Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Archaeological landscapes of the Lower Mainland, British Columbia: a settlement study using a Geographic… Beattie, William Grant 1996

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1996-0089.pdf [ 3.75MB ]
JSON: 831-1.0058369.json
JSON-LD: 831-1.0058369-ld.json
RDF/XML (Pretty): 831-1.0058369-rdf.xml
RDF/JSON: 831-1.0058369-rdf.json
Turtle: 831-1.0058369-turtle.txt
N-Triples: 831-1.0058369-rdf-ntriples.txt
Original Record: 831-1.0058369-source.json
Full Text

Full Text

ARCHAEOLOGICAL LANDSCAPES OF THE LOWER MAINLAND, BRITISH COLUMBIA A Settlement Study using a Geographic Information System by WILLIAM GRANT BEATTIE B.A., The University of British Columbia, 1990 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ARTS in THE FACULTY OF GRADUATE STUDIES (Department of Anthropology and Sociology) We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA December 1995 © William Grant Beattie, 1995 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver, Canada DE-6 (2/88) ABSTRACT A selection of archaeological sites located in the Lower Mainland area of coastal British Columbia is used to test the use of the landscape in evaluating certain statements about prehistoric human settlement for the region. In total, 62 sites were chosen which show at least some evidence of occupation or limited activity use during the Marpole period (2400-1400 BP). The main problem in describing settlement strategies for this area is the inability to infer site use from subsistence or lithic assemblages recovered through either excavation or surface collection. This lack of information for the majority of the sites makes it impossible to use such traditional approaches to settlement analysis as determining sociopolitical relationships between locations or describing a seasonal round or movement from one type of site to another. An alternative approach to circumventing this dilemma is to use the landscape in the immediate vicinity of each site as a source of new data that can be used to determine some of the criteria involved in making the settlement choice. A Geographic Information System was used to gather information about the terrain, both generally for the region and near the sites and then comparing this local profile with the other sites across the region. This information is presented in a series of tabular and statistical displays that allow evaluations, not only about settlement choices inherent in the collection of sites but also about the efficacy of the landscape approach in archaeological research. The non-randomness of site location decisions is demonstrated along with evidence that certain landscape features were favoured over others. It is also apparent that a preliminary site typology can be affirmed as a result of certain types of sites selecting for specific terrain elements. iii TABLE OF CONTENTS Abstract ii Table of Contents iii List of Tables iv List of Figures v Acknowledgement vii Introduction 1 Settlement and Landscape 1 Landscape Analysis and Geographic Information Systems 5 The Study Area 9 Cultural Background 9 Methods Model 13 Hypotheses 13 Data Sources 14 Data Summary and Analysis 16 The Variables 18 Clustering and Multidimensional Scaling 32 Summary 43 Conclusion 44 Bibliography 46 Appendices A. Site Database Record Form 56 B. Sites used in the landscape study 57 C. Variable Measurements 59 D. Vegetation Measurements 61 E. Soil polygons with soil type and area 63 F. Vegetation Codes 65 G. Soil Codes 67 LIST OF TABLES 1. Comparative Statistics - Actual Sites and Random Points (Distance measurements) 2. Comparative Statistics - Actual and Random Sites (ELEVATION & SLOPE) 3. Topographic and Vegetation Data resequenced LIST OF FIGURES 1. Map of study area 2 2. Storage of spatial information in a GIS 6 3. Storing data layers in a GIS 7 4. Raster vs. Vector in a GIS 8 5. NEAREST SITE 19 6. SHORE 20 7. STREAM 20 8. Vegetation Polygons intersected at DgRr 1 21 9. Vegetation area comparisons 22 10. Vegetation profile including intersected water 23 11. Site size by Vegetation type 24 12. Soil Area comparisons 25 13. Site size by soil type 26 14. ELEVATION and SLOPE 27 15. ELEVATION values with "flat" sites removed 27 16. Normal Probability Plot for actual ELEVATION values 28 17. SLOPE values with "flat" sites removed 28 18. Comparison of ASPECT values 29 19. Chi-square statistic for ASPECT 30 20. DEPTH in back of actual sites and random locations 31 21. BREADTH in front of actual sites and random locations 31 22. Normal Probability Plots for BREADTH and DEPTH 31 23. Cluster diagram of the topographic variables 35 vi 24. Multidimensional scaling plot of the topographic variables 36 25. Cluster diagram of the vegetation variables 38 26. Multidimensional scaling plot of the vegetation variables 39 27. Cluster diagram of the combined variables 40 28. Multidimensional scaling plot of the combined variables 42 ACKNOWLEDGEMENT R.G. Matson inspired this research many years ago when he commented that archaeological sites in this part of the world often seemed to be near the base of steeply rising headlands. This sparked my interest in looking at the landscape as a lasting set of clues to ancient human occupation. To him and to Prof. David Pokotylo as my other thesis committee member I owe a great debt, not only for their guidance but also for their patience during this overlong process. I wish to also thank Kathryn Bernick for pointing me towards the area surrounding the mouth of the Fraser River as an area of study. Her knowledge of the region and her subsequent help with new data on sites were a great benefit. I am greatful as well to Morley Edridge of Millenia Research for providing me with up-to-the-minute material from his recent survey work oh the Fraser. I relied extensively on the Department of Geography and a group of people there who helped me with computer access and the navigation of the confusing byways of Geographic Information Systems. Prof. Brian Klinkenberg was most helpful in arranging my use of the comuter lab facilities and obtaining data for the research. Graduate Student Catherine Griffiths was most gracious with her time in answering my many banal questions. Vincent Kujala was always available to help me with system problems. My colleagues in the Department of Anthropology and Sociology were very supportive over the years. I want to mention particularly fellow classmates Andrew Mason and Vicki Feddema. Doug Brown and Warren Hill were a great help at a particularly bleak time at the end of the project. Like most archaeology students here at UBC, I too have many fond memories of the help and the kindness of Joyce Johnson. In the course of this research I have felt the loss of both my parents, Robert Graham and Katherine Grace Beattie. My thoughts were often of them as I tried to bring this work to completion. And finally I thank my dearest friend and companion, Lalya Wickremasinghe. Her presence in my life through all these years has been my inspiration and my source of strength. 1 Introduction This study uses a Geographic Information System to extract information from the physical landscape surrounding archaeological sites. The assertion here is that by making a series of measurements of the way the local landscape is connected to archaeological locations, statements can be made about how people came to choose these particular spots for living and working. A region surrounding the mouth of the Fraser River in southwestern British Columbia (see Figure 1) provides a test case for this proposition. Within this area, a set of sixty-two archaeological sites has been selected which show evidence of human activity during the Marpole period (2400 to 1400 BP). Since the main research focus is on landscape and not specifically on Marpole, the settlement questions addressed here have been prompted by the overall study of subsistence and settlement and, therefore, remain general and preliminary. Three basic hypotheses guide the research. First, landscape data should show that the sites are not randomly distributed over the terrain. Secondly, specific landforms will be located close to sites because they are desirable features to be near; and thirdly, distinctions will be possible between types of sites based on greater or lesser diversity of the local landscape. The Geographic Information System, (GIS) has been used here to generate the large amount of data necessary for a meaningful analysis. A major problem in an approach such as this is the difficulty in acquiring the information using traditional methods. Individual site reports, produced by almost as many authors as there are sites, cannot provide a consistent or comprehensive collection of data. Many measurements used in this present research were never a part of the original field documentation. Trying to extract this material manually from paper maps would be far too time-consuming, inaccurate, and in many cases impossible. A short background examination of approaches to settlement patterns in archaeology begins this discussion. Next, it is shown how landscape and a GIS can generate new information. This is followed by a description of the study area and the methods used to examine it. The data that came out of the study are then examined and finally the results are summarized. Settlement and Landscape Settlement pattern studies can be generally lumped into two broad approaches. The first takes a socio-political view and looks at individual buildings and settlements and then attempts to link these localised phenomena across a broader territory or region (Trigger 1968). These approaches often attempt to look at the nature and use of 2 Figure I: Map of the Study Area with site locations buildings, religious activities, relationships to agriculture, as well as broader city-town-village-hamlet-farm connections. Classic efforts of this type in the Americas include Gordon Willey's Viru Valley study (1953), his work in the Maya Lowlands (1989), and the large scale projects in the Oaxaca Valley (Blanton et al. 1981) and the Valley of Mexico (Sanders etal. 1979). The second approach, embedded in the cultural ecology paradigm, looks at relationships between human actors and various environmental elements to which they have adapted. A cycle of economic activity, often in the form of a seasonal round, is deemed to be the main determining factor behind why particular locations are chosen over others (Dewar & McBride 1992:228). Subsistence analysis and the remains of various technologies used to exploit the resources provide the data used to describe these patterns. Optimal foraging (Winterhalder & Smith 1981), catchment area (Vita-Finzi & Higgs 1970; Flannery 1976), and mobility strategy studies (Binford 1980, 1982; Chatters 1986) would all fit this category. Several projects in the American Southwest would come under this approach (Plog 1971, 1978; Plog & Hill 1971; Euler & Chandler 1978; Matson et al. 1988). Most forays into settlement description for the Northwest Coast and interior British Columbia follow these methods as well (Pokotylo 1978; Thompson 1978; Matson etal. 1979; Pomeroy 1980; Hobler 1983; Lepofsky 1985). Both approaches rely heavily on knowledge of the contents of sites which they attempt to place into the settlement system, mainly as a means of determining site type and the related activities performed there. "Archaeologists must begin their analyses on materials remaining at archaeological sites" (Binford 1982:5). This reliance is fine in an ideal world of unlimited time and money but in reality it often means using a very small number of sites to explain very complex living systems. One way around this dilemma has been to suggest a greater reliance on surface remains (Schlanger 1992; Ebert 1992; Camilli & Ebert 1992). However, this requires a particular setting with open, sparsely populated, arid areas which is not possible in most situations in the Pacific Northwest where the very extent of known sites is often impossible to determine because of extensive plant cover or partial destruction from modern development. These schemes have, in fact, been accused of following a simple functionalist perspective that selects for areas of open and obvious prehistoric evidence (Wheatley 1993:137). Even when sites are excavated, the researcher's particular agenda may not involve a search for information that would place the particular site into a settlement scheme. Trigger (1968) may place dwelling structures at the top of his list of insights into settlement but this does not help in a region such as the one being looked at here. After almost 50 years of archaeological research in the Lower Mainland and with records on over 200 sites, we have a 4 grand total of one probable house structure having been excavated and that single find was made purely by chance during a project with a completely different research focus (Matson & Coupland 1995:161; Matson etal. 1991). To these problems we can add the factors of a region presently covered by the third largest urban concentration in Canada. What sites we do know of are almost entirely the results of serendipitous discovery and any future survey would be hampered by the presence of hundreds of kilometers of concrete, asphalt, landfills, sea and river dykes, and modern housing. If traditional ways of studying prehistoric land use place impossible demands on the archaeological record in most areas and have limited application in a coastal rainforest environment, it may seem folly to even attempt such a study. This is a depressing notion given that the Fraser is recognized as "one of the greatest producers of salmon in the world" (Kew 1992:178) and at the very heart of the so-called Developed Northwest Coast ethnographic pattern (Matson & Coupland 1995). One potential way around some of these problems is to approach the subject from a new perspective, that of landscape. Despite the busy efforts of over 150 years of Euro-Canadian occupation, post-contact activity has still not managed to entirely change every shoreline, obliterate every freshwater stream or flatten every hillside and headland. These elements still remain as sources of information on the sorts of terrain and environment that prehistoric users of the area adapted to. The use of landscape here as a means of focusing research and grouping various types of data requires definition. In much of the literature on settlement landscape has acted as little more than a synonym for similar terms. In one of the pioneering attempts at using the terrain surrounding sites as a source of information it is referred to as the "natural environment" (Plog 1971:47). Other terms include "settlement space" (Wood 1978), an area of "economic zonation" (Binford 1982), a site's micro-, meso-, and macro- environments (Butzer 1982:39), "off-site archaeology" (Foley 1981) and so on. Recently an entire volume devoted to its use as a concept in archaeological research (Rossignol 1992; Rossignol & Wandsnider 1992) simply employs the term without actually defining it, but using it as a source of "ecological and geological system variables" (p.4). In a somewhat puzzling discussion, one writer sees it as a term "surprisingly difficult to define" (Roberts 1987:78) and then vaguely refers to it as "the assemblages of real-world features - natural, semi-natural, and wholly artificial" (ibid. 79). Finally, from a post-processualist perspective, landscape is a story: "the world as it is known to those who dwell therein, who inhabit its places and journey along the paths connecting them" (Ingold 1993:156). In this study landscape is defined on a slightly more practical basis: a heterogeneous land area containing a number of important visible elements that influenced humans in choosing to interrupt their movement over the earth's surface and to persist in that place long enough to leave a material record. This definition utilizes ideas taken from geography as well as landscape ecology. Landscape is first of all something we can see with our eyes, beginning as a source of information and then as a familiar set of information for the human viewer. A standard distinction that is made is between natural and cultural landscapes (Haggett 1965:11; Naveh & Lieberman 1994:11; McGarigal & Marks 1993) with the cultural landscape further partitioned as managed (eg. a pasture), cultivated (agriculture), suburban (town and country mixed), and urban (cities) (Forman and Godron 1986:286). Under this distinction modern-day Vancouver comprises almost entirely a cultural landscape with only the larger landforms and subsurface geology still unchanged. For the work described here the landscape is regarded for the most part as a natural one, with the archaeological site representing only a small remnant of the cultural landscape that it once would have formed. In other words, the attempt here is try and view the landscape as the first settlers might have, albeit in a very simplified form. By extracting a small number of elements from this complex aggregate it is hoped that some fundamental statements about the prehistoric use of the landscape can be made. What is largely omitted here are the usual components of a settlement system analysis - the resources, the technologies, and the costs of operating within the system (time and distance involved in acquiring resources). This exclusion is generally because of the non-existence of these sorts of information for the majority of Lower Mainland sites. Landscape Analysis and Geographic Information Systems As noted above, the use of the surrounding landscape as a source of quantifiable information first achieved a highpoint in the early 1970's with the work of the Southwest Archaeological Group (SARG). Starting with the question "Why did the prehistoric inhabitants of the Southwest locate sites where they did?" (Plog 1971:45), their goal was to explain the distribution of archaeological sites using variables taken from both the natural and social landscapes. These would include water and soil resources, plant and animal communities, and various measured social distances between sites (Plog 1971:47-48). While these pioneering efforts made great strides in developing fieldwork methodologies (Plog & Hill 1971) and survey techniques (Mueller 1975), the ambitiousness and complexities of the problems they set for themselves may have proved too overwhelming for subsequent workers. 