cluster analysis. Each color represents one of five clusters. 104 Understanding the relationship between the var iables in each cluster and how those variables relate to the act of food storage is the next step. Further interpretation of the analysis results indicates a distinct grouping related to geographic location. Th is is not entirely surprising since some of th e variables were environmental and similar locations will have similar environments. Table 5.2: K - means clusters organized into their respective groups. Group 1 Group 2 Group 3 Group 4 Group 5 COMANCHE GILYAK AINU_HOKKAIDO VEDDAH CARIBOU - INUIT PLAINS - OJ IBWA NGANASAN BOTOCUDO YARURO - PUME INGULIK - INUIT BLACKFOOT TUBATULABAL EASTERN - POMO YUROK NETSILIK - INUIT OJIBWA - KITCHIBUAN ATSUGEWI POMO - SOUTHERN CHINOOK MICMAC YAVAPAI - YUKI - PROPER QUINAULT WEAGAMON - OJIBWA UTE - TIMANOGAS POMO - NORTHERN MAKAH KASKA U TE - WIMONANTCI KLAMATH BELLA - COOLA TAHLTAN UINTAH - UTE NOOTKA TLINGIT INGALIK ASSINIBOINE CHUGASH NASKAPI RAINY - R. - OJIBWA ALEUT LABRADOR - INUIT NO. - SAULTEAUX INLAND WATERWAY MACKENZIE - INUIT PEKANGEKUM - OJIBWA ROUND - LK. - OJIBWA NIPIGON - OJI BWA MONTAGNAIS HOLIKACHUK NORTON - SOUND - INUIT KOBUK - INUIT KOTZEBUE - SOUND - INUIT GREAT - WHALE - INUIT N0ATAK - INUIT NUNAMIUT - INUIT SIVOKAMIUT - INUIT POINT - HOPE - INUIT COPPER - INUIT UTKUHIKHALINGMIUT AI VILINGMIUT - INUIT BAFFIN - ISLAND - INUIT TAREUMIUT - INUIT 105 However, this also indicates that decisions to store are tightly linked with environmental variables. Another variable that correlates groups is the percentage of fishing. This is most like ly tied to the need for an abundant and predictable seasonal resource. While it is possible to analyze each cluster and pick out the related var iables, the data is best served by contextualization. The next section will detail ethnographic background on se veral of the culture groups in the dataset. 5.6 Variation in Food Storage Unfortunately, the act of food storage is an uncommon practice for hunter - gatherers (although that is what makes the case study of northern lower Michigan interesting), and on th ose occasions when it is utilized, it is often over looked in the ethnographic record. The accounts that do mention food storage do so only in passing reference, and only a handful provide any level detail. Due to this scant evidence it is difficult to get a complete picture of food storage in hunter - gatherer groups, but the ethnographic record can provi de some insight. Though the information is scant, an attempt was made to produce a representative sample of hunter - gatherer food storage around the world. 5.6.1 Gilyak The Gilyak (Nivkh), at the time of their study were a semi - sedentary fishing group wi th a population near 6,000 (Black 1973). They lived along the lower Amur River in the far eastern reaches of modern day Russia. Because they inhabit a large area all along the Amur River, the environment ranges from tundra in the northern regions to mixed forests in the southern reaches of the Amur River (Black 1973). 106 supplemented with gathering and land mammal hunting (Black 1973). Their settlement pattern consisted of seasonal movements between summer and winter residences, with logistical hunting/gathering forays. Winter dwellings were inhabited October through April and revolve d around the hunting of seals and land mammals like fox. Summer villages were inhabited May through September and focused on fishing, specifically for Salmon. Gilyak use the summer months to prepare supplies for the oncoming winter. Once caught, fish are processed and hung on drying racks. Once a supply has been established, they are taken to the store houses at the designated winter sites (Schrenck and Nagler 1854). The bilberry ( Vaccinium vitis ), the fruit of the wild rose ( Rosa cinnemonea ), and Gilyak b erry ( Empertrum nigrum ) are also collected in surplus and stored for winter (Schrenck and Nagler 185 4). In the fall, Gilyak women take logistical collecting trips to gather berries. These berries are not processed, they are loaded into the storehouses wher e they will freeze for winter use (Schrench and Nagler 1854). Additionally, the Gilyak collect sea a lgae ( Laminaria esculenta ) to dry for winter storage. 5.6.2 Tubatulabal From the Great Basin of California, the Tubatulabal relied on a residential movem ent pattern with semi - permanent winter villages. Their economy primarily revolved around the collection of acorns and pi on nuts. Second to this was fishing, leaving the hunting of wild game as the least important economic activity (Voegelin 1903). The seasonal movements of the Tubatulabal began in the summer when family groups moved to higher elevations to hunt, fish, an d g ather nuts. In the late fall they would move west to collect the yearly acorn harvest. Once a supply of acorns had been harvested, they moved back down to their winter hamlets. 107 Tabatulabal primarily stored pi on nuts and acorns. Acorns, once collec ted, were sun - dried then placed in elevated granaries located near the acorn collecting grounds. There is also mention of secondary acorn caching located at their hamlet houses. Acorn caches located at the acorn grounds were reused from one year to the nex t. Pi on nuts were extracted from pi on cones by placing them on a brush fire, and once opened, the nuts then had to be pried from the cones. Similar to acorns, the pi on nuts were first sun - dried, but unlike acorns, they were stored in a stone - lined ca che l ocated near the pi on gathering locations. Both the acorn and pi on caches would be visited throughout the winter to load up on a supply of acorns or pi on nuts for the hamlet (Voegelin 1903). Juniper, gooseberries, and manzanita berries were also store d for winter use. To effectively store berries, they first had to be pounded to remove liquids, then shaped into cakes and sun - dried (Smith 1944). Fish, the second staple food commodity, was also stored for winter use. Multiple ethnographers (Sm ith 1944; V oegelin 1903) claim that food stores were so well planned for winter, that winter itself was a season of relaxation. There is no evidence of failed caches or inadequate stores of food. 5.6.3 Quinault Located along the Quinault River in north west Washin gton, the Quinault were primarily 800. The spring and fall salmon runs were the primary economic activities, and on which hinged all other activiti es. Permane nt villages were located at prime salmon fishing locales, but community movement, in the form of logistical hunting, often took place in the summer, between salmon runs, or during the winter if the salmon catch was disappointing (Olson 1936). Se mi - permanen t camps in the mountains were inhabited by individual families. At these camps, the men would hunt for elk, bear and deer, while the women and children collected berries. 108 Storage for the Quinault, like most other things, revolved around salmon . Successfu l spring and fall salmon runs were absolutely necessary for an adequate supply of stored food for the winter. If it was a poor year for salmon, winter stores were supplemented with land mammals caught during the rainy season. Processed fish were preserved through smoking over a fire; this process took a week to complete. Other fish that were stored included halibut, rock cod, and bass. Berries (black huckleberries, red huckleberries, and blackberries) were also a common foodstuff that was stored. Berries we re dried, and often made into cakes before storing. Stored foodstuffs did not seem to have its own location, the ethnographies only state that once processed, food was stored in the houses. 5.6.4 Pomo The Pomo of California are made up of thr ee geograph ical groups; Eastern Pomo, Southern Pomo, and Northern Pomo, each adapting to their particular environment. The Eastern Pomo resided near Clear Lake and relied on hunting and gathering in the hills to the east, the Northern Pomo traversed betwee n the marin e resource on the Pacific coast and the redwood forests of the north, while the Southern Pomo adjusted to a riverine environment along the hinged on wa ter availab ility. Acorns are the most discussed foodstuff that was stored. Acorn caches were built outside houses and at central locations. Blackberries, elderberry are dried and stored for winter. Venison was the only game meat that was dried and stored . Fish, spe cifically salmon, was oftentimes smoked and stored for winter. 109 5.6.5 Copper Inuit The climate in this region is quite hostile, with temperatures of 40 summer season and plummeting to - These temperatures are often made worse by the constant pummeling of the high speed easterly winds and the westerly winds that often are accompanied by storms. The Copper Inuit practice seasonal movement organized by hunting. During the winter, the time of aggregation, groups gathered in large snow - house settlements along the coast (or even on the ice) to hunt seals through the ice (Collignon 1993). As seals wer e depleted in one area, settlements moved down the coast to new grounds. Settlements were moved n umerous times throughout the winter (Collignon 1993). During the short summer months, groups dispersed and headed inland to hunt caribou and fish in the lakes and rivers. Food caches were strategically placed along the landscape. When dispersing for the hunting territory (Jenness 1922). Seal blubber was a staple food th at was stored in caches. During the season of seal hunting, families saved the majority of the bl ubber extracted from seals and set it aside. Before families disperse inland for the summer, the blubber is cached away in stone - lined caches and not returned to until autumn, the season of scarcity. The Copper Inuit consider it a cultural taboo to take bl ubber inland for the summer months (Stefansson 1914). 5.6.6 Ingalik The Ingalik, though descriptive of an overarching cultural group, generally occupy the lo wer Yukon River in Alaska. The surrounding environment, though harsh throughout the year, consist s of spruce - birch forests rich in flora and fauna. The Ingalik economy revolves around the 110 Yukon River fishing. Winter villages are set away from the river and are usually larger than summer villages. Summers are spent in smaller villages along the river, at key fishing locales. In addition to summer and winter sites, logistical mobility was utilized for caribou hunting. Salmon were caught in surplus with gill n ets and was the primary food source for most of the year. The caches are constructed roughly 5 feet above ground and are 8 feet on each side and take a month to construc t (VanStone 1979). While the men use caches to store their hunting and fishing gear, it is the wo caches are also used, primarily by women, for cold storage of foodstuffs such as be rries, fern roots, rhubarb, wild fowl, essentially any foodstuffs that is not dried. Platform cac hes are used during logistical hunting forays; food is left to be collected on the return trip (VanStone 1979). 5.6.7 Ojibwa - Round Lake The Round Lake Ojib wa resided near Round Lake, Ontario. Falling along the 45 th parallel, Round Lake has a comparable climate to that of northern lower Michigan, making the Round Lake Ojibwa, and their storage practices important to consider. Assuming an environmental role in the decisions behind food storage, comparing similarly structured groups, in similar environment s will be useful. At the time of study, the Round Lake Ojibwa group was estimated to have a population of less than 300 people that inhabited an area of 5,00 0 square miles. The seasonal movement round revolved around the village (Rogers 1962). Summers we re spent at the Village along Round Lake; summer began in June and lasted until late August or early September. Summer is considered a time of relaxation broke n up by occasional fishing. In the fall, before freeze - up, there is an intensification of fishing (Rogers 1962). This fishing ensures a winter supply of food. 111 Winters are spent at established, dispersed hunting camps consisting of 3 to 4 families. Througho ut the winter, logistical hunting trips are utilized where men leave the winter camp and bring th e collection of game back (Rogers 1962). During the late winter, fishing is intensified, requiring a group effort to set the nets under the ice. Groups return to the summer village once the ice begins its breakup. Fish are dried in smoke lodges and hung outside on cache racks. Blueberries, gooseberries, and June berries were collected, dried and stored in birch bark mukuks (Densmore 1974). Food that was prepar ed during the summer and fall was stored in subterranean pits and cached on the way to the winter camps. Ethnographic accounts (Densmore 1974) describe a strategic placement of caches at spring extraction sites, such as the sugar bush where maple sap was h arvested in early spring. One narrative describes a winter with scarce hunting success, and fear of starvation, but never mentioned opening the spring food cache early. Other risk management strategies, including exchange, were implemented before resorting to opening the cache. 5.6.8 Discussion This brief background on hunter - gatherer groups that utilize food storage elucidates several significant aspects that condition why they employ such storage. First, though this is a representative sample of instan ces of global food storage, most instances occur in North America. T his is partially a sampling issue due to the copious amount of data collected by the Bureau of Indian Affairs and the Bureau of American Ethnography in the 19 th and 20 th centuries, but it also speaks to the variables at play in the selection of food storag e as a risk management strategy. 112 Kelly (1995) found that food storage is almost non - existent south of 28 degrees North latitude; therefore , hunter - gat herer groups in those areas relied o n indirect storage strategies or other risk management options. This latitudinal constraint relates to the seasonality of food storage and the seasonal abundance of specific plants and foods (Soffer 1989; Testart 1982 ). T ropical areas lack strong seasonal shifts, and the plants there do not have a period of seasonal abunda nce. Resource choices often revolve around the variables of abundance, reliability, and predictability. Food storage requires a seasonal ly abundan t food resource that provides storable nut ritional benefits. Additionally, the resource must be predictable ; a n abundant mast every five years is not reliable (Soffer 1989 ; Rowley - Co nwy and Zvelebil 1989 ). The structure of the seasonally abundant resource is also integral. There must be an efficie nt means to collect the resource en masse ; whether it be with new te chnology or increased labor (Soffer 1989; Testart 1982 ). Finally, in order for food storage to be a reliable , risk - reducing mechanism, storage tech nology is required (Soffer 1989; Testart 1982 ). The six cultural groups discussed fit this pattern of stora ge use. These groups each reside in regions with strong seasonal shifts and take advantage of seasonal abundance of particular resources. The Gilyak rely on the seasonal abundance of salmo n runs, while the subsistence economy of the Tubatulabal revolves ar ound acorns. These resources are also predictable, making them ideal for storage. Groups like the Ingalik relied on the technology of the gill net to mass capture salmon, other groups, lik e the Inuit created stores of seal blubber through aggregation and g roup labor. There is substantial variation in the methods and practice of food storage amongst these groups. Strategy is evident in the placement of the food storage containers. The Tub atulabal utilize two levels of caching location to maximize their ha rvest; whereas the Ingalik each had 113 individual food caches to ensure their survival. Some groups, like the Inuit, only relied on a single resource, seal blubber, for their storage, while o ther groups, like the Gilyak store multiple foodstuffs from fish to algae. It is important to note that all of these groups, with the possible exception of the Inuit, practice limited residential mobility. In most cases, there are only between two and four residential moves per year. This may seem like an obvious revelatio n, but it reinforces previously stated studies (Rowley - Conwy and Zvelebil 1989; Testart 1983) that seasonal gap storage is only compatible with low mobility. Though food storage is a ris k management strategy for each of these groups, it is tightly intertwined with other strategies. For example, during the winter the Inuit would move down the shoreline when resources became depleted, and if the fall salmon run w as poor for the Quinault, th ey diversified their resources by broadening their game selection for winter hunting. It was also evident while evaluating the ethnographic records for these groups, that the act of storing food was, not only culturally ingrai ned, but linked to multiple other aspects of their daily lives. The two cases mentioned, the Inuit taboo of taking stores of food inland and the refusal of the Ojibwa to open a spring cache early, are just a couple of examples, but they indicate the deep - s eated awareness of the absol ute need for food during specific seasons. If an Inuit group used their blubber reserve during the summer in the inland, then they would starve upon their return trip in the fall. Likewise, for the Ojibwa spring was a time of en vironmental scarcity and a t ime when gardens were planted. There is no time to collect or hunt additional foods because the season is too busy with preparation for the remainder of the year; therefore, if a group consumed their spring cache in the winter, it would lead to a series of events that doomed them for the year. Another Ojibwa example (though not specifically Round Lake Ojibwa) explains that when game meat is cached on scaffolding, away from camp, it is taboo to 114 66). These examples exhibit the cultural role of food storage. Most importantly, through this ethnographic research, I never came across a reference to a failing in the food storage itself. There were a few scenarios where starvation was discussed, but it was due to poor planning in the amount of food being set aside, not due to a cache spoiling or rotting. Food storage was not a risky endeavor in itself, the primary risk was in food supply, which was counteracted through other risk management strategies. 