Journal of World Prehistory

, Volume 19, Issue 1, pp 47–118

The Archaeology of the Plateau of Northwestern North America During the Late Prehistoric Period (3500–200 B.P.): Evolution of Hunting and Gathering Societies


    • Department of AnthropologyThe University of Montana
  • James C. Chatters
    • AMEC Earth and Environmental Inc.
  • Michael Lenert
    • Department of AnthropologyUniversity of California
  • David S. Clarke
    • Department of AnthropologyThe University of Montana
  • Robert C. O'Boyle
    • Department of AnthropologyThe University of Montana

DOI: 10.1007/s10963-005-9001-5

Cite this article as:
Prentiss, W.C., Chatters, J.C., Lenert, M. et al. J World Prehist (2005) 19: 47. doi:10.1007/s10963-005-9001-5

The Plateau of northwestern North America offers one of the world's most important records of hunter–gatherer cultural diversity and evolutionary process. During the late prehistoric period, Plateau hunter–gatherers participated in a wide variety of mobile and sedentary mobility regimes, maintained diets emphasizing anadromous fish, roots, and larger game animals, and held patterns of social organization spanning egalitarian through ranked societies. In this paper, we provide a broad overview of the final 3500 years of human occupation on the Plateau that includes two primary goals. First, we provide a new chronology of late prehistoric cultural change and stability that integrates data from the Northern or Canadian Plateau and the Southern or Columbia Plateau. Second, we offer new ideas concerning the emergence, dispersal, and diversification of Plateau hunting and gathering societies. We close with recommendations for Plateau archaeology in the 21st century.


PlateauLate Prehistoric periodHunter–gatherersSocial organization


The Plateau of northwestern North America is an important context for the study of variability in hunter–gatherer behavior and cultural dynamics. At initial European contact, indigenous hunter–gatherer peoples of the Plateau, despite a substantial degree of linguistic homogeneity in many areas, participated in a wide range of socioeconomic systems that ranged from mobile foraging with egalitarian social relations to semi-sedentary collecting with ascribed social inequality (Prentiss and Kuijt, 2004a). The ancient history of the Plateau is probably complex, featuring population movements, in situ cultural evolution, and cultural transmission processes. Plateau archaeologists have sought to explain variability in subsistence intensification (Chatters, 1989a, 1995a; Galm, 1985; Kuijt, 1989; Lepofsky and Peacock, 2004; Thoms, 1989), mobility strategies (Chatters, 1987, 1995b; Chatters and Prentiss, 2005; Lohse and Sammons-Lohse, 1986; Prentiss and Chatters, 2003a,b; Prentiss and Kuijt, 2004b), technological organization (Andrefsky, 1995; Hayden et al., 1996a; Magne, 1985; Prentiss, 2000), social complexity (Hayden, 1994, 1997a, 2000a; Hayden and Adams, 2004; Prentiss et al., 2003, 2005a,b; Prentiss and Kuijt, 2004b), exchange (Blake, 2004; Hayden and Schulting, 1997), and conflict (Cannon, 1992; Chatters, 2004; Kent, 1980). With rare exceptions (e.g. Prentiss and Chatters, 2003a,b; Hayden and Schulting, 1997) these studies have tended to focus on particular sites or subregions within the broader Plateau. Similarly, the Plateau is characterized by multiple cultural chronologies organized by subregion (Ames et al., 1998; Andrefsky, 2004; Goodale et al., 2004; Pokotylo and Mitchell, 1998; Rousseau, 2004a; Stryd and Rousseau, 1996) with only one short regional synthesis (Chatters and Pokotylo, 1998).

There are three reasons why the time is right for a more comprehensive regional synthesis of the archaeology of the Plateau. First, in order to address questions of significance to the widest audience, Plateau researchers require a synthesis that provides access to the entire region. It is abundantly clear that change on the Plateau was rarely the result of entirely localized developments (e.g. Chatters, 1995b; Prentiss and Kuijt, 2004c; Rousseau, 2004a). Therefore, hypothesis development is impeded by lack of access to materials and patterns from adjacent subregions. This issue is made all the more critical by the fact that much of the primary literature in Plateau archaeology is in the form of not easily accessible cultural resource management reports and graduate theses. Second, substantial new data have accumulated even since the Chatters and Pokotylo synthesis of 1998 (e.g. Prentiss and Kuijt, 2004c), making a broad new synthesis even more critical. Finally, new research suggests that processes of culture change often did act on a regional scale or at least had regional impacts (Goodale et al., 2004; Kuijt, 2001; Kuijt and Prentiss, 2004). If we can recognize regional scale evolutionary processes, then we should also seek explanations on the same scale. This can only follow from a broad and updated synthesis.

In this paper, we seek to accomplish two major goals. First, we offer an updated synthesis of the Later Holocene (post 3500 b.p.) archaeology of the Plateau region. In this synthesis we focus on variability in four major areas: subsistence, settlement, technology, and social organization. Where appropriate, we also comment on exchange and conflict. Second, we pose new hypotheses regarding the evolutionary origins and diversification of late period socioeconomic patterns on the Plateau. We focus on three critical problems in our closing discussion: subsistence and demographics, emergence of new socioeconomic strategies, and the development of social status inequality.


For those unfamiliar with the Plateau, we provide a short overview of Plateau environments, paleoenvironmental changes since 4000 cal. b.p., and current views of culture history. This section is intended as a brief review; more detailed assessments are available elsewhere (Chatters, 1998; Prentiss and Kuijt, 2004a). We close the section with an overview and justification for our new synthetic view of the region's prehistoric chronology.
Fig. 1.

Map of the Plateau region.

Contemporary Environment

The Plateau (Fig. 1) is a diverse landscape lying between the Cascade/Coast Ranges to the west and Rocky Mountains on the east. It is bounded on the north by the Subarctic spruce forest of central British Columbia and on the south by the Basin and Range province of the western United States. The Plateau is further divided into a Northern, or Canadian Plateau and Southern, or Columbia Plateau. While this division is in part arbitrary, based on a contemporary political boundary (49th Parallel), there are some important physiological distinctions between the two areas. The Canadian Plateau, inclusive of the Fraser, Thompson and Okanagan drainages and adjacent mountains tends to be topographically more diverse than the Columbia Plateau, featuring narrow valleys and steep mountains in many areas. In contrast, much of the interior Columbia Plateau consists of rolling hills and broad plains, interrupted periodically by deep canyons, buttes, and mesas (particularly on its west side). Andrefsky (2004) and Rousseau (2004a) divide the Columbia and the Canadian Plateau at the Okanagan Highlands of north-central Washington and southern British Columbia. Due to the combination of topographic and seasonal weather regimes, the Plateau features what may be the most diverse array of plant communities in North America. The Cascade/Coast Ranges produce a substantial rain shadow effect that produces highly arid conditions across many areas of the interior Plateau. This is particularly evident across much of the Columbia Basin where vegetation ranges from sagebrush and greasewood steppe to open grasslands. On the Canadian Plateau, this pattern of highly xeric vegetation is found only in more narrow canyon and valley bottoms where the more dominant pattern is one of mesic and subalpine forests. An interior rainforest is found in southeastern British Columbia between the Columbia River valley and the Purcell Mountains, extending southward into the panhandle of Idaho. Food resources for hunter–gatherers included a wide variety of animal and plant species. Of critical importance to nearly all Plateau peoples were the combination of anadromous fish (especially salmon), geophytes or roots, berries, and ungulates. Despite relatively low populations of ungulate species, the massive salmon runs and productive geophyte patches along the Columbia and Fraser drainages set the economic stage for persistence of large and densely packed villages in the most productive segments (Fig. 2). Limited availability of salmon resources required peoples to maintain substantially higher mobility and smaller group size on the eastern margin of the Plateau.
Fig. 2.

Map of the Plateau region showing major sites and localities mentioned in text 1, Bridge River; 2, Keatley Creek; 3, Bell; 4, Lochnore-Nesikep Locality; 5, Kamloops Locality [Van Male, EeRb3; Harper Ranch, Curr]; 6, Scowlitz; 7, Slocan Narrows; 8, Vallican; 9, Arrow Lakes Locality [Deer Park, Inonoaklin]; 10, Kettle Falls Locality [Ilthkoyape]; 11, Chief Joseph Dam Locality; 12, Fish Hatchery; 13, Lenore; 14, Hatwai; 15, Alpowai Locality; 16, Hatiupuh; 17, Marmes Rockshelter; 18, Harder; 19, Miller; 20, Umatilla; 21, Wildcat Canyon; 21, The Dalles Locality [Five Mile Rapids, Wakemap Mound, Mack Canyon]; 23, Wells Reservoir Locality; 24, Avey's Orchard.


Our understanding of change in ancient environments has been steadily increasing over the past 20 years. As summarized in detail by Chatters (1995a, 1998), we recognize several major fluctuations in paleoclimates and associated subsistence resource distributions after 3500 cal. b.p. The period of ca. 4200–2400 cal. b.p. (peak Neoglacial period) is generally considered to have been cool and wet as indicated by alpine glacial advances (Burke and Birkeland, 1983; Reyes and Clague, 2004), expanded forest cover (Bennett et al., 2001; Mack et al., 1978a,b,c; Mehringer, 1985), reduced forest fire frequency (Hallett et al., 2003; Hallett and Walker, 2000), reduced sediment content in river systems (Chatters, 1995a), and greater numbers of salmon in annual spawning runs (Chatters et al., 1995a). Conditions became increasingly dry after 2400 cal b.p., followed by a pattern of slowly increasing moisture between 1800 and 1100 cal. b.p. Fires were frequent in many areas at this time, glaciers receded but then began to slowly expand in some areas (Reyes and Clague, 2004), forests declined then expanded (Chatters and Leavell, 1995), and somewhat fewer spawning salmon are likely to have made it into the Plateau (Chatters et al., 1995a) despite evidence for booming fish populations in parts of the northeastern Pacific (Finney et al., 2002). Cool and moist conditions at ca. 1100–1200 gave way to a new round of droughts between 900 and 600 cal. b.p., with more frequent fires (Chatters, 1996; Chatters and Leavell, 1995; Hallett et al., 2003), glacial recession (Reyes and Clague, 2004), more frequent, severe floods (Chatters and Hoover, 1992) sediment choked river systems (Chatters, 1998), and general decline in forest cover around the Plateau (Chatters, 1998; Hallett et al., 2003). After 500 cal. b.p., cool and moist climates returned with associated improved salmon spawning conditions. Fisheries data, however, suggest below average marine productivity in the eastern Pacific (Finney et al., 2002).