6 Their results often tend to close on a note of "more work is needed" (Plog 1978; Euler & Chandler 1978) and something of a hiatus followed. Since then there have been a few small-scale attempts at integrating landscape variables into research (Schermer & Tiffany 1985; Kellogg 1987; Snyder 1991; Ellis & Waters 1991) plus one large study in a European setting that looked at complex Iron Age spatial patterning (Marquardt & Crumley 1987). Most recently, however, there has been a strong renewal of interest, largely as a result of new computer technology known as Geographic Information Systems. These systems can be roughly characterised as putting printed map data into a computer where it can be subsequently viewed and manipulated. What they in fact entail, however, is something far more complex than the simple scanned-in images of a drawing or desktop publishing program. A GIS retains information on how various map parameters, (Jines, points, text, and shapes), are defined in space (coordinates) and how they relate to each other. This information is in turn linked to attribute information stored in a separate database or data bases. Because the system operates as a relational database, all of this information can be brought together as standard map images either on a computer monitor's screen or as plotted paper output or it can be selected or manipulated in various different or new forms. Querying capabilities allow for a wide range of statistical and summary outputs. Figure 2 illustrates how a simple square would be stored in a GIS. A(x!,y!) B(x2,y2) SQUARE 1 C(x 3,y 3) D(x4,y4) Figure 2: Storage of spatial information in a GIS. Each of the four points A,B,C,D can be located in real-world terms (eg. as UTM coordinates); they form four distances (AB, AC, BD, CD) which in turn form an areal shape ("SQUARE 1") with its own label. Because each line is stored in terms of its relation both to the outside as well as the inside of the figure, a query could provide details on the figure's neighbours as well as its area and perimeter. From an attribute database special information can be stored. If this were a building lot in a municipal setting, for example, its assessed value and address might be external variables. As mentioned, the figure could be displayed on a screen or sent out to a printing device (Aronoff 1989). The real power of a GIS is in the way it can store multiple sets of data in layers, all of which can be related to each other thereby creating new forms of data. Figure 3 is a simplified illustration of how water, elevation and cultural intrusions are represented by separate layers. A more sophisticated query, using all three layers could provide the elevation for the building plus its distance to the nearest stream, for example. Figure 3: Storing data by layers in a GIS (from Gaffney & Stancic 1991:25). A final aspect of how a GIS works is the way it actually encodes the data. This can take two forms, vector and raster. In a vector-based situation, the data are stored much as they are shown in Figure 3, as a series of points, lines, and text identifiers, all linked within a coordinate system. In a raster system the entire area of the map is defined in terms of individual pixels which vary in size according to the fineness of grain required. Figure 4 shows a comparison of the two methods. (In this example three landscape features, water, plants and buildings, are shown as being in the same layer). Some systems are based entirely on one or the other of these two storage methods. Others, such as the system used in this study, TerraSoft, employ both. Although raster-style storage is very useful for areal based analysis, it can require tremendous amounts of storage to maintain all of its information. 8 THE RASTER AND VECTOR DATA MODELS 1 1 1 1 1 1 1 1 1 1 1 <00 200 300 400 500 X -AX IS B. RASTER REPRESENTATION C. VECTOR REPRESENTATION Figure 4: Raster vs Vector in a GIS (from Aronoff 1989:164). GIS has found many uses in municipal settings and among resource applications, especially forestry. Whether or not it merits such kudos as "the biggest step forward in the handling of geographic information since the invention of the map" (Gaffney & Stancic 1991:13) remains to be seen. Much of GIS use so far by archaeologists has been partly driven by the novelty of it. Some projects have not involved much more than developing a familiarity with the systems and their capabilities (Green 1990; van Leusen 1993; Ruggles et al. 1993). Others have attempted to apply GIS to some of the earlier forms of settlement research such as catchment area analysis (Hunt 1992) and predictive modeling (Maschner & Stine 1995). Attempts have been made to carry out the sorts of analyses that were suggested by the people involved in the SARG projects. These involve trying to describe optimal locations on the basis of comparisons with soil quality (Gaffney & Stancic 1991) as well as aspects, distances to water, and separation between sites over the territory (Brandt et al. 1992; Kvamme & Jochim 1990; Kvamme 1992; Savage 1990). A collection of essays and reports from 1990, Interpreting Space: GIS and archaeology (Allen et al.), acts as an early signpost to these developments. 9 The Study Area This area (see Figure 1) has been and continues to be dominated by the Fraser River, one of the great rivers of the world. At present, with a length of 1370 km, it drains over one-quarter of the province of British Columbia (Clague etal. 1983:1314). For the last 9000 years, following the opening of its basin after the last glaciation (11,000 years ago), the Fraser River has been building its delta (presently about 1000 square kilometers in area) out into the Strait of Georgia, currently at a rate of between 12 and 30 million tons of sediment per year (Williams & Roberts 1989:1658). Sea levels have been stable, fluctuating no more than 2 m in the last 5000 years (Clague etal. 1983:1325). In the last 2300 years the delta has extended itself about 5.4 km (2.4 m per year) (Williams & Roberts 1989:1665), about half the length of Lulu Island and all of Sea Island (see Figure 1). Boundary Bay would not have been affected by this recent growth since there is no indication of the Fraser entering it in the last 5000 years (Clague et al. 1983:1325). One consequence of this delta growth may have been the abandonment within the last one thousand years of two of the major archaeological sites in the area, the Marpole site (DhRs 1) and the Glenrose Cannery site (DgRr 6) (Burley 1979:557, 1980:58). The Fraser River historically had and continues to have the status as "the greatest producer of salmonine fishes of any single large river in the world" with an annual production of about 14 million salmon (Northcote & Larkin 1989:172). Cultural Background We have evidence of human use of the mouth of the Fraser dating back 8500 years BP at the Glenrose Cannery site, DgRr 6 (Matson 1976) but the principal focus here is on the occupation from 2400 to about 1400 years ago, the so-called Marpole period. There is some overlap with the subsequent Late or Strait of Georgia period which dates from about 1400 BP to 200 BP. The culture associated with Marpole was first described by Charles Borden based on materials recovered from a large midden site within the Vancouver city limits - DhRs 1 (Borden 1950, 1970). This work was further refined by Mitchell (1971) and Burley (1979,1980). It has been described as displaying the Developed Northwest Coast Pattern of winter villages with large plank houses, dependence on stored fish, usually salmon, abundant sophisticated art, and ascribed status, as well as seasonal moves to specialized resource procurement locations (Matson & Coupland 1995:224). 10 Archaeological research on Marpole sites in this area beganwith excavation at the Marpole midden at the turn of the century (Hill-Tout 1895; Smith 1903) but modern approaches and techniques only appear after the Second World War, work that for the period 1946 to 1970 is embodied in the efforts of one man, Charles Borden. Most of the sites used in this study were initially visited and reported on by Borden and many of them were in fact excavated by him. Following his efforts, archaeology in the Lower Mainland has been conducted in three main areas. Academic research, sometimes inspired by impending site destruction, has been carried out at the Marpole site (DhRs 1) (Burley 1979), Glenrose Cannery (DgRr 6) (Matson 1976, 1981), St. Mungo Cannery (DgRr 2) (Calvert 1970), Beach Grove (DgRs 1) (Matson et al. 1980), Point Grey (DhRt 5) (Coupland 1991), and Crescent Beach (DgRr 1) (Ham 1982; Matson et al. 1991). Salvage activity has been conducted throughout the area, often producing very important and, in some cases, the only documentation (Archer 1972; Bernick 1989; Ham et al. 1986; Grabert & Spear 1976; Areas Consulting 1991). As well there have been a number of surveys of the area. None of these has been of the probabilistic variety and most have been conducted for various heritage conservation and monitoring purposes (Kenyon 1953; Kidd 1964; Cranny & Bunyan 1975; Bussey 1985; Ham 1987, 1988; Ham et al. 1978; Bernick 1991; Eldridge & Mackie 1993b). In essence then, the archaeological record for the mouth of the Fraser is by and large the product of serendipitous discovery and it has rarely been on any particular investigator's agenda to relate particular sites to the specific settlement of the Fraser Delta region. The usual result is to assign a type of activity to the site which in turn is linked to the broader pattern of the seasonal round. As Matson and Coupland point out "Northwest Coast archaeology has traditionally been 'single site oriented'" (1995:313) and Thompson's 1978 work on the Skagit River delta of Northwestern Washington stands as the only attempt at describing settlement patterns in the Gulf of Georgia area. The last source of information reviewed here is the ethnographic research conducted in this area. The Mouth of the Fraser is the traditional territory of the Central Coast Salish people, particularly the Downriver Halkomelem speakers of the Lower Mainland (Suttles 1990). They include the Musqueam in the Vancouver city area, the Tsawwassen to the south at Point Roberts, the Kwantlen and Katzie further inland on the Fraser, and the Nikomekl in the area of the South Fraser extending down into the United States. In addition to the Halkomelem, Squamish speakers to the immediate north have had regular influence over Burrard Inlet. The region has seen a 11 number of ethnographic investigations carried out, including major ones by Suttles (1951, 1955, 1987), Barnett (1955), and Duff (1952). The focus on settlement and its relationship to the surrounding terrain, however, remains sketchy. Barnett makes a few general points: Villages were located near the mouths of rivers on sheltered bays or inlets out of reach of storms and marauding strangers .... interest centered upon beach sites conveniently located with respect to gathering and hunting grounds .... the village groups occupying simultaneously or in turn several traditionally assigned spots for their hunting, gathering and wintering activities. The winter villages are always to be regarded as the foci of these pursuits. (1955:18-19). Suttles is more direct when he describes the shore as a critical element in the Salish landscape, "the midline in the life of man" (1951:46). A few other small hints can be culled from these works. In the summer the local soil was not a very important consideration for dwelling structures but in winter solid supportive soils were needed for the big plank houses (Suttles 1951:46). A prairie was important to have close to a winter village for camas and other bulbs (Suttles 1951:59). Other "nice to have" features included mudflats, eel grass beds, and protection from northeast and southwest winds. Fishing locations and root and clam beds were, in many instances owned by certain family groups (Suttles 1951:56). Cedar was critical for both its wood and its bark for use in building, basketry, netting, and a variety of other household items while Douglas fir was employed apparently only as a fuel (especially its bark) (Suttles 1951:48). Both Barnett and Suttles pretty much avoid the topics of sewage and garbage disposal and the subsequent possibility for local pollution. Suttles makes one minute mention of a location translated as "feces net", stating that "the name comes from the fact that the beach was used as a latrine" (1951:207). He does not expand on this further so we can only speculate that this might have been the standard sewage disposal method. Barnett makes one mention of ritual disposal of animal and salmon bones (1939). Leonard Ham, citing an American ethnographic example from Oregon, says the shell refuse was always placed in specific locations around permanent settlements and, using the Beach Grove site (DgRs 1), speculates that these may have been between and behind the houses (1982:160-162). In his view the beach/shoreline represents a critical interface for daily activities. In another archaeological setting further to the south on San Juan island, the habitation area was on the highest, best drained part of the site and the refuse was dumped into the intertidal zone adjacent to it (Whittaker & Stein 1992). 12 The last issue to be discussed here is the apparent regular movement from location to location. "Nature is exploited by the intensive use of specific places at specific times in the year... the yearly round is fairly rigidly determined" (Suttles 1951:50); or as an archaeologist has put it "We all know the people moved around a lot and that they did so as a means of acquiring various resources from various places during the year" (Mitchell 1983:97). It is important to emphasize that Suttles was describing the lifeway of a particular group of Salish who were located in more of an island setting (the Haro and Rosario Straits south of the area being looked at in this study). In the case of the Lower Mainland, researchers are not so sure. There is a hint of a less mobile pattern in the above quote from Barnett of simultaneous occupation, ie. not seasonal abandonment; plus Mitchell states that "Most of the Musqueam people remained in residence at their principle villages ... from fall to spring. Some villagers seem also to have stayed there through the summer." (1983:101). Finally, the travel distances involved here do not necessitate the wholesale movement of people from one resource area to another. Ethnographic records of canoe travel suggest that no location in the study area would be more than a single day's journey (Croes & Hackenberger 1988:35; Sproat 1868[1987]). The sense from this is that many sites may not have had to accommodate most or all of a village on a particular visit. Short term usage by smaller numbers may have been a more typical pattern. What this says in terms of site structure is that the larger village sites would have had extensive year-round use while more limited activity locations would have seen more of camp or work station type visits. This problem of mobility and the image it presents of land use has recently led to renewed warnings on the way the ethnographies are used. Ford cautions that, even with good data from the site, we cannot easily determine a seasonal use or abandonment (1989). Bernick and Wigen present a similar case where a focus on one resource (Chum salmon) may have skewed the entire sense of how a particular site was used (1990). One study where the author had good data on site locations as well as prime salmon harvesting areas could not conclusively link the two (Hobler 1983). Finally, some of the mobility that was described by Suttles and Mitchell may have been the result of the movement of European colonisers into the region. A recent investigation of Kwakwaka'akw post-contact settlement shows what a radical impact it had on a stable sedentary lifeway (Galois 1994). A study on the Dena'ina of Alaska also relates how disease, white settlement, and the introduction of the trading post as a new economic focus greatly disrupted the old settlement structure (Ingold 1992:792). 