5.7 Understanding Hunter - gatherer Movement Tiered systems of risk management strategies are the most effective way to mitigate risk and uncertainty. The data thus far indicates a correlation between hunter - gatherer groups that utilize food storage and their system of seasonal mo vement. The question arises, what is the movement threshold for the utilization of food storage and where do the hunter - gatherers of the Late Woodland in Michigan fall along that threshold? Table 5.3 shows the seven cultural g roups previously discussed, and their corresponding number of moves per year and total distance moved per year. Additionally, based on Binford estimate the seas onal movement for the people of the Late Woodland. Using the average of three weather stations (Pellston, Cheboygan, and Petoskey) in northern lower Michigan, I was able to create a movement model for the region of study, labeled Inland Waterway. 115 Table 5.3: Ethnographic groups and their respective movement. Data collected from Binford 2001. N AME A REA KM 2 # of Residential Mov es/yr Total Dis. Moved PCA cluster INGALIK 690 4 64 1 GILYAK 76.7 2.03 46.26 2 OJIBWA 142.9 17 364 2 TUBATULABAL 58 9 110 2 EASTERN - POMO 7.4 3 36 3 POMO - SOUTHERN 20.3 1.65 36.27 3 POMO - NORTHERN 31 1.37 7.9 3 QUINAULT 28.3 2 12 4 INUIT 2210 12 385 5 INLAND WATERWAY 398 7 153 3 Considering the information in T able 5.3 , it is clear that the Inuit are the outliers. This group moves several times per ye ar for a total distance of almost 400 miles. However, as previously mentioned, the Inuit have two primary residential moves, the inland during the summer and the coast during the winter. It is t he residential moves along the coast, following seals, that so mewhat skews their total movement in relation to understanding food storage. The practice of food storage for the Inuit is best understood as a caching at fall sites. The other outlier is the s edentary by 1900 resulting in zero residential moves per year; however, the ethnographic data from eHRAF indicates a continued reliance on a seasonal movement strategy well into the 20 th century . Using EnvCalc (Binford and Johnson 2014), number of moves an d total distance moved were estimated, and those estimations are exhibited in Table 5.3. Unfortunately the estimations are quite high (17 moves per year totaling 364 miles) compared to the ethno graphic data which would indicate 2 - 3 residential moves per ye ar totaling no more than 200 miles (Rogers 1962). Because my determination of the Round Lake Ojibwa as food storers was based on the 116 ethnographic data, it seems only appropriate to continue to u se the eHRAF data to reconsider the Ojibwa movement. This dat aset clarification, modifying Ojibwa movement and contextualizing Inuit movement, shifts the output and a clear pattern of movement and food storage emerges; food storage as a primary risk manag ement strategy versus food storage as a secondary risk managem ent strategy. Groups like the Gilyak, Tubatulabal, Quinault, and Pomo (specifically Eastern Pomo) use seasonal gap storage as a primary risk management strategy. These groups collect and process food in their respective season of plenty and rely solely on this stored food during the lean season. Other groups, including the Inuit, Ingalik, and Ojibwa use stored food as an emergency food source during a very narrow and specific time of need. Those who primarily practice seasonal gap storage can limit their m ovement because they have created a resource rich microenvironment; therefore, these groups only move a couple of times throughout the year. Whereas the secondary seasonal gap storers using cac hing to create a short - term resource patch; therefore, these g roups still utilize movement as their primary risk management strategy. I am positing that hunter - gatherers use one of two patterns for seasonal gap storage, reliant storage that relies solely on stored food the entirety of the lean season, and redundant storage, which is used as a supplement during shortages or as an emergency measure. These two seasonal gap storage practices can be analyzed using a Front - back loading model. Reliant seasonal g ap storers front - loaded storage during the season of abundance , allowing them to evenly distribute resources and in turn, created a season of leisure. Redundant seasonal gap storers created just enough stored food to survive, but they have to procure food all year round. 117 Consideration of foodstuffs stored can add an other level of interpretation. Reliant seasonal gap storers seem to rely on foodstuffs that require time consuming processing; whereas redundant seasonal gap storers tend to select foodstuffs that entail minimal upfront processing. For example, the Gilyak primarily store salmon, requiring time intensive processing for storage, while the Inuit store seal fat, which is simply cut up and stored frozen (with the helpful hand of the frigid climate). When considering the people of the northern lower Michigan, th ey seem to straddle these two storage strategies. If they w ere reliant seasonal gap storers then there should be evidence for large amounts of storage at or near winter sites. But, if they practiced redundant seasonal gap storage then the evidence should i ndicate a back - loaded resource strategy. A further analysis of their storage practices, compared to those of the above - mentioned hunter - gatherer groups is required. Situating the hunter - gatherers of northern lower Michigan within their social and climatic environment will contextualize their decisions regarding th eir storage strategies. 5.8 Conclusion This chapter has used the ethnographic record to analyze the storage practices of hunter - gatherers. Combining two datasets, EnvCalc and eHRAF, a series of variables were selected for examination. After performing a K - means and PCA analysis, a series of relationships were plotted for these variables. The results indicated five separate clusters with related data. A further ethnographic analysis was conducte d on the five clusters to determine the relationship of the variables to the act of storage. Data on the specific storage practices of the Gilyak, Tubatulabal, Pomo, Ingalik, Ojibwa, and Inuit were collected. A clear pattern linking mobility, storage strat egy, and foodstuffs stored emerged from the data. An additi onal 118 consideration of the movement practices of these hunter - gatherer groups allowed for a focusing of the pattern. It was proposed that these differing storage strategies be termed reliant and r edundant . Reliant seasonal gap food storage is a primary ri sk management strategy, that is, all other risk avoidance is planned around the success of this strategy. Additionally, reliant storage is a front - loaded mechanism that often leaves a season of lei sure. Redundant seasonal gap storage is a secondary risk ma nagement strategy, or a back - up strategy if things go awry. This storage strategy is constructed in such a way that it minimizes up - front costs by selecting resources that are efficient to process and store, thereby mitigating additional risk. The role o f food storage in the Late Woodland of northern lower Michigan will be analyzed in Chapter six. Contextualizing food storage in the specific environment of Michigan and providing additional cultura l background will aim to elucidate the research question. U nderstanding whether seasonal gap storage was used as a reliant or redundant strategy and laying out archaeological signatures for each strategy, will aid in the interpretation of the current settl ement models for Michigan. 119 6.0 MODELING HUNTER - GATHERER FOOD STORAGE Thus far this dissertation has discussed how food storage is used as a risk management strategy, used actualistic experimental archaeology to evaluate the risk that might accrue to su bterranean food storage, and has laid out a pattern of food storage evident in ethnographically recorded hunter - gatherer groups. Now I turn my attention back to the Michigan case study, discuss the archaeological evidence for food storage, summarize the re search that has been conducted on food stor age in Michigan, and evaluate best fit for the current models explaining settlement and subsistence patterns in northern lower Michigan during the Late Woodland period. 6.1 Food Storage in Michigan The State of Michigan Archaeological Site Files record approximately 90 sites (see Table 6.1 for a full list of sites) across Michigan with reported Late Woodland storage features. This inventory does not include storage features that are included as part of a larger site but not mentioned in the State files. Only a fraction (9.7%) of these 90 storage feature sites have been examined beyond initial identification. Early archaeological excavations in Michigan revealed a series of storage features in the Missaukee Rese rve as early as the 1920s (Greenman 1927). These excavations detailed storage pits with a diameter of 5 feet (1.7 m), extending 3 to 6 feet (1 to 2 meters) in depth, and producing minimal, if any, cultural material (Greenman 1927). In the ensuing 90 years, research on storage features has not gaine d much traction, and only a handful of additional storage feature sites have been excavated. Sites with excavated storage features include Juntunen (McPherron 1967), the Ranger Walker II site (Branstner 1991), the Skegemog Point site (Hambacher 1992), 120 the Point (Howey and Parker 2008), Grapevine Point (Howey and Parker 2008), West Edge (Howey and Frederick 2016), the Gorge (Howey a nd Frederick 2016), 20OT283 (Hambacher et a l. 2015), 20OT3 (Hambacher et al. 2015), Fletcher (Lovis 1985), and Marquette Viaduct (Lovis et al. 1996) (see Appendix C for descriptions of excavated storage features) . There are some additional historic storage features that have been excavated (Dunham and Branstner 1995; Albert and Minc 1987); however, the above list focuses on those features that correlate to the Late Woodland. Site Number Site Name Excavated Yes/No Additional Info 20AA32 No 35 pits 20A A39 N/A 20AA52 No 11 pits 20AA54 Black River Cache N/A 55 pits, 2 miles from a camp near Lake Huron 20AA76 Shattuck N/A 4 - 5 pits, not confirmed 20AA77 M. Cornelius N/A 10 pits, not confirmed 20AA84 Yes 24 pits, O'Shea and Milner 1983 20AA92 S tewart 2 N/A 20AA114 Cornelius N/A O'Shea and Milner 194 survey 20AN45 Yes - 1 historic record and excavation of pits next to a mound 20AR447 No 3 pits, maybe historic 20AR546 No 1 pits, near to 20AR547 which has 5 pits 20AR547 No 5 pits 20C H381 N/A 3 pits, maybe historic 20CH468 No 9 pits 20CH476 No 5 pits 20CH477 No 15 pits 20CN40 Hippler No 15 pits 20CN41 Lathers No 30 pits, Colonial Point 20CN42 O'Neil No 7 pits 20CN45 Young Oak No 7 pits 20CN63 Yes Table 6.1 : Recorded Storage Feature (cache pit) sites in Michigan . 121 Site Number Site Name Excavated Yes/No Additional Info 20CN63 Yes 25 - 30pits 20CN63 Yes 20 pits 20CN63 Yes 60 pits 20CN70 N/A 9 pits, Howey 20CN71 N/A 3 pits, Howey 20CN75 N/A 1pit 20CN76 N/A 1 pit 20CN77 N/A 1 pit 20CN78 N/A 1 pit 20CN79 N/A 1 pit 20CN81 N/A 1 pit 20CN82 N/A 1 pit 20CN83 N/A 2 pits 20CN84 N/A 4 pits 20CN85 N/A 1 pit 20CN86 N/A 4 pits 20CN88 N/A 1 pit 20CN89 N/A 1 pit 20CN90 N/A 2 pits 20CN91 N/A 2 pits 20CN92 N/A 1 pit 20CN93 N/A 1 pit, eas t of 20CN92 20CN94 N/A 1 pit, north of 20CN76 20DE436 N/A 4 pits, 90m north of CR499 20DE438 N/A 9 pits, lithics, near 20DE108, metal kettle also found ( could be maple sugaring) 20DE611 N/A 4 pits 20EM111 Probed 20GT59 Murray N/A 12 pits, Juntunen and Traverse ware, carbonized fruit seeds, nuts 20GT125 Yes - 1 7 pits, found carbonized fruit seeds, nuts, and medicinal plants 20IS200 N/A 8 pits 20LK133 N/A 42 pits 20MA13 No 20MA14 N/A 20MA16 N/A 20MA17 N/A 20MA17 N/A 20MA49 N/A 20MA50 N/A Table 6.1: ( ) 122 Site Number Site Name Excavated Yes/No Additional Info 20MA52 N/A 20MD72 N/A 20MN31 Porter Creek Cache Pits Yes 109 pits, .5 miles from known campsite 20MN45 N/A 7 pits, near 20MN309 20MN163 N/A pi ts near mound feature 20MN164 N/A no info 20MN216 N/A "numerous pits" 20MT62 N/A 20MT114 N/A 20MT115 N/A 20MT125 N/A 20NE89 N/A Holman, burial mound 20NE13 N/A 20NE14 N/A 20NE18 N/A Mounds 20NE72 N/A Holman, ha bitation site 20OA9 N/A Earthworks 20OA25 N/A Mounds 20OA121 N/A 20OT283 Yes 10 excavated 20OT3 Yes 6 excavated 20WX134 N/A Holman 20WX205 N/A 20WX206 N/A On the surface these features consist of a symmetrical circular d epression, 1.0 to 1.5 meters in diameter and anywhere between 10 and 60 cm deep. The center of the depression is often spongy, due to the increased infill of organic material. Once excavated, these features range from 80 cm to 2 meters in depth. Most featu res reveal a simple soil profile, indicating a single episode of use (Dunham 2000); however, recent excavations of features in the Inland Waterway posit episodes of reuse over a 30 - year perio d (Howey and Frederick 2016). Though Table 6.1: ( ) 123 few in number, the excavated storage features can further inform on the use of food storage as a risk management strategy. In addition to archaeological research, food storage has been well documented in the ethnographi c record throughout the Great Lakes region (Dunham 2000). The ear liest account of food storage practices in Michigan was recorded by Charlevoix (1923) in 1721 during his ed with - long stay with the Chippewa in 1763 - 1764, where he describes their seasonal mobility pattern, is another useful record of historic food storage practices. Likewise, th e account of John Tanner (1956), captured as a boy by the Ottawa, describes 30 years of Ojibwa lifeways in the late 1700s to the early 1800s. Frances Densmore, an ethnographer who studied the Ojibway in the early twentieth century, also adds insight into t ethnohi story on Ojibway and wild rice also discusses the practice of storage. Though these sources describe lifeways of hunter - gatherer and mixed economy groups using food storage throughout the Gr eat Lakes region, caution should be exercised before literally ap plying these accounts to the settlement and subsistence systems of the Late Woodland period. The devastating effects of European contact dramatically interrupted the cultural practices of native groups living in the Great Lakes region. Disease led to depop ulation, and a shift in survival strategies created new cultural practices and entrenchment in European economies (Trigger 19 68 ; Cleland 1992). Therefore, it may be unlikely that the settlement and subsistence strategies described in the historic and ethn ographic accounts were a literal continuation of practices in the Late Woodland period. However, a glimpse at food storage practices used in the historic period can provide in sightful information on successful decision - 124 making in a similar environment as th at experienced by earlier counterparts. These accounts inform on the foods that were often stored, storage techniques and technologies, the correlation between storage and sea sonal movement, and how this all operated in the larger system of risk management . Laying out specific aspects of storage features found in northern and north central Michigan is an integral step in understanding the larger role of food storage. 