Lost Worlds:Locating submerged archaeological sites in Southeast Alaska An Archaeological Settlement Model from 10,500 to 16,000 calendar years BP (cal yr before present) Kelly Rose Bale Monteleone NSF Office of Polar Programs # 0703980 & 1108367
Study Area is in SoutheastAlaska’s AlexanderArchipelago, specifically Princeof Wales Island.Haida Gwaii to the south isalso important for thisresearch.These areas consist of theNorthern Northwest Coast ofNorthern America and theNorthwest Coast cultural area
HypothesisThe archaeological record of Southeast Alaskaextends to areas of the continental shelf that weresubmerged by post-Pleistocene sea-level risebetween 16,000 and 10,600 cal BP.
Theory• High Level Theory – Landscape• Middle-Range Theory – Archaeological Settlement Models• Low Level Theory – GIS (Geographic Information System) – Underwater Archaeology
Landscape theory provides a theoretical frameworkwhereby the research focus is appropriate for an area thatis larger than an archeological site. It facilitates analysisat multiple scales, incorporating regional geomorphologyand actualistic studies (e.g. site formation processes andethnoarchaeology) to answer questions regarding land-use, settlement patterns, and other spatially relatedquestions(Anschuetz et al. 2001, Bender 2002, Casey 2008,Kantner 2008, Rossignol and Wandsnider 1992).• Seascapes (Bjerck 2009, McNiven 2008:150,Van de Noort 2003: 405).• Non-sites (Dunnell and Dancey (1983)
• Evolutionary ecology is “the application of natural selection theory to the study of adaptive design in behavior, morphology, and life history” (Cannon and Broughton 2010: 1).• Landscapes are the context in which decisions or behavior choices are made. These choices affect an individuals‟ survival and reproductive successes (Bird and Codding 2008: 396, Johnson 1977: 479, Kantner 2008: 61).
Archaeological Settlement Models• A model can be regarded as a collection of irregular polygons, mapped onto a landscape, indicating locations that are „favorable‟, „likely‟, or „probable‟ to contain an archaeological site of the type being modeled (Kvamme 2006: 27).• Human uses of space can be viewed in terms of a subset of environmental variation. Even culturally determined variability can be mapped using environmental variables, though this must be tested and supported in each case (Kvamme 2006: 14) .
Archaeological Settlement Models as Middle Range Theory1. It is unambiguous. 1. At times of lower sea- levels, people would have2. It provides plausible cause lived on the continental and demonstrable effects shelf. not based on simple 2. Late-Pleistocene/early- correlation. Holocene sea-level rise3. It follows uniformitarian 3. Where people live in assumptions association to coastlines can be uniformitarian in nature4. It is independent of general 4. This theory is independent or high-level theories. of landscape theory(Verhagen and Whitley 2012: 64-47)
Low-level TheoryGIS Underwater Archaeology• GIS is a tool that • Underwater archaeologists researchers use to investigate paleo- investigate, store, analyze, landscapes and other and visualize spatial submerged environments phenomena such as artifact (Flatman 2003, Parker and site distributions. 1999).These are method level theory that each have their own biases andassumptions
This modelThe model incorporates both inductive, utilizingknown archaeological site data, and deductive,utilizing anthropological theory and theethnographic record, types of modeling (Verhagenand Whitley 2012). The scale of measurement forthis analysis are interval or ratio, and both theanalytic (archaeological) and the systemic(dynamic living system) contexts were analyzed todevelop the model (Kohler 1988: 35-37, Schiffer1972).• 10 m resolution for the model – 5 m resolution for the DEM
NWC research questions1) The origins and settlement history of NWC people are important and is specifically relevant to this research.2) There is significant archeological variability in material culture from north to south along the NWC, specifically with respect to the presence and absence of early microblade technology, projectile point types, and baskets and other non-lithic artifacts.3) Issue related to variability within this region is the development of the NWC cultures (Suttles 1990a). This topic includes research into the timing and development of long distance trade, subsistence strategies, and the transition from chipped to ground stone tools (Moss 2004: 185-187).4) The cultural chronologies of the NWC are regionally variable. Different researchers have focused on different aspects of the archaeological and ethnographic records when developing different chronologies (next slide). Fedje and Mackie‟s (2005) chronology is utilized for this research (highlighted in pink).