13 One implication from all of this is that the presence of a big and important resource (salmon) may have determined the overall general location of people in the mouth of the Fraser region, but this initial motivation is superceded at some point by the choice of specific locations based on optimal landscape elements. Methods Model The model employed here states that site location is the outcome of a choice made by actors operating within the settlement system (i.e., it is not a random point on the landscape). We gain information about that choice by attempting to describe the settlement space (Wood 1978). This can be done by first determining the values for a select set of environmental and topographic variables related to those site locations and then comparing these values to see what patterns (if any) exist among them. A comparable set of random points is used as a check on such patterns. The objective here is to relate fundamental landscape information to settlement choice and from that information develop basic descriptions of Marpole period land use for the mouth of the Fraser region. This is somewhat in contrast to much of the current activity in settlement pattern studies which involve site prediction, a harkening back to the enthusiasm for predictive modeling in the 1970's and early 80's (Williams et al. 1973; Carr 1985; Parker 1985). Some of this renewed interest is directly related to the powerful data capturing abilities of GIS computer software plus demands being placed on cultural resource managers by governments and, in turn, major resource companies (Warren 1990; Carmichael 1990; Altschul 1990; Eldridge & Mackie 1993a). For this reason, this study does not provide a detailed profile of the entire landscape itself. It is used instead as a source of new variables and new attributes that can add to our knowledge of archaeological sites. (For an example of trying to describe the broader landscape in terms of chunks of empty, non-site spaces to be compared with chunks containing sites, see Maschner& Stine 1995). Hypotheses A number of interesting problems specific to Marpole settlement have been raised. Ham (1982:359) states that "all of the major archaeological sites in this area are either multicomponent, or consist of a complex of sites from different time periods in close proximity to each other". In contrast, Croes and Hackenberger (1988) felt that a shift had taken place during this period away from marine shores to more riverine and estuarine areas. Thompson (1978) saw the adaptation as still fairly diffuse during Marpole with, as yet, no particular focus on any resource or geographic region. As already mentioned in the Introduction, however, it was felt that in this study, with the landscape being primary, questions should be of a more "generic" nature. Three hypotheses have been formulated to guide the research process. The first looks at the issue of randomness. A comparison of the landscape variables will demonstrate that archaeological sites in this study are not randomly arrayed across the region. This basically involves comparing variable values at both actual and random locations. In some cases, the overall background is looked at, particularly for the two sets of environmental variables, vegetation and soils. The second hypothesis looks at local topography for clues as to differences and similarities among the sites. Since similar solutions were used to overcome similar topographic restrictions, some variables will show that sites are found on or near certain landforms throughout the study area, specifically on or near the lower levels of a rising headland. This primarily involves looking at variations in elevation within the immediate site area. As well, since a freshwater stream and distance to a major shoreline are considered critical, these distances are measured as well. Again comparisons are made with the set of random points. A most basic attempt at a site typology is made using these data, trying to distinguish extended use habitation sites and more limited use sites. Habitation sites, deemed to be at "the center of exploitative activities in general" (Plog & Hill 1971:13), will display greater diversity of environmental and topographic variables. This again involves a comparison between actual site locations and random coordinates. The presence or character of certain variable values should show habitation sites sharing specific attributes in a combination not found at other site locations. Data Sources Sites: Within the study area, information on over 180 prehistoric sites was reviewed for possible inclusion in an inventory of Marpole occupation and use locations. (See Appendix A for a sample of the form used in this selection). The sites were chosen from this period because they are the the best represented in the archaeological record for the Lower Mainland. A small sample, for example using the preceeding St. Mungo or Locarno Beach 15 period sites, would not have provided a sufficient test case for the landscape analysis. This review involved checking Archaeological Site Inventory Forms completed by investigators in the region over the past fifty years. These forms are part of the initial site designation process required by the Provincial Government (Apland & Kenny 1989). Wherever possible other sources of information were also used. These included university undergraduate reports and graduate theses, consultant reports, survey reports, field notes, journal articles, and books. From this material, 62 sites were chosen as being either definitely in use during the Marpole period or highly likely to have been so. This number is in rather marked contrast to the usual interpretation for this region. Out of the 40 sites that Matson and Coupland are willing to designate as having a Marpole component (1995:Table 7-1, p.202) only 17 are in the area reviewed here. Some sites included here would fit better in the subsequent Late period but owing to the problems of dating the end of Marpole, it was thought that their inclusion here was justified (Matson & Coupland 1995:218). Other sites which we know very little about are included because of the presence of diagnostic artifacts, the site being of large size, or close physical association with known Marpole sites. Appendix B lists these sites along with the justification for their inclusion. Random Locations: Using a random number generator (SYSTAT Data 1992) and the UTM coordinate range for the area under study, 300 point locations were produced. These in turn were automatically plotted on the map by the GIS software. At this point it was decided to restrict the map area to a buffer zone 2 km inland from all marine and riverine shorelines as none of the selected sites were outside this strip (see Figure 1). All random points that fell outside of it were removed from the map. As a result, 80 random points were retained. Maps: The base maps for the Canadian locations at a scale of 1:20,000 were Terrain Resource Information Management (T.R.I.M.) maps produced by the British Columbia Ministry of Lands and Parks in both digital and hardcopy form. American map information was derived from US Department of the Interior Geological Survey maps. Soils information was taken from surficial geology maps (Vancouver, New Westminster) produced by the Geological Survey of Canada (1978). Vegetation cover came from Vegetation of the Southwestern Fraser Lowland, 1858-1860 (North et al. 1979). Since many of the old streams are now confined to sewer pipes if they exist at all, information on them was taken from the vegetation map, Waterways ofBurnaby (City of Burnaby 1993), and Vancouver's Old Streams (Harris 1989). Because the T.R.I.M. maps use the new North American Datum (NAD 83), 16 all the other maps which used NAD 27 had to be adjusted to match it. This was necessary because when a map is digitized for use in a GIS it must be registered in real space, in this case to coordinates in the Universal Trans Mercator (UTM) projection system. With the NAD 83 correction there was a slight shift in these coordinates of about 95 metres on the Easting and 200 metres on the Northing; therefore, no map drawn according to NAD 27 quite matched the T.R.I.M. maps. All such non-T.R.I.M. registration points had to be fed into a "geographic calculator" for correction (The Geographic Calculator Version 2.0 (1993)). Initially climate, in the form of wind, precipitation, and temperature data (Oke & Hay 1994), was to be used in this study. The lack of differentiation in these values across the region, however, led to their deletion. Data Summary and Analysis The Geographic Information System used here was TerraSoft, a software product developed in Nanaimo, British Columbia and used extensively in forestry applications. Once all digitizing was completed two further elements were added to aid in collecting information about the local area surrounding each actual site location and random point. This was done by drawing a 50 and a 500 meter buffer around each point. The 500m buffer was used to look at local vegetation, soil types, and slope and elevation values (see Figure 8). The 50 m buffer was used to determine an aspect value for each site (the general direction it faces, sometimes referred to as exposure). As already discussed, although the map used here was created as a vector file the TerraSoft system has the capability to convert vector data to raster. In order to determine elevation, slope, aspect and viewshed information a raster file was necessary. A fifty meter pixel size was used and, because of the number of processing tasks involved and the time required to perform them, the overall map was broken down into smaller raster maps. For example, a map of this size (2500 km )^ would involve approximately 550,000 pixels if converted to raster in its entirety, and, since some of the routines involve checking every pixel in the map (viewshed for example), the processing becomes onerous. The goal at this point was twofold - to summarize the spatial data and to attempt some analysis of this material. In their present state of development, GISs are most noted for their summarization abilities. Spatial information can be manipulated to produce new variables or forms of data. Visual displays, either on computer monitors or hardcopy graphic images, can answer many basic inquiries. Since most of the raw data is stored in a 17 series of relational databases, this can be retrieved in tabular form with some basic summary statistics. The analytical power of these systems is, however, quite limited. The investigation of patterns, seeking relationships between patterns, and the modeling of such relationships remains outside of the GIS itself (Bailey 1994). In the case of this exercise this further manipulation of the data has been done using the statistical software package SYSTAT for Windows (1992). * The first set of information was taken directly from the map displayed on the monitor using the distance measuring facility. These were distances from both actual and random site coordinates to the closest shore, stream, and nearest site (See Appendix C). Again using the vector map files, all vegetation and soil polygons within the 500m buffer of each active site and random point along with their intersected areas was determined (Appendices D and E). Figure 8 illustrates how this measurement was done. For many sites, close proximity to either a marine or riverine shoreline often reduced the total land area for the buffer, therefore this "aquatic" area was also determined. Using separate rasterized segments of the larger map, minimum, maximum, and mean elevation as well as slope values were calculated for the 500m buffered area around each actual site and random point location. This was done by recording elevation and slope of each pixel in the buffered area, passing this information into a database (dBASE III+) and then into SYSTAT where these values were weighted by their total areas within the buffer. (See Appendix C). For example, if from the GIS we learn that 40 of the pixels within the buffer show an elevation of 17 m, 20 show 14 m, 10 show 1 lm, and 30 show 9 m, each pixel type is weighted by the total as follows: (40 X 171 + (20 X 14") + (10 X 1 n + GO X 91 = an average of 13.4 m for the buffer. 100 Since aspect was determined in a similar way, it was felt that the variations present in a 500m buffered area would be too great and ultimately meaningless, therefore, a 50m buffer was used instead. This resulted in the manual averaging of at most 4 pixel values (degrees from north) to determine a typical "most viewable" direction for each site (Appendix C). The final set of information drawn from the GIS came from the viewshed analysis. This type of measurement is seen as being a truly original capability of a GIS (B. Klinkenberg, personal communication). Only a computer can perform the millions of calculations required to examine the elevation value of each pixel in every direction from a point on a map to determine if or at what point the view is obstructed. This viewshed capability, however, is not used here to measure what an individual might be able to see from a vantage point located at a site. This has proven fruitful in some situations where sufficient elevation can be achieved or where there is little or no tree cover (eg. Kvamme 1992:77). The presence of extensive forests in the Lower Mainland area, however, would prevent much lateral or landward viewing. At the same time, with most sites being on or very near a shoreline, good views out over the water would be present in most cases. Instead, the viewshed analysis was used here as a measure of "openness" around a site, of how much relatively unobstructed movement would be possible inland from the shore, as well as laterally to either the left or the right, while remaining in view of the site. Hills, cliffs, bluffs, and headlands would act as natural boundaries (although not insurmountable ones) to easy movement in the immediate area of the site. The assumption here is that habitation sites would have broader and deeper viewshed areas compared to limited activity sites. (The exception being sites located in totally open environments such as the Fraser Delta). This information was added to the other distance measures (see Appendix C). The Variables Distance Measurements The three distance measurements, to the nearest site, marine or riverine shore, and freshwater stream, were examined using box-and-dot plots and stem-and-leaf diagrams (see Figures 5, 6, and 7). This was done for both actual sites and for the random points. Despite the appearance of obvious dissimilarities between the actual and random values, a further comparison was done using a Mann-Whitney test (SYSTAT Statistics 1992:480). Both sets were melded into one file with each record identified as either "a" for actual or "r" for random and then the resulting data was tested to see if it could form a single population. As can be seen from Table 1, none of the three variables show any similarity between the actual and random samples. Table 1: Comparative Statistics - Actual Sites and Random Points (Distance Values in Meters). NEAREST SITE STREAM SHORE Mean — Actual 1386 441 219 Random 1796 943 794 Median — Actual 881 248 95 Random 1648 699 741 Mann-Whitney 1805 1704 1231.5 Probability 0.00 0.00 0.00 19 From the box plots, stem-and-leaf diagrams, and the statistical results a number of observations can be made. Looking first at the NEAREST SITE variable, we see that with the randomly generated points there is a near normal distribution of intersite distances (based on a visual examination of Figure 5a and 5c as well as the closeness of the mean and median values in Table 1). Random Actual - i 1 1 1 1 r-H * • MINIMUM IS: LOWER HINGE IS: MEDIAN IS: UPPER HINGE IS: MAXIMUM IS: 133. 627. 881. 1887. 6731. 0 11112222333344 OM 56666666666777788688 1 01222244 IH 667888 02333 567 4 66 —OUTSOE VALUES-8 3 7 MINIMUM IS: 138. LOWER HINGE IS: 1192. MEDIAN IS: 1648. UPPER HINGE IS: 2471. MAXIMUM IS: 5740. 0 1122444 0 555588999 1H 00111122222333333344444 1M 655566668888899 2H 0000000122444 2 56666666889 3 00000333 3 4 0 •••OUTSOE VALUES— 5 7 1000 2000 3000 4000 5000 6000 7000 Nearest Site (in meters) a) b) Actual c) Random Figure 5: NEAREST SITE. From this, then, we can assume that they provide a reasonable representation of the background environment. The actual site locations, in contrast, however, show something different. The median value is opted for here, rather than the mean, because of skewness in both Figures 5a and 5b. This value, 881 m, is almost half of the one determined for the random points, 1648 m, and indicates that, where sites are found, we can expect at least one other site within the space of a kilometer or less. Also we can state that, whereas random points are arrayed in a more even fashion across the landscape, actual sites tend to be found in groups. A probable explanation for the outliers among the actual sites (see Figure 5b) is that they are in areas where survey work is spotty or where most sites are deemed to have been obliterated by modern development. The situation for the other two distance variables is, if anything, even more striking, especially with the SHORE values (Figure 6). Here, while the random points fall close to the midpoint of the 2 km buffer, (a median 741 m in from the shoreline), the actual sites are located almost entirely within 200 m of it (a median value of 95 m). With the STREAM values (Figure 7), despite the presence of a greater number of outliers (Figure 7b), we can still see that the majority of actual sites are within 500 m of fresh water versus 1200 m for the general landscape (Figure 7c). 20 Random Actual . . j . . 8 83 « t • « #• o M I N I M U M I S : 6 . LOWER H I N G E I S : 4 5 . M E D I A N I S : 9 5 . U P P E R H I N G E I S : 2 2 0 . M A X I M U M I S : 1 5 7 6 . M I N I M U M I S : LOWER H I N G E I S : M E D I A N I S : U P P E R H I N G E I S : M A X I M U M I S : 0 . 