6.1.1 Landscape Distribution of Storage Features Storage features rarely occur in isolat ion, they are often accompanied by tens or even hundreds of other storage features. For example, 20OT283 (Hambacher et al. 2015) has a reported 29 storage features, w hile other sites, like the Gorge (Howey and Parker 2008) have upwards of 220 pits. Large c lusters of storage features are often set away from habitation sites, 198 9) is reportedly some distance away from a known habitation site, similar to the storage f eatures at Pine Point (Howey and Parker 2008) where the closest habitation site is almost a mile away. According to the ethnographic record, small stores of food were kept at summer sites. These stores often consisted of seeds for planting, and food from t he late summer harvest that was cached just before leaving the summer village (Densmore 1979; Tanner 1956). Storage features that are spatially associated with a habi tation site, or that are in close proximity to habitation site, are less common overall an d exhibit fewer storage features than sites that lack habitation. For example, Grapevine Point (Howey and Parker 2008), an inland habitation site location along Dougl as Lake, has around 15 storage features within a few hundred yards of the campsite (Howey and Frederick 2016). There is an apparent correlation of smaller clusters or storage features, or even individual storage features located at habitation sites. It sh ould be noted that there is a possible sampling error in this aspect of the pattern; habit ation sites which happen 125 to include storage features rarely mention the storage features in the site file, it is sites that consist only of storage features that are labeled as such in the State files. Therefore, storage features in conjunction with habita tion sites are most likely underrepresented in the State Files of Michigan. Storage features are also associated with Late Woodland mounds and earthworks (Greenman 19 27; Holman and Krist 2001). A system of tiered storage is also evident in the landscape p attern pertaining to storage. Dunham (2000) argued for a model of storage use that consisted of cached food at primary habitation locales, such as summer villages or prime fishing spots, and task - specific storage locales at extraction sites like maple suga ring camps or berry patches. Howey and Frederick (2016) also argue for a tiered system of storage, includ ing the construction of at least three different types of dis persed storage facilities. In o ne type , groups placed smaller clusters of cache pits on s uitable landforms in the vicinity of their residential sites . In the second type , they placed a larger food store in nondescript location removed from (and unappealin g for) residential occupation but in a position that could be monitored from a residential site, thereby protecting the stores from outsider intrusion/theft. These large number of storage containers were also constructed where micro - environmental features of the landscape may have increased preservation success. Th er e, constituting the third type, they constructed a very large storage facility in a location logistically accessible to multiple family groups. Holman and Krist (2001) also see a pattern in storage use, though similar to Dunham (2000), they argue the pattern is related to the sea sonal distribution of resources and movement. Pits are often located in areas where there is seasonal overlap in resource location, that is, resources that are availa ble in the fall, then again in the same location during spring (Holman and Krist 2001). Th 126 gathering location on the way to and from winter hunting grounds (Holman and Krist 2001). In addition to locales with seasonal overlap, storage is also located at strategic settin gs where flexibility in decision - making (Holman and Krist 2001). In additio n to the consideration of resource distribution for the location of food storage features, there is also evidence for purposeful selection based on microenvironmental features such as soil. Previous research (Howey and Frederick 2016) on the numerous stora ge features that surround Douglas Lake, a northern outlet to the Inland Waterway, shows a less than coincidental placement of storage features in soil series that are classified as excessively drained. Of the 67 clusters of storage features recorded around Douglas Lake (including Pine Point and Grapevine Point), 88% were located in Rubicon sands (excessively drained), which make up less than 30% of the surrounding soil s (Howey and Frederick 2016). Likewise, storage features excavated at 20OT283 also had a s urrounding soil matrix of 86.9% sand (Hambacher et al.2015). Additionally, a Compound Topographic Index (CTI) was performed at Douglas Lake (Howey 2015). The CTI, whi water accumulatio features on well - drained soils was statistically significant when compared to a random d istribution of points across the same landscape (Howey 2015). This implies a conscious sel ection of well drained soils for storage; soil that will prevent water infiltration, thereby preventing rot. 127 6.1.2 Movement and Food Storage The pattern of food storage features can also be analyzed within the context of seasonal movement. Historic s ources detail how f ood storage was practiced in conjunction with seasonal movement strategies. During the early fall, food supplies collected throughout the summe r months were transported, on the way to fall sites, to be stored at spring sites, often maple sugaring locales (Densmore 1979; Holman and Krist 2001). Foodstuffs (blueberries and acorns) were also collected at fall locations and taken along for winter use or cached at known locales (Densmore 1979; Tanner 1956). This stage in the seasonal movement pattern corresponds with the seasonal districts described by Holman and Krist (2001) and the task specific storage locales mentioned by Dunham (2000). Winter was a time of hunting, which required increased logistical mobility and flexibility to successfull y hunt game. This flexibility in where and when to move led to a storage practice that cached food along known routes, rather than at specific habitation locales (Tanner 1956; Quimby 1966; Holman and Krist 2001). During maple sugar season in early spring, the Ojibwe would open stores that were deposited in the fall that contained ; note i ncorporation of non - local introduced domesticates into the local system ). And as soon as the O jibwe group arrived at their summer camp the women and children began to collect berries and dry them for winter use (Densmore 1974; Densmore 1922). Though not d irectly supported by the ethnographic record, it has also been argued (Howey and Parker 2008) that rather than aiding in seasonal movements, storage features located in overlapping seasonal resource locales are evidence of prolonged residential stays. For example, Grapevine Point, an intensively occupied Late Woodland site along Douglas Lake has ev idence for hundreds, potentially thousands of storage features within one mile (1.6 km) of the 128 habitation site (Howey and Frederick 2016). The storage features th at have been excavated reveal fall resources, including acorns and berries, along with spring resources which include terrestrial tubers (the trout lily; Howey and Parker 2008). Additionally, faunal remains (specifically bear) lead to the conclusion that t his area was potentially used from fall through spring (Howey and Parker 2008). This model doe s not argue for sedentism, but for a new system of movement, which when accompanied with storage allowed for more flexible options. 6.1.3 Evidence for Foodstuffs Stored There have only been a handful of Late Woodland storage features that provide evidenc e of the types of food stored (see Howey and Parker 2008; Howey and Frederick 2016; Hambacher et al. 2015). Most of our knowledge of commonly stored foodstuffs co mes from the ethnographic/ethnohistoric record. Densmore (1929) provides detail of the food st rice, sugar, and vegetables were cached in separate pits . Seed potato and seed corn were stored in a was stored in bags and dried fish 40). While this information is useful in understanding storage, many of the listed foods, like potatoes and wild rice, would not have been utilized during the Lat e Woodland. It is my opinion that the selection of foods stored and the processing required fo r said foods was integral in the utilization of food storage as a risk management strategy. Though it would be difficult to argue that additional foods were not s tored, I believe there were one or two resources that tipped the scales in favor of food stora ge. Excavated storage features have evidence for storage of foods including raspberries, cherries, elderberries, blueberries, hazelnuts, and acorns (Howey and Fr ederick 2016; Dunham 2000; Hambacher et al. 2015). Information collected from the ethnographic and ethnohistoric records can inform on the behavioral patterns found in the archaeological record. Many of the 129 ethnographic documents support the archaeological findings in terms of the utilization of food storage within a system of seasonal movement. Ho wever, the historic system of seasonal rounds and food storage revolved around the European fur trade. Groups made economic decisions based on the surplus of furs and a reliance on an ever - encroaching European market (Cleland 1992). Storage in the Late Woo dland could not rely on a post - European contact market system, they utilized storage and seasonal movement as a mechanism for survival. The next section will disc uss how aspects of storage and movement can inform on the use of storage as a risk management system and its operation during the Late Woodland. 6.2 Late Woodland Settlement and Subsistence Changes As Chapter 2 reveals, the Terminal Late Woodland saw a shift in settlement and subsistence in northern lower Michigan. This shift is evidenced by an increase in community size, increasingly restricted territories, demarcation of social boundaries, incorpor ation of cultigens into diet, and a proliferation of food storage. Three differing models, each presenting an explanation for these cultural shifts, were laid out. An evaluation of the current understanding of food storage will lead to a better understandi ng of how these three models might account for food storage. Though none of the models directly account for food storage, except to say that it was used, we can potentially take what we know about food storage and evaluate each model for best fit. The pr evious section discussed the general characteristics of food storage features found in northern lower and central Michigan and used the ethnographic record to explain how these features might have been used in conjunction with seasonal movement. The resear ch presented provides strong evidence for a risk management system that advantageously used seasonal 130 movement and strategically placed storage pits. A flexible system of storage was revealed that indicated a tiered system that created storage at residentia l sites, task - specific sites, and at regional locales where course - of - action decisions could be made. There is evidence, in terms of drainage, for p urposeful planning and placement of the storage features to reduce risk and prevent rot. The admittedly scan t amount of foodstuffs that have been recovered from storage features indicates storage of predominately wild plant resources. There is little evide nce for storage of maize or storage of meat/fish, though the ethnographic/ethnohistoric record does state th at these items were stored throughout the historic period. The pattern of food storage in northern lower Michigan is consistent with a redundant s torage system, a system that mitigates risk, but does not reduce movement or lead to a season of leisure. Th is system did not produce enough stored food to secure them through the winter, instead the stored food supplemented a system that revolved around s easonal movement and variable seasonal resource capture or potential. However, it can be argued that the evi dence from the Inland Waterway may suggest that at times, this redundant system of storage bordered on a reliant system where increased storage offs et the need to move every season. 6.2.1 Fitting Food Storage into Michigan Models The Inland Shore Fisher y model (Cleland 1982) argues that the invention of the gill net allowed for an intensification of deep water fishing. This intensification and incr eased reliance in fishing led to a need to secure productive fishing locales. This demarcation of fishing lo cales resulted in increased territorialism and an intensification of resource extraction within the demarcated areas. The fall timing of the fishing and the surplus of fish resulted in a storable resource right before winter. In this model, the abundance o f storage is linked directly to the increased reliance on gill nets. 131 What I categorize collectively as the Ecosystem and Land - Use Model (Lovis et a l. 2001; Holman and Lovis 2008; Dunham 2014) argues for a flexible resource system in the late Late Woodland . Multiple fall resources like wild rice, acorns, maize, and fall spawning fish were harvested to off - set winter shortfalls. This model also asserts that the shift in late Late Woodland site location can be attributed to the flexible system of resource ext raction. Based on this model, the shift in resource selection showed a proclivity towards storable fall resources. The abundance of storage features in the late Late Woodland is explained as the result of a shift in resources to acorns, wild rice, and maiz e (where possible). changes observed during the Late Woodland were brought about by the introduction of maize. Though not necessarily cultivated in tensively , maize cultivatio n became a sufficiently significant activity that it resulted in economic and social changes. According to this model, the popularity of maize culti vation was played out on the coasts where areas of productive agriculture showed increased territoriality an d created a dichotomy between the inland foragers and coastal farmers for maize cultivation and showed intensification of resource extraction. Food storage in the late Late Wood land, according to this set of models, can be seen as the result of intensification of resource extraction (of not only maize) due to increasing for malization of territories. More food needed to be extracted from ever - decreasing territories and exploitatio n areas. Each of these models posit a justification, though ephemerally, for the proliferation of food storage in the late Late Woodland. However, none of these models consider the role of storage in itself. It is clear there was a shift in the settlemen t and subsistence patterns in the late Late Woodland, and all of the models explain that the shift in resource selection to foods that 132 happen to be storable resulted in the use of storage. They each hypothesize a different trigger for the change in economi c strategy, from fall spawning fish to maize, but the correlation between perfectly storable foodstuffs and the proliferation of storage is generall y an afterthought for each model. The lack of consideration for the prominent role of food storage in the changes evident during the Late Woodland requires a new model. I argue that food storage was the driving factor, the main innovation and motivation, for the changes in settlement and subsistence evident in the late Late Woodlan d. The following section lays out my proposed model: The Seasonal Gap Storage Model. 6.3 Modeling Late Woodland Seasonal Gap Storage in Northern Lower Michigan Chapter Five analyzed several ethnographically recorded hunter - gatherer groups that use seasonal gap storage and determined the generalized pattern of Reliant and Redundant storage strategies. The current available information for food storage in the Termi nal Late Wood land of northern lower Michigan was evaluated under the patterns of food storage and was determined to be a redundant strategy, that at times bordered on reliant. Though it is useful to understand how food storage was utilized, the main resear ch question i s, Why was there a shift in the risk management strategy during the Terminal Late Woodland? The remainder of this chapter addresses this question by presenting a model for the ubiquity of seasonal gap food storage in northern lower Michigan. 6.3.1 Late W oodland Setting Archaeological evidence for the Late Woodland of northern lower Michigan indicates a dramatic population increase. This inference is supported by an increase in the number of Late 133 Woodland sites compared to Middle Woodland sites, and Late Woodland sites th at are larger and more densely occupied compared to those of Middle Woodland sites (McPherron 1967; Howey 2006; Holman and Brashler 1999; Lovis 1973). Ceramic traditions with distinct, homogenous styles are also evident across northern Mic higan and the Upp er Peninsula of Michigan (McPherron 1967; Lovis 1973; Brashler 1981; Hambacher 19 92 ). From Juntunen ware in northern lower Michigan to Traverse ware in central western Michigan, these ceramic traditions have rigid boundaries and indicate i ncreasingly restr icted territories. Culminating in the Terminal Late Woodland, these changes in community size and area demarcation would have been noticeable across Michigan. At the local level, groups would have had additional mouths to feed, with less territory , and p otentially fewer social networks within which to do so. Using ceramic styles to get a rough estimate of territory, we can see that Juntunen ware extends from Houghton Lake up to Naomikong Point on the southern shore of Lake Superior, and e ven north along t he eastern Lake Superior basin. The straight - line distance between the two sites is 177 miles (285 km), though it is unlikely these two territory extremes represent a single group. Rather, seasonal rounds occurred on an east - west axis, bet ween Lake Michiga n/Huron coastal resources and interior resources. Territory size can be estimated using a generalized territory shape and known habitation sites. Hunting and gathering movement patterns commonly take the shape of an ellipse (Ellis 2011), t herefore a series of ellipses were created to estimate the territory size of hunter - gatherers using northern lower Michigan. 134 Figure 6.1 : Estimated hunting territory for a hunting - gathering group inhabiting northern lower Michiga n. The red dot indicates the farthest northern point for Juntunen ware. Googlemaps.com At the extreme, the ellipse that covers a majority of northern lower Michig an would have an area of 7,959 miles ( r x r the territory, and well beyond the range of known hunter - gatherer territories. Though Juntunen ware extends south to Houghton Lake, its presence is ephemeral. The g eographic area that has the strongest representation of Juntunen ware is the Inland Waterway , north through the Straits of Mackinac. More than likely, the Straits acted as a natural boundary and signified a change in territory. To predict a more accurate s cenario, an ellipse that encompasses the Inland Waterway, southward to Charlevoix, was calcu lated (see Figure 6.1). This area is roughly 530 km 2 , which, according to Table 5.3 in Chapter 5, is smaller than the Ingalik, but much larger than the Round Lake O jibwa. 135 Envcalc program it was estimated that a group living in the Inland Waterway would have to move residentially seven times a year, for a distance totaling 153 km to adequately sustain their population. C omparatively, the Ingalik, who inhabit an area of 690 km 2 , would only have to move four time s, totaling 64 km a year, and the Round Lake Ojibwa, with an area of 143 km 2 , would require 17 moves, 364 km, to sustain their respective populations. Similar to th e environment inhabited by the Round Lake Ojibwa, resources in the Inland Waterway are separ ated from season to season, requiring a large territory, and multiple residential moves to access four seasons of resources. Due to these environmental constraints, it would not take a dramatic population increase to put stress on this settlement and subsi stence system. 