NWC Chronology Ames andTime 14C Years Davis (1990) Moss (1998) Fedje & Mackie (2005) Maschenr (1999) AD 2000 Late 1000 BP Late Pacific Late Late Developmental 1500 BP 2000 BP Middle The Middle 2500 BP Developmental Middle Pacific Developmental (Marpole) 3000 BP Stage Transitional Early Middle (Locarno 3500 BP Developmental Beach) 4000 BP Early Pacific Early (Charles) 5000 BP Transitional 6000 BP Early Coastal 7000 BP Biface Tradition 8000 BP Paleomarine & Archaic Early The Lithic Stage 9000 BP NWC Microblade 10,000 BP Tradition 11,000 BP
NWC culture has anemphasis on salmonharvesting, permanentvillages or towns, andsocial stratification withhereditary slavery .
NWC culture is also known for theirwoodworking technology, twinedbasketry decorated with falseembroidery or overlay, and basketryhats. Other uniformities through the NWC include a lack of pottery and footwear, uses of plank houses, woodworking technology, and a heavy dependency on fish (especially salmon).
Pre-9000 cal BParchaeological sites• K1 Cave on Haida Gwaii is the oldest (table on next slide)• Namu on the bottom of the map is the only mainland site• 49PET408 and Chuck Lake are the only pre- 9000 cal BP archaeological sites within the study area
Mean ofRegion Site Component Calibrated Age Ranges Ground Hog Bay 2 Lower 11,528 Hidden Falls1 10,157SE Alaska 49 PET 408 (On Your Human Remains1 10,207 Knees Cave) Bone Tool 12,129 Chuck Lake Loc 1 (midden) 9,204 K1 Cave1 12,650 East 9,906 Lyell Bay South 9,483 Echo Bay1 9,916 Richardson1 10,442Haida Gwaii Arrow Creek 21 10,584 Gaadu Din Cave 11 12,683 Gaadu Din Cave 21 12,480 Werner Bay 12,481 Kilgii Gwaii1 10,511BC Mainland Namu (ElSx1) 11,0491 Average of several mean calibrated age ranges
Sea-Level Curve Global Average Haida Gwaii Study Area Haida Gwaii
Map is to orient yourselfbefore videoThe 2 rectangles areShakan Bay in the northand the Gulf of Esiqubel inthe southThere are also pinkishpentagons on the largertownsClips show how much landwas available to pastpeoples that is not easilyavailable forarchaeological survey.
Blue lines are the maximum extent (the last glacial maximum) based on Cararra et al (2003, 2007). Green area is what they describe as refugia, unglaciated areas that supported flora and fauna through the last glacial maximum) Study area was deglaciated by around 14,000 cal BPLegend$ Communites Hessuer 1960 - Glacial Refugia probable late Wisconsin Cordilleran ice marginal position probable pre-late Wisconsin Cordilleran ice marginal position LGM based on Carrara et al. 2003 Possible refugia - Carrara et al. 2007 Modern Alaskan Glacier
Points used to create theDEM using ESRI‟s ArcGISInverse Distance WeighttoolGreen is land. Note thevariability in density ofpoints used to create theDEM at 5 meter resolution.Data was compiled from• NOAA hydrologic surveys• USGS topographic DEM• Multibeam sonar data was purchased from SciFish Inc.
DEM generated for thisproject over NOAA chartwith matching contourlines (solid are NOAA,dotted are DEM).The contours are similar.Differences are inlocations where there isno NOAA data point.The NOAA data pointswere included in thegenerated DEM.
A) Multibeam datacollected for Shakan Bayin 2012.B) Difference betweenDEM and multibeam at10 m resolution.
Aspect SlopeThese are important variables when locating a settlement (or a camp site).
Water FeaturesStreams were generatedby Andrew Wickert, agraduate student at UCBoulder using GRASS 7.0.Lakes were created usingArcGIS’s basin fill andrepresent depressionsthat were likely wetland,marshes, or lakes.Color dots arearchaeological sites.
Each variable was buffered at 50, 100, 1500, 2000, and 3000 meters to create ranked locations based on distance to resources.Streams Lakes Tributary Junctions
Sinuosity 3 km Sinuosity values were classified based on statistical analysis of archaeological site locations .3 km bufferL = Length along the coastLd = Linear They were then buffereddistance following the same method as the water variables.
Percentages used in ArcGISweighted overlay to createhigh potential maps.
Several models werecreated and tested todetermine which wasmost effective.Orange is ModeratelyHigh Potential and redis high potential forarchaeological sites.
The model (and all thepreceding variables) weregenerated in 500 yearintervals from 10,500 to16,000 cal yr BP.To create a final result, theresults were mergedproducing the maps onthe right.