2 2 9 . 7 4 1 . 1 3 8 6 . 2 2 2 5 . 500 1000 1500 2000 2500 Nearest Shoreline (in meters) a) 0 00000000334558 OH 01222222233344444444 1 0111124 CM 55666778899999 2H 23378999 1 0011244 3 689 1 699 4 779 2H 012 5 234467 2 778 7M 123556789 3 12 8 12335 3 7 9 258 4 2 10 0 4 6 11 59 —OUTSOE VALUES— 12 456 5 4 13H 579 6 1 14 0112569 6 78 15 5 9 3 16 0116 14 4 17 6 15 7 18 36 b) Actual 19 20 S256 c) Ra 22 2 Figure 6: SHORE. Rarvjom Actual -i r - M I N I M U M I S : LOWER H I N G E I S : M E D I A N I S : U P P E R H I N G E I S : M A X I M U M I S : 0 0 0 0 0 0 1 2 4 4 OH 5 5 6 6 6 9 9 9 1 0 1 2 2 1 5599 2 M 0 0 0 1 2 3 2 67869 3 12 3 5 6 8 6 9 4 2 3 3 4 H 67 5 0 0 1 6 2 0 . 9 3 . 2 4 8 . 4 6 7 . 2 9 3 1 . M I N I M U M I S : 0 . LOWER H I N G E I S : 2 6 8 . M E D I A N I S : 6 9 9 . U P P E R H I N G E I S : 1 2 2 8 . M A X I M U M I S : 3 9 2 2 . 5 6 6 8 7 0 1000 2000 3000 Nearest Stream (in meters) 4000 7 12 14 9 16 1 "OUTSIDE VALUES— 17 18 2 5 2 9 a) b) Actual 0 1 2H 3 4 5 6M 7 8 9 10 11 I2H 13 14 17 19 21 22 24 25 26 28 29 32 36 39 00OOOI1157788999 1136 34899 3344666899 245 344 049 002688 12577 59 2357 16679 6 56 127 25 8 6 5 179 5 5 0 7 7 2 c) Random Figure 7: STREAM. Even with actual site values showing discrepancies via outliers we can still conclude that interesting patterns are displayed here. There is definite selection for nearness to shorelines and streams and also a tendency for the sites to not stand isolated on the landscape but rather to form clusters or groups. Vegetation and Soils These variables describe the local environments of the sites. However, they present some limitations. The vegetation map produced by North et al. (1979) does not cover the entire area looked at in this study, extending north only up to latitude 49° 15' and south to the US border. Thus, only 34 of the actual sites and 50 of the random locations could be looked at. The soil maps were more extensive but soil types that matched the Canadian system were lacking for the four US sites and they were excluded. The other note of caution is that the vegetation data are for the mid-19th century and can at best be an estimate of the plant communities of two thousand years ago. As already mentioned, a circular 500m buffer was drawn around each site location (see Figure 8). This was not meant to indicate anything like an original site area but rather to capture a space in which the local site environment could be explored. In the GIS this buffer was made to intersect information from other parts of the database. This is the so-called layering capability of a GIS where a site buffer "layer" is placed over a vegetation or soil "layer" and, like a cookie cutter, removes the local information. For each layer two sorts of information result, first, the number and identity of vegetation or soil areas within the buffer, and, second, their areas. Figure 8: Vegetation polygons intersected at DgRr 1 - Crescent Beach. In analysing the results for these variables three assumptions were made: 1. the actual sites would not parrot the background environment 2. the sites would align with certain soil and vegetation types more than others and not in proportions identical to the general landscape 3. particularly in the case of vegetation, they would show greater numbers of types in the immediate vicinity of a site than the random locations. Figure 9 compares both sets of locations to the general background environment. (The background here is limited to the 2km strip around all shorelines from within which actual sites and random locations were taken). Both 22 sets of site locations bear a reasonable resemblance to the overall vegetation landscape. The discrepancies are discussed for each variable. 5 0 4 0 3 0 ± 2 0 10 LJ Li__ j j L U L L a)Actual vegetation Type 5 0 40 3 0 0_ <0 20 10 -i 1 1 r • rn Vegetation Type b)Random Locations 2 3 4 5 6 Vegetation Type c)Background Environment 1. Grasses 2. Shrubs 3. Shrubs/Moss 4. Woodland 5. Scrub Forest 6. Conifer Forest (Cedar) 7. Conifer Forest (Douglas Fir) Key to types Figure 9: Vegetation Area Comparisons. Vegetation: In addition to the seven general plant groups used by North et al. (1979), (see Appendix F for the complete list within the groupings), the area of any marine or riverine water intersected by the site buffers is also shown in Figure 10 (as "Type 8"). As expected from the previous analysis of distance to nearest shoreline, the actual sites show three times the water area of the random locations. This fits well with what we know of the importance of the resources from the marine foreshore in the lifeway. The first noticeable discrepancy between the actual sites and the background environment (Figures 9a and 9c) is that the actual sites are half as likely to encompass the grass communities, vegetation type 1. With the two communities of smaller tree types, vegetation categories 4 and 5, actual locations are almost the opposite of the background environment. The "woodland" community is almost double what might be expected, and "scrub forest" less than half the proportion in the general landscape. The woodland group is typically made up of maple, alder, cottonwoods, and some willow. Alder particularly is the first tree to grow in recently vacated areas (for example, following a fire) (Slaymaker et al. 1992:34). This may indicate local clearing practices around sites either for wood products or the maintenance of a prairie environment. 1 2 3 4 5 6 7 8 Vegetation Types (Water included) 1 2 3 4 5 6 7 8 Vegetation Types (Water included) a)Actual b)Random Figure 10: Vegetation profile including intersected water One of the problems with presenting data in the manner of Figures 9 and 10 is that it does not tell us very much about how locally complex the vegetation really is. Is it a matter of the entire buffer being dominated by a particular plant community or are many represented at each location? In the case of the actual sites 76% (26 out of 34) had 3 or more communities within their 500m radius. Figures 11a and b try to illustrate this. They show the actual numbers and sizes of the communities intersected in the two situations. As a result, we see that in an actual site setting there is a finer grained selection of plant groups; the polygon areas are smaller but also more numerous. A random location tends to be dominated by a single plant type. The 50 random point buffers intersected a total of 24 127 plant polygons, an average of 2.5 per site; the 34 actual sites intersected 131 polygons for an average of 3.9 per site. Somewhat puzzling here are the slight discrepancies between the random locations and the background environment, especially in the case of vegetation types 5 and 7. The assumption here is that this is a product of sampling error. Vegetation Type Vegetation Type a)Actual b)Random Figure 11: Site size by Vegetation Type. At this point two general statements are possible about vegetation. There is indication that with the actual sites there is a selection for more arboreal surroundings, particularly the large conifer communities (vegetation types 6 and 7). There is also evidence for a more diverse vegetation mosaic within the immediate area of actual sites as was seen in Figure 11. A closer examination of this diversity will be presented below. Soils: Although we can assume that soil type is a less critical variable for a non-agricultural people, it can play a factor in the placement and the type of housing. The analysis here was carried out in the same fashion as with vegetation, again using the 500m buffers. (See Appendix G for a detailed list of the soil types). The results show that the actual site locations follow fairly closely the distribution of soils in the general landscape (Figures 12 a-c). A major discrepancy occurs with soil group 02, bog, swamp and shallow lake deposits, which is fairly self-explanatory. There is a slight tendency to select more for marine shore deposits (soil group 03) which is again to be expected with the close association between actual sites and shorelines as already discussed.The high correlation with the general landscape on the Fraser River sediments (soil group 05) was a bit surprising but these soils do intrude well into the Point Roberts and Marpole areas, both with a fair number of sites. A higher than expected 25 presence of Pre-Vashon deposits (soil group 10) is also interesting since these are quite rare as surface soils but where they do occur, on the outer edge of both Point Grey and Panorama Ridge, we do have a number of sites. 01 02 03 04 05 06 07 08 09 10 11 12 13 Soil Types a)Actual Locations 01 02 03 04 05 06 07 08 09 10 11 12 13 Soil Types b)Random Locations c CD O (5 a ,o 01 02 03 04 05 06 07 08 09 10 11 12 13 Soil Types c)Background Environment 01 Landfill 02 Bog, swamp, shallow lake deposits 03 Marine shore sediments 04 Lowland and mountain stream deltaic, channel fill 05 Fraser River sediments 06 Marine shore & fluvial sand 07 Capilano sediments - raised marine, deltaic & fluvial deposits 08 Vashon Drift & Capilano sediments 09 Vashon Drift 10 Pre-Vashon deposits 11 Undivided Pre-Vashon deposits 12 Tertiary 13 Pre-tertiary Key to Types Figure 12: Soil Area Comparisons. Looking at the second part of the display of this data (Figures 13a and b), it is harder to discern distinct patterning. Almost in opposition to what was observed with vegetation, the actual sites tend to intersect larger soil polygon areas than the random locations do. One example is the Fraser River sediments (soil type 5). For random locations the polygons appear to be almost half the size of those for the actual sites. Part of this is due to the overall 26 larger size of the soil polygons; in this case there are more sites (n=55) intersecting fewer polygons (n=l 18) for an average of 2.0 per site. There is also, however, no reason to suspect that the original settlers would have looked for diversity in the local soils if their primary consideration was for solid and stable foundation for dwellings. 1 2 3 4 6 6 7 B 9 10 11 12 13 Soil Types a)Actual 01 02 03 04 OS 07 08 09 10 11 12 13 Soil Types b)Random Figure 13: Site size by Soil Type. Topographic Variables This section again uses the method of looking at the contents of a circular buffer around each site to gather certain generalised information. Without knowing the center or extent of the sites it is unreasonable to expect that an elevation, slope, or aspect value taken at just one single point truly represents the situation at that setting. Once an average value for these variables was obtained for each actual site and random location, the procedure was much the same as for the three distance measures already discussed. Figure 14 provides a basic graphic comparison of the actual and random elevation and slope values. One problem with these particular measurements, however, is the high number of random point locations located in flat regions, particularly in the Fraser Delta Lowlands. 27 of the 80 random locations fall into this category whereas only 7 of the 62 actual sites do so. While this immediately tells us something about the actual site choices that were being made, it also may unduly bias further understanding about the overall elevation, slope, and aspect. For this reason all "flat" locations were removed from both datasets. Figures 15 and 17 show the results including stem-and-leaf diagrams. Random Actual Random Actual 100 200 300 Mean Elevation (meters) 400 27 10 20 Slope (in percent) a) b) Figure 14: ELEVATION and SLOPE - all sites and random points included. Looking at the ELEVATION variable first (Figure 15), we can see that for the random locations (or the general non-flat background landscape), the situation is quite skewed because of the inclusion of a few high elevation locations. The actual distribution, however, approaches an almost normal shape. Random Actual M I N I M U M I S : LOWER H I N G E I S : M E D I A N I S : U P P E R H I N G E I S : M A X I M U M I S : 3 . M I N I M U M I S : 0 . 12 LOWER H I N G E I S : 1 1 1 9 . ' M E D I A N I S : 2 9 . 3 4 . U P P E R H I N G E I S : 5 6 5 8 . ' M A X I M U M I S : 3 0 9 . 0 100 200 300 400 Mean Elevation (in meters) - Flat' sites removed 0 2334 0 0011245788999 0 5677899 1H 23356788 1H 001111223344 2M 336799 1M 56889 3 2269 2 01144 4 2349 2 556789 5H 2345679 3H 234 6 3 6899 7 47 4 0334 8 0 4 78 9 45 5 3 10 9 5 58 11 8 •••OUTSIDE VALUES 22 9 23 6 30 9 a) b) Actual c) Random Figure 15: ELEVATION values with "flat" locations removed. In fact, when these actual elevation values were further analysed using a normal probability plot (SYSTAT Graphics 1992:232) in which the actual elevations are compared with a computer generated mathematically normal distribution, the values do approach a fairly straight line (Figure 16). The other important thing to note is the much 28 Random Actual narrower range of values as compared to the random locations (especially evident on the stem-and-leaf diagrams in Figures 15b and 15c). 10 20 .30 40 Elevation 50 60 Figure 16: Normal Probability Plot for Actual ELEVATION values. The comparison for the SLOPE variable (Figure 17) shows a greater similarity between the two sets of locations, although the actual sites do show slightly higher values. M I N I M U M I S : 1 . M I N I M U M I S : 0. LOWER H I N G E I S : 3 . LOWER H I N G E I S : 3 M E D I A N I S : 7 . M E D I A N I S : 5. U P P E R H I N G E I S : 1 2 . U P P E R H I N G E I S : 8 M A X I M U M I S : 2 2 . M A X I M U M I S : 2 9 . 0 1111111 0 0001111111 OH 222222223333 0 4444455 OM 66677 0 88899 1H 0000111 1 22 1 4555 1 66677 1 2 2 2 OH 22222222233333 0M 4444445 0 666777 OH 888899 1 00 1 222 1 1 6 •••OUTSIDE VALUES" 1 79 2 8 0 10 2 0 3 0 Slope (in percent) - Flat' sites removed a) b) Actual c) Random Figure 17: SLOPE values with "flat" sites removed. Both ELEVATION and SLOPE were put through the same routine used in the distance analysis above where random and actual locations were combined into a single dataset and then compared to see if they could be interpreted as being part of the same population. This time the probability is somewhat higher (see Table 2) with the slope values statistically similar enough to suggest that, for actual sites, they conform to the background 29 environment. The fact that they are about 2 percent (4.6 vs. 6.6) higher, however, is important to note. In an area that receives a modern average annual precipitation of 1167 mm (Oke and Hay 1994:29) good drainage would be an important factor for any location with longterm use. Table 2: Comparative Statistics - Actual Sites and Random Points (Elevation Values in Meters; Slope Values are Percents). Elevation Slope Mean — Actual 23.0 7.8 Random 47.9 6.1 Median — Actual 19.2 6.6 Random 29.2 4.6 Mann-Whitney 1108.5 1711.0 Probability 0.045 0.08 The third topographic variable to be examined is ASPECT and again the so-called "flat" locations were eliminated. It should be noted that two of the random points with very low average elevations showed an aspect value of zero and were dropped from the dataset. Figure 18 shows the results. It can be seen that the actual sites 40 i 1 1 1 1 1 1 1 1 1 40 a ^ Actual Sites (flat' sites removed) b) Random Locations ('flat' locations removed) Figure 18: Comparison of ASPECT values. show a marked tendency for a southwestern aspect in comparison to the random locations. In other settings, the American Southwest for example, exposure to the sun was deemed to be an important factor in choosing a location, for both long-term and short-term occupation. In a study conducted in Utah, Haase was able to demonstrate a strong preference for a southern exposure (Haase 1985:71-75). A chi-square test was performed on the information shown in Figure 18 to see if something similar might exist within this study area (see Figure 19). If we accept a critical 30 value of 0.05 the result shown here falls within the accepted limits and does not represent a significant deviation from a normal population (3.209 falling within the range of 0 to 3.841). However, if the critical value is extended to 0.10 we do have a significant deviation (3.209 being outside the range of 0 to 2.706). Although 0.05 is often used in archaeology as a benchmark it is by no means an outside limit (Shennan 1988:69). The result here, therefore, remains in somewhat of a grey area; further analysis using additional site settings would be helpful. Actual Random Total Southwest 17 8 25 Other 38 42 80 Total 55 50 105 Pearson Chi-square 3.209 DF 1 Prob 0.073 Figurel9: Chi-square statistic for ASPECT Viewshed and "Openness" As already discussed, the capability of the GIS to generate information about viewable area around a particular point on a landscape was considered of questionable value in a heavily treed setting like the one being looked at here. Nevertheless, two variables were defined for this analysis. The first was DEPTH - the distance inland from the site to the point where the viewshed stopped. The second is BREADTH - the lateral distance to the left and right of the site location showing again the extent of the viewshed analysis in these directions. Any outward spread of the viewshed, usually over open water, was considered meaningless here since it often meant extending to the edge of the map itself. Both variables were viewed with box plots and stem-and-leaf diagrams (Figures 20 and 21) and, although the actual sites showed that both variables were skewed towards high values, a probability plot check showed that the distribution was reasonably normal (Figure 22). The stem-and-leaf diagrams for the actual sites (Figures 20b and 21b), relay some interesting information on the general shape of the landscape. We can see from DEPTH (Figure 20b) that space, in terms of distance back from the shoreline, was limited to approximately 1/2 km for most sites in this region. At the same time lateral space (BREADTH) was not usually a problem with from 1 to 2 km usually available on either side of a site (Figure 21). 