6.3.2 Seasonal Gap Storage Model An increased population would be seen at the local level through increasingly strained resource districts that failed to adequately produce enough food for the community. Referring back to the Z - score mode l, discussed in Chapter 3, the subsistence strategy of the early Late Woodland became too risky; the variance for return did not produce enough to outweigh the risk (Edwards 2017). A new, or secondary, risk management strategy was required to compensate fo r the reduced resources and increased risk. The p eople of the Inland Waterway could increase their movement to compensate for reduced resources, but due to the increased population across Michigan, this was still a risky strategy. They could also engage in exchange, for which there is local evidence (see 1989; Howey 2012), but they would have to have a reliable resource with which to exchange. The disruption from the increasingly demarcated territories would have resulted in fewer, less reliable so cial networks on which to fall back on (Howey 2012) . Diversification of resources was 136 a third option, and more than likely this happened to a certain degree since we have evidence for an increased utilization of acorns. However, because resources, like aco rns, have a short harvesting season, storage was re quired to fill the seasonal gap. 6.3.3 Delayed Returns and Back - loaded Foodstuff The decisions behind the specific resources collected is another important consideration in understanding the selection o f food storage. Optimal Foraging T heory (OFT) posit s that human behavior is rational, and that decision - making will be based on the most energy efficient food selection (Kelly 1995). Using a Front - Back loadi ng model (Tushingham and Betting er 2013) to deter mine whether a resource requires more energy to for age and collect (front - loaded) or to process (back - loaded) can help explain decisions of food selection by hunter - gatherers. Like Bettinger et al. (1997), Tushingham and Bettinger (2013) find that because acorns are a back - loaded food resource (more energy to process) they are often chosen over other resources, even though acorns provide less total nutrition. Dunham (2014) argues that there is an increase in acorn use at late Late Woodland sites, based on floral assemblages and lipid residues (see also Ko oiman 2018 for supporting evidence from Drummond Island ). Change in cooking practices from stewing to boiling, as evidenced by reduction in temper size and patterning on the cooking vessels, also alludes t o an increase in consumption of starchy foods like acorns or wild rice (Kooiman 2018). Additionally, late Late Woodland sites in the Upper Peninsula of Michigan show a change in location to mixed pine habitats, which have a greater abundance of oak trees ( Dunham 2014). Sites were being selected for their p roximity to acorns. This evidence suggests a late Late Woodland adaptation towards the selection of acorns. 137 Increased use of acorns implies a diversification in diet. Though nutritious, acorns are bitter in taste, and require leaching before consumption, making them a back - loaded resource. The Tubatulabal and Pomo of California both utilized acorns. These ethnographic accounts indicate acorns were easily collected, by women, and immediately stored, with n o additional processing, in above ground caches. Co llection and storage of acorns in a climate like other hand, is humid and cooler (especially in October), and wou ld therefore require additional processing if not j udicious selection before storing the acorns. MSUSStoRE (see Chapter 4) recreated an acorn processing pit, like the one found at the Green Site (see Appendix Two). Features, like the processing pit, indi cate an additional upfront cost of acorn storage. I f not put in a dry and warm environment, acorns will quickly begin to mold, as evidenced by MSUSStoRE. Additionally, if the acorns are not thoroughly dried, they will mold while in the subterranean storage pit. Therefore, parching the acorns is required be fore storage. This extra processing step indicates that acorn harvesting in Michigan came with additional upfront costs that cannot be deferred. Though this extra cost does not negate the efficiency of ac orn storage, it does indicate purposeful decision - m aking. Food storage in decision to invest additional labor into the upfront costs of a back - loaded resource allude s to a stressed resource subsistence system. Season al movement was no longer a reliable risk management system. In contrast to acorns, a back - loaded food, storing fish is a front - loaded task. As discussed in the ethnographic section, the processing requi red before storing fish is time - consuming and 138 labor intensive. Though the archaeological evidence suggests an intensification of deep water fishing during the late Late Woodland (Cleland 1989; Smith 2001), there is little evidence for the storage of fish. Only a single excavated pit at Grapevine Point (How ey and Frederick 2016) produced a calcined bone fragment, which was believed to be sturgeon. Floral evidence from other excavated storage features indicate storage of nuts, and berries. I would argue tha t foodstuffs, such as acorns, were selected over fi sh for storage primarily because of their back - loaded processing. In the case of risk - management, acorns were Collection and parching of acorns was not nearly as labor inte nsive as catching and processing a surplus of fish. 6.3.4 Mass Capture and Selectionist Models Though considered a back - loaded resource, acorns re quire additional processing costs when utilized in humid that explain the intensification of acorns, the Mass Capture Model (Madsen and Schmitt 1998; Ugan 2005) and the Selectionist Mod el (Church and Nass 2002). It should be noted that aco rns were used and stored centuries earlier at sites in the southern half of lower Michigan, like Fletcher (Lovis 1985). The Fletcher site in the Saginaw Valley has storage features, with evidence of a corns, that date to the Middle Woodland period. Unlike n orthern lower Michigan, southern lower Michigan has dense concentrations of oak forests; acorns are abundant. Northern lower Michigan has acorns, but not nearly in the concentration of southern lower Michigan. The density of acorns in the southern half of Michigan makes acorns an obvious resource choice; however, this is not the case for northern 139 lower Michigan. According to OFT, acorns would have been seen as a low - ranking resource in northern Michiga n. One advantage that acorns have as a resource is tha t they can be mass collected. Acorns have a short but intense season where a single oak tree can produce upwards of 10,000 acorns. The high - volume productivity of oaks allows for a mass harvesting. Th e ability to mass collect a resource increases the ranki ng of that resource (Madsen and Schmitt 1998). However, in the setting of the late Late Woodland two changes needed to occur for acorns to be utilized as a resource, a) instability in subsistence prac tices due to increasing population, and b) a technology that accommodated the mass capture of acorns. The technology of storage allowed for acorns to gain fitness and increase their ranking as a resource. According to the Selectionist Model , a low - ranked resource can gain fitness when a technology that accommo dates said resource is created (Church and Nass 2002). When food storage, or the storage container itself is considered a technology then it is possible to link the intensification of acorns with the proliferation of food storage. Subterranean food storage containers were a technology that accommodated the mass capture of acorns resulting in a high - ranked resource. 6.3.5 The Role of Women in Resource Selection The agency of women should also be noted when discussing resource selection. Whelan et al. (2013 ) analyzed the opportunity costs of long distan ce foraging utilized in logistical mobility and found that the costs were high for women with small children. Residential mobility is more cost effective since the foraging distance is shorter and more conveni ent to bring children. The Me - Wuk of Sierra Nev ada were forced to transition to a logistical mobility system, requiring women to somehow offset their dramatically increased opportunity costs (Whelan et al. 140 2013). The decision to forage for the back - loaded resource of acorns was chosen over the front - lo aded resource of pinon nuts (Whelan et al. 2013). Once collected, the opportunity costs of acorns is greatly reduced because the remaining costs can take place at the base camp where childcare is easier (Whela n et al. 2013). The Late Woodland of northern l ower Michigan displays a decrease in residential mobility. The changing settlement system required a change in subsistence. Ethnographic/ethnohistoric evidence suggests that the tasks of gathering and storage was organized and performed by women. Acorns were likely selected ove r other resources for their back - loaded requirements and storability. It is more than likely that the additional tasks for storage were carried out by women and children and perhaps older individuals. Harvesting acorns and berries and drying the surplus ov er a fire before storage were menial tasks that could be accomplished between or in tandem with other tasks. The act of storing food was a planned process, but it was more than likely pla nned around already existing fall activities. Initially, storage was a supplement to seasonal movement rounds, but its proliferation quickly expanded. 6.3.6 Storage Features as Emplaced Facilities An increase in the number of territories , with consequent decreases in size, during the Late Woodland period would have result ed in a weakening of social relations and networks. Hunter - gatherers rely heavily on social networks for information exchange (Whallon 2006). Information regarding, for example, productiv e hunting grounds or bumper crop acorn locations would require dissem ination for year to year survival. Hunter - gatherers c ould not survive on guesswork, they needed social networks, or social mechanisms to exchange necessary information (Whallon 2006; Howe y and Frederick 2016). 141 The socio - economic changes that occurred in the late Late Woodland period weakened the existing mechanism for information exchange, therefore a new mechanism, in the form of mplaced f acilities was created (Howey and Frederick 20 16). Subterranean food storage features can be seen as Frederick 2016:39). In the Late W oodland landscape of northern lower Michigan, subterran ean food storage features acted as, not only a planned resource patch, but also a physical mechanism for decision - making emplaced at a specific point on the landscape. The tiered placement of storage f eatures across the landscape disseminated information t hat allowed for further decision - making regarding movement (Howey and Frederick 2016). 6.4 Summary During the late Late Woodland, effects of increased regional population were experienced at the lo cal level through additional mouths to feed but also fr om resources extracted in a more restricted area , affecting social networks and information exchange. Storage of acorns and berries initially supplemented the diet during the lean seasons, but this soo n turned to reliance on storage during the late winter months. Acorns, a back - loaded resource, required minimal processing and could easily be collected by woman - directed task groups. Once utilized as a primary risk management strategy, storage pits were p laced around the landscape to bolster resource patches along seasonal movement rounds. Cache pits were strategically placed so they would be encountered on the route back from the spring maple sugar site or sites with seasonal overlap, and along resource e dges where vital decision - making regarding seasonal mov ement occurred. Storage features also took on the role of information exchange. As emplaced facilities, 142 subterranean storage features held, and disseminated pertinent information regarding the landscap e. Based on the archaeological and ethnographic infor mation, the hunter - gatherers of the Inland Waterway used a system of redundant Seasonal Gap Storage. In this system they set aside enough resources to bridge the lean season resource lull but could not sustain themselves solely on the food stored. In areas like Douglas Lake, the evidence suggests that at times this storage system co uld be successful enough to use reliant season gap storage; whereby enough surplus food was collected that the hunter - gathe rers could remain sedentary, or nearly sedentary during good years. 143 7.0 CONCLUSION AND FUTURE RESEARCH Previous research and i nterpretation on the seasonal resource extraction patterns and mobility of the Late Woodland of northern lower Michigan, wh ile recognizing the role of storage, and changes during the course of the regional Late Woodland period, did not distinguish well bet ween earlier and later Late Woodland decision - making strategies. The result of the current analysis reveals that prevailing models in fact underestimated both the role and impact of subterranean food storage; a regional form of what is termed Seasonal Gap Storage. This research has approached the topic of storage and emphasized its importance during the later Late Woodland per iod through a tripartite design; first conducting replicative experimental archaeology (MSU SStoRE) on food storage to understand the level of risk (risk prone vs. risk averse) and energy expenditure, second by analyzing hunter - gatherer food storage practi ces from around the world for patterns of storage use, and finally by evaluating the environmental and social dynamics occurring in t he late Late Woodland of northern lower Michigan. The aggregate results of these three facets of the research indicate that incorporation of subterranean food storage during the late Late Woodland period was utilized as a risk management strategy, and that its implementation altered both settlement and subsistence practices. The first of the research dimensions explored was focused on replicative actualistic experimentation designed to evaluate issues of energy expenditure, efficiency, timing, and relativ e risk in the storage of various common and storable foods. MSU SStoRE provided evidence that subterranean food storage can be a risk averse risk management strategy. Though there is an inherent learning curve to properly reduce said risk, once mastered, t he technology of food storage is reliable and efficient. MSU SStoRE, the first well - controlled experiment of its kind, prov ed that wild foodstuffs could be processed and then effectively and safely stored up to 144 six months. Though other experiments (Arzigia n et al. 2007; Reynolds 1974; Cunningham 2005; Grooms 1996) have shown the utility of food storage when storing maize and o ther domesticates, MSU SStoRE assessed the potential risk of wild foodstuffs (non - domesticates) storage. When placed in the context of assessing risks associated with storage, MSU SStoRE provided evidence that filled the information gaps left by the ethn ographic/ethnohistoric documents. By determining that subterranean storage of wild foods is risk averse, that risk measurement could then be input into a Z - Score model to compare with other subsistence and risk - management strategies practiced by hunter - gat herers globally. With the experimental results in hand, a second major facet of the research was to situate the late Late Woodland of northern lower Michigan into the larger context of hunter - gatherer societies that employed food storage. An ethnographic ally/ethnohistorically derived dataset of hunter - gatherer groups that use food storage as a risk - management strategy was then compile d and evaluated. Using eHRAF and EnvCalc (Johnson and Binford 2014), 62 hunter - gatherer groups were determined to have prac ticed seasonal gap storage as a risk management strategy, and of those 62, seven groups were evaluated in greater detail. Comparison, detailed evaluation, and statistical analysis of these seven hunter - gatherer groups revealed a pattern in food storage pra ctices; reliant seasonal gap storers, those that consume only food stored during the lean season, and redundant seasonal gap storers, those who use food storage as a supplement during shortages or as an emergency measure. Data for the hunter - gatherers of n orthern lower Michigan during the late Late Woodland period was then statistically analyzed to determine if it fit one or the other o f the two patterns. The storage pattern for the late Late Woodland period of northern lower Michigan revealed a risk mana gement strategy that used redundant storage for most years but had the 145 ability to convert to reliant storage when necessary. The Seas onal Gap Storage Model was laid out to explain why storage became the selected risk management strategy during the late Lat e Woodland period. The Seasonal Gap Storage Model argues that an increase in population was felt at the local level and began to str ess the existing subsistence strategy (McPherron 1967; Howey 2006; Holman and Brashler 1999; Lovis 1973). Environmentally, the Medieval Climatic Optimum created instability in lake levels (Lovis et al. 2012), i.e. lake elevations in the Michigan - Huron lake basin fell below those of MCO levels, resulting in increased unpredictability in near coastal resources. This combination of factors, increased population and resource instability, created a subsistence strategy that no longer mitigated risk; the variance for return did not produce enough to outweigh the risk (Edwards 2017). Therefore, a new risk management strategy was