Throughout this paper we discuss environmental correlates with cultural patterns and events. As evolutionists, we do so under the assumption that the natural (and social) environment establishes selective conditions that influence evolutionary trajectories. Despite the desire of some anthropological archaeologists to see agent-based manipulation of belief systems as the primary force in cultural change (e.g., Pauketat, 2001; Thomas, 1991; Tilley, 1996), environmental conditions have significant impacts on cultural systems, particularly those of northern hunter–gatherers who are highly dependent on careful scheduling to harvest critical subsistence resources (Chatters, 1995b; Fitzhugh, 2003; Jones et al., 1999; Lepofsky et al., 2005; Mason and Gerlach, 1995). It is a major mistake to assume that the interactions between human populations and their environments play anything other than a major role in socioeconomic change in human societies (Redman, 1999).

Frameworks of Culture History

Many frameworks have been created for defining periodicity in Plateau prehistory. If we combine the Canadian and Columbia Plateau areas (Chatters, 1995b; Richards and Rousseau, 1987) we can recognize at least 14 distinct chronologies, defined for specific drainages or segments of major systems (e.g. Chief Joseph Dam, Kettle Falls, Mid-Fraser, South Thompson, etc.). Reasons for this complexity vary (Bicchieri, 1975; Lyman, 1985) but this proliferation of chronologies is largely a byproduct of how archaeological research has progressed within the region. As noted by Chatters (1995b) late period Columbia Plateau archaeology has primarily been pursued through large cultural resource management (CRM) contracts associated with major dams and reservoirs (Campbell, 1985; Chance and Chance, 1982; Chatters, 1986; Schalk, 1983a,b). Although dam construction has played little role on the Canadian Plateau, much research here too has been conducted in a CRM environment (Mohs, 1982; Richards and Rousseau, 1982; Wilson, 1992) with similar results. These projects have tended to produce an enormous amount of information useful for building local chronologies. Lack of access to unpublished material, tight reporting deadlines, and differing views on chronology may have contributed to reluctance on the part of many researchers to offer more synthetic views.

Despite the Balkanization of the Plateau chronology, there are some good syntheses on a subregional scale (Table I). Andrefsky (2004) divides the late Columbia Plateau chronology into Middle (5000–2000 b.p.) and Late Archaic periods (2000–200 b.p.). Chatters (1995b) also divides the late period Columbia Plateau chronology into two major periods termed Pithouse II (3500–2400 b.p.) and the Winter Village or Ethnographic Adaptive pattern (2400–200 b.p.). Rousseau (2004a) identifies change on the Canadian Plateau at the same point in time as Chatters but also subdivides the latest time block. Consequently, this chronology includes the Shuswap (3500–2400 b.p.), Plateau (2400–1200 b.p.), and Kamloops (1200–200 b.p.) horizons. Finally, Goodale et al.(2004) outline a similar sequence in the Upper Columbia drainage area that includes Collector I (4000–2000 b.p.), Collector II (2000–600 b.p.), and Collector III (600–200 b.p.).
Table I.

Subregional Cultural Chronologies of the Plateau

Columbia (Andrefsky, 2004)

Columbia (Chatters, 1995a,b)

Canadian (Richards and Rousseau, 1987)

Upper Columbia (Goodale et al., 2004)


Kamloops Horizon

Collector III


1200–200 b.p.

600–200 b.p.

Late Archaic


Plateau Horizon

Collector II

2000–200 b.p.

2400–200 b.p.

2400–1200 b.p.

2000–600 b.p.


Pithouse II

Shuswap Horizon

Collector I


3500–2400 b.p.

3500–2400 b.p.

4000–2000 b.p.

Middle Archaic

Pithouse I


5000–2000 b.p.

5000–3500 b.p.

Table II.

Summary of Cultural and Environmental Chronology for the Late Holocene on the Plateau



Climate period



3500–2400 b.p. (3600–2500 cal. b.p.)

Neoglacial (cool/moist)

First Classic Collectors


2400–1700 b.p. (2500–1800 cal. b.p.)

Transition to little climatic optimum (warm/dry)

Classic Collector throughout Plateau; Complex Collectors in Lower Columbia and Lower Fraser


1700–1200 b.p. (1800–1250 cal. b.p.)

Early little climatic optimum (warm/moisture increasing)

Complex Collectors in Mid-Fraser; Classic Collector population decline in north; rising numbers in the south


1200–700 b.p. (1250–650 cal b.p.)

Peak little climatic optimum (warm/dry)

Mid-Fraser Complex Collector villages at peak size and complexity early; Classic Collector populations peak, stabilize, and drop in south; Classic Collector populations generally rise in north


700–200 b.p. (650–250 cal. b.p.)

Little Ice Age (cooler/greater moisture)

Large villages abandoned in Fraser and Upper Columbia areas; Some large villages occupied in south; Classic Collector population expansion in north, but eventually reduced numbers throughout region

Table III.

Key Dated Housepit Sites by Time Period on the Plateau

Period I

Columbia Plateau: Ahsahka, Chief Joseph Dam group (six sites including 45OK11 and 45DO372), Clearwater Fish Hatchery, Hatwai, Three Springs Bar, Wells Reservoir Group (45DO72, 45OK74)


Upper Columbia: Arrow Lakes group (Cayuse Creek, Deer Park, Inonoaklin), Slocan Narrows


Canadian Plateau: Bell, Lochnore Creek, Mitchell, Van Male

Period IIa

Columbia Plateau: Alpowai locality, Lenore, Lyons Ferry Fish Hatchery, Mack's Canyon, Umatilla, Wildcat Canyon


Upper Columbia: Vallican


Canadian Plateau: Bell, Curr, Harper Ranch, Kamloops Reserve 1, Keatley Creek, Seton Lake

Period IIb

Columbia Plateau: Alpowai locality, Harder, Hatiupuh, Lenore, Miller, Paquet Gulch, Umatilla, Wildcat Canyon


Canadian Plateau: Bell, Bridge River, Curr, Gibbs Creek, Keatley Creek, Seton Lake

Period IIc

Columbia Plateau: Alpowai locality, Avey's Orchard, Chief Joseph Dam group (45OK2 and 45OK288), Ferry Canyon A, Hatiupuh, Lenore, Mack's Canyon, Three Spring Bar, Umatilla, Wildcat Canyon


Upper Columbia: Ilthkoyape, Slocan Narrows, Vallican


Canadian Plateau: Bell, Cow Springs, Harper Ranch, Keatley Creek, Seton Lake

Period IId

Columbia Plateau: Ferry Canyon B, Miller, Three Springs Bar, Umatilla, Wells Reservoir (45DO372)


Upper Columbia: Ilthkoyape, Slocan Narrows, Vallican


Canadian Plateau: Bridge River, Curr, Harper Ranch, Keatley Creek

One of our goals in this paper is to offer a new framework in which to make sense of the sometimes bewildering array of period, phase, horizon, and tradition designations. In order to accomplish this we identify a chronology based on variation in major cultural evolutionary patterns (Tables II and III). This is not a traditional culture history in the sense of North America's culture history paradigm (Lyman et al., 1997) where periods are based on groupings of normative culture traits, usually stylistically defined artifact types. We recognize that cultural evolution was continuous on the Plateau, at varying levels. However, to make sense of it on such a grand scale we need to subdivide the sequence into periods that can be understood in reference to major historical events and patterns. While we do outline groups of artifacts typical for each time frame, we assert that they are not the defining characteristics of each period. Rather, they are the byproducts of the human actions that maintained the structures of the cultural systems that we describe. Fortunately, some artifacts are somewhat time sensitive (e.g. projectile points) and can be used with caution as horizon markers. Since we are concerned with the sequence of historical events and processes, all radiocarbon dates are presented in their calibrated form. All researchers agree that a distinctive pattern (Period 1) can be recognized between 3600 and 2500 cal. b.p. that is characterized by small housepit hamlets and villages with fewer than 10 housepits that were used contemporaneously (within the limits of our ability to distinguish ages) and a salmon-oriented delayed-return economy (collectors per Binford, 1980 [see also Prentiss and Chatters, 2003b)]. It is clearly an adaptation to seasonal resource restrictions brought on by peak Neoglacial cool and moist conditions (Chatters, 1995b). We label the socioeconomy of this pattern “Classic Collector” because these are the original Plateau collectors who persist in many areas into Period II despite significant changes during that time in other areas of the region. We do not intend the term “Classic Collector” to be used as an essentialist type or entity. Rather we use the term heuristically to describe a structurally distinct class of hunter–gatherer adaptive organization.

Period II is longer and more complex, as it includes persistence of the former pattern and innovation or expansion of new variants of that design into the region. The major distinction between Periods I and II lies in the emergence and expansion of the corporate group household that originated on the Northwest Coast slightly before 2000 cal. b.p. (Lepofsky et al., 2005). The corporate group household is usually found in large and more permanent villages in settings with highly concentrated resources. As we discuss in more detail later, this strategy permitted its owners to out-compete its less well-organized neighbors where resources were concentrated and needed protection from competitors while some group members simultaneously utilized other more spatially disjunct resources. Some villages that contained these corporate groups also developed institutionalized status inequality expressed on an inter-individual and, more commonly, inter-household basis (e.g. Hayden, 1997a). These “Complex Collectors” (another heuristic concept) emerged and periodically expanded and retracted across multiple short intervals between 2500 and 250 cal. b.p., correlating closely with climatic changes. On the basis of these fluctuations, we subdivide Period II into four subperiods (Table II). Period IIa (2500–1800 cal. b.p.) includes the initial emergence of the Complex Collector strategy on the Northwest Coast while peoples of the interior experienced drought-induced resource stress and population decline. Period IIb (1800–1250 cal. b.p.) is characterized by the appearance of the new Complex Collector strategy into the Mid-Fraser area. Period IIc (1200–650 b.p.) was unstable, with fluctuating initial climatic conditions leading to droughts. The Complex Collectors of the Mid-Fraser and Upper Columbia systems experienced a population boom during the middle to late portion of this time. Period IId (650–250 b.p.) marks the development of Little Ice Age climatic conditions. The large villages associated with the Complex Collectors collapsed and a florescence of Classic Collector strategies occurred on the Canadian Plateau. Late in Period IId, many formerly large villages re-coalesced, but typically on smaller scales than were seen in Periods IIb and IIc. Complex Collectors expanded into the Lower Columbia area.