31 Random Actual 500 1000 Depth (in Meters) 1500 M I N I M U M I S : 7 1 . LOWER H I N G E I S : M E D I A N I S : 4 1 8 . U P P E R H I N G E I S : M A X I M U M I S : 1 4 8 9 . 0 7778 1 4556788 2H 0001568 3 244777 4M 001133679 5H 02334567 6 114 7 1 8 1366 9 3 10 1 11 7 •"OUTSIDE VALUES" 13 29 14 8 2 1 1 . 5 9 9 . a) b) Actual Random Actual 2000 4000 6000 8000 Breadth (in Meters) a) 1000 M I N I M U M I S : LOWER H I N G E I S : M E D I A N I S : U P P E R H I N G E I S : M A X I M U M I S : 0 1234 0 88999 1H 0022444 1 66778888 2M 00012 2 557789 3 012 3 679 4H 023 4 77 5 004 5 6 6 23 6 7 34 1 4 4 3 . 2 1 2 8 . 4 0 1 8 . 9 9 7 0 . "OUTSIDE VALUES" 9 9 b) Actual M I N I M U M I S : LOWER H I N G E I S : M E D I A N I S : U P P E R H I N G E I S : M A X I M U M I S : 0 079 1 036 2 12778 3H 35569 4M 002244577 5 11234 6 46 7 1 8H 679 9 66889 10 5 11 0 12 13 5 14 25 3 3 1 . 4 6 7 . 8 7 8 . 1 4 5 2 . c) Random Figure 20: DEPTH in back of actual sites and random locations (meters). 1 9 5 . M I N I M U M I S : LOWER H I N G E I S : M E D I A N I S : U P P E R H I N G E I S : M A X I M U M I S : 0 234 0 6677 1H 01244 1 5899 2 14 2M 5667899 3 0012 3 66788 4H 00 4 5578 5 4 5 6 3 6 88 7 2 7 6 c) Random Figure 21: BREADTH in front of actual sites and random locations (meters) 3 i 1 1 1 1 1 3 2000 4000 6000 8000 10000 Breadth in front of site 500 1000 Depth back of site Figure 22: Normal Probability Plots for BREADTH and DEPTH. 2 4 8 . 1 4 5 9 . 2 8 5 9 . 4 0 1 5 . 7 6 5 0 . 1500 32 When compared to the set of random locations, however, there is no discernible difference between the two. Again a Mann-Whitney exercise was performed and for both the BREADTH and DEPTH variables the probability is good that both active and random situations comprise the same population (DEPTH probability is 0.246; BREADTH is 0.468). Clustering and Multidimensional Scaling Background Up to this point the analysis has used the site database in its entirety to show that it does not represent a random distribution of points across a landscape and that it can be seen as displaying broad patterns related to an adaptation to the physical and biotic features of the surrounding terrain. The next question is whether or not within this collection of sites further patterning can be detected, specifically whether we can discern any groupings or clusters of sites that might aid us in classifying the sites on the basis of the landscape elements already described (Hypothesis 3). Cluster analysis is most effective as a research tool when three different processes take place (Everitt 1993). First there should be some prior indication that there is the possibility that groups or clusters exist within the data. The box-and-whisker plots, stem-and-leaf diagrams, and bar charts have already shown diverse ranges in the values of the various variables, suggesting further patterns between the various sites. Secondly, a clustering technique must be chosen which best suits the data being analysed. This in turn involves a number of decisions as to how the data are to be viewed, compared and manipulated. The third step is to somehow confirm that the results of the clustering are showing valid groups and not simply arcane products of the algorithms employed, "artifacts" as one writer describes them (Everitt 1993:142). This may primarily involve the analyst's own intuition regarding the material but certain other data display techniques such as multidimensional scaling can also be employed. If the clustering is successful in splitting the actual sites into two or more groups, some means of testing whether or not it validates Hypothesis 3 is still required. As has already been discussed, one of the great frustrations with the Lower Mainland sites is the vagueness surrounding their original function. In some cases, however, the original investigators have gone out on an intuitive limb and actually assigned a site type. Therefore, those sites that have shown some reasonably good evidence for prolonged habitation are being used as a test of the clustering exercise. These include the major Point Roberts sites, Beach Grove (DgRs 1), Brandrith Park (DgRs 4), Whalen Farm (DfRs 3) and Tsawwassen (DgRs 2), the Musqueam sites (DhRt 1,2,3,4), the two major Fraser River sites, Glenrose (DgRr 6) and St. Mungo (DgRr 2), the Marpole site (DhRs 1), the Locarno Beach site (DhRt 6), one of the Stanley Park sites (Lumberman's Arch, DhRs 2), one of the Burrard Inlet sites (Belcarra Park, DhRr 6) and the two American sites on Birch Bay (45-WH-9,45-WH-l 1). Crescent Beach (DgRr 1), although usually designated a limited activity site, is also included here because of its extensive use over millennia. The clustering was performed three times (see Figures 23, 25, and 27). The first run used all 62 sites and their associated topographic variables: NEAREST SITE, STREAM, SHORE, ELEVATION, SLOPE, ASPECT, DEPTH, and BREADTH. (ASPECT in this case is the numeric value assigned by the GIS, in the range of 0 to 8). In the initial attempts the NEAREST SITE and ASPECT variables created particular problems. NEAREST SITE tended to pull all isolated sites out as a group. Since this is felt to be more a product of a gap in the record than a reflection of past reality, it was decided to not use it. The ASPECT variable has already been looked at above and may be reflecting more the general orientation of the landscape rather than a particular choice (see pp. 29-30 and Figure 19). In the clustering and scaling analyses it tended to sequence the data in a way that proved confusing and finally inappropriate for the analysis (grouping sites first according to orientation rather than according to more meaningful variables). It was subsequently dropped from the variable set. The second clustering routine used only the 34 sites with vegetation data. This first required a choice as to how these variables were structured. As shown in Appendix D, the vegetation content of the 500m site buffers was displayed in two separate matrices. The first, Table 1, showed the actual plant communities encountered around each site plus the area of that plant community. The maximum number of plant communities, seven at DgRs 4, determined the width of the matrix. The second display (Table 2) records the information differently; here the individual plant communities are assigned to one of seven broader types plus intersected water - type 8 (see Appendix F). The assumption here is that a site could potentially have all seven types within its environs, although in reality this does not occur with the sample used here. Both matrices were fed into the cluster routine with the results from using the Table 1 matrix proving unsatisfactory. Clustering the data in this form simply grouped sites according to numbers of plant types without saying very much about the vegetation itself. The second matrix, which included the area of intersected water as an eighth type, was more productive and the results are retained here. 34 Finally, a third cluster exercise was done combining the eight vegetation variables with the six topographic variables used in the first clustering operation. All three routines used standardised values for the variables in order to allow each to contribute equally to the calculations (Matson & True 1974:55; Pokotylo 1978). (The mean for the particular variable is subtracted from each observation and the result is then divided by the standard deviation for that variable [SYSTAT Data 1992:111]). As part of the clustering process SYSTAT requires a "distance metric" to compute similarity between cases. Two coefficients are provided for continuous scales, Euclidean distance and Pearson product moment correlation, (SYSTAT Statistics 1992:25), and for the analysis performed here Euclidean distance was chosen. The Pearson coefficient tends to average out all the variable values for each case (Everitt 1993:43) and because of the diversity of measurements in this study, this was not considered desirable; if a particular variable was responsible for making a case a member of a cluster this was deemed to be important information. Although SYSTAT provides a number of clustering or "linkage" methods, (single [nearest neighbour], complete [furthest neighbour], centroid, average, median, and Ward's [SYSTAT Statistics 1992:27]), it was decided to use Ward's. This was partly due to its preferential use in archaeological applications (Matson & True 1974; Pokotylo 1978; Shennan 1988) as well as its recommendation for providing the most homogenous clusters (Shennan 1988:217; Everitt 1993:65). As part of the confirmation step mentioned above, two techniques were employed. First multidimensional scaling was done to try and replicate the groupings of cases produced by the clustering. If there were major discrepancies, these would be investigated further. SYSTAT provides only non-metric multidimensional scaling (SYSTAT Statistics 1992:109) and, although up to five dimensions can be chosen, the default of two dimensions was used here. The second confirmation technique simply involved re-ordering the various data matrices in the same sequence as the clustering and multidimensional scaling diagrams and examining the results (Pokotylo 1982). This proved to be most beneficial in explaining what was behind the various groupings as well as confirming that both statistical procedures had divulged meaningful patterns in the data. Results A cluster diagram and a multidimensional scaling plot were produced for each of the three groups of variables (Figures 23 through 28). In each case the "habitation" sites are identified with a triangle. Although a DgRr 7 DgRr11 DhRt13 DgRr 9 DgRs 9 DgRs 11 ~ DhRr 8 DhRt 6 DhRr 20 DgRr 6 DhRt 17 DhRs 7 DgRr12 DhRt15 DhRt14 DhRr16 DhRt 5 DhRr15 _ DhRr17 DhRt22 DhRt 36 DhRq 12 DgRs17 DhRq 13 | DgRr 24 DhRr18 45-WH-17 DhRs16 DhRq 21 DgRr 1 DhRs 1 A A A A A A A A A A A A A DhRr 9 45-WH-1 45-WH-11 DhRq 19 DhRs 4 DhRs 21 DhRs19 DhRt 18 DhRt 10 DhRs 6 DgRs 7 DfRs 3 DhRq 1 DhRr 75 DhRr 3 DhRr 6 DgRs 2 DhRs 5 DgRs 15 DgRs 1 DhRt 74 DgRs 30 DgRs 4 DhRt 4 DhRt 2 DgRs 36 DhRt 3 DhRt 1 45-WH-9 DhRs 2 DgRr 2 Figure 23: Cluster Diagram of the topographic variables 36 number of tables of the re-ordered data were generated, only one is included here and will be discussed further on (see Table 3). Looking first at the topographic variables (Figure 23), we can see two large clusters, A and B, with the majority of the habitation sites in cluster B (15 out of 17). The main variables determining this split are ELEVATION and SLOPE which gives an initial indication of a preference for moderate elevations and reasonably flat areas. Cluster A has an average elevation of 39.8 m above sea level with a slope value of 13.5 degrees whereas the remaining 15 habitation sites show average elevation to be only 14.9 m with an average slope of 4.8 degrees. (These calculations were made from a table of the variables resequenced in the order of the cluster diagram). The other cluster with distinct elevation and slope values is the one containing the so-called "flat" sites (cluster Al). 2 O <n 0 z LU Q -2 DhRt17 DhRr18 DgRr 6 DhRr 20 DhRt 5 DhRs 7 DhRt 6 DhRr16 DhRr15 DhRt 14 DhRt15 DhRr17 DgRs 11 DgRr 1 2 - ^ C ) DhRr 8 -^AOP DgRr 1 ^yf7 45-WH-17/U/ D h R s 1 6 x DhRt 13 DgRs 9 DgRr 9 DgRr 11 DgRr 7 DhRq 21 DgRr 2 DhRr 9 45-WH-1 DfRs 3 DhRs 2 DhRt 1 45-WH-9 45-Wh-11 DhRt18 DhRs 6 DhRt10 DgRs 7 DhRt 2 DgRs 36 DhRt 3 DhRt 4 DhRq 19 . ^ DgRs 30 V \ D g R s 4 • ^ ^ D h R s 2 1 DhRs 4 DhRs 19 DhRt 74 DgRs 1 DhRs 5 DhRq 12 DhRq 13 DgRr24 DgRs 17 -2 -1 A = Habitation Sites 0 1 2 DIMENSION 1 O = Limited activity sites, petroglyphs.unknown, etc. Figure 24: Multidimensional scaling plot of topographic variables Not as recognizable in the cluster diagram but something which was very apparent in the multidimensional scaling plot (Figure 24), was the importance of a broad open area in front of the habitation sites, represented here by the variable BREADTH. This realization was partly the result of the scaling being prone to produce a seriation using 37 at least one of the variables. As already mentioned, this tendency to re-order the cases from low to high and then back down to low on the strength of the values of a particular variable initially led to the discarding of the ASPECT variable. For the remaining topographic variables, however, the seriation effect actually had the benefit of revealing a pattern that was not clear in the cluster diagram or, for that matter, in the initial analysis of the viewshed variables, BREADTH and DEPTH (pp. 30-32). While the average value for BREADTH for the entire set of 62 sites is 2488 m, this average goes up to 5263 m for the 12 habitation sites in the arc on the upper right side of the plot (Figure 25). Already, then, we have seen three distinct characteristics that seem to have been taken into account when choosing habitation locations: a moderate elevation, generally flat surroundings, and above average openness in front of the site. It must be pointed out, however, that these elements do not indicate rare locations in the landscape generally. A quick check with the random locations revealed that an almost identical proportion of the terrain held the same range of elevation and slope values (21.3% of random locations versus 22.6% of the actual site locations). Upon reflection, these may all seem to be commonsense attributes to select for when choosing an appropriate location for longterm use but they are still very hard to pick out without the aid of the clustering and scaling exercises. Turning to the vegetation variables, the clustering diagram (Figure 25) does not provide any quick insights into the distribution of the sites. Again two large clusters were produced (A and B) but with approximately equal numbers of habitation sites in each (4 and 6). When looked at individually, the smaller sub-clusters form mostly around particular plant communities. What is interesting in these sub-clusters is the tendency of the habitation sites to have a good volume of cedar in their immediate vicinity. Both the sub-clusters A2 and B2 which contain seven out of the ten habitation sites have a reasonably high concentration of this plant type (between 15 and 52 hectares out of a possible 78.6 hectares within the 500m buffer). This trend is repeated on the multidimensional scaling plot (Figure 26) with the sites forming an arc in the upper half of the figure (from DhRsl on the right through to DgRs 9 on the left) all showing this preference for cedar within the buffered area. [ADgRs 1 DgRs 7 DgRr 7 DgRr12 DgRr 11 DgRr 9 ADgRs 2 DgRs 11 DhRt 17 , DgRs 9 ADgRr 6 DhRt 15 DhRt 14 IADgRr 1 DgRs15 DhRq 12 :—| DhRq13 H. DgRs17 r [^DgRs 4 DhRq 21 1 DgRr 24 1 DhRr 75 1 _ A DgRr 2 1 DhRt 74 A DhRt 3 1 A DhRt 4 ^~ DhRq 19 DhRt 36 DhRt 22 DgRs 30 DgRs 36 ADfRs 3 ADhRs 1 DhRs 19 Figure 25: Cluster Diagram of the vegetation variables. 39 O co z UJ Q -1 h DhRt 74 DhRr 75 DgRr 2 DgRr 1 DgRr12 DgRr 7 DgRr 11 DgRs 7 DgRr 9 DgRr 6 DhRt 14 DhRt 15 DhRt 17 DgRs 11 V \ DgRs 2 O DgRs 9 DhRt 4 DhRt 3 DhRs 19 DhRs 1 DhRt 22 DhRt 36 DgRs 1 DfRs 3 A / DgRs36 DgRs 30 DgRs 4 DgRr 24 DhRq 21 DgRs15 DhRq 12 DgRs 17 DhRq 13 - 2 - 1 0 1 2 DIMENSION 1 A= Habitation Sites 0= Limited activity sites, petroglyphs, unknown, etc. Figure 26: Multidimensional scaling plot of vegetation variables. The final clustering and scaling routine, using the combined set of 14 variables, produced the most satisfying results in terms of identifying those landscape elements most closely linked to habitation locations. Here the cluster diagram (Figure 27) shows three clusters with the third one, cluster C, containing nine of the ten habitation sites. This group again contains those sites with moderate elevation and slope plus a wide open area in front (see also the boxed area of the third section of Table 3). Heavily treed environments tend to prevail (both the cedar and Douglas fir communities) but, interestingly, often in combination with a large segment of grassland which is reminiscent of the ethnographic contention for the importance of a prairie setting nearby (see page 11). Cluster A, with only the one habitation site, represents the least desirable elements for long term use - high, sloping terrain with the plant community dominated by Douglas fir (see shaded areas in top section of Table 3). It is in this group that we find the largest values for intersected water and the shortest distances to the shoreline (boxed areas of top section of Table 3), both indicative of the site being hemmed in by the rising headland in back and the water in front. 40 DgRr 7 DgRr 11 DgRr 9 DgRs 9 DgRs 11 DgRr12 DhRt 14 DhRt 15 ADgRr 6 DhRt 17 DhRt 22 DhRt 36 DhRq 19 DhRq 12 DhRq 13 DgRs17 DgRr 24 DhRq 21 DhRs 19 ADhRs 1 ADfRs 3 DgRs 36 DgRs 30 ADgRs 1 DgRs 7 ADgRs 2 DhRt 74 ADhRt 4 ADhRt 3 ADgRr 1 DhRr 75 ADgRr 2 ADgRs 4 DgRs 15 Figure 27: Cluster diagram of the combined topographic and vegetatiion variables. Table 3: Topographic and Vegetation data resequenced to follow the order of Figure 27. 