requi red, a redundant seasonal gap storage strategy. MSU SStoRE revealed the timing for subterranean storage correlates closely with bot h the fall fish spawn, and the spring run of maple sap. This storage period, November to March (in northern Michigan), not only corresponds with the timing for extraction of ethnographically documented resources (wild berries, acorns, fish, and maple syrup ), it is also climatically (temperature, precipitation) the best time for storage. According to MSU SStoRE, storing before October, or after mid - March increases the risk of spoilage (rot) due to rainy weather patterns. During the late Late Woodland, after the implementation of the gill net, both the fall spawn of large deep - water fish, and the spring maple sap run are times of natural community aggregation; multiple bands coming together to harvest a resource that allowed capture of a surplus beyond immedia te needs. With a larger workforce at hand, a side task of acorn collection could have been accomplished with only minor lab or costs. Decision - making, based on labor costs for 146 collecting and foraging, is best understood as a gendered division of labor; hunt er - gatherer (Egan 1993:27). Therefore, it was women that made the strategic decision to set aside additional food and create redundant storage as a risk management system. According to the data captured by MSU SStoRE, acorns and wild berries require a certain level of processing before successful storage. Though still considered a back - loaded resource, the climate of Michigan required a period of d rying to ensure a successful storage episode, by reducing the likelihood of sprouting, mold, and insect infestation, and th erefore reducing the risk in acorn storage. The fact that subterranean storage in northern lower Michigan is met with more risk tha n storage in other climates, such as California with its less variable conditions, indicates an extreme need for redundant seasonal gap storage. During the late Late Woodland period oak trees were not prevalent in northern lower Michigan as they were in ot her parts of Michigan. Acorns were not a ubiquitous resource and they required both pre - processing and post - processing, and yet evidence indicates that acorns were the main resource selected for storage. I argue that acorns gained fitness because of their storability, which was established as early as the Archaic period in areas just south of the Inland Waterway. Other resourc es, like fish, required both additional work to harvest and process before storage, and therefore more wasted labor costs (energy sin k) if the storage failed. Acorn collection and processing could have been easily folded into (most likely by women) an exis ting subsistence system where the risk of storage failure did not outweigh the return. Egan (1993) explains that for a given year, in a hunter - gatherer band inhabiting a temperate environment, women would have been responsible for one 147 third (33.3%) of the dietary decisions. However, this percentage would have been increased when redundant seasonal gap storage was utilized. The spatial organization of the system was also of utmost importance. Storage facilities, subterranean cache pits, were strategically positioned on the landscape in coordination with seasonal movement. A flexible system of storage was revealed that indicated a tiered system that created storage at residential sites, task - specific sites, and at regional locales where course - of - action deci sions could be made. Purposeful planning and placement of the storage features, particularly in combination with the presence and selection of well - drained soils, reduced risk and prevented spoilage. The more residentially mobile and immediate return - orie nted subsistence system of the early Late Woodland was supplanted by a more flexibly mobile, back - loaded resource subsistence system in the late Late Woodland. I am arguing that food storage was a primary factor in this economic shift. Food storage used as a risk management strategy in a pattern of redundant seasonal gap storage, where food is stored as a supplement and/or emerge ncy contingency, led to an increase selection and utilization of fall subsistence resources. Fall foods were selected for their st orability; the technology of food storage was the trigger for subsistence changes. 7.1 Research Results With the data generated from this study incorporated into the above summary, this new model remains to answer the initial questions that guided the research. Research Question #1: How much risk is involved in the act of food storage? Expectations: Comparative analysis of global food storage practices, along with data from experimental research (see Chapter Four), will be used to evaluate risk. I exp ect to find that when 148 food storage is used as a primary risk buffering strategy the storage pits are risk averse. It is no t until the technology of subterranean food storage is reliable, or presents minimal risk, that food storage is utilized. I also expe ct to find that in hunter - gatherer communities where food storage is known to be less reliable, it is not used as a primary risk management strategy. Outcomes: Though this question is broad, it can be evaluated based on the data collected through eHRAF a nd EnvCalc. In nearly all of the ethnographic and ethnohistoric documents that discuss the use of food storage by hunter - ga therer groups, there is never a mention of risk. Ethnographic and ethnohistorical documents revealed that when food storage is used a s a primary risk management strategy, there is minimal concern for risk, or for the outcome of a failed storage episode. Th is broad observation leads to the conclusion that once the technology and technique of food storage is developed or acquired, it is a reliable risk management strategy. Additionally, MSU SStoRE revealed that subterranean storage in Michigan can be a reliab le, effective, and efficient method once the proper technique is acquired. Research Question #2: What was the range of variation of hunter - gatherer and mixed horticulturalists food storage technologies as ethnographically documented around the world? And how does this variation determine whether food storage is a primary risk - management strategy? Expectations : I expect to find that h unter - gatherers storage practices follow similar patterns based on differing causal or p rimary decision variables, i.e. groups in similar climates will utilize food storage in a similar manner. Outcomes: Although the i ssues at play in the use of subterranean food storage around the world vary, there exists a standard set of factors that in fluence the use and decision - making behind food storage. Some of the factors are attributable to latitude and ipso facto - climate, 149 s easonal resource availability, and resources that are collectable en masse . Deciding which type of storage to utilize is heavily dependent on the types of foodstuff available, climate, and existing technologies i cal storage is almost non - existent south of 28 degrees North latitude; therefore , hunter - gat herer groups in those areas relied on indirect storage strategies or other risk management options. This latitudinal constraint relates to the seasonality of food s torage and the seasonal abundance of specific plants and foods (Soffer 1989; Testart 1982 ). Temperate areas lack strong or abrupt seasonal shifts, and the plants there do not have a period of seasonal abundance. Food storage requires a seasonal ly abundan t food resource that provides storable nutritional benefits. Additionally, the resource must be predictable ; an abundant mast every five years is not reliable (Soffer 1989 ; Rowley - Co nwy and Zvelebil 1989 ). The structure of the seasonally abundant resource s usceptible to storage is also integral to its success . There must be an efficient means to collect the resource en masse ; whether it be with new technology or increased labor (Soffer 1989; Testart 1982 ). Finally, for food storage to be a reliable, risk - re d ucing mechanism, a well established, tested and reliable storage technology is required (Soffer 1989; Testart 1982 ). If these factors are present, then food storage is a reliable mechanism for risk management. Taking these factors into account, a pattern in the practice of food storage emerged. Hunter - gatherers tend to use food storage in one of two ways, reliant or redundant. Reliant seasonal gap storers collect and store enough food during the season of plenty so during the season of scarcity they rely s olely on the collected food; so much so that the lean season is also the leisure season. Whereas the redundant seasonal gap storers collect and store jus t enough food so as to supplement times of scarcity or to use in times of emergency. It is only the rel iant storers that use food storage as a primary risk - management strategy. 150 Research Question #3: Was subterranean food storage selected as the primary ri sk Expectations: I expect to fi nd that by the late Late Woodland period, the technology and methodology for subterranean food storage had been enhanced, leading to reliable food preser vation, and therefore was selected over other risk management strategies. Outcomes: Subterranean food storage in northern lower Michigan had the potential to be a primary risk management strategy during the late Late Woodland; however, evidence suggests it was likely used in conjunction with multiple risk management strategies, such as seasonal movement a nd exchange. It is hard to argue that subterranean food storage was the primary risk management strategy, mainly because the pattern of use in northern l ower Michigan alludes to a redundant seasonal gap pattern. Though, some evidence, like that from Dougla s Lake (Howey and Parker 2008), indicates an occasional use of reliant seasonal gap storage. Redundant seasonal gap storage emerged as a risk management strategy in the late Late Woodland period because the existing subsistence system became risk prone du e to multiple factors, including changes in costal environments, as well as alteration in temperature, and possibly too the effects of demographics. Beca use of the enhanced flexibility that comes with food storage, the decision - making by female collectors, and the ability of storage to be incorporated into seasonal movement, the practice of storage gained fitness. 7.2 Summary of Research Food storage is a behavioral mechanism used to even out the food supply throughout the year. The abundance of food during the seasons of plenty (i.e. summer and fall in Michigan) are collected and set aside for seasons of scarcity. Food storage, when analyzed as a ris k 151 management strategy, is one mechanism used to reduce risk. Other mechanisms include movement (or mobi lity), exchange, and diversification. These strategies are used in a hierarchical system, where one is considered less risky than the next; however, risk management strategies are rarely used in isolation. Risk management strategies are often intertwined a nd used in conjunction with one another. This is evidenced in northern lower Michigan by the correlation of subterranean storage and seasonal movement ro unds; storage was staged at multiple locations to ensure success throughout the seasonal stops. Underst anding why food storage became a popular risk management strategy during the late Late Woodland was the focus of this research. Multiple lines of eviden ce were used to evaluate the proliferation of subterranean storage features in northern lower Michigan during the late Late Woodland period (AD 1000 - 1600). Though the practice of food storage can be found in other regions of Michigan during earlier periods (see Lovis 1985), it is the foodstuff being stored (specifically acorns) and the degree to which it is being stored that makes for an interesting case in northern lower Michigan. MSU SStoRE evaluated the efficiency and risk for food storage of wild foodst uffs in northern lower Michigan and determined the practice to be risk averse. This series of experimen ts also allowed for a calculation of labor costs required to build a storage feature, collect and process foodstuffs, and to cache and un - cache stored fo od. It was determined that this series of events was primarily a back - loaded process. As such, it posed less risk than a front - loaded process due to the minimal upfront labor costs of back - loaded foodstuffs, like acorns. While MSU SStoRE was conducted to evaluate the potential for risk in subterranean food storage, further evaluation was necessary to under standing how the behavior behind food storage manifested in hunter - gatherer communities. Ethnographic/ethnohistoric research was conducted using eHRAF an d EnvCalc (Johnson and Binford 2014) to determine patterning in the practice of 152 food storage. Food stor age variables were selected and analyzed in R ( R Core Team 2017 ) resulting in a behavioral dichotomy, reliant and redundant seasonal gap storage. Once the risk of food storage and the behavior behind the use of food storage was understood, Optimal Foragin g Theory (OFT) was then used to decipher the why; Why would the hunter - gatherers of northern lower Michigan select food storage? The Front - Back Loading model (Tushingham and Bettinger 2013), the Mass - capture model (Madsen and Smith 1998; Ugan 2005), and t he Selectionist model (Church and Nass 2002) were all applied to understand how food storage gained such fitness during the late Late Woodland period. Archaeological evidence for many of the storage features that have been excavated indicates the storage o f acorns. Though evidence for storage other foodstuffs, such as raspberries, blackberries, and possibly fish has been found, acorns ar e the most consistently cached food. Acorns are easy to collect, can be collected in large quantities, can be easily store d (once dried), and provide a nutritional advantage during the winter. It is argued that though acorns were not as prevalent in northe rn lower Michigan as they were in other parts of Michigan, nor were they as expedient to process in the humid climate of M ichigan, acorns were a back - loaded resource that gained fitness when food storage became a dominant practice. Subterranean food storag e was the necessary technology, the key factor responsible for propelling the fitness and desirability of certain foodstuf fs, like acorns. Acorns are one of the more obvious choices for a storable resource, but according to the Seasonal Gap Storage model, the success in the storage of back - loaded resources leads to an increased reliance on redundant food storage. This increa sed reliance, in turn, leads to an additional investment in subterranean food storage; evidenced by storage of front - loaded resources such as raspberries, blackberries and likely fish. More time and labor costs were being 153 invested in other mass - capture res ources because food storage became a reliable, risk - averse risk management strategy. 7.3 Future Research It is evident that there are still several avenues that can be pursued to allow better understanding of the complexities of subterranean food stor age in northern lower Michigan. This research was the first of its kind to consider food storage as a prime mover during the late Late Woodland period. Previous research on storage and storage features has delved into the mechanics and utilization of food storage; this research is the first to consider the why; Why was there a proliferation of storage during this time? Though my model, t he Seasonal Gap Storage Model, elucidates this question, there is still much to be learned. As Appendix B explains, not e very surface depression is a storage feature and not every storage feature is evidenced by a surface depression. Further research need s to be done on ground - truthing surface depressions and excavating storage features. There are several assumptions that re volve around storage features that need to be confirmed before the Seasonal Gap Storage Model can be further tested. Appendix B propos es a new method for ground - truthing surface depressions, that will ideally lead to a more accurate understanding of the qu antity and spatial distribution of storage features that exist across northern lower Michigan. In addition to simple ground - truthing of storage features, a greater number of features need to be systematically excavated and analyzed to strengthen the stor age features dataset. Currently, only a minuscule percentage of storage features have been excavated, and even fewer have been systema tically sampled for content, such as macro and microbotanicals. Much of the Seasonal Gap Storage Model was built off the d ata collected from only a handful of storage 154 features; additional data collection through systematic excavation could better test the model, potentially strengthen the model, or potentially lead the model in a new direction. Calculating the number of cac he pits used in a given season is also a necessary avenue of inquiry to evaluate their use and prominence. Though hundreds of storage features have been recorded archaeologically, we cannot assume that a) storage features were a one - time use phenomenon, an d b) that all forms of storage were subterranean. Beyond the scope of this dissertation is the consideration of other storage structur es, i.e. scaffolding. This is a well - documented storage practice; however, it is difficult to uncover archaeologically. It would be a worthwhile endeavor to perform additional comparisons of hunter - gatherer groups that utilize food storage and determine pr oclivity for scaffolding versus subterranean storage. Another avenue for future research would be additional replicativ e storage experiments. Experiments that test the reuse of storage features would aid in the above question of total number of storage features used in a given year. If it was commonplace to reuse an existing cache pit then we need to further understand the archaeological signs of reuse and what that alludes to in the larger settlement and subsistence system. MSU SStoRE was the first of its kind to evaluate subterranean storage of wild foodstuffs, but further controlled experimental testing should be done t o understand the potential for risk in the storage of other foodstuff like fish or wild rice. These are both prominent fall resources that more than likely would have been stored. Several ethnographies and ethnohistories make mention of the storage of fish and wild rice, but was the storage of fish and rice as risk averse as the storage of acorns and berries? Though acorns are easily co llected well into October and November, wild berries are generally ripened by August, leaving a gap of time between berry season and the fall fish spawn. 