In this section, we seek to review basic archaeological data reflecting on settlement and mobility, subsistence, technological organization, and social organization during the Late Holocene of the Plateau. While we ask much of the record, we also realize that the data on which we base our discussion is highly variable in quality and detail. Following Chatters (1987), we attempt to assess six dimensions of settlement/mobility behavior including type, stability, frequency, scheduling, demography, and range. By type we mean residential versus logistical mobility. It is well known that after 3500 cal. b.p. all Plateau societies were logistically organized collectors (Chatters, 1995b; Kuijt, 1989; Prentiss and Kuijt, 2004c). Thus, we expect to recognize indicators in settlement patterns that include base camps, field camps, locations, and caches (per Binford, 1980, 2001). Stability is manifest in the chronological consistency of land-use patterns. Highly stable systems produce geographically redundant settlement patterns, more permanent houses and cemeteries. Less stable systems leave more varied occupational records, ephemeral residential facilities, and normally do not have formal cemeteries. Mobility frequency is a byproduct of duration of occupation associated with residential and logistical camps. We assess frequency by examining seasonality of occupation and duration of individual occupations. Demography tells us about human population size and structure during site occupations (e.g. Hassan, 1981) and among occupations throughout an annual cycle. We explore variability in demography through studies of site size and house floor configurations (e.g. Hayden, 1997a,b; Hayden et al., 1996b). Scheduling is a critical component of hunter–gatherer adaptations (Chatters and Prentiss, 2005). It is measured through assessments of relationships between activity variation, demography, and seasonality. Mobility range is a measure of the geographic extent of a mobility system. It is best measured using artifact raw material types that have known sources (e.g. Andrefsky, 1995; Hayden et al., 1996a).

We use the terms “forager” and “collector” extensively throughout the paper, relying on Binford's (1980) original definitions. In doing so, we recognize that Binford intended these distinctions to mark ends on a continuum of variability in hunter–gatherer mobility and subsistence behavior. As pointed out elsewhere (Chatters, 1987, 1995b; Prentiss and Chatters, 2003b), however, we also recognize that the structures of human socioeconomies may lock groups into one strategy or the other [or in other unique configurations such as the Clovis “high tech forager” strategy (Bamforth, 2002)]. The actual behavioral manifestations of that strategy and its seasonal application may vary widely, though, depending on social and ecological contexts; archaeologists have become increasingly adept at recognizing this kind of variation (Fitzhugh and Habu, 2002).

In this study we are interested in two major aspects of subsistence, emphasizing predation and consumption strategies. Consumption strategies may be interpreted as delayed or immediate return consumption, depending on the presence of storage features and anatomical distributions of faunal remains that indicate processing for storage (well-preserved salmon assemblages dominated by post-cranial elements, for example). By predation we mean the actual procurement strategies employed by past human populations. Where data permit, we explore prey specialization versus diversification using previously published measures of assemblage evenness and richness (e.g. Burns, 2003; Chatters, 1995b). By monitoring changes in prey spectrum (richness) and mode of procurement (evenness) through time we are able to examine patterns of intensification. Subsistence intensification is a process of change in subsistence behavior whereby we recognize an increase in energetic return in reference to land area or labor input (Morrison, 1994). Intensification can be adaptive or maladaptive. Higher returns per unit of labor can be highly adaptive and are usually accompanied by technological innovation (e.g. fishing nets) that permit, in the latter case, large quantities of an otherwise low-ranked resource to be harvested in aggregate, thereby raising its ranking compared to other resources (Butler and Campbell, 2004; Chatters, 1995a). In contrast, a population might produce more food per unit of land, but they might have to work harder, an ultimately maladaptive action that stresses both individuals and their resource base. An agricultural example of this is Boserup's (1966) cropping frequency model whereby more food is produced by an increasingly higher annual labor cost. For hunter–gatherers, the latter process may include harvesting progressively smaller food resources that require higher processing times or resources located at greater distances from settlements and thus requiring higher pursuit and transport costs (Broughton, 1994).

The quality of data on plant use on the Plateau varies more widely than any other available information (Lepofsky and Peacock, 2004). While most excavation reports list animal remains, far fewer provide adequate analysis and discussion of botanicals. Further, paleoethnobotanical research has concentrated in only select locales, notably the Mid-Fraser Canyon and immediate environs (including the Hat Creek and Botanie Valleys of British Columbia) and the Calispell Valley of northeastern Washington. Given the lack of intensive studies elsewhere, we are forced to rely on data from these studies as proxies for the larger region. Most of this information comes from roasting pits in field camps for plant processing, so we cannot fully reconstruct geophyte use from residential contexts such as housepit deposits where other resources were also consumed.

Winter-village collector systems are typically characterized by technological strategies that seek to extend the intra-annual utility of subsistence resources (e.g. Binford, 1977, 1979; Hayden et al., 1996a; Torrence, 1983, 1989). People used winter down-time to manufacture a wide range of tools that enabled them to mass-harvest resources during brief, productive seasons (Chatters, 1987). One indicator of this, in winter villages, is a consistently high ratio of small, informal flake tools to more-formalized bifaces and unifaces, and frequent appearance of various ground-stone tools such as abraders (Hayden et al., 1996a; Prentiss, 2000). The former are often highly fragmentary from reuse and trampling on house floors and result in part from the production of the complex wood, bone, and hide gear critical for spring and summer resource procurement (e.g. traps, nets, composite fish spears). Variability occurs, however, in the role lithic tools play in various resource procurement and processing strategies. Eastern Plateau collectors, for example, tended to favor hunting to a much higher degree than those of the west (Goodale et al., 2004; Roll and Hackenberger, 1998). Consequently, lithic assemblages tend to display higher frequencies of curated hunting gear (projectile points, hide scrapers, bifacial knives). Late period villagers in lower, western portions of the Fraser and Columbia drainages engaged in far more substantial wood working operations than many of those to the east because they constructed larger houses and possibly even wooden sculptures of household crest animals (Teit, 1900). Finally, we examine variability in social organization, emphasizing two issues. First, we ask how were household labor groups were organized within villages. In order to assess this, we rely on the record of activity differentiation seen in the spatial distribution of artifacts and features on housepit floors (Chatters, 1989a,b; Hayden, 1997a,b; Hayden and Spafford, 1993). The most complex societies of the Plateau featured multiple family units [corporate groups (Hayden and Cannon, 1982)] placed around edges of housepit floors, conducting redundant activities associated with routine food preparation and tool production. We can also recognize varying degrees of specialization between groups on these floors, supporting the inference of a diversified economy (e.g. Hayden, 1997a). Smaller and less-complex households tended to subdivide floor space by activity type rather than residence group. As we argue later, the advent of the co-residential corporate group marks a major threshold in Plateau prehistory. Second, we examine the record for indicators of variation in status organization. Hayden and colleagues have defined substantial evidence for inequality in the Mid-Fraser area on the basis of household archaeology and burials (Hayden, 1997a; Schulting, 1995). Hayden (1997a) used variation in utilitarian and prestige items between household domestic units and between houses to successfully argue for status inequality at Keatley Creek. On the basis of child burials with elaborate grave goods, Schulting (1995) goes so far as to argue for the presence of ascribed status inequality in richer resource areas after 1500 b.p. While we occasionally discuss burial data as a partial indicator of variation in social status, we also recognize that this requires caution. Few burials have been systematically excavated for any time period, particularly on the Canadian Plateau, so sample sizes are too small to support meaningful generalizations. We also recognize that it can be tricky to interpret burial contexts; mortuary treatment and grave goods might reflect individuals' positions in life (Schulting, 1995), but they can be ambiguous in their messages, especially when small samples are involved (Goldstein, 1980; Hodder, 1982; Hofman, 1986).

Population Dynamics

There have been a number of attempts to model population dynamics on the Plateau using changes in radiocarbon date frequencies as proxies for population size. Richards and Rousseau (1987) made an early attempt, looking at differences between the Mid-Fraser Canyon and other areas of the Canadian Plateau. This was the first formal recognition that the dynamics of Mid-Fraser populations may have differed from those elsewhere, especially after 2000 b.p. Peacock (1998) produced another plot of date frequencies, this time combining all Canadian Plateau dates. Lenert (2001) plotted housepit floor dates comparing Mid-Fraser and other Canadian Plateau distributions. The most sophisticated approach to this problem was taken by Chatters (1995b, 2004) who used a correction factor to overcome the expected loss of sample material with the passage of time.

In this paper we update those chronologies using calibrated data. As Chatters' data is recent and comprehensive, we focused on collecting Canadian Plateau data for an updated look at this region. We collected 260 Canadian Plateau (Lillooet and Fraser/Thompson regions) dates from CRM reports, academic reports, dissertations, published materials (Richards and Rousseau, 1987; Stryd and Rousseau, 1996), and the Canadian Archaeological Radiocarbon Database. We calibrated dates using the Radiocarbon Calibration Program CALIB REV4.4.2 (Stuiver and Reimer, 1993) which uses calibration data provided by Stuiver et al. (1998). Dates were calibrated to 95.4%—2 sigma—calibrated age ranges. We then calculated means for each calibrated date range. These calibrated means were then counted at 20-year intervals for plotting distributions. When multiple mean dates from a site fell within the same interval, the means were “stacked” and counted as 1 dated component for that particular interval. For example, if we were plotting calibrated means at 20-year intervals and if the Bridge River site contained date means of 1121, 1131, 1137, these dates were “stacked” by counting them as a single representative dated component for the 20-year interval of 1120–1139. Canadian Plateau dates (separating Mid-Fraser from other Canadian Plateau dates) are presented in Fig. 3. The Columbia Plateau plot (Fig. 4) provides a calibrated version of data outlined by Chatters (1995b), which differs in being an integral of the probability curves for all dates, represented as cumulative numbers of dates per 20-year interval. We refer to the results of these data plots throughout the rest of the paper.
Fig. 3.

Plot of trends in calibrated radiocarbon dates from the Canadian Plateau.
Fig. 4.

Plot of trends in calibrated radiocarbon dates from the Columbia Plateau (adapted from Chatters, 1995a,b).