41 Site ID Stream Shore Eleva-tionSlope Brea- Depth VI V2 V3 V4 V5 V6 V7 V8 dth DgRr 7 2931 21 55.0 16.4 1020 205 0.0 0.0 0.0 0.0 0.0 0.0 40.6 38.0 DgRr 11 2558 22 48.4 17.8 329 189 0.0 0.0 0.0 6.6 0.0 0.0 26.3 45.7 DgRr 9 1854 20 43.4 14.2 1212 340 0.0 0.0 0.0 13.7 0.0 0.0 28.0 36.9 DgRs 9 1611 43 38.5 12.2 2506 195 0.0 0.0 0.0 0.0 0.0 14.4 29.5 34.7 DgRs 11 1274 113 44.4 15.1 4083 370 0.0 0.0 0.0 0.0 0.0 19.3 28.1 31.2 DgRr 12 500 6 36.2 11.9 450 194 0.0 0.0 0.0 0.0 0.0 3.2 29.9 45.5 DhRt 14 10 41 43.7 10.9 1464 113 0.0 0.0 0.0 0.0 0.0 28.0 21.(1 27.2 DhRt 15 291 25 39.9 9.9 1464 97 0.0 0.0 0.0 0.0 0.0 28.1 22* 27.9 A DgRr 6 212 45 34.1 10.3 3000 116 0.0 0.0 0.0 0.0 0.0 28.4 13.2 37.0 DhRt 17 66 93 25.0 6.6 3132 181 0.0 0.0 0.0 0.0 0.0 18.0 27.9 32.7 DhRt 22 0 328 0.0 0.0 0 328 38.0 33.7 0.0 0.0 0.0 0.0 0.0 6.9 DhRt 36 0 276 0.0 0.0 0 276 36.4 26.9 0.0 0.0 0.0 0.0 0.0 15.3 DhRq 19 436 108 7.2 3.5 2845 546 0.0, 26.6 0.0 0.0 0.0 22.7 0.0 29.3 DhRq 12 220 220 0.0 0.0 0 220 0.0 0.0 39.5 0.0 0.0 11.1 8.3 19.7 DhRq 13 96 96 0.0 0.0 0 96 0.0 0.0 38.0 0.0 0.0 0.0 15.0 25.6 DgRs 17 352 90 0.0 0.0 0 90 0.0 0.0 MM 9.6 0.0 0.0 0.0 29.1 DgRr 24 0 34 0.0 0.0 0 34 0.0 0.0 0.0 25.0 0.0 20.1 0.0 33.5 DhRq 21 271 119 14.3 5.5 1282 534 3.8 0.0 0.0 30.4 0.0 9.5 8.2 26.7 DhRs 19 69 273 14.8 4.3 1893 922 22.7 0.0 0.0 5.8 22.2 12.4 0.0 15.5 A DhRs 1 386 219 11.5 3.7 1658 565 23.8 4.5 0.0 9.0 12.0 12.0 0.0 17.3 A DfRs 3 129 286 18.1 5.9 4719 864 63.0 0.0 0.0 4.6 0.0 2.2 0.0 8.8 DgRs 36 399 541 7.3 1.4 6225 1038 50.1 0.0 0.0 11.2 0.0 0.0 16.5 0.0 DgRs 30 516 887 10.6 1.2 6330 1411 51.4 0.0 0.0 3.7 0.0 0.0 23.5 0.0 A DgRs 1 92 937 10.5 2.1 2128 1314 4.9 0.0 0.0 16.7 0.0 0.0 56.9 0.0 DgRs 7 565 35 2.9 1.5 3953 1048 0.0 0.0 0.0 7.0 0.0 0.0 39.4 32.2 A DgRs 2 772 192 24.8 9.9 2023 1518 4.0 0.0 0.0 0.4 o.o" 15.0 35.6 23.6 DhRt 74 204 879 27.0 4.4 4219 1092 0.0 0.0 0.0 0.0 0.9 51.3 26.1 0.0 A DhRt 4 118 611 11.7 1.8 5053 1142 25.6 0.0 0.0 0.0 3.2 49.8 0.0 0.0 A DhRt 3 421 420 11.8 2.8 4741 705 29.1 0.0 0.0 0.0 2.4 32.5 9.4 5.2 A DgRr 1 383 190 20.9 6.8 2035 569 2.6 0.0 0.0 0.0 0.0 36.9 20.5 18.6 DhRr 75 23 96 24.6 8.8 3625 908 0.0 0.1 0.0 0.0 0.0' 53.0 0.0 29.2 A DgRr 2 200 140 19.2 7.0 9970 607 0.5 0.0 0.0 0.0 0.0 47.2 7.5 30.9 A DgRs 4 431 1440 21.3 1.7 7450 2376 0.1 0.0 0.0 42.3 0.0 27.0 9.2 0.0 DgRs 15 1494 1576 0.0 0.0 0 1576 0.5 0.0 78.0 0.0 0.0 0.0 0.0 0.0 Note: The final site in the table, DgRs 15, is erroneously included in the third group because of an extreme DEPTH value. It belongs to the central group. 42 CM z O to z UJ Q -1 DhRt17 DgRr 6 DhRt 14 DhRt 15 DgRs 11 DgRs 9 DgRr12 DgRr 7 DgRr 11 DgRr 9 DgRr 2 DhRr 75 DgRr 1 v DgRs 2 A . DgRs 7 v \ DhRt 74 / DhRt 4 ^ / / DhRt 3 DgRs 4 DgRs 1 DgRs 30 DgRs 36 DfRs 3 DgRs 15 DhRs 19 DhRs 1 DhRt 22 DhRt 36 DgRr 24 DhRq 21 DhRq 19 DhRq 12 DhRq 13 DgRs 17 - 2 - 1 0 1 2 DIMENSION 1 A= Habitation sites O = Limited activity sites.petroglyphs, unknown, etc. Figure 28: Multidimensional scaling plot of topographic and vegetation variables combined. The multidimensional scaling plot (Figure 28) this time follows the cluster diagram much more closely. One interesting addition to the insights already provided by the first and second cluster exercises is the appearance of the DEPTH variable as an important factor in habitation choice. All of the sites in the upper group on the plot, which contains eight of the ten habitation sites, show much higher than average values for DEPTH (1122 m versus 342 m -see also boxed segments on lower section of Table 3). Again we have the somewhat unexpected demonstration of the importance of openness for long term site use. One other notable pattern in this plot is the strong grouping of low elevation sites in the lower left corner. All eleven of these sites share high concentrations of the first four vegetation types (stepwise shaded area of middle section of Table3), partly attributable to their common landscapes but nevertheless showing the value of vegetation in forming the groups we have seen here. 43 Summary The goal of this study was to employ a series of measurements taken from the landscape surrounding a group of archaeological sites and from those measurements make some statements about why those particular locations were chosen for occupation or other related activities. The first hypothesis simply stated that these locations were not randomly distributed in space. This was easily demonstrated by the distribution of most of the variables. NEAREST SITE, STREAM and SHORE for the actual sites all showed distinctive differences from the randomly plotted points. Of particular note was the importance of a close by shoreline (a median value of 95 m for actual locations versus 741 m for the random ones), thereby confirming the point Suttles (1951:46) had made in his ethnographic research. Nearby vegetation showed a definite pattern for a more diverse plant population within the immediate area of a site. Among the topographic variables (ELEVATION, SLOPE, and ASPECT), we again saw distinct patterns among the actual locations. In the case of ELEVATION, the sites were in a much tighter range; SLOPE showed a slight preference for higher values. The weakest differences were with SOILS, ASPECT, and, initially, the two viewshed variables, BREADTH and DEPTH, although interesting information was still produced. Hypothesis two is closely related to the first one and can be regarded as being accepted. Settlement to Marpole people meant not having great distances between neighbouring activity and occupation areas. A freshwater stream was always important, preferably within a walking radius of 250 meters. The shore truly was a "midline", always present, always a basic component of everyday life. This did not, however, directly translate into living on the beach. On average, living and working in this region meant staying about 20 m above sea level. A slightly higher than average slope was also a practical consideration. The direction a location faced was apparently, however, pretty much determined by the terrain. Vegetation was definitely selected for, partly for a varied range of plants within the immediate vicinity and also for stands of the large conifers, cedar and Douglas fir. The tendency for modern industrial-agricultural societies to destroy plant diversity in settlement areas tends to bias our thinking towards other cultures. It would make eminent sense for any group of people contemplating a settlement location to choose diversity even if the result of their lifeway might often result in its destruction. For the early Salish settlers it may have been more possible to preserve that local diversity than it would have been for more agricultural land users. The results for the soil analysis were not as clear-cut, but commonsense choices for solid ground and away from swampy boggy areas were demonstrated. The viewshed analysis was equivocal when it was initially looked at 44 as support for Hypothesis 2 since there appeared to be little difference between the general landscape and the local landscapes of the sites. It did, however, help to disprove a personal notion that rising headlands were somehow preferred in close proximity to the actual sites. Instead they appear as problems that simply had to be dealt with by the original settlers. The third hypothesis was by far the most contentious in this study with all of its implications of "predicting the past". Yet the use of landscape did provide some valuable insights into the problem. From the clustering and scaling routines it was apparent that when a major occupation was planned, a number of elements were looked for. The site was definitely not to be placed directly on the beach but rather, on average, about 20 m above sea level. A moderate slope to the land was required and an overall sense of open space was important, with a broad expanse in front and a greater than average depth behind the site. Although it may be stating the obvious in a region such as this, it can also be said that there was a preference for large tree types, especially cedar. In the case of the sites presumed to have had limited use, two general groups emerged, each aligned with two broad features of the terrain in the study area. This is most easily seen in Table 3 with one set of sites situated on narrow strips of land between the shore and the backing headlands; the other is found in flat open settings close to, or on, the river delta. The near absence of habitation sites within these two groups again demonstrates the key role of landscape in not only the choice of site location but also for the type of activity that took place there. Conclusion One of the chronic problems in the study of Northwest Coast archaeology is that "little settlement pattern work has been carried out" (Matson & Coupland 1995:313). An underlying objective of this study has been to investigate the applicability of using a theoretical construct, landscape, and, on a more technical level, a set of computer-based tools in a Geographic Information System to make statements about settlement using a collection of Marpole period sites as a test case. From the landscape a range of variables were used for analysis. When elements as different as local vegetation, elevation, soils, and distances to various features in the surrounding area help to show distinct patterns, it demonstrates how useful the landscape can be as a supplement to our often paltry archaeological data base. As has been already stated, up to this point, very little has been said either by ethnographers or archaeologists about why particular locations were chosen over others. 45 In the Summary section those landscape features that were important in the selection of sites were identified. Using these measurements from a large number of sites allows us increased confidence in making statements about site location in the area surrounding the mouth of the Fraser. How rigorously we apply these statements specifically to the Marpole period remains a problem, however. They can certainly be used in a general descriptive sense and thus provide us with insight into the way this land was used two thousand years ago. We cannot, however, extend this to say that this represents a Marpole pattern somehow distinct from a Locarno Beach pattern or a St. Mungo pattern. The landscape approach is certainly capable of contributing to such conclusions, but only after a comparative analysis beyond the scope of the test case approach employed here. The use of the GIS has, in many ways, made this analysis possible. The volume and the variety of information employed here present horrendous data management problems and many of the variable values would be impossible to obtain otherwise. For example, elevation and slope averages and areas from the 500 m site buffers would be most difficult to obtain manually and the viewshed data would simply not be available. When a high number of sites is factored in, anything beyond a dozen or so locations would involve an impractical amount of time and effort to obtain even a part of the information presented here. Once the critical components of the landscape are stored in the computer, any number of repeated queries and analyses are possible without a great deal of cost to the analyst. A major caveat must be made, however. Geographic Information Systems are not "user friendly" computer packages. With a project such as the one described here, the user requires a good working knowledge of all aspects of the system. This entails setting up the working environment, digitizing maps, correcting errors in the resulting spatial data, developing useable databases from that data, generating new forms of data from the various point, line, and text features that were taken from the maps, performing various analytic functions, and finally producing the sorts of outputs required to answer the research questions that initiated the process in the first place. Much time, effort, and money can be spent in trying to acquire pre-digitized maps. The maps that formed the base of the work done here, although provided free in this instance, could cost over seven thousand Canadian dollars. The basic knowledge required to use such a system, therefore may place prohibitive demands on the time, patience, and pocketbook of the researcher. Nevertheless, it can be a powerful and sophisticated tool in the study of human settlement space. 46 Bibliography Abbott, Donald. 1962 Preliminary report on the Beach Grove site DgRs 1. Report on file with the Archaeology Branch, Victoria, BC. Allen, Kathleen M.S., Stanton W. Green & Ezra B.W. Zubrow eds. 1990 Interpreting Space: GIS and archaeology. London, Taylor and Francis. Altschul, Jeffrey. 1990 Red flag models: the use of modelling in management contexts. In Allen et al. pp.226-238. Apland, Brian & Owen Beattie. 1972 Preliminary report on salvage archaeology at the Barnet site (DhRr 16). Report on file at the Archaeology Branch, Victoria. Apland, Brian & Ray Kenny eds. 1989 British Columbia Archaeological Impact Assessment Guidelines. Victoria, Ministry of Municipal Affairs, Recreation & Culture. Areas Consulting Archaeologists, Ltd. 1991 Archeological investigations at Tsawwassen, B.C. Vol I. Report on file, Archaeology Branch, Victoria. 1993 1993 Archaeological investigations at the Locarno Beach Site (DhRt 6), Vancouver, B.C. Report on file, Archaeology Branch, Victoria. Archer, David J.W. 1972 Preliminary report on salvage excavations at Musqueam North-East: DhRt 4. Ms. on file, Laboratory of Archaeology,. Univ. of British Columbia, Vancouver. Aronoff, Stanley. 1989 Geographic Information Systems: a management perspective. Ottawa, WDL Publications. Bailey, Trevor C. 1994 A review of statistical spatial analysis in geographical information systems. In S. Fotheringham & P. Rogerson eds. Spatial Analysis and GIS. London, Taylor & Francis. Barnett, Homer G. 1939 Culture Element Distributions, IX.Gulf of Georgia Salish.t7«i'vers/rv of California Anthropological Records 1(5):221-295. 1955 The Coast Salish of British Columbia. Westport, Conn., Greenwood Press. Bernick, Kathryn. 1989 Water Hazard (DgRs 30) Artifact Recovery Project Report. Report on file, Archaeology Branch, Victoria 1991 Wet Site Archaeology in the Lower Mainland Region of British Columbia. Vancouver, copyright Kathryn Bernick. Bernick, Kathryn & Rebecca J.Wigen. 1990 Seasonality of the Little Qualicum River wet site. Northwest Anthropological Research Notes 24(2):153-159. 47 Binford, Lewis R. 1980 Willow smoke and dogs' tails: hunter-gatherer settlement systems and archaeological site formation. American Antiquity 48(l):4-20. 1982 The archaeology of place. Journal of Anthropological Archaeology 1:5-31. Blanton, R.E., S.A.Kowalewski, G.Feinman, & J.Appel. 1981 Ancient Mesoamerica: a comparison of change in three regions. Cambridge, Cambridge University Press. Borden, Charles. 1950 Preliminary report on archaeological investigations in the Fraser Delta region. Anthropology in British Columbia 2:13-27. 1958 Unpublished field notes for the Brandrith Farm site, DgRs 4 on file at the Archaeology Reading Room, University of British Columbia. 1970 Culture history of the Fraser-Delta region. BC Studies 6 and 7: pp 95-112. 1976 A water-saturated site on the Southern Mainland Coast of B.C. In Dale Croes ed. The Excavation of Water Saturated Sites. Mercury Series, Paper No. 50, Ottawa, National Museums of Canada. Brandt, Roel, Bert J. Groenewoudt, and Kenneth Kvamme. 1992 An experiment in archaeological site location: modelling in the Netherlands using GIS techniques. World Archaeology 24(2):268-282. Broderick, Michael & Valerie Silvertz. 1979 Preliminary report on archaeological investigations of the Pitt River site (DhRq 21). Report on file at the Archaeology Branch, Victoria. Burley, David. 1979 Descriptive Report on the 1973 Archaeological Excavations at the Marpole Site (DhRs 1), Vancouver. Report on file in the Archaeology Lab, UBC. 1980 Marpole: Anthropological reconstructions of a Northwest Coast culture type. Dept. of Archaeology, Simon Fraser University publication No. 8, Burnaby. Bussey, Jean. 1985 Archaeological Components, Vancouver Heritage Inventory Phase II. Vancouver, Vancouver Museum. Butzer, Karl W. 1982 Archaeology as Human Ecology. Cambridge, Cambridge University Press. Calvert, Gay. 1970 The St. Mungo Cannery site: a preliminary report. BC Studies 6-7:54-76. Camilli, Eileen L. & James Ebert. 1992 Artifact reuse and recycling in continuous surface distributions and implications for determining land use patterns. In Rossignol & Wandsnider, pp. 113-136. Carmichael, David. 1990 GIS predictive modelling of prehistoric site distributions in central Montana. In Allen et al. pp.216-225. Carr, Christopher ed. 1985 For Concordance in Archaeological Analysis. Prospect Heights, Waveland Press. 48 Charlton, Arthur S. 1972 Noon's Creek and Belcarra: a preliminary report on excavations near Port Moody. In Roy L. Carlson ed. Salvage 71: Reports on salvage archaeology undertaken in British Columbia in 1971. Burnaby, Simon Fraser University. 1974 Archaeological investigations at the Cates Park site DhRr 8. Report on file with the Archaeology Branch, Victoria. 1980 The Belcarra Park site. Burnaby, Dept. of Archaeology, SFU Publication 9. Chatters, James C. 1987 Hunter-gatherer adaptations and assemblage structure. Journal of Anthropological Archaeology 6:336-375. City of Burnaby. 