155 Therefore, questions remain as to the timing of storage. It is likely that there were several storage episodes throughout the year to accommodate different resources. To accurately analyze the inherent risk of storage of all foodstuffs, more experiments should be conducted. 156 APPENDICES 157 A PPENDIX A : The Green Site 158 The Green Site In November 2015 an archaeological feature was discovered and excavated at the Green site near Perry, Shiawassee Co unty, Michigan . The Green site is unusual in that it is not a larger or even smaller and more temporary occupation locale, but comprises a single, burned, circular feature, 130cm in diameter, consisting of hundreds of carbonized acorns and fragments of car bonized wood. I t appears to be the outcome of a single behavioral event. The feature was discovered by the property owner, Mr. Green, while excavating a foundation with a back hoe; therefore, while the depth of the original feature is unknown, based on th e surrounding e levations, soil profiles and stratigraphy, it was most likely a pit dug to a depth of one meter (Frederick and Albert 2016). Importantly, there were no signs of the pit visible above the depth at which it was first found and reported. Moreo ver, local surv ey of the surrounding property on two occasions revealed no additional evidence of occupation that might be associated with the feature. Ecologically, the Green site is located at the juncture of two major land cover types; mixed oak savann a, and oak - hick ory forest, with minor associations of various wetland communities (Comer and Albert 1997; https://mnfi.anr.msu.edu/data/veg1800/shiawassee.pdf ). The association with prox imal oak communities is not surprising given minimal transport costs. The feature was Excavated in quadrants. Based on the limited internal stratigraphy it was discovered that the charred acorns rested on top of a layer of carbonized wood, most likely the fuel bed for the fire. The individual loads of acorns placed in the pit could also be readily discerned. The underlying sands upon which the fuel bed and acorns were in contact were occasionally oxidized to a pink color and fused. During the archaeologica l excavat ion, the majority of the carbonized acorns were collected, along with almost total samples of the 159 carbonized wood. No additional cultural material other than a single firecracked rock was discovered within the pit. Further analysis was conducted on the ac orn and wood samples. Acorns and pieces of charcoal were separated from the soil matrix first by judicious screening through ¼ inch /5 mm mesh screen, and then by hand picking . While the acorn pieces were being separated from the soil matrix by han d, they w ere also being sorted into four categories mostly whole acorn (shell and nut meat mostly intact), partial acorn (half or less than half of the shell and nut meat intact), nut meat without shell, and shell pieces. The acorns in each of the four c ategories were weighed using a n Ohaus TM Scout Pro XP - 401 digital balance . To determine the species of acorn recover ed from the Green site feature , we analyze d the height:breadth ratio of the intact acorn s . Red oak acorns will generally have a height:bread th ratio close to one (1) , while white oak acorns will have a ratio higher than one , because their height is greater than their breadth, i.e. they are longer rather than round . Using this criterion , the height and breadth of all the acorns that were sorted into the using Mitutoyo TM CD - digital calipers. The ratio was then calculated. Based on the ratios it was determined that the carbonized acorns were white oak. The carbonized wood was also identified as white oak by Dr. Fr ank Telew ski ( Quercus alba ; Frederick and Albert 2016). The 14 C age of an intact acorn shell and meat sample resulted in a radiocarbon age of 1240 ± 70 B.P. 13 C = - 24.9 , cal AD687 - cal AD783: Stuiver and Reimer etc.). Hypothesized Feature Function After excavation, comparative literature search and further lab analysis, it was hypothesi zed that the feature at the Green site was an acorn proce ssing pit. These features were 160 used to dry acorns to prolong their shelf - life during storage. Drying acorns is a necessary first step before caching. Features like the one discovered at the Green sit e that inform on the technology for this processing behav ior have only been identified at a handful of other Midwestern sites (Bruhy et al. 1999 at the Butternut Lake Inlet site; Connor 2004 at the Walnut an d therefore are still poorly understood. As part of an on going series of experiments conducted to understand the technology of food storage, the MSU Subterranean Storage Research Experiment (or MSU SStoRE), an actualistic experiment was implemented to repl icate the technology of acorn processing. 161 A PPENDIX B : The Ferrell Ridge Site 162 The Ferrell Ridge Site The Ferrell Ridge Site (20CN111) is located on a small spit of land to the west of Mullet Lake, along M - 27 ( Indian River, Mullet T wp . T 36N , R 2W , Section 31 , SW 1/4 of SW 1/4 of SE 1/4, Cheboygan County , MI , coordinates N 45.467391 W 84.602228 ). Archaeological survey and test excavations were conducted between May 27 th - June 10 th under DNR Permit AE 2016 - 01 by a Michigan State University field crew under the direction of Kathryn Frederick. Ferrell Ridge was originally identified in the Fall of 2014 as potentially containing hundreds of surface depressions, believed to be remnant s of prehistoric subterranean storage features. This hypothesized prese nce and abundance of subterranean cache pits was based on characteristics of surface depressions on the forest floor, particularly a size of about one meter in diameter, a discernible d epression in the center, and a spongy or uncompacted center assumed to result from refilling. The goal of the field season was to a) ground truth as many of these possible features as feasible, b) collect samples from any confirmed subterranean storage fea tures, and c) if successful in identifying a caching site to create a d etailed map of the site using a Total Station. The ground - truthing of a field of potential subterranean storage features was the first known attempt at this type of confirmatory researc h. The Ferrell Ridge site (Figure 1) spans an area situated on a spit o f land approximately ¼ mile/402m long (west - east) and 200ft/61m wide (north - south). Located in a secondary growth mixed forest consisting of both coniferous (pines) and deciduous (oak) trees ( http://geo.msu.edu/extra/geogmich/vegetation.html ), it has a nat urally undulating surface, with many surface hummocks and depressions. The site is surrounded by lowland swamp, and a pond to the south. The soil on the spit of land consists of Rubicon sands, leading into Au Gres sands. 163 The soil to the north of the site i s Roscommon muck, and Histosols to the south ( https://websoilsurvey.nrcs.usda.gov/app/WebSoilSurvey.aspx ). A ppendix Figure 1 : USGS Indian River Quadrangle; Red circle indicates lo cation of Ferrell Ridge Site. 164 Potential Storage Pit Investigations During the two - week field season ten of the surface depressions were formally cross - sectioned, following procedures established by Howey (Howey and Frederick 2016). In an effort to survey and test a greater number of surface depressions a procedure using a soil auger, suggested by Dr. Randy Schaetzl, was also employed in tandem with the formal excavations. This method required an auger with a 4 inch/10 cm bucket to probe the center of the s urface depressions. The purpose was to identify any anomalies (burie d horizons, or organics) in the soil strata. This methodology cannot confirm the presence of a storage feature, but it can provide evidence for soil anomalies. All formal excavations were fenced during excavation, and all excavations were backfilled and re stored upon completion. Of the ten formal cross - sections, only one surface depression was confirmed as a likely prehistoric subterranean storage feature (Appendix 1). An additional 20 surf ace depressions were tested using an auger, and only one of the 20 t ests had a positive result (Appendix 2). Due to the negative data, initial plans for mapping the landscape of the storage features was reevaluated. With a total station location of the si ngle confirmed storage feature, the negative surface depressions, an d the auger tests, were mapped. Total Station Datum: N 45.468459 W 84.603450 Total Station back - site: N 45.468460 W 84.603443 While the resulting field data was wrought with negative res ults, it still proved very insightful for the overall research. Previous research (Howey et al. 2016) indicated that these subterranean storage features are densely packed around certain areas of the Inland Waterway, but the results of this excavation impl y that not every cluster of surface depressions is cultural . Although there is a known habitation site approximately one mile from the Ferrell Ridge site, 165 this small spit of land was apparently no t adequate for the storage needs of the prehistoric peoples. The next step will be to compare the micro - environment of Ferrell Ridge to sites with known subterranean features in order to better understand landscape selection. The results of this brief fiel d season caution that surface depressions need to be more ri gorously ground - truthed before they can be deemed archaeological features. Currently, I am working with Dr. Schaetzl to create protocols for the sampling procedures and ground truthing of subterra nean storage features. Description of Tested Surface Depres sion Features Pit Feature 1: N 45.468134 W 84.603309 This surface depression was randomly selected for excavation. The surface depression measured 80cm in diameter, and was 15cm deep at its most depressed point. The feature was excavated by laying a nort h - south transect and excavating the west side of the transect; exposing the east profile of the feature. All soil was screened through ¼ inch/6mm archaeological screen. Feature 1 was only excavate d to a depth of 40cm below surface before it was determined to be the result of natural formations. There was no evidence of a storage feature, more likely it was a burned stump. No artifacts were discovered. Pit Feature 2: N 45.468229 W 84.603297 This su rface depression was randomly selected for excavation. The s urface depression measured 85cm in diameter, and was 12cm deep at its most depressed point. The feature was excavated by laying a north - south transect and excavating the west side of the transect; exposing the east profile of the feature. All soil was scre ened through ¼ inch/6mm archaeological screen. A dense gravel layer was encountered 85cmbd. Feature 2 was excavated to a depth of 95cm bd. The profile consisted of a slight pit depression, but it bottomed out at 51cmbd, too shallow for a 166 storage feature. S oil profile did not indicate any cultural activity, no buried horizons or organics indicating disturbance beyond the surface depression. No artifacts were discovered. Pit Feature 3: N 45.467205 W 84.602345 This surface depression was randomly selected for excavation. The surface depression measured 90cm in diameter, and was 20cm deep at its most depressed point. The feature was excavated by laying a north - south transect and excavating t he west side of the transect; exposing the east profile of the feature. All soil was screened through ¼ inch/6mm archaeological screen. Feature 3 was excavated to a depth of 150cm bd. Feature was full of highly (almost vitrified) wood and dense roots. Dr. Randy Schaetzl (soil scientist) concluded that Feature 3 was most likel y a tree fall pit where the stump then burned and fell back into the pit. This was not a cultural feature, it was the result of a natural phenomenon. No artifacts were discovered. Pit Feature 4: N 45.467166 W 84.602324 This surface depression measured 91c m in diameter, and was 25cm deep at its most depressed point. This depression was selected because it was one of the deepest depressions, though it was not symmetrical. The feature was excavated by laying a north - south transect and excavating the west side of the transect; exposing the east profile of the feature. All soil was screened through ¼ inch/6mm archaeological screen. This depression was surrounded by large White Pines, leaving the sandy soil extremely dry. At 134cm bd there was a compact stratum o f reddish (7.5yr 4/4) sand followed by a fine gravel lens. Final depth of excavation was 147cm bd. Soil profile did not indicate any cultural activity, no buried horizons or organics in dicating disturbance beyond the surface depression. No artifacts were d iscovered. Pit Feature 5: N 45.466347 W 84.600107 167 This surface depression measured 100cm in diameter, and was 14cm deep at its most depressed point. This depression was selected because it was one of the widest and most symmetrical depressions. The feature was excavated by laying a north - south transect and excavating the east (large root on the west side) side of the transect; exposing the west profile of the feature. All soil was screen ed through ¼ inch/6mm archaeological screen. This depression was surrou nded by large White Pines, leaving the sandy soil extremely dry. At 110cm bd the north side of the excavation was a compact strata of reddish (7.5yr 4/4) while the south side was loose sand (10yr 5/6) Final depth of excavation was 124cm bd. Dr. Schaetzl co ncluded that this feature was the natural result of an ancient tree fall, which he estimated being upwards of 2000 years old. No artifacts were discovered. Pit Feature 6: N 45.466415 W 84.600028 This surface depression was randomly selected for excavation. The surface depression measured 80cm in diameter, and was 20cm deep at its most depressed point. The feature was excavated by laying a north - south transect and excavating the west side of the transect; exposing the east profile of the feature. All soil wa s screened through ¼ inch/6mm archaeological screen. Dark, organic feature, filled with roots, very distinct from the surrounding soil matrix, but too narrow to be a storage feature. Fe ature 6 was excavated to a depth of 150cm bd. Dr. Schaetzl concluded th at Feature 6 was most likely a tree fall pit. This was not a cultural feature, it was the result of a natural phenomenon. No artifacts were discovered. Pit Feature 7: N 45.466382 W 84 .600053 This surface depression was randomly selected for excavation. T he surface depression measured 82cm in diameter, and was 10cm deep at its most depressed point. The feature was excavated by laying a north - south transect and excavating the west side o f the transect; exposing the east 168 profile of the feature. All soil was screened through ¼ inch/6mm archaeological screen. Dark, organic feature, filled with roots, very distinct from the surrounding soil matrix, but too shallow to be a storage feature. Fea ture 7 was excavated to a depth of 160cm bd. Dr. Schaetzl concluded tha t Feature 7 was most likely a tree fall pit. This was not a cultural feature, it was the result of a natural phenomenon. No artifacts were discovered. Pit Feature 8: N 45.468470 W 84. 603491 This depression was excavated because of its proximity to a prop osed parking lot. The surface depression measured 90cm in diameter and 15cm deep at its most depressed point. The feature was excavated by laying a north - south transect and excavating the west side of the transect; exposing the east profile of the feature. All soil was screened through ¼ inch/6mm archaeological screen. Feature 8 was excavated to a depth of 147cm bd. The profile clearly indicated a pit feature, much different than previous profi les. The storage feature measured 110cm in depth and 115cm in di ameter. The soil profile showed a stark contrast between the pit fill and the surrounding C - horizon, with a layer of organic material. No cultural material was found within the feature. Dr. Sc haetzl examined the feature and explained that though there was no pedogenesis, it was too deep to be a natural phenomenon. Additionally, the feature was too conical and too deep to be a tree throw. Though Dr. Schaetzl could not give 100 percent certainty that this feature was a storage pit, it had many similarities to other pit features found in the Inland Waterway. Soil samples were collected from the feature and will be analyzed by Dr. Schaetzl. Pit Feature 9: N 45.468499 W 84.603520 This surface depre ssion measured 70cm in diameter, and was 10cm deep at its most d epressed point. This depression was selected due to its immediate proximity to Pit Feature 8, the likely 169 storage feature. The feature was excavated by laying a north - south transect and excavat ing the east (large root on the west side) side of the transect; exposing the west profile of the feature. All soil was screened through ¼ inch/6mm archaeological screen. Excavated to a depth of 90cm bd. Profile indicated the depression was asymmetrical wi th the E Horizon dipping on the south side of the profile (not f ollowing the surface depression. Feature 9 is most likely not a cultural phenomenon, evidenced by t he typical soil sequence and consistent soil profile. Two small flakes (lithic debitage) were discovered while screening. Pit Feature 10: N 45.468481 W 84.603411 This surface depression measured 70cm in diameter, and was 15cm deep at its most depressed po int. This depression was selected due to its proximity to Pit Fe ature 8, the likely storage feature. The feature was excavated by laying a north - south transect and excavating the west side of the transect; exposing the east profile of the feature. All soil was screened through ¼ inch/6mm archaeological screen. Excavate d to a depth of 100cm bd. The profile was similar to Feature 9 in that it was a consistent depression with typical stratigraphy. Feature 10 is most likely not a cultural feature. No artifacts were discovered. 