Use of radiocarbon date frequencies as markers of population dynamics does offer some challenges. It can never be an absolute measure of population size, but it does offer an at least ordinal assessment of population dynamics. Factors other than population might affect distributions. Higher group mobility could produce more sites and thus more dates for examples. On the Columbia Plateau, however, a reduction in the frequency of residential mobility during the Middle Holocene (from the Cascade Phase to Pithouse I) is inversely correlated with rapid increases in numbers dated sites (Chatters, 1995b). The same phenomenon is recognizable on the Canadian Plateau when stable villages replace the more ephemeral residential hamlets after about 2000 cal. b.p. (Richards and Rousseau, 1987; Rousseau, 2004a). A counter argument to the latter case might be that research on big sites leads to more dates, thus biasing the record. This problem is at least partially overcome when each site is only counted once per time interval (Lepofsky et al., 2005). If anything, this will tend to underestimate population size as it does not take into account variability in the sizes of groups that occupied the sites. Similarly, concentrated research in one area could lead to regional sampling bias. Fortunately, this problem has been at least partially overcome by extensive cultural resource management work throughout many portions of the Plateau that otherwise have not been the subjects of problem-oriented research. Consequently, we make note in this paper of places in the chronology where dates seem insufficient as population proxies. Otherwise, we assume that the radiocarbon date proxy provides our best current indicators of relative variation in population dynamics.

Period I: Classic Collectors (3600–2500 cal. B.P.)


Between 5300 and 3600 cal. b.p., housepit dwellers of the Columbia Plateau [Pithouse I (Columbia Plateau)] and the Baker site on the Canadian Plateau placed residential sites in ecotones with optimal access to a variety of resource patches, made little use of logistical mobility or storage, harvested an exceptionally wide array of food resources, and organized stone tool production around a variety of formally shaped bifaces, unifaces, and blades (Chatters, 1986, 1995b; Chatters and Prentiss, 2005; Prentiss and Kuijt, 2004c). This low-mobility forager pattern rapidly declined after 4100 cal. b.p. and, as we have detailed elsewhere (Chatters and Prentiss, 2005; Prentiss and Chatters, 2003a,b), was replaced by the collector strategy, which spread into the interior northwest after ca. 3600 cal. b.p. as Pithouse II and the Shuswap Horizon. The organization of the new communities was distinctly different.

Pithouse II and Shuswap horizon peoples were logistically organized collectors. Settlement patterns include residential bases in the form of hamlets and small villages of pithouses, associated with a variety of field camps and procurement locations (Chatters, 1995a; Rousseau, 2004a). Storage facilities in the form of cache pits are found in residential sites. Residential settlements typically include small groups of housepits, rarely exceeding ten. It is not clear in many cases whether even these small sites were the result of accretion or simultaneously occupied housepits. Regardless, occupation sizes were small. One possible exception could be the Slocan Narrows village in southeast British Columbia where two parallel rows of large housepits, some dating to ca. 2700–3000 cal. b.p., are found on opposite sides of the Slocan River (Prentiss et al., 2001). Full confirmation of this pattern as a Period I phenomenon will have to await further dating. Residential hamlets are located in contexts that provide access to specific resources such as migrating salmon. Field camps are also systematically placed to emphasize game, roots or riverine resources such as mollusks (Chatters, 1989a; Lohse and Sammons-Lohse, 1986).

Period I housepits have a number of commonalities in basic structure but vary widely in size (Figs. 5 and 6). Typical houses range from 5 to 12 m in diameter (Chatters, 1986; Campbell, 1985; Sappington, 1994; Turnbull, 1977; Wilson, 1980), although much larger examples are known from the Mid-Fraser and Upper Columbia areas of British Columbia. Sanger (1970) documented small to medium (7–12 m) Shuswap horizon houses in the Lochnore-Nesikep locality of the Mid-Fraser canyon. Stryd (1973) documented a possibly much larger house (about 16 m) at the Bell site dating to Shuswap horizon times. Prentiss et al. (2001; see also Goodale et al., 2004) described two very large houses (17 and 22 m diameter) dating to nearly 3000 cal. b.p. at the Slocan Narrows village. Despite varying in size, Period I housepits are typically associated with very little rim midden, which are collections of artifacts, faunal remains and particularly floral materials, deposited around the perimeters of pithouses through interior hearth and floor clean-up, re-roofing events (old roof material placed out on rim), and outside activities (e.g. Hayden, 1997a). Floors are typically thin, with highly variable degrees of debris accumulation. Sites from Wells Reservoir and Chief Joseph Dam along the Mid-Columbia contain no evidence for sustained single occupations, but do have evidence for frequent reoccupations in the form of multiple floors, numerous artifacts, and scuffed (re-arrangement of archaeological materials by foot traffic [e.g. Chatters, 1987)] feature material (Campbell, 1985; Chatters, 1986, 1995b). The Slocan Narrows houses contain little floor or rim midden and exceptionally few artifacts. The Arrow Lakes sites also have thin floors and only moderate numbers of artifacts and floor features (Turnbull, 1977). Mid-Fraser and South Thompson sites appear to have thin floors and little rim material, but there are substantial numbers of features and artifacts on the floors (Richards and Rousseau, 1982; Sanger, 1970; Stryd, 1973).
Fig. 5.

Map of the Van Male site (EeRb10), a small Period I village (adapted from Richards and Rousseau, 1982).
Fig. 6.

Plan map of a housepit floor, Van Male site (adapted from Richards and Rousseau, 1982).

Fully excavated house floors consistently contain a single, usually central hearth, indicating residents lived as a single economic unit; repeated domestic units, which appear in Period II are not evident. Given the large sizes of some Upper Columbia and Mid-Fraser houses, it is possible that some groups temporarily explored aspects of corporate group organization [co-residential, multi-family groups acting as economic units (Hayden and Cannon, 1982)]. The information on artifact and feature patterning on house floors is insufficient, however, for us to determine if redundant residential units exist in the households or if there is any variation in the activities of household families.


There are relatively little zooarchaeological data on Shuswap horizon predation practices (Rousseau, 2004a). Nevertheless, Rousseau (2004a) suggests that diets included a broad range of resources, with an emphasis on salmon, mussels, and various medium to larger mammals (e.g. beaver, deer, elk). There is little evidence for intensification of geophytes or other plant resources (Lepofsky and Peacock, 2004; Peacock, 1998). In contrast, there is extensive direct evidence for Pithouse II subsistence behavior on the Columbia Plateau. Chatters (1995a,b), and Butler and Campbell (2004) demonstrate a pattern of subsistence, characterized by comparatively low evenness and richness, clearly indicating resource specialization emphasizing salmon and large ungulates. Chatters also describes predation tactics that include targeting of freshwater mussels. Relying on extensive data from the Calispell Valley project, (Andrefsky, 2004; Andrefsky et al., 2000) and Thoms (1989) document a process of camas (a geophyte requiring wet meadows) intensification during Period I on the Columbia Plateau. Chatters (1995a), see also Chatters and Prentiss (2005), argues that a pattern of “technogenic” subsistence intensification is recognizable during this period, associated with technological innovations imported from the Northwest Coast.


Period I technologies include a core and flake based lithic technology and a group of bone and antler tools that can be traced to the Locarno Beach Phase of the Northwest Coast (Fig. 7). Nearly all Shuswap and Pithouse II sites are dominated by a lithic technology organized around production of flake tools from small to medium sized prepared cores. Prentiss (1993, 2000) calls this the winter-village lithic technological organization, which he first identified in the Period Ic villages of the Mid-Fraser on the basis of ethnographic research and a study of lithic artifacts from the final floor of Housepit 7 at Keatley Creek (see also Hayden et al., 1996b). Of particular importance to Hayden et al.(1996a) are the expedient block core, biface, portable long-use, bipolar, and groundstone cutting strategies. Hayden et al.(1996a) suggest that small cores were shaped during visits to quarries embedded into food gathering trips during the autumn season, though this would be difficult to confirm. These cores, they suggest, transported to the semi-subterranean houses and used throughout the winter as a source of flakes for expedient tools (expedient block core strategy) and more formally designed cutting, drilling, and scraping tools (portable long-use tools). By late winter, with raw material running short, remaining larger flakes, tools, and exhausted block cores were split using bipolar methods (bipolar strategy). Meanwhile, various groundstone tools (e.g. adzes, saws) were created and used for heavy-duty tasks such as woodworking and sawing nephrite. The economic logic behind this pattern of organization was to use winter “down time” (e.g. Binford, 1979) to prepare a wide variety of tools and clothing for intense use during the busy warm season. This strategy, with local variations, appears to dominate assemblages throughout Periods I and II throughout the Plateau region.
Fig. 7.

Artifacts of Period I: (a–d) stemmed projectile points; (e) key-shaped uniface; (f) bone bead; (g) bone bracelet or pendant fragment; (h) bone disk beads; (i) non-toggling harpoon tip; (j) bone bi-point; (k) nephrite adze; (l) zoomorphic hand-maul (Van Male site); (m) antler wedge; (n) portion of a toggling harpoon valve; (o) toggling harpoon tip.

During Period I, cores and tools are typically derived from raw material sources close to the winter villages, which leads to considerable variability in material quality. Richards and Rousseau (1987) note that raw materials used during the Shuswap horizon are often of lower quality than those used before and after this period. A similar observation has been made for the Lower Snake (Leonhardy and Rice, 1970) and Mid-Columbia regions. The inference that can be drawn from this pattern is that Period I peoples moved within small resource-gathering ranges (Chatters, 1995a).

Period I exhibits a high degree of variability in projectile point styles (Fig. 7). Stemmed points are common in both the north and south, but stem morphology ranges from contracting to expanding, split base. Rousseau (2004a) observes stylistic similarity between split-base stemmed points of the Plateau and those from the McKean complex of the Great Plains region (Frison, 1991). Chatters and Prentiss (2005), and Prentiss and Kuijt (2004c) argue for similarities between Plateau contracting-stem points and similar variants found in the Locarno Beach Phase of the Northwest Coast (Matson and Coupland, 1995). Bi-pointed or leaf-shaped projectile points are also found throughout the Plateau during this period. Sites on the Columbia Plateau [Three-Springs Bar, Wells Reservoir, Chief Joseph Dam, Arrow Lakes (Upper Columbia)] have produced a number of point styles not typically found to the north, including side and corner-notched variants (Chatters, 1986; Daugherty et al., 1967; Lohse, 1985; Turnbull, 1977). Despite high variability in their blade dimensions, Period I point types have substantial hafting areas, suggesting their use as tips for atlatl darts.