1993 Waterways of Burnaby: our natural and cultural heritage. Map from Heritage Advisory Committee, Environment and Waste Management Committee. Clague, John J., John Luternauer, & Richard Hebda. 1983 Sedimentary environments and postglacial history of the Fraser Delta and lower Fraser Valley, British Columbia. Canadian Journal of Earth Sciences 20:1314-1326. Coupland, Gary. 1991 The Point Grey Site: a Marpole spring village component.Canadian Journal of Archaeology 15:73-96. Cranny, Michael & Don Bunyan. 1975 Report on the archaeological survey of the north side of the Fraser River. Report on file, the Arcaeology Branch. Victoria, BC. Croes, Dale & Steven Hackenberger. 1988 Hoko River archaeological complex: Modelling prehistoric Northwest Coast economic evolution. In Barry Isaac, ed., Research in Economic Anthropology, Supplement 3, pp. 19-85. Greenwich, Conn., JAI Press Inc. Dewar, Robert E. & Kevin A. McBride. 1992 Remnant settlement patterns. In Rossignol & Wandsnider. Duff, Wilson. 1952 The Upper Stalo Indians. Anthropology in British Columbia Memoir No. 1. Victoria. Ebert, James. 1992 Distributional Archaeology. Albuquerque, U. of New Mexico Press. Eldridge, Morley & Alexander Mackie. 1993a Predictive Modelling and the Existing Archaeological Inventory in British Columbia. Report on File, Archaeology Branch, Victoria. 1993b Fraser River Intertidal Wet Site Survey. Report on File, Archaeology Branch,Victoria. Ellis, G. Lain & Michael R. Waters 1991 Culture and landscape influences on Tucson Basin Hohokam settlement. American Anthropologist 93:125-137. Euler, Robert C. & Susan M. Chandler 1978 Aspects of prehistoric settlement patterns in the Grand Canyon. In R.C. Euler & G.J. Gummerman (eds.), Investigations of the Southwestern Archaeological Group: an experiment in archaeological co-operation. Flagstaff, Museum of Northern Arizona. 49 Everitt, Brian S. 1993 Cluster Analysis, 3rd ed. London, Edward Arnold. Flannery, Kent V. ed. 1976 The Early Mesoamerican Village. Orlando, Academic Press. Foley, Robert. 1981 Off-site archaeology: an alternative approach for the short-sited. In Ian Hodder, Glynn Isaac, & Norman Hammond eds., Pattern of the Past. Cambridge, Cambridge University Press. Ford, Pamela. 1989 Archaeological and ethnographic correlates of seasonality: problems and solutions on the Northwest Coast. Canadian Journal of Archaeology 13:133-149. Forman, R.T. & M. Godron. 1986 Landscape Ecology. New York, John Wiley. Gaffney, Vincent, & Zoran Stancic. 1991 GIS Approaches to Regional Analysis: A case study of the island of Hvar. Ljubljana, Znanstveni institut Filozofske fakultete. Galois, Robert. 1994 Kwakwaka'wakw Settlements, 1775-1920: A geographical analysis and Gazeteer. Vancouver UBC Press. Geographic Calculator, The 1993 Version 2.0. Gardiner, Maine, Blue Marble Geographies. Geological Survey of Canada. 1978 Surficial Geology, Vancouver, British Columbia (Map 1486A) and Surficial Geology, New Westminster, British Columbia (Map 1484A). Ottawa, Dept. of Energy, Mines and Resources. Grabert, G.F. & C.E.Larsen. 1975 Marine transgressions and cultural adaptation:preliminary tests of an environmental model. In Wm. Fitzhugh ed., Prehistoric Adaptations of the Circumpolar Zone. The Hague, Mouton Publishers. Grabert, G.F. & R.L.Spear. 1976 Archaeological Investigations at Birch Bay State Park, 1976. Bellingham, Dept of Anthropology, Western Washington State College. Green, Stanton. 1990 Sorting out settlement in southeastern Ireland: landscape archaeology and geographic information systems. In Allen et al. pp.356-363. Haase, William R. 1983 Pueblo II and Pueblo III Settlement Patterns on Cedar Mesa, Southeastern Utah. Unpublished M.A dissertation, Washinton State University. Haggett, Peter. 1965 Locational Analysis in Human Geography. London, Edward Arnold (publishers) Ltd. 50 Ham, Leonard C. 1982 Seasonality, Shell Midden Layers, and Coast Salish Subsistence Activites at the Crescent Beach Site, DgRrl. Unpublished PhD Dissertation, Vancouver, Univ. of British Columbia. 1987 An Archaeological Heritage Resource Overview of Richmond B.C. Report on File, Archaeology Branch, Victoria, B.C. 1988 An Archaeological Resource Overview of the University Endowment Lands, Vancouver, British Columbia. Report prepared for the Musqueam Band Council, Vancouver, BC. Ham, Leonard C. & Owen Beattie. 1976 Report to the ASAB on disturbed human remains salvaged from DhRt 1. Report on file at the Archaeology Branch, Victoria. Ham, Leonard C , Michael Broderick, Stephanie Yip, & Rose Gindl. 1978 The Evaluation of Archaeological Sites in the Greater Vancouver Regional District: Proposals for development. Report on file, Archeology Lab, UBC. Ham, Leonard C , Arlene Yip, Leila Kullar, & Debbie Cannon. 1986 A Charles Culture Fishing Village. Report on file, Archaeology Branch, Victoria. Harris, Gerry. 1989 Vancouver's Old Streams. Vancouver, Vancouver Public Aquarium Association. Hill, Beth & Ray Hill. 1974 Indian Petroglyphs of the Pacific Northwest. Seattle, University of Washington Press. Hill-Tout, Charles. 1895 Later prehistoric man in British Columbia. Transactions of the Royal society of Canada, 2nd series, 1(2):103-122. Hobler, Philip. 1983 Settlement location determinants: an exploration of some Northwest Coast data. In R.J. Nash ed., 77ie Evolution of Maritime Cultures on the Northeast and the Northwest Coasts of North America. Burnaby, B.C., Simon Fraser University. Hunt, Eleazer. 1992 Upgrading site-catchment analyses with the use of GIS investigating the settlement patterns of horticultualists. World Archaeology 24(l):283-308. Ingold, Tim. 1992 Foraging for data, camping with theories: hunter-gatherers and nomadic pastoralists in archaeology and anthropology. Review essay. Antiquity 66(252):790-803. 1993 The temporality of landscape. World Archaeology 25(2):152-1T'4. Kellogg, Douglas. 1987 Statistical relevance and site locational data. American Antiquity 52(1): 143-150. Kenyon, Walter A. 1953 An archaeological survey of the Lower Fraser from Chilliwack to the Gulf of Georgia. Report on file at the Archaeology Reading Room, University of British Columbia. Kew, Michael. 1992 Salmon availability, technology, and cultural adaptation in the Fraser River Watershed. In Brian Hayden ed. A Complex Culture of the British Columbia Plateau. Vancouver, UBC Press. 51 Kidd, R.S. 1964 Lower Fraser archaeological site survey: report on the field season of 1963. Report on file at the Archaeology Reading Room, University of British Columbia. Kvamme, Kenneth 1992 Terrain form analysis of archaeological location through Geographic Information Systems. In Gary Lock & Jonathan Moffett eds. Computer Applications and Quantitative Methods in Archaeology 1991. Oxford, BAR International Series 5577. Kvamme, Kenneth & Michael Jochim 1990 The environmental basis of Mesolithic settlement. In C. Bonsall (Ed.) The Mesolithic in Europe: Proceedings of the third international symposium, Edinburgh 1985. Edinburgh, University of Edinburgh Press. Lepofsky, Dana Sue. 1985 An Integrated Approach to Studying Settlement Systems on the Northwest Coast: The Nuxalt of Bella Coola, B.C. Unpublished M.A. dissertation, Univ. of B.C. Marquardt, William & Carole Crumley. 1987 Regional Dynamics: Burgundian landscapes in historical perspective. Orlando, Academic Press. Maschner, Herbert & Jeffrey Stine. 1995 Multivariate approaches to site location on the Northwest Coast of North America. Antiquity 69:61-73. Matson, R.G. 1976 The Glenrose Cannery Site. Mercury Series, Paper No. 52. Ottawa, National Museums of Canada. 1981 Prehistoric subsistence patterns in the Fraser Delta: the evidence from the Glenrose Cannery site. BC Studies 48:64-85. 1992 The evolution of Northwest Coast subsistence. In Barry Isaac, ed., Research in Economic Anthropology, Supplement 6, pp. 367-428. Greenwich, Conn., JAI Press Inc. Matson, R.G. & D.L. True. 1974 Site relationships at Quebrada Tarapaca, Chile: a comparison of clustering and scaling techniques. American Antiquity 39(1):51-74. Matson, R.G., L.C. Ham, & D.E.Bunyan. 1979 Prehistoric Settlement Patterns at the Mouth of the Chilcotin River, B.C. Report on file at the Archaeology Laboratory, University of British Columbia. Matson, R.G., Deanna Ludowicz, & William Boyd. 1980 Excavations at Beach Grove in 1980. Victoria, Heritage Conservation Branch. Matson, R.G., William D.Lipe, & William R. Haase IV. 1988 Adaptational continuities and discontinuities: the Cedar Mesa Anasazi. Journal of Field Archaeology 15(3):245-264. Matson, R.G., H.Pratt, & L.Rankin. 1991 1989 and 1990 Crescent Beach Excavations, Final Report Vancouver, Laboratory of Archaeology, UBC. Matson, R.G. & Gary Coupland. 1995 The Prehistory of the Northwest Coast. San Diego, Academic Press. McGarigal, Kevin & Barbara Marks. 1993 FRAGSTATS: Spatial pattern analysis program for quantifying landscape structure. Corvallis, Oregon, Oregon State University. 52 McMillan, A. 1971 Excavation at Pigeon Cove, DhRr 9, Summer 1971. Report on file at the Archaeology Branch, Victoria. Mitchell, Donald. 1971 Archaeology of the Gulf of Georgia area, a natural regiion and its culture types. Syesis 4(1). 1983 Seasonal settlements, village aggregations, and political autonomy on the central Northwest Coast. In Elisabeth Tooker, ed., The Development of Political Organizationn in Native North America. Washington, Proceedings of the American Ethnological Association. Monks, G. 1976 Quantitative Comparison of Glenrose components with the Marpole Component from Site DhRt 3. In Matson 1976 pp. 267-280. Mueller, James W. ed. 1975 Sampling in Archaeology. Tucson, Univ. of Arizona Press. Naveh, Zev & Arthur Lieberman. 1994 Landscape Ecology: theory and application. New York, Springer-Verlag. North, M.E.A., M.W. Dunn, & J. Teversham. 1979 Vegetation of the South Fraser Lowland, 1858-1880. Map from Lands Directorate, Environment Canada. Northcote, T.G & P.A.Larkin. 1989 The Fraser River: a major salmonine production system. In D.P.Dodge ed., Proceedings of the International Large Rivers Symposium. Canadian Special Publications in Fisheries and Aquatic Sciences No. 106. Fisheries and Oceans Canada. Oke, Tim & John Hay. 1994 The Climate of Vancouver. B.C.Geographical Series, No.50. Vancouver, Dept. of Geography, UBC. Parker, Sandra 1985 Predictive modelling of site settlement systems using multivariate logistics. In Christopher Carr (Ed.), For Concordance on Archaeological Analysis. Kansas City, Westport Publishers, Inc. Patenaude, Valerie C. 1985 The Pitt River Archaeological Site DhRq 21: A Coast Salish seasonal camp on the Lower Fraser River. 2 Vols. Archaeology Branch, Victoria. Percy, R. 1977 The Liquid Air salvage project. Report on file at the Archaeology Branch, Victoria. Plog, Fred. 1971 Some operational considerations. In G.J. Gummerman (Ed.): The Distribution of Prehistoric Population Aggregates. Prescott, Prescott College Press. 1978 An analysis of variability in site locations in the Chevron Drainage, Arizona. In R.C. Euler & G.J. Gummerman (eds.), Investigations of the Southwestern Archaeological Group: an experiment in archaeological co-operation. Flagstaff, Museum of Northern Arizona. Plog Fred & James N. Hill. 1971 Explaining variability in the distribution of sites. In G.J. Gummerman (Ed.). The Distribution of Prehistoric Population Aggregates. Prescott, Prescott College Press. 53 Pokotylo, David L. 1978 Lithic Technology and Settlement Patterns in Upper Hat Creek Valley, B.C. Unpublished Ph.D. dissertation Univ. of British Columbia. 1982 A multivariate analysis of inter-assemblage variability in the Shield Archaic Tradition. In Margaret Hanna & Brian Kooyman, eds. Approaches to Algonquian Archaeology. Calgary, The University of Calgary Archaeological Association. Pomeroy, A. 1980 Bella Bella Subsistence and Settlement. Unpublished Ph.D. dissertation, Simon Fraser University. Post, Gerhard & Stuart Van Male. 1963 A preliminary report of an archaeological survey of the south and southeast shores of Burrard Inlet, British Columbia, Canada. Report on file at the Archaeology Reading Room, University of British Columbia. Roberts, B.K. 1987 Landscape archaeology. In J.M.Wagstaff ed. Landscape and Culture: geographical and archaeological perspectives, pp. 77-95. Oxford, Basil Blackwell. Rossignol, Jacqueline. 1992 Concepts, methods and theory building: a landscape approach. In Rossignol & Wandsnider, pp. 3-19. Rossignol, Jacqueline & LuAnn Wandsnider. 1992 Space, Time, and Archaeological Landscapes. New York, Plenum Press. Ruggles, Clive, David Medyckyj-Scott, & Alun Gruffydd. 1993 Multiple viewshed analysis using GIS and its archaeological application: a case study in northern Mull. In Jens Andresen, Torsten Madsen, & Irwin Scollar, eds. Computing the Past: computer applications and quantitative nethods in archaeology. Aarhus, Aarhus University Press. Sanders, W.T., J.R.Parsons, & R.S.Santley. 1979 The Basin of Mexico: Ecological processes in the evolution of a civilization. New York, Academic Press. Savage, Stephen. 1990 Modelling the Late Archaic social landscape. In Allen et al. pp. 330-355. Schermer, Shirley J. & Joseph Tiffany. 1985 Environmental variables as factors in site location: an example from the Upper Midwest. Midcontinental Journal of Archaeology, Vol 10 (2):215-240. Schlanger, Sarah. 1992 Recognizing persistent places in Anasazi settlement systems. In Rossignol & Wandsnider, pp. 91-112. Shennan, Stephan. 1988 Quantifying Archaeology. Edinburgh, Edinburgh University Press. Slaymaker, O., M.Bovis, M.North, T.R.Oke, & J.Ryder. 1992 The primordial environment. In Graeme Wynn & Timothy Oke, eds. Vancouver and its Region. Vancouver, UBC Press. Smith, Harlan. 1903 Shell heaps of the Lower Fraser River, British Columbia. Publications of the Jesup North Pacific Expedition 2(4). Memoirs of the American Museum of Natural History 4(4): 133-191. Snyder, Sandra L. 1991 Site location analysis in the Central Oregon Cascade Range. Northwest Anthropological Research Notes, 25(1):117-137. 54 Sproat, Gilbert. 1868 Boatmanship. In Tom McFeat ed. Indians of the North Pacific Coast (1987), pp.17-21. Struthers, April. 1971 Preliminary report - DhRr 17. Report on file at the Archaeology Branch, Victoria. Sutherland, Eileen. 1975 Report of an excavation at DgRs 11 English Bluff, Tsawwassen, 1969 and DgRs 9, Tsawwassen Beach, 1970, Archaeological Society of B.C. Report on file at the Archaeology Branch,Victoria. Suttles, Wayne 1951 Economic Life of the Coast Salish of Haw and Rosario Straits. New York, Garland. 1955 Katzie ethnographic notes. Anthropology in British Columbia Memoir No. 2. Victoria. 1987 Coast Salish Essays. Vancouver, Talonbooks. 1990 Central Coast Salish. In Wayne Suttles (Ed.), Handbook of North American Indians, Volume 7: Northwest Coast. Washington, Smithsonian Institution. SYSTAT for Windows, Version 5 Edition. 1992 Data. Evanston IL., SYSTAT, Inc. 1992 Graphics. Evanston IL., SYSTAT, Inc. 1992 Statistics, Evanston IL., SYSTAT, Inc. Thorn, Brian. 1992 Archaeological investigations at the Whalen Farm site (DfRs 3), 1949-50: Re-contextualizing Borden's Whalen Farm. Unpublished Honours paper, Vancouver, Univ. of British Columbia. Thompson, Gail. 1978 Prehistoric settlement changes in the Southern Northwest Coast: a functional approach. University of Washington Reports in Archaeology 5. Seattle. Trigger, Bruce. 1968 The determinants of settlement patterns. In K.C.Chang ed., Settlement Archaeology. Palo Alto, National Press. UBC Books n.d. Unpublished field notes and artifact catalogues on file at the Archaeology Reading Room, University of British Columbia. van Leusen, P. Martijn. 1993 Cartographic modelling in a cell-based GIS. In Jens Andresen, Torsten Madsen, & Irwin Scollar, eds. Computing the Past: computer applications and quantitative nethods in archaeology. Aarhus, Aarhus University Press. Vita-Finzi, C. & E.S. Higgs. 1970 Prehistoric economy in the Mount Carmel area of Palestine: site catchment analysis. Proceedings: Prehistoric Society 36:1-37. Warren, Robert E. 1990 Predictive modelling of archaeological site location: a case study in the Midwest. In Allen et al. pp.201-215. 55 Wheatley, David. 1993 Going over old ground: GIS, archaeological theory and the act of perception. In Jens Andresen, Torsten Madsen, & Irwin Scollar, eds. Computing the Past: computer applications and quantitative nethods in archaeology. Aarhus, Aarhus University Press.. Whittaker, Fran H. & Julie K. Stein. 1992 Shell midden boundaries in relation to past and present shorelines. In Julie Stein ed. Deciphering a Shell Midden. San Diego, Academic Press Inc. Willey, Gordon R. 1953 Prehistoric Settlement Patterns in the Virii Valley, Peru. Washington, Bureau of American Ethnology, Bulletin No. 155. 1989 Settlement pattern studies and evidences for intensive agriculture in the Maya Lowlands. In C.C.Lamberg-Karlovsky ed., Archaeological Thought in America. Cambridge, Cambridge University Press. Williams, Harry & M. Roberts. 1989 Holocene sea-level change and delta growth: Fraser River Delta, British Columbia. Canadian Journal of Earth Science 26:1657-1666. Williams, Leonard, David Hurst Thomas, & Robert Bettinger. 1973 Notions to numbers: Great Basin settlements as polythetic sets. In C.L Redman ed., Research and Theory in Current Archaeology, pp. 215-237. New York, John Wiley. Winterhalder, Bruce & Eric Alden Smith. 1981 Hunter-Gatherer Foraging Strategies: Ethnographic and archaeological analyses. Chicago. U. of Chicago Press. Wood, John J. 1978 Optimal location in settlement space: a model for describing settlement strategies. American Antiquity 43(2):258-270. 