Kvamme’s gain is a ratio based statistic used to evaluate archaeological predictive models. The values range from -1 to 1. A value greater than 0.5 or 50% indicates a positive gain or a useful model. A value between 0 and 0.5 has no predictive utility. A negative value means the low potential areas are more likely to produce archaeological sites than the high potential areas or a reverse gain. Gain Predictive Utility Model Data Set Statistic (gain)Known sites are the Weight 8archaeological sites used Known sites 0.9446 Positivefor the model generation. 3 2012 survey 0.9053 Positive Random locations 0.2270 None2012 survey is 9 locations Known sites 0.9967 PositiveGPS within the study area, 4 2012 survey - Noneonly half of these are new. Random locations 0.8542 PositiveThe focus of this survey Known sites 0.9479 Positivewas on 7000 to 10,500 3+4 2012 survey 0.9049 Positiveyear old sites and does not Random locations 0.2401 Nonefit with this model.1000 random locations were also tested. The results have a positive gain value.
Moran‟s IMoran’s I tests spatial autocorrelation. Thehypothesis is that there is no spatialcorrelation in the data. The differencebetween The Global Morans I or spatialautocorrelation tool, measures spatialautocorrelation based on feature locations andfeature values simultaneously. Given a set offeatures and an associated attribute (value 3 or4), the tool evaluates whether the patternexpressed is clustered, dispersed, or random.Moran’s I was calculated at 250, 500, 1000,1500, 2000, and 3000 m bands. There aresome missing values for the 3000 m band. Thisis the same “memory error” that wasencountered at the larger scale and is likelydue to the complexity of the polygonscompared at 3000 m. The results indicate thatthe models are clustered. The need for 3000 mdistance band indicates that the data is not asclustered as the results indicated since somepolygons did not have any neighbors within2000 m.
2 Surveys2010 – Side scan sonar2012 – Multibeam sonar & sub-bottom profilingSediment samples and ROVvideo were collected both years
Density of multibeam points per 1m2. Purple is less 80 pts (range form 0 – 401 pt / m).The density of points is too low to locate the intended archaeological targets.
2010 Side Scan Anomalies 4 – two rectangular features 5 - shipwreck 3 – possible weir
Side-scan image of shipwreck.Based on location, it isassumed to be the Restless.
A fish weir is a lowstone or wood wallthat traps fishbehind it when tideis out. This meansdinner is easy tocatch.
Raised semi-circular features and two depressions. Image on the left is the originalsonar and on the right is the sonar image with the anomalies depicted in white .Possible weir structure.
Location in ShakanBay of anomalyreconstructed to52 m, the depth ofthe anomaly.
Zoomed to area of sonar anomaly (now looking from the north). Notethe bay to the west; this would be a good location for a settlement orcamp.
Topographic locations of DanishMesolithic settlements based onfishing model (Fisher 1995: 374,2004:32, Benjamin 2010: 257). A)Narrow islet connecting largebodies of water. B) Between asmall island and mainland. C) andD) At the tip of a headland. E andF) at the mouth of a streamLocation in Shakan Bay is anexample of diagram A.
This is the same reconstruction with the possible weir locationreconstructed from the sonar.Nothing was located in 2012 using the multibeam sonar or remotelyoperated vehicle using video.
Sub-bottom from Shakan Bay indicate possible river channel.Sub-bottom imagefrom Shakan Bay witha depression similarto archaeologicalpits.
Cut wood from VV25-20 at Shakan Bay Anomaly three. Radiocarbon dating returned a “modern” result.ROV image of stick picked up atShakan Bay anomaly three. Piece of rounded wood recovered from Shakan Bay seven (VV-26-06). Natural piece of wood.
Discussion• Third iteration of the model – Archaeological site location modeling is an iterative process.• The multibeam survey of Shakan Bay identified unknown information about the geology for the bay and region. Along the western side of the surveyed area is a fault ridge, a large raised mound that is present as a linear feature in the multibeam data.• Model Resolution – 5 m, 10 m, 50 m – 5 m did not improve Gain values and 50 m was not useful – (some archaeological models are produced at over 1 km which is too low a resolution)
Discussion• Implications for SE Alaska and Northern NWC • Model can extend chronology • It would support local oral traditions of local antiquity• Coastal Migration Hypothesis and the First Americans • Locating a submerged archaeological site will provide support for the coastal migration of the First Americans to the New World– No confirmed archaeological sites have been located at this time
Photo: Forest Haven (Sealaska Hertigate Institute Intern and Tsmisian Native) andKelly Monteleone wet screening samples in Shakan Bay.
Acknowledgements• NSF – Office of Polar Programs – NSF award # 0703980 & 1108367• Maxwell Museum of Anthropology, UNM• University of New Mexico (UNM) and University of Colorado (INSTAAR)• Sealaska Heritage Foundation• Residents of Southeast Alaska• Dissertation Committee• E. James Dixon, Andrew Wickert, Mark Williams, Amalia Kenward, Michael Grooms, Travis Shinabarger, Jason Brown, Lee Drake, Nick Jarman, William Taylor