A variety of other lithic and bone tool types is recognized on the Plateau during Period I (Fig. 7). Split cobble or cobble spall scrapers (scrapers created on flakes bearing all or nearly 100% dorsal cortex from outer surfaces of large cobbles) are found in some sites, such as Three-Springs Bar (Daugherty et al., 1967). Rousseau (1992) documents “key-shaped formed unifaces” on the Canadian Plateau and argues that these were used for preparing atlatl darts. Convex-edged “end scrapers” are found throughout the Plateau, but are particularly common to the north (Rousseau, 2004a). Although end scrapers are typically considered to be hide working tools, Rousseau (2004b) suggests that some may have been used in fish processing. Groundstone tools include cylindrical pestles primarily on the Columbia Plateau (Chatters, 1986; Sappington, 1988; Turnbull, 1977), hand mauls generally associated with the Canadian Plateau (Richards and Rousseau, 1982), and occasional adze bits of nephrite (Grabert, 1968; Richards and Rousseau, 1987; Turnbull, 1977). Bone tools from the Plateau include awls, needles, points, bi-points, wedges, leister tips, and harpoon valves (Campbell, 1985; Daugherty et al., 1967; Richards and Rousseau, 1982, 1987). As Chatters and Prentiss (2005) point out, the harpoon parts and wedges are virtually identical to those of the central Northwest Coast at ca. 3000–4000 cal. b.p. (Matson and Coupland, 1995).

Social Organization

Period I housepit floors illustrate a consistent pattern of household spatial organization featuring central hearth features and peripheral distributions of cache pits and post-hole features (Chatters, 1995b; Richards and Rousseau, 1982). There is no sign of multiple domestic units on these floors. Rather, variation in household debris seems to be the result of the often-discrete spatial segregation of activities of single domestic units. Several extremely large Period I housepits have been identified in the Canadian Plateau and Upper Columbia subregions (Goodale et al., 2004; Rousseau, 2004a; Sanger, 1970) and it is possible that multi-family units (possibly corporate groups) could have occupied these places. These features have not been completely enough excavated, however, for us to evaluate this possibility. Likewise, there is no evidence for differences in prestige or status between households (Chatters and Pokotylo, 1998; Rousseau, 2004a).

Although there is evidence for exchange (Ames et al., 1998; Richards and Rousseau, 1987), there is little to suggest a well developed prestige exchange network of the kind envisioned by Hayden and Schulting (1997) for later times. Burial evidence, although very rare and poorly reported, suggests little socioeconomic differentiation between individuals during Period I (Schulting, 1995).


Period I hunter–gatherers of the Plateau were logistically organized collectors with the ability to target, intensively harvest, and store a variety of foods, particularly emphasizing salmon and large ungulates in the major river valleys. A delayed-return economy permitted these groups to successfully endure the long Neoglacial winters. Population proxies suggest a high degree of success marked by rising populations in the south (Fig. 4) and stability in the north (Fig. 3). Social groups were rarely very large and rim middens and floor debris in many housepit sites indicate shorter residential stays than are characteristic of later periods. Multiple reoccupations in many sites do, however, suggest some degree of stability in mobility strategies, and use of predominately local lithic materials suggests small mobility ranges. Socioeconomic egalitarianism seems to have characterized social relations.

Period IIa: Emergence of Complex Collectors (2500–1800 cal. B.P.)


Period IIa is typified by continuation of the Classic Collector strategy in almost all areas of the Plateau. Two important changes were, however, beginning. The first of these was an independent evolutionary process underway on the Northwest Coast that appears to have led to the emergence of more Complex Collectors in the lower valley and delta of the Fraser River, although current evidence indicates that those changes did not appear on the Plateau until the beginning of Period IIb. The second pattern of change is not associated with the organization of Plateau societies, but with changing demographic patterns and increases in mobility range. Chatters (2004; but see Lyman, 2000) documents a decline in human populations on the Columbia Plateau indicated by reduced frequencies of radiocarbon dates between ca. 2300 and 1800 cal. b.p. (Fig. 4). This decline is particularly evident along rivers and at some important root gathering areas. He also notes that evidence for bow and arrow use, aggregated settlements, and increased signs of violence appear at this time on the Columbia Plateau. Conversely, populations on the Northern Plateau population appear to have increased at this time (Fig. 3). These developments are associated with an increasingly dry climate and its associated impacts on food resources that likely included some unpredictability and declines in salmon runs and variable effects on productivity of different root resources and ungulate prey.

On the central Northwest Coast, populations increased rapidly after 2500 cal. b.p. and then peaked and rapidly declined or coalesced into larger local groups after ca. 2200 cal. b.p., hitting bottom by around 1800 cal. b.p. (Lepofsky et al., 2005). This period is generally known as the early Marpole Phase and is characterized by the emergence of the first large, multi-family, corporate-group households in the region (Matson and Coupland, 1995). The earliest example of these (ca. 2500 cal. b.p.) is found at the Scowlitz site in the Lower Fraser Valley of British Columbia (Lepofsky et al., 2000). Early Marpole villages are characterized by rows of rectangular houses positioned along river or beach fronts. Houses can be over 15 m long, featuring redundant domestic units, among which occur varying degrees of specialization in crafts and food production (Ames and Maschner, 1999; Burley, 1980; Chatters, 1989b; Matson and Coupland, 1995). From a labor standpoint, this strategy permitted a household economic unit to exploit multiple resources simultaneously. It would also have permitted simultaneous production of a wide variety of goods for domestic use or exchange. Most fundamentally, it would have held together a working group of families with complementary areas of expertise beyond the temporary alliances that are more typical of less-aggregated, generalized hunter–gatherers. Despite this complex labor organization, early Marpole probably did not feature ascribed status differentiation (Matson and Coupland, 1995). When placed in an environment where resources were predictable and clustered, but impermanent, this strategy would offer important competitive benefits over other, less-differentiated strategies (e.g. Classic Collectors).
Fig. 8.

Map of the Fish Hatchery site, a small Period IIa village (adapted from Schalk, 1983a).

While village sizes increased in some areas of the Plateau, house sizes generally declined (Fig. 8). Villages with more than 10 housepits (in some cases possibly greater than 50) appear during Period IIa in both the north and south. These larger clusters of houses are typically associated with optimal fishing places such as rapids or the lower reaches of major trunk streams. Larger sites along the Columbia River cluster around The Dalles and include Umatilla (Schalk, 1980), Wildcat Canyon (Dumond and Minor, 1983), and Mack's Canyon (Cole, 1967, 1969). Period IIa occupations are also found at the Vallican site, a large village in the Upper Columbia drainage (Mohs, 1982). Hayden (1997a, 2000b) asserts that there were Periods I and IIa occupations at the Keatley Creek village in the Mid-Fraser canyon of British Columbia, but reanalysis of the stratigraphy and radiocarbon dates from key excavations suggest that a densely aggregated village with its very large houses did not emerge at this site until ca. 1600 cal. b.p. (Prentiss et al., 2003). It does appear likely, though, that some smaller housepits were occupied at Keatley Creek during Period IIa. The nearby Bell site may also have experienced limited occupation during this time (Stryd, 1980). Rousseau (2004a: Fig. 1–10) provides a spectacular historical photograph of an exceptionally large housepit village (EeRb3) at the confluence of the North and South Thompson Rivers in Kamloops, British Columbia. He asserts that the village dates to Period IIa and early IIb, based on limited investigations (Wilson, 1980). Unfortunately, it has since been destroyed, preventing more extensive dating studies. None of these larger housepit villages contain firmly dated large houses (rim crest diameter exceeding 14 m). Small hamlets and villages containing mid-size to smaller houses (less than 14 m rim crest diameter) are also found throughout the Plateau including the Alpowai sites (Brauner, 1976), Lyons Ferry Fish Hatchery (Schalk, 1983a), Site 45DO372 of The Wells Reservoir Project (Chatters, 1986), Knight Creek (Hackenberger, 1993), Harper Ranch (Wilson, 1980), and Curr (Carlson, 1980).

Housepits from Period IIa generally have thin floors and relatively little midden development on the rims. Kamloops villages [EeRb3 (Wilson, 1980), Harper Ranch (Wilson, 1980), and Curr (Carlson, 1980)] have consistently very thin floors, limited rim middens, and cache pits, along with moderate numbers of artifacts and faunal remains (Richards and Rousseau, 1982, 1987; Rousseau, 2004a). A similar pattern is presented in the Mid-Fraser area at sites such as Bell (Stryd, 1973). Thin floors and highly variable debris accumulations are present in key Columbia Plateau villages such as Mack's Canyon (Cole, 1967, 1969), 45DO372 (Chatters, 1986), and Lyons Ferry Fish Hatchery (Schalk, 1983a). Data are insufficient for an evaluation of rim midden sizes at most sites, although rim middens at 45DO372 were limited.

Other data on site content and distribution clearly indicate that mobility was logistical. Chatters (1995a) documents a variety of sites including field camps on upland plains and river terraces. He notes that these sites show indicators of intensive, repeated use that, in some cases, spans all of Period II. Site types include hunting camps (faunal remains dominated by larger game such as deer or sheep; lithic artifacts dominated by bifaces and projectile points) and root processing locations [presence of root roasting pits, and varied lithic assemblages, but often including woodworking tools (digging stick manufacture and resharpening)]. Of particular interest is the brief increase in frequency of large game kills during Period IIa. A good example is the Tsulim site (Chatters et al., 1995b), ca. 2200 cal. b.p. bison kill located on the west-central Columbia Plateau.