170 Results of Auger Testing Auger Test 1: N 45. 468194 W 84.603445 - Negative Auger Test 2: N 45.468125 W 84.603299 Negative Auger Test 3: N 45.468331 W 84.603400 - Negative Auger Test 4: N 45.467992 W 84.603258 - Negative Auger Test 5: N 45.467885 W 84.603126 - Negative Auger Test 6: N 45.467825 W 84. 602896 Positive Auger Test 7: N 45.467697 W 84.602887 - Negative Auger Test 8: N 45.467186 W 84.602269 - Negative Auger Test 9: N 45.467072 W 84.602142 Negative Auger Test 10: N 45.466521 W 84.600953 - Negative Auger Test 11: N 45.466373 W 84.600874 Negative Auger Test 12: N 45.466784 W 84.600751 Negative Auger Test 13: N 45.466997 W 84.601041 Positive Auger Test 14: N 45.467103 W 84.601685 - Negative Auger Test 15: N 45.467221 W 84.602326 - Negative Auger Test 16: N 45.468294 W 84.603151 - Negat ive Auger Test 17: N 45.468237 W 84.603151 - Negative Auger Test 18: N 45.468524 W 84.603371 - Negative Auger Test 19: N 45.4684 91 W 84.603455 Negative Auger Test 20: N 45.468444 W 84.603499 - Negative 171 A PPENDIX C : Description of Excavated Storage Fe atures from 172 Excavated Storage Features from Lower Michigan's Late Woodland Period Site Feature Diameter at surface (cm) Max depth of pit on surface (cm) Max depth of pit subsurface (cm) Total est imated depth of pit (cm) Additional notes Juntunen Feature 4 61 NA NA 49 U - shaped, flat bottom Feature 39 116 NA NA 110 U - shaped, flat bottom Grapevine Point High Terrace Pit 1 148 17 82 99 High Terrace Pit 2 220 70 115 185 High Outwash Pit 1 200 55 85 140 Pine Point Trench 2 Pit A 135 21 85 106 Trench 2 Pit B 162 29 112 141 Trench 1 Pit A 280 42 115 157 Trench 1 Pit B 270 62 112 174 The Gorge Pit 1 230 18 118 136 West Edge Cluster 1 Pit A 200 27 116 143 Cluster 2 Pit A 225 51 11 2 163 Cluster 3 Pit A 210 40 85 125 Skegemog Point 180 30 20OT283 Feature 5, 7, 13, 22, 27, 28, 42, 44, 53, 58, 60, 64, 67, 80, 81, 82, 85, 96, 97, 98, 103, 104 80 - 200 20 - Apr 1m - 2.2m 100cm rounded bottom, Feature 17 210 NA NA 82 dates= AD 1520 (65), AD 1510 (13), AD 1540 (32), AD 1640 (64), AD 1550(104) Feature 29 90 NA NA 53 steeply sided, with cylindrical profiles Feature 65 200 NA NA 89 black charred, burned horizon on the bottom Feature 78 148 6 NA 94 depressions= 80 - 200cm in diamet er (most are 1 - 1.3m), 4 - 20cm (avg. 10cm) deep 173 Feature 79 NA NA NA 127 small quantities of microbotanicals= blackberry/raspberr y, acorn, black walnut, butternut, hazelnut, walnut, wild rice (3 pits) 174 REFERENCES 175 REFERENCES Albert, D. A. 1995 Regional Landscape Ecosystems of Michigan Minnesota, and Wisconsin: A Working Map and Classification. Gen. Tech. Rep. NC - 178. USDA Forest Service, North Central Forest Experiment Station, St. Paul , MN . Albert, D. and L. Minc 1987 The Natural Ecology and Cultural History of the Colonial Point Red Oak Stands. University of Michigan Biological Research Station, Michigan Department of N atural Resources, Lansing , MI . Anderson, D . G., and Gillam, J. C. 2000 Paleoindian C olonization of the Americas: I mpli cations from an E xamination of P hysiography, D emography, and A rtifact D istribution. American Antiquity 65(1): 43 - 66. Anderson, D. G., & Hanson, G. T. 1988 Early Archaic S ettlement in the S outheastern United States: A C ase S tudy from the Savannah River V alley. American Antiquity 53(2): 262 - 286. Arzigian, C . M., W. Holtz - Leith, and D. Martinek 2007 Pit Storage and Plant Foods: Ethnographic, Experimental, and Archaeological Evidence. Manuscript on file at Mississippi Valley Archaeology Center, University of Wisconsin - La Crosse . Barnes, B.V. and Wagner Jr, W.H. 1981 Michigan T rees. A G uide to the T rees of Michigan and the Great Lakes Region. Baxter, M.J. 2015 Notes on Quantitative Archaeology and R. Nottingham Trent University - 2015 ARCHEOLOGIA - Studi e Ricerche . Bermúdez, J.L. 2009 Decision T heory and R ationality . Oxford University Press , Oxford. Bettinger, R. L., M . Ripan, and H . McCarthy 1997 Central Place Models of Acorn and Mussel Processing. Journal of Archaeological Scienc e 24:887 - 899. Binford , L. R. 1980 Willow Smoke and Dogs' Tails: Hunter - gatherer Settlement Systems and Archaeological Site Formation. American Antiquity 45(1):4 - 20. Binford, L.R. 176 2001 Constructing frames of reference: an analytical method for archaeological theory building using ethnogr aphic and environmental data sets . Univ of California Press. Binford, L. R. and A. L. Johnson. 2014 Program for Calculating Environmental and Hunter - Gatherer Frames of Reference (ENVCALC2.1). U pdated Java Version, August, 2014. Bird, R. L. B. and D. W. Bird 1997 Delayed Reciprocity and Tolerated Theft: The Behavioral Ecology of Food - Sharing Strateg ies. Current Anthropology 38:49 - 77. Black, L. 1973 t Arctic Anthropology 10(1). [Madison]: 11 2. http://ehrafworldcultures.yale.edu/document?id=rx02 - 023 Boserup, E. 1993 The C onditions of A gricultural G rowth: The E conomics of A grarian C hange U nder P opulation P ressure . Earthscan Publications Ltd, London . Branstner, S. M. 1991 Decision - Making Proc esses in a Culture Contact Context: The Case of the Tionontate Huron of the Upper Great Lakes . Unpublished Ph.D. Dissertation, Department of Anthropology, Michigan State University , East Lansing. Brashler , J. G. 1981 Early Late Woodla nd boundaries and in teraction: Indian ceramics of southern lower Michigan Anthropological Series, Vol ume 3, N umber 3 , Publications of the Museum, Michigan State University . Brashler, J.G. and Holman, M.B. 1985 Late Woodland Continuity and Change in the Saginaw Valley of Mich igan. Arctic Anthropology , pp.141 - 152. Brenton, B. 1988 The Seasonality of Storage. In Coping with Seasonal Constraints, edited by R. Huss - Ashmore, pp. 45 - 54. University of Pennsylvania Museum of Archaeology, Philadelphia, PA. Brose, D. S. and M. J. Hamb acher 1999 The Middle Woodland in Northern Michigan. In Archaeology of the Great Lakes State, edited by J. R. Halsey, pp. 173 - 192. Cranbrook Institute of Science, Bulletin 64. Bloomfield Hills, Michigan. Brouwer, M.E . 177 2011 Modeling M esolithic H unter - G atherer L and U se and P ost - G lacial L andscape D ynamics in the C entral Netherlands . Unpublished Ph.D. Dissertation, Department of Anthropology, Michigan State University , East Lansing. Cannon, A. and D. Y. Yang 2006 Early S torage and S edentism on the Pacific Northwest Coast: A ncient DNA A nalysis of S almon R emains from Namu, British Columbia. American A ntiquity 71:123 - 40. Cashdan, E. A. 1990 Introduction. In Risk and U ncertainty in T ribal and P easant E conomies , edited by E . Cashdan, pp. 1 - 16. Westview Press, Boulder, CO. Charles, D.K. and J. Buikstra 2002 Sitting, Sighting, and Citing the Dead. In The Space and Place of Death , e dited by H. Silverman and D. Small. Archaeological Papers of the American Anthropological Asso ciation Vol. 11 , Arlington, VA. Chivis, J. 2016 The I ntroduction of Havana - Hopewell in W est Michigan and N orthwest Indiana: An I ntegrative A pproach to the I dentification of C ommunities, I nteraction N etworks, and M obility P atterns . Unpublished Ph.D. Disse rtation, Department of Anthropology, Michigan State University , East Lansing. Church, F. and Nass Jr, J. P. 2002 Central Ohio Valley D uring the Late Prehistoric Period: Subsistence - Responses to Risk. Northeast Subsistence - Settlement Change: AD 700 1300 , edited by J. Hart and C. Reith p.11. New York State Museum Bulletin 496. University if the State of New York. Cleland, C. E. 1966 The Prehistoric Animal Ecology and Ethnozoology of the Upper Great Lakes Region . Anthropological Papers No. 29. Museum of Anthropology, University of Michigan, Ann Arbor. Cleland, C. E. 1982 The Inland Shore Fishery of the Northern Great Lakes: Its Development and Importance in Prehistory. American Antiquity 47(4):761 - 784. Cleland, C. E. 1992 Rite ve Americans , pp. 74 - 127. University of Michigan Press, Ann Arbor. Cleland, C.E. and Ruggles, D.L. 1996 The Samels Field Site: An Early Archaic Base Camp in Grand Traverse County, Michig an. Investigating the A rchaeological R ecord of the Great Lakes S tate: E ssays in H onor of Elizabeth Baldwin Garland , edited by M. B. Holman, J. G. Brashler, and K. 178 E. Parker, pp. 341 - 382. New Issues Press, Western Michigan University, Kalamazoo, Michigan Cleland, C.E., Holman, M.B. and Holman, J.A. 1998 The Mason - Quimby Line Revisited. The Wisconsin Archeologist , 79 (1), pp.8 - 28. Collignon, B. 1993 Variations o f a Land Use Pattern: Seasonal Movements a nd Cultural Change Among t he Copper Inuit. Etudes Inui t Studies , pp.71 89. Available at: http://ehrafworldcultures.yale.edu/document?id=nd08 - 038 [Accessed August 8, 2018]. Colson, E. 1979 In Good Years and in Bad: Food Strategies of Self - Reliant Societies. Journal of Anthropological Research 35(1):18 - 29. Cowan, F. L. 2006 A Mobile Hopewell? Questioning Assumptions of Ohio Hopewell Sedentism. In Recreating Hopewell , edited by D.K. Charles and J.E. Buikstra, pp. 4 91 - 509. University Press of Florida, Gainesville. Cunningham P. 2005 Assumptive Holes and H ow to Fill Them: A series of Experiments Investigating the Pit Storage Potential of Hazelnuts and Acorns. EuroREA 2 : 55 - 66. Cunningham, P. 2011 Caching Y our S avings: The U se of S mall - S cale S torage in European P rehistory. Journal of Anthropological Archaeol ogy 30:135 - 144. Dekin, A.A. 1966 A F luted P oint from Grand Traverse County. Michigan Archaeologist , 12 (1) : 35. Densmore, F. 1929 Chippewa Customs . Smithsonian Institution Bureau of American Ethnology Bulletin 86. U.S. Gov ernment , Washington , D.C. Dens more, F . 1974 How Indians Use Wild Plants for Food, Medicine and Crafts (formerly Uses of Plants by Chippewa Indians). Dover Publications, Incorporated, New York. Originally published 1926/1927 (Uses of Plants by t he Chippewa Indians), Bureau of American Ethnology, Smithsonian Institution, Washington, D.C . Dunham, S.B. 2000 Cache P its: E thnohistory, A rchaeology, and the C ontinuity of T radition. Interpretations of Native North American L ife: M aterial C ontributions to E thnohistory , edited by M. Nassaney an d E. Johnson, p. 225 - 260. Society for Historical Archaeology and University Press of Florida, Gainesville, FL. 179 Dunham, S.B. 2014 Late Woodland S ettlement and S ubsistence in the E astern Upper Peninsula of Michigan . Unpublished Ph.D. Dissertation, Departmen t of Anthropology, Michigan State University , East Lansing. Dunham, S. B. and M. Branstner 1993 1992 Cultural Resource Surveys: Hiawatha Nationa l Forest . Great Lakes Research Associates, Inc., Williamston, MI. Edwards, R.W. 2017 The Canine Surrogacy Appr oach and Paleobotany: An Analysis of Wisconsin Oneota Agricultural Production and Risk Management Strategies. Unpublished Ph.D. Dissertation, Department of Anthropology, University of Wisconsin, Milwaukee. Egan, K.C. 1993 Hunter - Gatherer Subsistence Adap tation in the Saginaw Valley, Michigan. Vol. I&II. Unpublished Ph.D. Dissertation, Department of Anthropology , Michigan State University, East Lansing. Ellis, C. 2011 Measuring Paleoindian range mobility and land - use in the Great Lakes/Northeast. Journal of Anthropological Archaeology 30 (3) : 385 - 401. Ember, Carol R. 2007 Using the HRAF Collection of Ethnography in Conjunction With the Standard Cross - Cultural Sample and the E thnographic Atlas . Cross - Cultural Research 4 1 : 396 - 427. Emerson T. an d D. McElrath 2009 The Eastern Woodlands Archaic and the Tyranny of Theory, In Archaic Societies: Diversity and Complexity across the Midcontinent , edited by T. Emerson and D. McElrath . State University of New York Press, Albany, NY. Fitting, J. E. 1975 The Archaeology of Michigan: A Guide to the Prehistory of the Great Lakes Region . Bulletin 56. Cranbrook Institute of Science, Bloomfield Hills, M I . Fitting, J.E. and Cleland, C.E. 196 9 Late P rehistoric S ettlement P atterns in the U pper Great Lakes. Ethnoh istory , pp.289 - 302. Fitzhugh, B. 2001 Risk and I nvention in H uman T echnological E volution. Journal of Anthropological A rchaeology 20 (2) : 125 - 167. Frederick, K. and R. Albert 180 2016 Take Mine Well - done: A Late Woodland Acorn Charring Pit, Implications and Acorn Identification . Paper presented at the 2016 Graduate Students in Anthropology Symposium, East Lansing, MI. Frink, L. 2007 Storage and S tatus in P recolonial and C olonial C oas tal W estern Alaska. Current Anthropology 48(3):349 - 374 . F unk, C. 2010 Yup'ik Eskimo Gendered Information Storage Patterns. In Information and Its Role in Hunter - Gatherer Bands , edited by R. Whallon, W. A. Lovis and R. K. Hitchcock, pp. 29 - 58. Cotsen Ins titute of Archaeology Press, Los Angeles , CA. Greenman , E. 1927 Museum Report: Archaeological Investigations in Emmet County, Michigan. July 22 30, 1927. Unpublished manuscript on file at the Museum of Anthropology, University of Michigan, Ann Arbor . Grooms, T. C. 1996 . Unpublished Department of Anthropology, University of Cincinnati . Halstead, P. and J. O'Shea 1989 Introduction: Cultural Responses to Risk and Uncertainty . In Bad Year Economics: Cultural Responses to Risk and Uncertainty , edited by P. Halstead and J. O'Shea, pp. 1 - 8. Cambridge University Press, Cambridge. Hambacher, M. J. 1992 The Skegemog Point Site: Continuing Studies in the Cultural Dynamic s of the Carolinian - Canadian Transition Zone. Unpublished Ph.D. Dissertation, Department of Anthropology, Michigan State University , East Lansing . Hambacher, M. J., and M. B, Holman. 1995 Camp, Cach e and Carry: The Porter Creek South Site (20MN100) and Cache Pits at 20MN31 in the Manistee National Forest. The Michigan Archaeologist 41(2 - 3):47 - 94. Hambacher, M, J. Roberston, and R. Schaetzl 2015 Late Prehistoric Food Choices in the Upper Great Lakes Region: Evidence from 20OT283 and 20OT3 in the Lower Gr and River Valley of Michigan. Paper presented at the Society for American Archaeology Annual Conference, San Francisco, CA. Hambacher, M. J, K. Hagenmaier, K. E. Parker, K. C. Egan - Bruhy, T. J. Martin, R. Schaetzl, J. Ladwig, L. Scott Cummings, M. Malain ey, C. T. Espenshade, and T. Figol 2016 Phase II and Phase III Archaeological Investigations at 20OT344, 20OT283, and 20OT3 for the US - 31/M - 231 Holland to Grand Haven Bypass Project, Ottawa County, Michigan. Commonwealth Cultural Resource Group, Inc., Jack son, MI. 181 Hansson, S.O. 1994 Decision Theory . Department of Philosophy and the History of Technology, Royal Institute of Technology (KTH). Hart, J. P. 1993 Monogahela Subsistence - Settlement Change: The Late Prehistoric Period in the Upper Ohio River Valley . Journal of World Prehistory 7:71 - 120. Hart, J. P. 2014 A Critical Assessment of Current Approaches to Investigations of the Timing, Rate, and Adoption Trajectories of Domesticates in the Midwest and Great Lakes. Midwest Archaeological Conference Inc. Oc casional Papers, No.1. Hart, J. and W. Lovis 2013 Reevaluating What We Know About the Histories of Maize in Northeastern North America: A Review of Current Evidence. Journal of Archaeological Research 21(2):175 - 216. Henrikson, L. S. 2003 Bison F reezers and H unter - G atherer M obility: Archaeological A nalysis of C old L ava T ube C aves on Idaho's Snake River Plain. The Plains Anthropologist , 263 - 285. Holman, M. 1984 The I dentification of Late Woodland M aple S ugaring S ites in the U pper Great Lakes. Midcontinent al Journal of Archaeology , 63 - 89. Holman, M. B. and J. Brashler 1999 Economic, Material Cultural, and Trade in the Late Woodland Lower Peninsula of Michigan, In edited by J.R. Halsey, pp. 31 - 38. Cranbrook Institute of Sc ience, Bloomfield Hills, MI. Holma n, M. B. and R. G. Kingsley 1996 Territoriality and Social Interaction During the Early Late Woodland Period in Southern Michigan. In Investigating the Archaeological Record of the Great Lakes State: Essays in Honor of El izabeth Baldwin Garland , edited by M. B. Holman, J. G. Brashler, and K. E. Parker, pp. 341 - 382. New Issues Press, Western Michigan University, Kalamazoo, M I . Holman, M. B. and F. J. Krist 2001 Late Woodland Storage and Mobility in Western Lower Michigan. Wisconsin Archeologist: Papers in Honor of Carol I Mason 2:7 - 32. Holman, M. B. and W. A. Lovis 2008 The Social and Environmental Constraints on Mobility in the Late Prehistoric Upper Great Lake s Region. In The Archaeology of Mobility: Old World and New Wo rld Nomadism, edited by H. Bernard and W. Wandrich. Cotsen Institute of Archaeology, 182 UCLA, Los Angeles , CA. Howey, M.C. 2006 Ritual, R esources, and R egional O rganization in the Upper Great Lakes, AD 1200 - 1600 Unpublished Ph.D. Dissertation, Department of Anthropology, University of Michigan, Ann Arbor. Howey, M.C. 2012 Mound Builders and Monument Makers of the Northern Great Lakes, 1200 - 1600 . University of Oklahoma Press , Norman, OK. How ey, M.C. 2015 Geospatial L andscape P ermeability M odeling for A rcha eology: A C ase S tudy of F ood S torage in N orthern Michigan. Journal of Archaeological Science , 64 : 88 - 99. Howey, M.C. and O' S hea, J.M. 2006 Bear's J ourney and the S tudy of R itual in A rchaeology. American Antiquity 71 (2) : 261 - 282. Howey, M.C. and Parker, K.E . 