56 Appendix A TDHNTIFICATION Site # Name LOCATION Latitude: begin Longitude: begin SITE DATABASE RECORD FORM end end Rating SIZE Current Estimate <500 M 2 500 - 10000 M 2 10000 - 20000 M 2 20000 M 2 + TYPE (Shell Midden AGE Radiocarbon Judgemental CULTURE Late Marpole Locarno St Mungo Old Cord Date Artifacts Judgement COMMENTS / REFERENCES 57 Appendix B: Sites used in the landscape study. Site ID Site Name (Location) Site Culture Basis For Reference Type Inclusion 45-WH-1 Cherry Point Midd Marpole radiocarbon Grabert & Larsen 1975 45-WH- 9 Birch Bay State Park HAB Marpole radiocarbon Grabert & Spear 1976 45-WH-11 Birch Bay HAB Marpole artifacts Grabert & Larsen 1975 45-WH-17 Semiahmoo Spit LAS Marpole radiocarbon Grabert & Larsen 1975 DfRs 3 Whalen Farm HAB Marpole radiocarbon Thorn 1992; Matson & Coupland 1995 DgRr 1 Crescent Beach LAS Marpole radiocarbon Ham 1982; Matson 1992 DgRr 2 St Mungo Cannery HAB Marpole radiocarbon Calvert 1970; Ham et al 1986 DgRr 6 Glenrose Cannery HAB Marpole radiocarbon Matson 1976, 1981 DgRr 7 (White Rock) Petr association Hill & Hill 1974 DgRr 9 Ocean Park Beach Petroglyph Petr association Hill & Hill 1974 DgRr 11 Thousand Steps Petroglyph Petr association Ham 1982 DgRr 12 Sunburst Petroglyph Petr association Ham 1982 DgRr 24 Annacis Island Midd Marpole site size Eldridge & Mackie 1993b DgRs 1 Beach Grove HAB Marpole radiocarbon Abbott 1962; Matson etal. 1980 DgRs 2 Tsawwassen Midd Marpole radiocarbon Areas 1991 DgRs 4 Brandrith Farm Park HAB Marpole artifacts Borden 1958 DgRs 7 (Point Roberts) LAS site size site inventory form DgRs 9 Tsawwassen Beach LAS Late artifacts Sutherland 1975 DgRs 11 English Bluff Midd Marpole artifacts Sutherland 1975 DgRs 15 Blundell Rd. Basketry Isol Marpole radiocarbon Bernick 1991 DgRs 17 Tlekines HAB Late site size Eldridge & Mackie 1993b DgRs 30 Water Hazard Unkn Marpole radiocarbon Bernick 1989 DgRs 36 (Point Roberts) Unkn Marpole radiocarbon site inventory form DhRq 1 Noon's Creek LAS Marpole artifacts Post & Van Male 1963; Charlton 1972 DhRq 12 (Pitt Meadows) HAB Marpole artifacts site inventory form DhRq 13 (Pitt Meadows) LAS Marpole artifacts site inventory form DhRq 19 Mary Hill Site Unkn Late radiocarbon Eldridge & Mackie 1993b DhRq 21 Pitt River Midd Marpole radiocarbon Broderick et al. 1979; Patenaude 1985 DhRr 3 (Burrard Inlet) Midd site size site inventory form DhRr 6 Belcarra Park HAB Marpole artifacts Charltonl972, 1980 DhRr 8 Cates Park Midd Marpole artifacts Charlton 1974 DhRr 9 Pigeon Cove LAS Marpole artifacts, burial McMillan 1971 DhRr 15 Trading Beach LAS Marpole burial mounds site inventory form DhRr 16 Barnet Site LAS Marpole artifacts Apland & Beattie 1972 DhRr 17 Caraholly Midden LAS Marpole artifacts Struthers 1971 DhRr 18 Cove Cliff LAS Marpole artifacts site inventory form DhRr 20 Berry Creek Unkn burial mounds site inventory form DhRr 75 skwekwtexwqen Isol Marpole artifact Eldridge & Mackie 1993b DhRs 1 Marpole HAB Marpole radiocarbon Burley 1979 DhRs 2 Lumberman's Arch HAB Late artifacts site inventory form DhRs 4 (Stanley Park) Midd association site inventory form DhRs 5 Lost Lagoon Midd site size site inventory form DhRs 6 Second Beach Midd association site inventory form DhRs 7 Third Beach Midd site size site inventory form DhRs 16 (North Vancouver) LAS site size site inventory form I 58 DhRs 19 Liquid Air Midd Marpole artifacts Percy 1977 DhRs 21 (Stanley Park) Midd Marpole artifact site inventory form DhRt 1 Male Village (Musqueam W) HAB Late artifacts Ham & Beattie 1976 DhRt 2 Stselax (Musqueam E) HAB Late radiocarbon Borden 1970 DhRt 3 Musqueam N. ("Old") Midd Marpole radiocarbon Monks 1976 DhRt 4 Musqueam NE Midd Marpole radiocarbon Borden 1976; Archer 1972 DhRt 5 Point Grey LAS Marpole radiocarbon Coupland 1991 DhRt 6 Locarno Beach Midd Marpole radiocarbon Borden 1970; Areas 1993 DhRt 10 Wallace St. Unkn artifacts site inventory form; UBC Book A24 DhRt 13 Canoe Run LAS association site inventory form; UBC Book N16 DhRt 14 (Point Grey) Midd Marpole artifacts UBC BookN16 DhRt 15 (Point Grey) Midd Marpole artifacts UBC Book N16; Eldridge i Mackie 1993b DhRt 17 Booming Grounds Midd association Eldridge & Mackie 1993b DhRt 18 (Point Grey) Midd association site inventory form DhRt 22 Mah Farm Midd Marpole site size Ham 1987 DhRt 36 Tait Farm LAS Late radiocarbon Ham 1987 DhRt 74 Mc Cleary Golf Crse. LAS Marpole burial mound site inventory form Appendix C: Variable measurements. 59 Site ID Site Name (Location) Near Stream Shore Elevation Asp- Slope Breadth Depth Site ect 45-WH- 1 Cherry Point 5379 152 71 12.4 SW 4.4 2979 542 45-WH- 9 Birch Bay 2 3627 122 120 4.9 NW 1.6 5646 665 45-WH-11 Birch Bay 3627 196 48 6.5 SE 2.8 2571 455 45-WH-17 Semiahmoo Spit 4835 109 104 16.6 N 8.0 250 282 DfRs 3 Whalen Farm 2759 129 286 18.1 N 5.9 4719 864 DgRr 1 Crescent Beach 880 383 190 20.9 NW 6.8 2035 569 DgRr 2 St Mungo Cannery 789 200 140 19.2 NW 7.0 9970 607 DgRr 6 Glenrose Cannery 789 212 45 34.1 NW 10.3 3000 116 DgRr 7 (White Rock) 1772 2931 21 55.0 S 16.4 1020 205 DgRr 9 Ocean Park Beach Pet 882 1854 20 43.4 w 14.2 1212 340 DgRr 11 Thousand Steps Petr 882 2558 22 48.4 s w 17.8 329 189 DgRr 12 Sunburst Petroglyph 880 500 6 36.2 w 11.9 450 194 DgRr 24 Annacis Island 2330 0 34 0.0 0 0.0 0 0 DgRs 1 Beach Grove 884 92 937 10.5 N 2.1 2128 1314 DgRs 2 Tsawwassen 531 772 192 24.8 N 9.9 2023 1518 DgRs 4 Brandrith Farm Park 1216 431 1440 21.3 N 1.7 7450 2376 DgRs 7 (Point Roberts) 784 565 35 2.9 E 1.5 3953 1048 DgRs 9 Tsawwassen Beach 400 1611 43 38.5 SW 12.2 2506 195 DgRs 11 English Bluff 400 1274 113 44.4 SW 15.1 4083 370 DgRs 15 Blundell Rd. Basketr 1887 1494 1576 0.0 0 0.0 0 0 DgRs 17 Tlekines 1887 352 90 0.0 0 0.0 0 0 DgRs 30 Water Hazard 370 516 887 10.6 NE 1.2 6330 1411 DgRs 36 (Point Roberts) 370 399 541 7.3 NE 1.4 6225 1038 DhRq 1 Noon's Creek 640 51 312 28.8 SW 8.3 3715 881 DhRq 12 (Pitt Meadows) 627 220 220 0.0 0 0.0 0 0 DhRq 13 (Pitt Meadows) 627 96 96 0.0 0 0.0 0 0 DhRq 19 Mary Hill Site 2500 436 108 7.2 SW 3.5 2845 546 DhRq 21 Pitt River 2211 271 119 14.3 SE 5.5 1282 534 DhRr 3 (Burrard Inlet) 1421 48 61 26.4 SW 11.9 3250 895 DhRr 6 Belcarra Park 1296 471 15 32.4 W 17.5 1616 1504 DhRr 8 Cates Park 1804 623 63 21.4 S 11.4 1898 481 DhRr 9 Pigeon Cove 640 368 25 9.0 N 4.9 4356 584 DhRr 15 Trading Beach 670 93 27 47.9 S 15.5 882 172 DhRr 16 Barnet Site 1421 281 64 40.1 N 16.9 1422 469 DhRr 17 Caraholly Midden 2354 467 48 53.4 S 22.4 195 253 DhRr 18 Cove Cliff 1296 42 46 15.6 SE 6.6 1885 255 DhRr 20 Berry Creek 670 200 34 25.2 S 12.2 2087 412 DhRr 75 skwekwtexwqen 6731 23 96 24.6 SE 8.8 3625 908 DhRs 1 Marpole 663 386 219 11.5 S 3.7 1658 565 DhRs 2 Lumberman's Arch 741 508 52 13.3 E 3.9 7394 1449 DhRs 4 (Stanley Park) 161 318 23 3.0 SE 2.1 1769 642 DhRs 5 Lost Lagoon 697 683 375 13.2 SW 2.9 996 1545 DhRs 6 Second Beach 697 7 98 8.7 SW 2.8 2229 810 DhRs 7 Third Beach 1129 0 44 33.5 W 10.6 821 248 DhRs 16 (North Vancouver) 3430 282 50 18.2 SW 7.2 1045 553 DhRs 19 Liquid Air 663 69 273 14.8 SW 4.3 1893 922 DhRs 21 (Stanley Park) 161 236 48 3.0 SE 2.1 1769 667 DhRt 1 Male V i l l . (Musqm W) 347 158 201 5.0 SW 2.7 5436 470 DhRt 2 Stselax (Musqueam E) 133 193 461 11.7 W 1.8 5053 992 DhRt 3 Musqueam N. ("Old") 347 421 420 11.8 W 2.8 4741 705 60 DhRt 4 Musqueam NE 133 118 611 11.7 W 1.8 5053 1142 DhRt 5 Point Grey 2652 65 77 39.8 N 15.3 932 424 DhRt 6 Locarno Beach 1242 324 143 29.2 NE 10.6 1866 575 DhRt 10 Wallace St. 248 260 86 9.1 N 3.3 2720 947 DhRt 13 Canoe Run 2300 1773 83 58.3 SW 16.0 931 242 DhRt 14 (Point Grey) 281 10 41 43.7 s w 10.9 1464 113 DhRt 15 (Point Grey) 281 291 25 39.9 SW 9.9 1464 97 DhRt 17 Booming Grounds 1003 66 93 25.0 s w 6.6 3132 181 DhRt 18 (Point Grey) 248 53 183 12.5 N 4.1 2720 1044 DhRt 22 Mah Farm 1646 0 328 0.0 0 0.0 0 0 DhRt 36 Tait Farm 1646 0 276 0.0 0 0.0 0 0 DhRt 74 McClearyGolf Crse. 2027 204 879 27.0 s w 4.4 4219 1092 61 TJ C co Q i D i ID ca H ca o o t H O O \ O O O i t ~ ; i n i n q > o o N h M o d p K c o ' v d i o i o m d t n d r i T f co co cn C N O « O O r > - v q c n r > - ^ c n > O C N O c N O \ r ^ C > C O O rto'oi6deAin'^«iHt~inindr>r>N*ind (S N N M w -H CN CN cn ri a o 60 i 2 _ ca e o 60 o I* o u c o 60 ca |CN e o 60 ca n o eo >> Q i o q o q q q q q c ^ O P ' - ; e O O O i S O j d d d d d d d d d d d d d d d d d d d d d d d d d d d d d d d d d d 60 O O O O O O O O C J V O O d c J d o o d d c J o d d e> © © © © © © d d d d d d d u u PQ Pu < Q < "©•"©""•©"""©"""©"""q""©"-©"""©-^^  o d d 6 d d d d d ^ 6 c d d 6 d d 6 i f i d d d d d ^ r < i i n d 0H U e o U Q do ca u o * o o d CN _ q q q q q w ^ q a \ q q q o o T f r T j - © c > © d d d d d d o H ^ o ( d d d d d n d d d d £ U U q—©—6o"T7T-©—©"c©^"©"c©J"©"-c3\ d o \ i f i r ^ d d d d d d d 60U J3 u u PQ fc < Q ca u 60 < < £ OH _rTTT~uT~©—q""T>'T5—v7j5~©^ ^ d ^ f > d m * d d d c n ^ o d d d a 6 d N o d o \ c n c n i n d i ^ d d c N t ~ d d d c n ' v d c N ' d U U Q 4= J5 u u u pa m LL, < < Q u U U CA u ^ fc E PH UH OH cn U Q U Q Q U 60 cA U Q J3 U 0-E fc U HI K U Q S3 U U v o t ^ i o c N r t q c N c S r t ^ q T ^ O T ^ c o i n v o T f » t > m d ^ ' c s n d i » i r i i » r > ' t o \ d —i rt CS rt _ i ( s | r t r t r t rtrt w ^ ^ r t q i ^ < o o q q c N v 6 v 6 0 v 6 q O c J \ r t CN CN —< rt CN CS — C S C S C S C S C N r t c O C S C S U U Sf f i fc < U U Q Q Q Q U U Q U U U U U U e o U Q M ffi fc ffi X ffi 2 S ? U Q U U U U OH 60 U U fc fc 2 eoQ P U 60 U fc u Q t r i d w i » d ^ ^ ^ i f i \ d ' H f r i a * r t h o \ ^ a a c o « vocscncscNrtrtcNCNolcNcNcrirtCNt>-cncncNcncncNcncNcNcNcNTfc^ U U U S ^ f c f c > PQ O H O H e o Q U U Q Q Q Q U Q Q ^ Q Q Q 6 E P- - U .O ,-N " t-H u -8 " PQ 2 eo oo i i r& <C U ~ ~ U « a ac S S fc ° ac 6 o 6 o U U U U Q F ^ e o U rt CN _ CN CN r~ n . ' H i n r - O V O N W O N ' H v ! t r t r t r t r t c n c n ^ ^ ^ c N t ^ CT CT CT CT rt o\ ^ TI- in CN vo Tf ~ rt rt rt CN m r--M H H H H H H H H M M M ^ M ^ ^ C A C A C A C T ' C T C T Q H k H C A C A w « H 4 - » 4 - » 4 - » 4 - » ' W 4 - i C 60 6 o e o e o e o 6 o e o e o e a ) 6 0 6 o e o e o 6 0 6 0 6 0 6 0 6 0 j C ^ £ ^ ^ ^ x i x ! ^ x : j 3 J 3 j - 4 3 j 3 Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q 62 co CO o o t o f f l o o o i ^ i n L o q t D O i N ^ t N q H q q t ^ — _j - J : - as w oi co os t> ui o i> i> CM «d in o o6c6o t>o6tcuo iocoocooc^ i - t c o c o c o c o T f T f c o CM co co eo CM H N N N N H H CM CM CO OLqiqcNcoocoajoascoc^ o o ^ c d o c d t d o S o c D k r i c r i c n c r i c o ' o CN rH T f CM CM CM LOCO CO CM CM (Ni-H i-H lt> CO q q i o f f l o o r ] d ri ci h d d co CM CM CM <N ^ cN*cNTf&c3e><N.HcScici a, ( N c d ^ c d d o o c o o o L o i ^ O T f ' o S o o rv CO Tf CM CM H (N i-H rH rH CM lOHHCO^NINH LO t s q c i c > c i i > t i o , o s ^ c o T f ' o ' o o o o o c x J O L o c r J o c M rH CN rH T f to ^ N q q q T i i S q o u o o o o i C o o ' L^OOOoicOrHOOOOOcJlLOOOO iH CO c 3 o o CM o o o o o o o o o o o o o o o o o o< = ' < - ' c ' CO 6 H Lft o O O T f O O O O O O C O C O O CO CN 6 ( O I O 6 6 6 6 6 6 C J I 6 H 6 6 6 U 5 6 - * H & & & c 6 ^ c 6 L ^ r H c 6 ^ & « S & T f e i d d d m d i o i N o i i o d d d o o f f i 60 C c o Q > CO CM co O O O O O O O T f T f O O O O O O r H O LO LO CM CM CM CN co co b >< •3 c CD &, !< u 00 CQ CO-HWC^I^ai^^^rHCMTfr^ai^^^^COrHrH P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P P 63 ed < p p p p p p p O T f c o p o p o O p p p p p p p O O O O © © © © © © © © © © © © © c n O O O O o ' o o ' o o o o o ' o o o o o o o ' o e ' o s eu J3 eu "ca cd cu cd g eu o, c o Oil o 'S oo u s CD o, D < a T3 cu o .a e CD a, a, < CU cu oo a o 60 o o 60 "5 Q 'oo u < > 00 O O O O O O O O O O O O O O O O O O O O O O O O O O O O O © © v o c N o o o c n © © © ' © © ' © © ' © © © © ' © © © © © © o o o o o o o o o o X> > £ -a u O O o p o o p o OS Tf © © CN © © © © © © p © © © © © © p © H © cd © CN © cn o © © © v i cn © © cn © © © © © © © © O © © © o © © ©* < Tf on t*H 60 cd" <*H > X) < < x> X) X) "o Pi P-U oo 00 u U U cn CN r~; as CN o oo vq i n © © oo p © © © © © p cn OS oo © OS CN © -p-•p- l-H cd © cn H H r-* CN 00 Tf o i n rH i n © © vd © © © © © © © © ' o CN © >n vd © © Tf CN 2 H - 1 H H •—' HH rH r—' r-H H^ < on <*H 60 60 cd" U > cd ed X> X) C3 < TJ cd cd cd > < -a < x> CU < U U U < Po 00 U r> l> ?• CU oo U oo P H > D oo > > > 00 o Tf r-H_ "~! - H o Tf m i n cn o © p CN C-; © •—' © CN cn cn i n HH © OO Tf © HH 00 cd © cn H H od O as r-H CN vd © o © v i CN © r-° © vd C-" i n CN od Os o od od ©' r-" cn OS cu » - H CN H * CN r~« H H »—• rH CN cn rH < CN C o TJ 60 cd" cd" U x> cd" cd cd cu X) >» T3 CJ •a > -a cd u < X) > < < CJ u > cd < CJ U <: x> <: u "o FM U 6 , U C H U ? I H oo _u P H 00 > CO P H > P H > 00 U - H > oo U oo > Tf i n CN "n CN Ov vq cn oo VO O i n v i t-; cn © OS Tf vq Tf © Tf cn VO vo OS Tf Tf cn cd cn cn vd Tf cn r~ in cn Ov CN* © 00 Os i n © OS i n vd Tf cn 00 CN od © Os © Tf vd t-° , ', od cu cn CN CN CN CN CN ~- "~1 Tf cn Tf CN Tf m Tf Tf Tf CN cn Tf Tf cn cn cn CN VO r—I <4H Cj" CtH^  C«H cd cd < « > > -a oo > eu eu O cd - 9 X ) X ) C t H i X)X)X)X1X) > - a x ) X i x ) y < ; < < < t ; o t ) 0 O c d U c d U U U U U u cd o > - V O r - ^ - C N ^ - T f ^ ^ r Z ^ - r J . - ' ^ ^ c2e5r2e^ e^ 5r2r2e^  50 ^ o 60 jso ob ob w> w> S S S S S x l S S S S S S S S x l S x l Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Q Tf i n p^H HM ^ ' ( ^ M MH ^ ' | MH m — „ — 6060606 606060606043 Q Q D Q Q Q Q Q Q * 64 P o o o o o o o o o o c J o o o o o Tf e o 60 "3 1 ro r_ O os vq O o r-H VO un © rn CN © © < co g Polygc 60 < -O 00 U >i OH > VCb OO CN VO —; Tf p CN CO CQ CD t~ 0 0 rH . CN iri od o CN !/-> cn < e Polygc < -q ni -q eo > y > y < OH > OH > 00 -O iu u b OS VO CN OS p vq Tf rH Cd oi o CN (N 0 0 CO vd rn CN CO r-H r -cn cn CN Tf < c Polygc -O cd X> U > U T3 RT > OH > U O tt, u H- Cb Q cu •*-» oo u - . m T f r - o o c N v o T r <4—t —^» ^ 4>J +-* -4—» H)_> O ^ C ^ O H O ^ O ^ O ^ O H O H Q Q Q Q Q Q Q Q Appendix F - Vegetation Codes (from North et al. 1979) PLANT CODE PLANT TYPES 1. GRASSES sg sw s Salt Marsh: saltgrass, saltwort, sedge br s ct Tidal Marsh: bulrush, sedge, cattails ct Freshwater Marsh: cattails g Prairie: grass g W hh ca Prairie grass with shrubs: grass, willow, hardhack, crabapple 2. SHRUBS ca crabapple W willow W ca hh r Mixed shrubs: Willow, crabapple, hardhack, rose 3. SHRUBS/MOSS It cb P Labrador tea: labrador tea, cranberry, salal, Pine cb P Cranberry marsh: cranberry, Pine m P Moss with scrub pine: sphagnum, scattered Pine, Hemlock 4. WOODLAND M Maple bottom: Broadleaf maple, vine maple, ferns, some Cedar A Alder bottom: Alder, Willow, ferns, some Cedar.Hemlock, Spruce Cw Mixed woodland: Cottonwood, Alder, Willow, crabapple A B Ch Mixed deciduous regeneration forest: Alder, Birch, Cherry, Willow 5. SCRUB FOREST W sk Willow scrub: Willow, Alder, skunk cabbage W A pv rc Scrub with herbs: Willow, Alder, Hazel, ferns, pea vine, red clover PLANT CODE P H C P 6. CONIFEROUS FOREST (CEDAR) C P H C A s k C H S W S C C H D C M 7. CONIFEROUS FOREST (DOUGLAS FIR) PLANT TYPES Pine scrub: Pine species Mixed scrub: Hemlock, Cedar, Pine, some Douglas fir, Alder, Cherry Mixed coniferous forest on organics: Cedar, Pine, Hemlock, Spruce Cedar swamp: Cedar, Alder, Willow, hardhack, skunk cabbage Mixed wet: Cedar, Hemlock, Spruce, Alder, Willow, Yew, some Cottonwood Spruce: Spruce, Willow, Alder, crabapple Spruce: Spruce, Cedar, some Hemlock, Broadleaf Maple Mixed Coniferous: Cedar, Hemlock, Douglas fir, Alder, Willow Slope: Cedar, Broadleaf maple, Hemlock, some Douglas fir, Alder D F C D Mixed coniferous: Douglas fir, Grand fir, Cedar, some Hemlock, Pine Douglas fir: Douglas fir, salal, oregon grape, some Cedar, hawthorn Appendix G - Soil codes (from Geological Survey of Canada 1978) SOIL CODE SOIL TYPE 01 LANDFILL SAa sand, gravel, till, crushed stone, refuse 02 BOG, SWAMP, SHALLOW L A K E DEPOSITS SAb lowland peat overlaying Fb,c; SAc lowland peat underlyingFb SAd organic rich sandy loam to clay loam SAe upland peat 03 MARINE SHORE SEDIMENTS SAf sand to sandy loam SAg sand to gravel 04 LOWLAND AND MOUNTAIN STREAM DELTAIC, CHANNEL FILL SAh lowland stream channel fill SAi mountain stream marine deltaic medium to coarse gravel SAj mountain stream channel fill sand to gravel 05 FRASER RIVER SEDIMENTS Fa channel deposits, fine to medium sand Fb overbank sandy to silt loam Fc overbank silty to silt clay loam Fd deltaic and distributary channel fill Fe estuarine, fossiliferous interbedded fine sand 06 MARINE SHORE & FLUVIAL SAND SA-C postglacial and pleistocene marine shore & fluvial sand SOIL CODE SOIL TYPE 07 CAPILANO SEDIMENTS Ca raised marine beach, poorly sorted sand to gravel Cb raised beach, medium to coarse sand Cc raised deltaic and channel fill Cd marine and glaciomarine stony to stoneless silt 08 VASHON DRIFT & CAPILANO SEDIMENTS VCa bedrock within 10m or less of the surface VCb bedrock more than 10m below the surface 09 VASHON DRIFT Va lodgement tilland minor flow till Vb glaciofluvial sandy gravel 10 PRE-VASHON DEPOSITS (usually intebedded) PV a Quadra fluvial channel fill b Quadra deltaic deposits c Quadra marine interbedded fine sand d Coquitlam till e Cowichan Head organic sediments f Semiahmoo glaciomarine, glaciofluvial sediments 11 UNDIVIDED PRE-VASHON DEPOSITS UPV till, glaciofluvial, glaciolacustrine, fluvial, marine, & organic deposits 12 TERTIARY T Tertiary bedrock - sandstone, siltstone, shale, conglomerate 13 PRE-TERTIARY PT Mesozoic bedrock including granitic & associated rock types 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items