The Period IIa subsistence strategy remains similar to that described for Period I. As indicated by the settlement pattern data, logistical organization continues, with emphases on specific resources such as salmon, medium to large mammals, and geophytes as targets for intensive harvest and, often, delayed consumption. The pattern of salmon intensification, recognized in Period I, continues in Period IIa. Many Columbia Plateau villages provide evidence for significant use of medium to larger mammals including bison (Lyman, 2004). Some of these include Bone-in-Throat (Schalk, 1983a), Lyons Ferry Fish Hatchery (Schalk, 1983a), Umatilla (Schalk, 1980), Alpowai locality (Brauner, 1976), and Mack's Canyon (Cole, 1967, 1969). Bison are also found at camps and kills including Tsulim (Chatters et al., 1995b), Bafus (Morgan, 1993), 45GA17 (Schroedl, 1973), and Ford Island (Fryxell, 1962). Chatters (2004) demonstrates that the highest incidences of bison on the Plateau correlate inversely with inferred human population sizes. He concludes that reduced hunting pressure may have played a fundamental role in the increased visibility of bison in Period IIa. Declining population size and a drier climate may also play a role in apparent declining use of camas on the Columbia Plateau. Goodale et al.(2004) summarize evidence from various large scale excavation projects (e.g. Andrefsky et al., 2000; see also Thoms, 1989) demonstrating a near collapse in camas processing at ca. 2000 cal. b.p. in the Upper Columbia area. This may reflect a marked decline in the productivity of wet-meadow adapted camas during this drought interval. In contrast, harvest and processing of dry-meadow adapted geophytes (Turner, 1992) seems to be on the rise on the Canadian Plateau during Period IIa (Lepofsky and Peacock, 2004; Peacock, 1998) in response to the same climatic stimulus. Human population dynamics appear to correlate positively with these patterns, since populations drop in the south but appear to rise in the north (Chatters, 1995b, 2004; Peacock, 1998; Rousseau, 2004a) suggesting that access to carbohydrates may have had a serious affect on viability of collector populations across the Plateau. As argued by Kuijt and Prentiss (2004), carbohydrates provided by root crops were critical nutrients to collectors dependent on lean protein for winter survival.

That foodstuffs were being processed for delayed consumption is evident in the appearance of extramural storage facilities, the earliest of which are dated at McGregor Cave on the Lower Snake River (Hicks and Morganstein, 1994). This use of storage features in caves appears to coincide with increased use of upland resources and greater mobility range, but may also suggest that people found it necessary to hide their foodstuffs while absent from their base camps (Chatters, 2004).


The winter village core–flake-dominated pattern of lithic technological organization, described for Period I, continued in Period IIa. There is a distinct decline in frequency of groundstone at many sites, particularly in the south. Overall, the quality of lithic tools appears to have improved as fine-grained raw materials from a wider geographic range were increasingly used. Despite the dominance of simple flake tools, the assemblages also include finely made bifacial knives, large corner- and basal-notched projectile points, and various unifacial scrapers including key-shaped, formed unifaces (Fig. 9). Chatters (2004) argues that bow and arrow technology appeared by approximately 2300 cal. b.p. on the Columbia Plateau, although it took several hundred more years before it became the dominant projectile technology across the region (Rousseau, 2004a). According to Rousseau, bow and arrow technology did not appear on the Canadian Plateau during Period IIa. However, extensive research outside the region suggests that distinction of arrowheads from dart points is not always easily done and remains a subject of continuing debate (Hughes, 1998; Shott, 1997; Thomas, 1978). Bone tools are not well documented for Period IIa, although the pattern associated with Period I remains recognizable (e.g. Chatters, 1986). In addition, Rousseau (2004a) notes the presence of digging stick handles on the Canadian Plateau, which should not be surprising, given the evidence for root intensification.
Fig. 9.

Artifacts of Period IIa: (a and b) corner-notched projectile points; (c) key-shaped uniface; (d) end scraper; (e) harpoon (leister) tip; (f) bone bi-point; (g) net weight; (h) bone bead; (i) bone “button” (Keatley Creek site); (j) digging stick handle.

Social Organization

There is no evidence for status inequality in Period IIa. House sizes are consistently small to moderate, with little evidence for occupancy by more than single-family groups (Goodale et al., 2004; Rousseau, 2004a). There is also little evidence for status distinctions in a burial record that consists of cremations and simple inhumations with limited grave goods (Schulting, 1995). Exchange networks on the Plateau appear to have expanded beginning in Period IIa, eventually forming what Hayden and Schulting (1997) call the Plateau Interaction Sphere (PIS). There is, however, little to suggest that any sort of elite-driven trafficking in prestige goods had developed (Rousseau, 2004a). Despite the lack of evidence for status differentiation during this time, it is interesting that larger villages began to form on the Southern Plateau. Chatters (2004) points out that not only do larger housepit aggregations begin in Period IIa, but we also recognize, for the first time, fortified encampments on mesas and caves with hidden food caches. Chatters (2004) explores the possibility that the bow and arrow may have facilitated an increase in the lethality of conflict and a heightened concern for defense during this time. Population decline also suggests that there was probably some economic stress affecting competing groups. Subsistence stress and threat of violence from competitors may have led to population aggregations in key resource contexts, such as large villages near prime fishing sites, and the use of fortifications for protection from raiding and to guard access to key food sources.


Period IIa was characterized by a persistence of the Classic Collector strategy throughout the Plateau. While some central Northwest Coast groups experimented with more complex multi-family corporate groups in resource-rich river valleys (developing into Complex Collectors), interior peoples typically favored single-family households. Despite periodic aggregations, there is substantial evidence to suggest that Period IIa collectors were at least as mobile as those of the previous Period I and probably had a larger mobility range. High quality lithic materials, decline in the intensity of root use in some areas, and limited occupational debris at many housepit sites suggests even the possibility of increased residential mobility. Warfare, at least in the south, appears to have favored village nucleation. There is insufficient evidence to evaluate aggregation patterns in the north.

Period IIb: Complex Collectors on the Plateau (1800–1250 cal. B.P.)


Late in Period IIa, Marpole Phase Complex Collectors of the central Northwest Coast appear to have entered something like an evolutionary bottleneck. Populations declined and villages were abandoned between 1800 and 2000 cal. b.p. (Lepofsky et al., 2005). After 1800 cal. b.p. Marpole had changed, and now featured ascribed status inequality, as indicated by burial practices and differences in the proportions of prestige goods within and between houses (Burley and Knusel, 1989; Lepofsky et al., 2005; Matson and Coupland, 1995). Marpole populations subsequently rose throughout the following 450 years. Meanwhile, significant developments were occurring on the Plateau. Complex Collectors appeared in the Mid-Fraser Canyon of British Columbia after 1800 cal. b.p. and persisted throughout this period. Classic Collector populations in the north appear to have peaked early but declined later, while those of the south apparently rebounded throughout the period. Climate was very dry at the beginning of Period IIb, but moisture slowly increased thereafter. The availability of camas appears to have increased in the south, while dry-meadow geophytes are likely to have gradually declined in the north as forests expanded into meadows. With rising marine productivity (Tunnicliffe et al., 2001) and wetter interior conditions, salmon populations in the Columbia and Fraser systems were probably high between 1600 and 1200 cal. b.p.
Fig. 10.

Map of the Bridge River site, a large Period IIb village.

In Period IIb, villages in the The Mid-Fraser Canyon were large, whether measured by numbers of houses or total, house areas (Fig. 10). Important examples include Bell (Stryd, 1973), Bridge River (Prentiss et al., 2004; Stryd, 1974), and Keatley Creek (Hayden, 1997a,b, 2000a; Prentiss et al., 2003, 2005b). Village size varies from the Bell Site, with 23 housepits and 8 “flats” (possible earthworks or filled-in housepits) to Bridge River with 80 housepits and Keatley Creek with approximately 115 [Hayden and Ryder (1991) estimate that probably no more than 50% of the Keatley Creek houses were simultaneously occupied]. More significant are the sizes of many individual housepits. Houses in the medium to large range (generally 10–20 m rim crest to rim crest) are common in these villages. Keatley Creek features several housepits that exceed 20 m in diameter. Mid-Fraser housepits typically contain numerous large cache pits and extensive rim middens. Floors at Keatley Creek and Bell that date to Period IIb tend to be thin (less than 20 cm) while those from Bridge River are highly variable, but can be thick (20–130 cm) due to multiple reoccupations and accretion of floor sediments. Debris accumulations on floors and in rim middens are often substantial (Hayden, 1997a; Prentiss et al., 2003).

Until recently, it has not been possible to accurately assess the role of housepit accretion in the formation of these villages. Recent high-precision dating of 55 housepits at the Bridge River site (Prentiss et al., 2005b) suggests that while the villages did form through an accretional process, the number of simultaneously occupied houses appears to have ranged from 15 to 40, supporting Hayden's (1997a) contention that some of the Mid-Fraser villages were indeed very large.

Population proxies suggest that Classic Collectors of the Canadian Plateau suffered a significant population decline during Period IIb (Fig. 3; see also Prentiss et al., 2005a,b; Rousseau, 2004a). This is illustrated by the abandonment of major villages such as Kamloops Reserve (EeRb3) after ca. 1600 cal. b.p. Subsequent housepit occupations are rare and contain only a few, small, briefly occupied houses, as illustrated by the Curr site (Carlson, 1980).

In contrast, Columbia Plateau populations appear to have rebounded from their pre-1800 cal. b.p. low, demonstrating steadily rising numbers of villages and other sites during Period IIb (Fig. 4; see also Chatters, 1995a,b, 2004). Particularly large numbers of sites are recognized in the Lower Columbia area and include Umatilla (Schalk, 1980), Wildcat Canyon (Dumond and Minor, 1983), and Paquet Gulch (Jenkins and Connolly, 1994). A rectangular housepit dating to early in Period IIb, measuring at least 12 m in length, was excavated in the Calispell Valley (Dohm, 2000). The Lower Snake River area appears to have experienced a rapid rise in numbers of housepit sites. Major Lower Snake housepit sites include Alpowai (Brauner, 1976), Hatiupuh (Brauner et al., 1990), Harder (Kenaston, 1966), Miller (Schalk, 1983b), and Lenore (Sappington, 1994). Likewise, there was an apparent overflow of population into Hells Canyon, a deep defile farther up the Snake River. The Upper Landing village, and larger houses at Tryon Creek (Hackenberger and Thompson, 1995) and Dry Creek (Chatters et al., 1995a,b) are occupied solely during this interval. Villages typically remain small, with most sites containing fewer than 10 housepits. Large villages such as Miller and Umatilla, do exist, but it is unclear if occupations during Period IIb contribute significantly to their overall size. Most sites show signs of frequent reoccupation in the form of multiple floors and numerous (and sometimes overlapping) internal pits. Most also tend to show somewhat higher occupational longevity in larger artifact assemblages and thicker rim middens than are found in earlier Columbia Plateau villages. Housepit rims at these sites are still relatively shallow when compared to the Mid-Fraser housepits at Keatley Creek and elsewhere, however. One exception is Tryon Creek, where House 1 middens are particularly extensive (Hackenberger and Thompson, 1995). Settlement patterns remain generally similar to those of Period IIa.