2008 Camp, Cache, Stay Awhile: Preliminary Considerations of the Social and Economic Processes of Cache Pits along Douglas Lake, Michigan. Michigan Archaeologist 54 (1 - 4) : 19 - 44. Howey, M.C. and Frederick, K. 2016 Immovable F ood S torage F acilities, K nowl edge, and L andscape in N on - S edentary S ocieties: P erspectives from N orthern Michigan. Journal of Anthropological Archaeology 42 : 37 - 55. Ingold, T. 1983 The S ignificance of S t orage in H unting S ocieties. Man 18(3):553 - 71. Jenness, D. 1922 Life o f The Copper Eskimos. Report o f The Canadian Arctic Expedition, 1913 - 1918 , vol. XII, pp.5 277, 9 plates. Available at: http://ehrafworldcultures.yale.edu/document?id=nd08 - 001 [Accessed August 8, 2018]. Jochim, M.A. 1976 Hunter - G atherer S ubsistence and S ettlement: A P r edictive M odel (No. 330.901 JOC. CIMMYT.). Kapp, R.O. 1999 Michigan Late Pleistocene, Holocene, and Presettlement Vegetation and Climate. In edited by J.R. Halsey, pp. 31 - 38. Cranbrook Institute of Science, Bloomfield Hi lls, MI. 183 Keen, R.A. 1993 Michigan Weather. American & World Geographic Pub., Helena, MT. Kelly, Robert L. 1983 Hunter - Gatherer Mobility Strategies. Journal of Anthropological Research 39:277 306. Kelly, Robert L. 1995 Egalitarian and Nonegalitarian H unter - Gatherers. In The Fo raging Spectrum: Diversity in Hunter - gatherer Lifeways , by R. Kelly, chap. 8 pp. 293 - 333. Percheron Press. Kelly R. and L. Todd 1988 Coming into the Country: Early Paleoindian Hunting and Mobility. American Antiquity, 532. Kin gsley, R. 1999 The Middle Woodland Period in Southern Michigan. In Past: The Archaeology of the Great Lakes State, edited by J. R. Halsey, pp. 148 - 172. Cranbrook Institute of Science, Bloomfield Hills, MI Kooiman, S. M. 2012 O ld Pots, New Ap proaches: A Functional Analysis of Woodland Pottery from Lake Illinois State University, Normal. Kooiman, S. M. 2018 A Multiproxy Analysis of Culinary, Techn ological, and Environmental Interactions in the Northern Great Lakes Region . Unpublished Ph.D. Dissertation, Department of Anthropology, Michigan State University , East Lansing Koldehoff, B. and J. Walthall 2009 Dalton and the Early Holocene Midcontinent : Setting the Stage, In Archaic Societies: Diversity and Complexity across the Midcontinent , edited by Emerson T. et al. State University of New York Press, Albany. Kuijt, I. 2009 What do we R eally K now about F ood S torage, S urplus, and F easting in Preagri cultural C ommunities? Current Anthropology 50:641 - 644. Larsen, C.E. 1985 Lake Level, Uplift and Outlet Incision: The Nipissing and Algoma Great Lakes. In Quaternary Evolution of the Great Lakes, edited by P. F. Ka rrow and P. E. Calkin, pp. 63 - 77. Special Paper 30. Geological Society of Canada, St. John's, Newfoundland. L ovis , W. 184 1974 Late Woodland Cultural Dynamics in the Northern Lower Peninsula of Michigan. Unpublished Ph.D. Dissertation, Department of Anthropol ogy, Michigan State University, East Lansi ng. Lovis, W. A. 1985 Seasonal Settlement Dynamics and The Role of the Fletcher Site in the Woodland Adaptations of the Saginaw Drainage Basin. Arctic Anthropology 22(2):153 - 170. Lovis, W. A. 1991 Screaming Loon: A Post - bow. Michigan Archaeologist 36(3 - 4):233 - 254. Lovis, W. A. 1999 The Middle Archaic: Learning to Live in the Woodlands. In Buried Past: The Archaeology of the Great Lakes State, e dited by John R. Halsey, pp. 125 - 146. Cranbrook Institu te of Science, Bulletin 64. Bloomfield Hills, M I . Lovis, W. A. , K. C. Egan, G. W. Monaghan, B. A. Smith, and E. J. Prahl. 1996 Environment and Subsistence at the Marquette Viaduct Locale of the Flet cher Site. In Investigating the Archaeological Record o f the Great Lakes State: Essays in Honor of Elizabeth Baldwin Garland , edited by M. Holman, J. Brashler, and K. Parker, pp 251 - 305. New Issues Press. Kalamazoo , MI. Lovis, W. A. and G. Bole 2002 L ate Gainey/Early Parkhill Phase Paleoindian Penetration Peninsula: A Paleoindian Preform from East Lake, Kalkaska County. The Michigan Archaeologist 48(3 - 4):225 - 243. Lovis, W. A., K. C. Egan - Bruhy, B. A. Smith, and G. W. Monaghan 2001 Wetlands and Emergent Horticultural Economies in the Upper Great Lakes: A New Perspective from the Schultz Site. American Antiquity 66(4):615 - 632. Lovis, W., R. Donahue, and M. Holman 2005 Long - Distance Logistic Mobility as an Organizing Pr inciple among Northern Hunter - Gatherers: A Great Lakes Midd le Holocene Settlement Systems. American Antiquity 704 : 669 - 693. Lovis, W. A. G. W. Monaghan, A. F. Arbogast, and S. L. Foreman 2012 Differential Spatial Preservation of Archaeological Sites in a Great Lakes Coastal Zone. American Antiquity 77(3):591 - 608. Madsen, D. B., & Schmitt, D. N. 1998 Mass collecting and the diet breadth model: A Great Basin example. Journal of Archaeological Science 25 (5) : 445 - 455. Malainey, M. , R . Przybylski and B . Sherriff 185 2001 Affect the Settlement and Subsistence Patterns of Hunter - Gatherers. American Antiquity 61(1):141 - 161. Marston J. M. 2011 Archaeological Markers of Agricultural Risk Manag ement. Journal of Anthropological Archaeology 30:190 - 205. Martin, S. R. 1985 Mode ls of Change in the Woodland Settlement of the Northern Great Lakes Region. Unpublished Ph.D. Dissertation, Department of Anthropology, Michigan State University, East Lansing. Martin, S.R. 1999 A Site for all Seasons: Some Aspects of Life in the Upper P eninsula During Late Woodland Times. In he Archaeology of the Great Lakes State, edited by J. R. Halsey, pp. 148 - 172. Cranbrook Institute of Science, Bloomfield Hills, MI . McPherron, A. 1967 The Juntunen Site and the La te Woodland Prehistory of the Upper Great Lakes Area. Anthropological Papers No. 30. University of Michigan Museum of Anthropology, Ann Arbor. Meltzer, David and Bruce Smith 1986 Paleoindian and Early Archaic Subsistence Strategies in Eastern North Ameri ca. In Foraging, Collecting, and Harvesting: Archaic Period Subsistenc e and Settlement in the Eastern Woodlands , edited by S. Neusius, p. 1 - 30, Occasional Paper 6. Southern Illinois University at Carbondale Center for Archaeological Investigations. Mithen , S. J. 1990 Thoughtful Foragers: A Study of Prehistoric Decision Making . Cambridge University Press, Cambridge. Milner, C. M. 1998 Ceramic Style, Social Differentiation, and Resource Uncertainty in the Late Prehistoric Upper Great Lakes. Unpublished Ph. D. Dissertation, Department of Anthropology, University of Michigan, Ann Arbor. 1998 The Socioeconomi c role of Late Woodland Enclosures in Northern Lower Michigan. In Ancient Earthen Enclosures of the Eastern Woodlands, edi ted by R. C. Mainfort and L. P. Sullivan, pp. 181 - 201. University Press of Florida, Gainesville. Morgan, C. 2012 Modeling M odes of Hu nter - Gatherer Food Storage. American Antiquity 77:714 - 736. 186 Naga oka, L. 2002 The Effects of Resource Depression on Fora ging Efficiency, Diet Breadth and Patch Use in Southern New Zealand. Journal of Anthropological Archaeology 21:419 - 442 . Naroll, R. 1964 American Anthropologist 66:863 - 67. Naroll, R. 1965 of Cross - Cultural Analysis, Social Research 32:428 - 51. Naroll, R. 1970 What Have We Learned from Cross - Cultural Surveys? American Anthropologist, 72(6): 1227 - 1288. Oetelaar, G.A. and Oetelaar, D.J. 2007 The N ew E cology and L andscape A rchaeology: I ncorp orating the A nthropogenic F actor in M odels of S ettlement S ystems in the Canadian P rairie E cozone. Canadian Journal of Archaeology/Journal Canadien d'Archéologie , pp.65 - 92. Oetelaar, G.A. and Meyer, D. 2006 Movement and Native American L andscapes: A C ompar ative A pproach. Plains Anthropologist 51 (199) : 355 - 374. 2001 Life, Death, and the Longhouse: A Gendered View of Oneota Social Organization. In Gender and the Archaeology of Death , edited by B. Arnold and N.L. Wicker. AltaMira Press, Walnu t Creek, C A. Olson, Ronald L. 1936 University o f Washingt on Publications i n Anthropology . The University of Washington , Seattle. http://ehrafworldcultures.yale.edu/document?id=nr17 - 001. O'Shea, J. 1981 Coping with Scarcity: Exchang e and Social Storage. In Economic Archaeology: Towards an Integration of Ecolo gical and Social Approaches , edited by A. Sheridan and G. Bailey, pp. 167 - 183. B.A.R, Oxford. 1989 The Role of Wild Resources in Small - Scale Agricultural Systems: Tal es from the Lakes and the Plains. In Bad Year Economics: Cultural Responses to Risk and Uncertainty, O'Shea, J. 2003 Inland foragers and the adoption of maize agriculture in the Upper Great Lakes of N orth America. Before Farming , 2003 (1) : 1 - 21. 187 Parker, K. 1996 T hree Corn Kernels and a Hill of Beans: The Evidence for Prehistoric Horticulture in Michigan. In Investigating the Archaeological Record of the Great Lakes State: Essays in Honor of Elizabeth B aldwin Garland , edited by M. B. Holman, J. G. Brashler and K. E . Parker , pp. 307 - 339. New Issues Press , Kalamazoo, MI. Prince, P. 2001 Dating and I nterpreting P ine C ambium C ollection S cars from T wo P arts of the Nechako R iver D rainage, British Columbia. J ournal of A rchaeological S cience 28 (3) : 253 - 263 . Quimby, G . 196 6 A Year with a Chippewa Family, 1764 - 1764. Ethnohistory 9(3):217 - 239. Raviele, M. 2010 Assessing Carbonized Archeological Cooking Residues: Evaluation of Maize Phytolith Taphonomy Experimenta l Residue Analysis . Unpublished Ph.D. Dissertation, Departme nt of Anthropology, Michigan State University, East Lansing. R Core Team 2017 R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. UR L https://www.R - p roject.org/ . Reynolds, P. J. 1974 Experimental Iron Age Storage Pits: An Interim Report. Proceedings of the Prehistoric Society 40:118 - 131. Robertson, J. W. Lovis, and J. Halsey 1999 The Late A rchaic: Hunter Gatherers in an Uncertain Environment. In Retrieving edited by John R. Halsey, pp. 125 - 146. Cranbrook Institute of Science, Bulletin 64. Bloomfield Hills, M I . Roger s, E.S. 19 62 The Round Lake Ojibwa (No. 1). Royal Ontario Museum. Rowley - Conwy, P. and M. Zvelebil 1989 Saving it for L ater: S torage by P rehistoric H unter - G atherers in Europe . In Bad Year Economics: Cultural Responses to Risk and Uncertainty , edited by P. Halstead and J. M. - 56. Cambridge University Press, Cambridge. Savishinsky, J. S. 1981 Mobility as an A spect of S tress in an Arctic C ommunity. American Anthropologist 73:604 - 618. Schaefer, J. M. 188 1969 Linked Pair Alignments for the HRAF Quality Cont rol Sample Universe. Behavioral Science Notes 4, no. 4: 2990320. Schaetzl, R.J. and Isard, S.A. 1991 The D istribution of Spodosol S oils in S outhern Michigan: A C limatic I nterpretation. Annals of the Association of American Geographers 81 (3) : 425 - 442. Sch aetzl, R.J., Krist Jr, F.J., Rindfleisch, P.R., Liebens, J. and T.E. Williams 2000 Postglacial L andscape E volution of N ortheastern L ower Michigan, I nterpreted from S oils and S ediments. Annals of the Association of American Ge ographers 90 (3) : 443 - 466. Schae tzl, R.J. and Weisenborn, B.N. 2004 The Grayling Fingers R egion of Michigan: S oils, S edimentology, S tratigraphy and G eomorphic D evelopment. Geomorphology 61 (3 - 4) : 251 - 274. Schaetzl , R.J., Knapp, B.D. and Isard, S.A. 2005 Modeling S oil T emperatures and the M esic - F rigid B oundary in the C entral Great Lakes R egion, 1951 2000. Soil Science Society of America Journal 69 (6) : 2033 - 2040. S c hrenk, L . I . and A . Nagler. 18 54 o f The Am Reisen Und Forschungen Im Amur - Lande In Den Jahren 1854 - 1856 . St . Petersburg: Kaiserliche Akademie der Wissenschaften. http://ehrafworldcultures.yale.edu/document?id=rx02 - 002. Seeman, M. F. and J. L. Branch 2006 The Mounded Landscapes of Ohio: Hopewell Patterns and Placements. In Recreating Hopewell, edited by D.K. C harles and J.E. Buikstra, pp. 491 - 509. University Press of Florida, Gainesville. Shott, M. and S. Wright 1999 The Paleo - Archaeology of the Great Lakes State, edited by J. R. Halse y and M. D. Stafford, pp. 59070. Cranbrook Institute of Science, Bloomfield Hills, MI. Simon, M. L. 2009 A Re gional and Chronological Synthesis of Archaic Period Plant Use in the Midcontinent. In Archaic Societies: Diversity and Complexity across the Mid continent. edited by Emerson T. et al. State University of New York Press, Albany. Simon, M. L., and K . E. Par ker 2006 Prehistoric P lant use in the American Bottom: New T houghts and I nterpretations. Southeastern Archaeology 25(2):212 - 257. Smith, Philip , E. L. 1972 Changes in Population Pressure in Archaeological Explanation. World Archaeology 4(1):5 - 18. 189 Smith, B. 2004 Michigan Basin. In An Upper Great Lakes Archaeo logical Odyssey: Essays in Honor of Charles E. Cleland , e dited by W. Lovis and C. Cleland (Vol. 1) Wayne State University Press, Detroit, MI. Smith, B. D. 2001 Low - Level Food Production. Journal of Archaeological Research 9(1): 1 - 43. Smith, C. S. 2003 Hunter G atherer M obility, S torage, and H ouses in a M arginal E nvironment: A n E xample from the M id - Holocene of Wyoming. Journal of Anthropological Archaeology 22 (2) : 162 - 189. Smith, B. D., & Yarnell, R. A. 2009 Initial F ormation of an I ndigenous C rop C omple x in E astern North America at 3800 BP. Proceedings of the National Academy of Sciences 10 ( 616 ): 6561 - 6566. Smith, C . R. 1978 Handbook o f North American Indians. California . Smithsonian Institution , Washington, D.C. http://ehrafworldcultures .yale.edu/document?id=ns22 - 005. Sof fer, O. 1989 Storage, Sedentism, and the Eurasian Paleolithic Record. Antiquity 63:719 - 732. Speth, J.D. 1983 Bison K ills and B one C ounts: Decision M aking by A ncient H unters . University of Chicago Press , Chicago, IL. S pielmann, K.A. 1986 Interdependence Among Egalitarian Societies. Journal of Anthropological Archaeology 5(4):279 - 312. Stefánsson, V. 1914 Stefánsson - Anderson Arctic Expedition o f The American Museum: Preliminary Ethnological Report. Anthropological Paper s o f The American Museum o f Natural History , Vol. XIV, p.395. Available at: http://ehrafworldcultures.yale.edu/document?id=nd08 - 002 [Accesse d August 8, 2018]. Stephens, D.W. and E . L. Charnov 1982 Optimal Foraging: Some Simple Stochastic Models. Behavior al Ecology and Sociobiology 10(4):251 - 263. Stopp, M. P. 2002 Ethnohistoric A nalogues for S torage as an A daptive S trategy in N ortheastern S ubarctic P rehistory. Journal of Anthropological Archaeolog y 21:301 - 328. 190 Styles, B.W., S. Ahler and M. Folwer 1983 Modoc Rock Shelter Revisited. In Archaic Hunters and Gatherers in the American Midwest , edited by J.L. Phillips and J.A. Brown, pp. 165 - 195. Academic Press, New York , NY. Surovell, T.A. 2000 Early Paleoindian W omen, C hildren, M obility, and F ertility. Amer ican Antiquity 65 (3) : 493 - 508. Testart, A. 1982 The Significance of Food Storage among Hunter - Gatherers: Residence Patterns, Population Densities, and Social Inequalities. Current Anthropology 23:523 - 537. Thompson V. and J. Turck 2009 Adaptive Cycles o f Coastal Hunter - Gatherers. American Antiquity 7 4(2):255 - 278. Trigger, B. 1968 The Detriments of Settlement Patterns. In Settlement Archaeology, edited by K. C. Chang, pp. 53 - 78. National Press, Palo Alto, CA. Tushingham, S. & Bettinger, R. L. 2013 Wh y F oragers C hoose A corns B efore S almon: Storage, M obility, and R isk in A boriginal California. Journal of Anthropological Archaeology 32 (4) : 527 - 537. Ugan, A. 2005 Does Size Matter? Body Size, Mass Collecting, and Their implications for Understanding Prehistoric foraging behavior. American Antiquity 70 : 75 - 9 . VanStone, J.W. 1979 Historic Ingalik Settlements Along t he Yukon, Innoko, a nd Anvik Rivers, Alaska. Fieldiana : Anthropology; Publication (Field Museum Of Natural History) , ( 72 ): 1296, pp.ix, 99. Available at: http://ehrafworldcultures.yale.edu/document?id=na08 - 008 [Accessed August 8, 2018]. Vennum, T. 1988 Wild R ice and the Ojibway P eople . Minnesota H istorical S ociety Press , St. Paul, MN. Voegelin, M. and E . Wheeler 1938 Anthropological Records . Berkeley: University of California Press. http://ehrafworldcultures.yale.edu/document?id=ns22 - 001. Ward, H. T. 1985 Social I mplications o f S torage and D isposal P atterns. In Structure and Process in Southeastern Archaeology , ed ited by R. S. Dickens and H. T. Ward , pp. 82 - 101 . University of Alabama Press , Tuscaloosa , AL. Wesson, C. B. 191 1999 Chiefly P ower and F ood S torage in S outheastern North America. World A rchaeology 13:145 - 64. Whallon, R . 2006 Social Networks and Information: Non - - Gatherers. Journal of Anthropological Archaeology 25:259 270. Whelan, C., A. Whitaker, J. Rosenthal, and E. Wohlgemuth 2 013 Hunter - American Antiquity 78(4). Winterhalder, B . 1986 Diet C hoice, R isk, and F ood S haring in a S tochastic E nviron ment. Journal of Anthropological Archaeology 5:369 - 392. Wi nterhalder, B . and C . Goland 1997 An Evolutionary Ecology Perspective on Diet Choice, Risk, and Plant Domestication . In People, Plants, and Landscapes: Studies in Paleoethnobotany , edited by K. J. Gremillion, pp. 123 - 160. The University of Alabama Press, Tuscaloosa. Yarnell, R.A. 1964 Aboriginal Relationships Between Culture and Plant Life in the Upper Great Lakes Region . Anthropological Papers, Museum of Anthropology, University of Michigan 23, Ann Arbor. Yerkes, R. W. 2002 Hopewell Tribes: A Study of Middle Woodland Social Organization in the Ohio Valley. In Archaeology of Tribal Societies , edited by W.A. Parkinson, pp. 227 - 245. International Monographs in Prehistory: Archaeo logical Series 15. Oxbow Books, Ann Arbor , MI.