Evidence for subsistence strategies on the Canadian Plateau is best known from the Mid-Fraser area for Period IIb. Data from the Keatley Creek site indicate intensive harvest of salmon, particularly sockeye (O. nerka). Dog remains frequently appear in housepit deposits in the Mid-Fraser area (Crellin, 1994; Crellin and Heffner, 2000; Burns, 2003; Rousseau, 2004a). Dogs appear to have served in a variety of capacities ranging from companionship, refuse disposal, and transport (Crellin and Heffner, 2000; Rousseau, 2004a) to food items (Burns, 2003) and possibly, victims of sacrificial rites (Hayden, 1997a). Faunal resource use at Keatley Creek's Housepit 7 shows a steadily increasing focus on mammals (particularly deer, beaver, and dogs) during the later portions of Period IIb and into Period IIc (Burns, 2003). Food-storage pits are typically found within house floors and between houses (Hayden, 1997a).

Lepofsky and Peacock (2004) demonstrate declining frequencies of earth ovens designed for geophyte cooking in both upland and lowland contexts on the Mid-Fraser area and nearby environs during Period IIb. There are two possible explanations for this phenomenon. First, intensely occupied Mid-Fraser villages are most numerous during the Period IIb suggesting that tethered villages may have overexploited root resources in their proximities (Kuijt and Prentiss, 2004). Hayden and Cousins (2004; see also Turner, 1992) disagree, however, stating that overexploitation would not have been possible. They base this conclusion on the ethnographic record (e.g. Turner et al., 1990) that suggests that foragers actively manipulated geophyte patches to maintain productivity. Second, steadily increasing moisture after ca. 1600 cal. b.p. may have reduced habitat for dry-meadow geophytes, thereby reducing the availability of this critical resource. Significant reduction in geophyte access may have played a role in population reduction on the Canadian Plateau outside the Mid-Fraser area.

Classic Collector populations of the Columbia Plateau appear to have expanded, perhaps stimulated by improved mammalian resources, salmon between 1600 and 1200 cal. b.p. and, likely, increasing camas production (Goodale et al., 2004; Thoms, 1989). Although salmon are virtually ubiquitous in Period IIb sites on the Columbia Plateau (Butler and Campbell, 2004, significant numbers of mammal remains are found in many sites including Umatilla, Wildcat Canyon, Harder, the Alpowai sites, and sites in Hells Canyon (Lyman, 1991). Bison predation declines with rising human populations, supporting Chatters' (2004) argument that bison populations on the Plateau were heavily affected by human predation. There is abundant evidence for food storage within housepit villages and a corresponding increase in cache pits within caves (Endacott, 1992; Hicks and Morganstein, 1994; Rice, 1969; Sheppard et al., 1987).


Lithic technology remains organized around the winter village core/flake model (e.g. Hayden et al., 1996a; Prentiss, 2000). Some variability is associated with changes in subsistence emphases. For example, Godin (2004) documents a greater wood working emphasis in early Period IIb lithic tools from Housepit 7 at Keatley Creek that changes to more frequent soft-material cutting (projectile points and biface knives) and scraping tools (various hide scrapers) by early Period IIc. This parallels the shift at this housepit from a salmon-dominated economy to one emphasizing mammals (Burns, 2003). Other examples of variation are associated with raw material access and tool production traditions. At the Bridge River site, also in the Mid-Fraser area, Prentiss et al.(2004) discovered a ground slate industry that is extremely atypical for Plateau lithic technologies. These tools appear to largely replace the more typical Plateau end scrapers and spall-based scrapers found at nearby sites such as Keatley Creek. We suggest that it may also reflect a tradition of ground-slate tool manufacture that originated in the Lower Fraser Valley.
Fig. 11.

Artifacts of Period IIb: (a and b) corner-notched projectile points; (c) basal-notched projectile point; (d) thin-stemmed projectile point; (e) nephrite adze (Bridge River site); (f) pipe fragment (Keatley Creek site); (g) biface; (h) ground slate knife with drilled hole (Bridge River site); (i) atlatl dart forshaft (Bridge River site); (j) bone awl; (k) stemmed bone projectile point (Bridge River site); (l) dentalium shell bead; (m) end scraper: (n) bone needle; (o) hand maul.

The denser populations of the Mid-Fraser and parts of the Columbia Plateau continued to favor fish and various plant foods. This is reflected in intensified production of ground stone tools (beyond the isolated slate example mentioned earlier) (Fig. 11). Hand mauls increase in frequency in The Dalles and Mid-Fraser areas. Mortars, pestles, net-sinkers, and abraders are increasingly common in sites throughout the Plateau (e.g. Hayden et al., 1996a; Kenaston, 1966). Zoomorphic sculptures in stone appear in The Dalles area in sites dating to Period IIb (Ames et al., 1998; Butler, 1959). A wide range of bone and antler tools is also found from this time across the Plateau. Mid-Fraser villages produced awls, points, needles, wedges, harpoon parts (leister tips), gorges, digging sticks, and digging stick handles (Richards and Rousseau, 1987). Awls, fish hooks, gorges, digging stick handles, matting needles, and composite harpoon heads are typical of the Columbia Plateau (Brauner, 1976; Dumond and Minor, 1983; Jenkins and Connolly, 1994).

The bow and arrow becomes fully established throughout the Plateau during Period IIb (Fig. 11). Arrow points appear on the Canadian Plateau as diminutive corner-notched and stemmed forms, often showing a high degree of resharpening (Rousseau, 2004a: Fig. 1.7). There is a dramatic decline in the frequency of key-shaped unifaces, paralleling the rise of bow and arrow technology, supporting Rousseau's (1992) contention that these tools were designed for dart manufacture. Although arrow points appeared on the Columbia Plateau during Period IIa, they become more widely distributed in Period IIb. Columbia Plateau arrow points are highly variable, but include corner and basal notch styles similar to those of the Canadian Plateau. Narrow-bladed, contracting stemmed or “pin stem” types appear in the Dalles area after 1450 cal. b.p. (Ames et al., 1998; Andrefsky, 2004; Dumond and Minor, 1983).

Social Organization

As outlined earlier, large and complex villages emerged in the Mid-Fraser Canyon area during Period IIb. These villages were eventually marked by a high degree of sophistication in labor organization and status differentiation. However, the basic socioeconomic structure of such villages in their earliest iterations is not entirely clear. The primary problem is that early Period IIb house floors have not been intensively investigated. The most thorough excavations of floors associated with large houses has been at Keatley Creek where all floors dated to Period IIc (Hayden, 2000b). A variety of other evidence does suggest, however, that complex forms of labor organization, similar to that of the early Marpole Phase design had emerged in the Mid-Fraser by ca. 1800 cal. b.p. Hayden et al.(1996b) provide provocative evidence for sustained control of lithic quarries or at least consistent use of specific quarry locales by households at Keatley Creek. Various researchers (Burns, 2003; Kusmer, 2000; Lepofsky et al., 1996; Lyons, 2003) provide evidence for complex food resource procurement operations in Mid-Fraser villages that likely required intra-household logistical mobility designed to collect simultaneously available resources from disparate areas. Finally, although no floors from medium or larger houses (e.g. greater than 10 m diameter) have been fully excavated, recent test excavations and geophysical investigations at the Bridge River site (Cross, 2004; Prentiss et al., 2004, 2005b) confirm that the layout of hearth features on the floors of several large houses (14–19 m diameter) dated ca. 1700–1800 cal. b.p. is very similar to that of the 800–900 cal. b.p. floor of Housepit 7 at Keatley Creek, which is thought to reflect a multi-family corporate group (Hayden and Spafford, 1993).

We think it highly likely that labor organization similar to Early Marpole had developed in the Mid-Fraser by 1800 cal. b.p., but it is not at all clear that a pattern of status inequality similar to that of the later Marpole phase on the coast (e.g. Hayden, 1997a) was present at this point. As Prentiss et al.(2005a) argue, a variety of data sets suggest that inequality developed, following the full establishment of aggregated Mid-Fraser villages. First, almost no artifacts that might be classified as prestige items (Hayden, 1998) are found in Keatley Creek deposits firmly pre-dating 1300 cal. b.p. and significant numbers do not appear until much later (Prentiss et al., 2003). Second, these items do appear in some frequency at the Bridge River site, but are more or less evenly distributed across all house sizes. Third, the salmon signature thought to be indicative of inequality in access to fish [Chinook (O. tshawytscha) and sockeye (O. nerka) salmon exclusively in large houses; pink/coho salmon in small, contemporaneous houses (e.g. Hayden, 1997a; see also Goodale, 2002)] is now in doubt given recent DNA studies of salmon from Keatley Creek (Speller et al., 2005). Fourth, extra-mural cooking features were apparently common prior to ca. 1300–1400 cal. b.p., but disappear after this point. Following Flannery (2002), inequality is often indicated by food hoarding, which means preparation and consumption out of sight of neighbors. If this is true, then the decline of outdoor food preparation could help indicate changes in social relations. Finally, Schulting (1995) does not recognize any signs of status inequality in the burial record of the Mid-Fraser area until after about 1300 cal. b.p. This does not mean that some incipient inequality might not have been present prior to 1300 cal. b.p. at least as associated with households' differential capacity to produce surplus food and non-food goods. Nonetheless, social relations between households appear to have been constrained by an ethic of egalitarianism.

Despite variable village sizes, house sizes on the Columbia Plateau are consistently small (typically less than 10 m diameter) during Period IIb, with rare exceptions (e.g. Dohm, 2000). Even in The Dalles area, where larger numbers of housepits have been recorded, there is no evidence for the unusually large houses found in the Mid-Fraser or Slocan areas. Excavations of house floors (e.g. Campbell, 1985 ; Draper, 1991; Dumond and Minor, 1983; Hackenberger and Thompson, 1995; Draper, 1991) do not support co-residential group sizes larger than extended single families. If corporate groups (sensu Hayden and Cannon, 1982) did exist, then they were organized differently, perhaps across multiple households. One 14-m diameter house, found at Dry Creek in Hells Canyon, may be an indicator of such an organization. Undated, but found in association with Period IIb houses at Dry and Tryon Creeks, this dwelling is marked by a very thin floor deposit, which may indicate it served as an assembly house for dispersed, cooperating families rather than as a dwelling.

Certainly, the settlement patterns associated with Columbia Plateau villages were stable, as indicated by hundreds of burials in cemeteries and numerous rock art panels (Keyser, 1992; Rice, 1978; Reid and Gallison, 1996). There is some evidence to suggest aspects of a prestige economy, at least on some parts of the region. Butler's (1959) elaborate stone sculptures from The Dalles suggest that socioeconomic conditions existed to support the surplus labor needed to produce these art forms. Hayden and Schulting (1997) demonstrate that The Dalles area sites accumulated a wide variety of prestige items including nephrite adzes, elaborately carved digging stick handles, and exotic objects of obsidian (e.g. Hayden, 1998).

Outside of the Mid-Fraser and possibly The Dalles, residential settlements show no sign of either corporate-group labor organization or status inequality. Housepit packing on islands and mesa-tops was likely for purposes of defense, as evidence for warfare increases during Period IIb (Chatters, 2004).


Period IIb was a critical time on the Plateau. Complex Collectors appeared in the Mid-Fraser area and possibly in the Lower Columbia around The Dalles. The bow and arrow spread throughout the region and may have played a role in stimulating nucleated settlements in many areas due to enhanced lethality of warfare and the need for defense. Classic Collector populations declined in the north, probably due to the combination of reduced access to salmon and various geophytes. In contrast, their populations rose in the south, as conditions improved for camas and other roots. Fortunes were to change again in the less-stable Period IIc.

Period IIc: Changing Fortunes (1250–650 cal. B.P.)


Climates associated with North America's west coast in the period of 1250–650 cal. b.p. are not known for their stability (Kennett and Conlee, 2002). Period IIc on the Plateau was no exception. Moist conditions prevailed ca. 1100–1200 cal. b.p., followed by a shift towards very warm and dry conditions of the Little Climatic Anomaly. This climatic “whipsaw” had strong effects on population and settlement dynamics throughout the Plateau. Most fundamentally, proxies for human populations continue to suggest an inverse relationship between northern and southern regions; while Columbia Plateau populations peaked early, stabilized somewhat, and subsequently declined (Fig. 4), the Canadian Plateau numbers first declined, then rose (Fig. 3).

The warm and dry conditions that dominated Period IIc do not appear to have favored success for the Complex Collectors of the Mid-Fraser area (Figs. 12 and 13); the Bridge River village, for example, was completely abandoned by 1150–1200 cal. b.p. (Prentiss et al., 2004). There is also only limited evidence for occupation of the Bell and Keatley Creek sites between 1000 and 1200 cal. b.p. (Lenert, 2001; Prentiss et al., 2003). Fortunes appear to have improved after 1000 cal. b.p., for we recognize intense occupations of very large housepits at the latter two sites dating in the 800–900 cal. b.p. range (Hayden, 1997a,b, 2000b; Prentiss et al., 2003). Following this resurgence, however, the Mid-Fraser was virtually abandoned for over a century. Hayden and Ryder (1991) blame the collapse on a local disaster blocking spawning salmon, but present evidence indicates that depopulation and abandonment could have been a region-wide phenomenon (Kuijt, 2001; Kuijt and Prentiss, 2004; Prentiss et al., 2003, 2005a,b; Rousseau, 2004a).
Fig. 12.

Map of the Keatley Creek site, a large Periods IIb and IIc village.

Complex–Collector-like patterns appear in two new areas at the end of Period IIb and beginning of IIc. Large circular, subrectangular, and rectangular houses (sometimes exceeding 15 m in maximum diameter) appear in the upper Mid-Columbia after 1450 cal. b.p. Ames et al.(1998) view these as multi-family or “communal” houses and suggest that they may have cultural origins to the north. Examples are found at Chief Joseph Reservoir (Campbell, 1985), Wells Reservoir (Grabert, 1968), and Rock Island (Galm and Masten, 1985). The Avey's Orchard site (Fig. 14) is of particular importance, demonstrating a ca. 1100 cal. b.p. long house with preserved structural elements and features that suggest simultaneous occupation by at least three domestic units (Galm and Masten, 1985).
Fig. 13.

Plan map of the Housepit 7 floor, Keatley Creek site (original map adapted from Prentiss, 1993).
Fig. 14.

Plan map of large rectangular housepit floor, Avey's Orchard site (adapted from Galm and Masten, 1985).

Several sites from the Upper Columbia area also contain evidence for the Complex Collector residential occupations. The Ilthkoyape site is a large village at Kettle Falls containing housepits up to 14 m in diameter dating as early as ca. 1200 cal. b.p. (Chance and Chance, 1982). Insufficient excavation was undertaken however, to determine if the largest houses were occupied by multi-family groups. Two large villages in the Slocan Valley may also contain evidence for Complex Collectors. The Vallican site is dominated by small to medium housepits (7–14 m diameter) of various shapes, including circular to rectangular designs, but also contains multiple earthworks that may have been house platforms (Mohs, 1982). Mohs (1982) argues that the larger houses, which date in the range of 700–800 cal. b.p., were occupied by multiple families. North of the Vallican site is the Slocan Narrows village, which contains more than 65 housepits of various ages and sizes (Goodale et al., 2004; Prentiss et al., 2001). The housepits of the late village (ca. 700–800 cal. b.p.) range from 7 to 16 m in diameter, with deeply excavated floors and massive, heaped rims. There is relatively little occupation debris, however, suggesting the occupations were not long term (Fig. 15). A 60-m long earthwork on the north side of the village core (apparently a defensive wall blocking access from the river) may date to this period (but possibly also to Period IId). Although the distributions of artifacts and feature distributions on the floors have not yet been intensively studied, housepit size and earthwork construction hint at the presence of Complex Collectors like those at Vallican.
Fig. 15.

Map of the Slocan Narrows site, a large Period IIc village (adapted from Prentiss et al., 2001).

Classic Collector populations of the Columbia Plateau appear to peak by 1100–1000 cal. b.p. and eventually decline. Most Hells Canyon sites, for example, show little or no activity during Period IIc. The Dalles area remains occupied during this period, but housepit villages contain evidence for changing residential patterns that are consistent with demographic decline. Small mat-lodges become the dominant residential structure at such sites as Wildcat Canyon, Umatilla, and Wakemap after 950 cal. b.p. (Butler, 1960; Dumond and Minor, 1983; Rice, 1971, 1978). Although houses are smaller, their frequencies remain high if not higher than during preceding occupational periods. Debris accumulations associated with individual house floors are not consistently high and intra-house cache pits are rare (Wildcat Canyon is an exception). Consequently, these villages appear to reflect more temporary aggregations (perhaps associated with the summer fishing season) than permanent winter settlements. Other housepit groups tend to be small (less than six houses) with floors less than 10 m in diameter. Some key Period IIc sites include Alpowai (Brauner, 1976), Curr (Carlson, 1980), several in the Chief Joseph locality (Campbell, 1985), Ferry Canyon “A” (Di Scipio, 1997), Hatiupuh (Brauner et al., 1990), Lenore (Sappington, 1994), and Three Springs Bar (Daugherty et al., 1967). Housepits at these sites typically have thin floors and little accumulation of rim midden. Debris accumulation ranges from low at the Chief Joseph Dam sites to moderate elsewhere. Many sites have cache pits and caching in caves continues. Mesa-top villages are relatively numerous (Chatters, 2004), as the use of upland areas for root resources continues. The overall pattern is structurally similar to that of the Classic Collectors of Period IIb.


As in Period IIb, the best subsistence information for the Mid-Fraser villages comes from the Keatley Creek site (Burns, 2003; Kusmer, 2000; Lepofsky et al., 1996; Lyons, 2003) and various studies of cooking ovens in its vicinity (Lepofsky and Peacock, 2004; Peacock, 1998; Pokotylo and Froese, 1983). The best evidence that families had differential access to salmon comes from this period. Berry (2000; see also Hayden, 1997a) documents a significant difference in frequencies of larger 3–5-year-old salmon versus small, 1–2-year-old fish between Housepits 7 and 3. The youngest floor of Housepit 7 is dominated by larger, older salmon, while Housepit 3 is dominated by smaller salmon. Hayden (1997a) interprets the pattern as evidence for control of optimal fishing spots by the occupants of Housepit 7. The dominance of larger salmon (which are probably Chinook) in Housepit 7 implies a possibly reduced emphasis on salmon during mid- to late-winter occupations of this dwelling, since these fish are harvested in the summer and do not last more than several months in dry storage due to high oil content (Kew, 1992; Romanoff, 1992). In her important analysis of change in predation patterns at Housepit 7, Burns (2003) convincingly argues that by Period IIc times, Housepit 7 occupants had altered their foraging priorities, replacing a strong signature of smaller salmon in the early strata (Period IIb) with a mammal-dominated pattern (supplemented primarily by Chinook and sockeye salmon) in Period IIc (Housepit 7, Rim 4: see Prentiss et al., 2003). Hayden (1997a) adds that large houses, such as Housepit 7, not only had more regular access to oily (and more tasty) salmon, but that they had the capacity for more food storage than other households. To Hayden (1997a), this and other evidence (Hayden et al., 1996a,b), indicate that foraging locales such as fishing sites or deer hunting ranges, were controlled by individual households.

Lepofsky and Peacock (2004) provide evidence that geophytes continued to be less import in winter villages during Periods IIb and c; roasting ovens remain uncommon in lowland settings. This pattern is particularly stark in the large villages like Keatley Creek, where the use of roots seems to have virtually disappeared during Period IIc. Geophyte intensification resumed at immediately before and after the time that the Mid-Fraser villages were abandoned. There are several possible explanations for this phenomenon. First, wet conditions late in Period IIb and early in Period IIc may have caused dry-meadow patches to shrink, forcing harvesters to take longer trips to obtain sufficient root supplies. This, in turn, would have reduced the incidence of local roasting events. Second, as argued by Kuijt and Prentiss (2004), the rise in populations that were tethered to large villages may have caused over-harvest of local geophyte patches, resulting in resource depression. Third, if fear of attack was a reality, then it may have been dangerous for small, vulnerable groups to spend long periods of time in the uplands, harvesting and roasting geophytes. Given the rising frequencies of deer bones (particularly limb elements) in late deposits at Keatley Creek, it does appear that at least some people spent considerable time hunting far from the winter village environs. Perhaps this was, in part, a defensive strategy. Ethnographically, males engaged in group hunts, during which they would be well defended by their own arms and numbers, while females, guarded by a few men, were the primary root gatherers. Groups made up primarily of females would have been more vulnerable to attack than primarily male groups. Finally, as outlined earlier, one manifestation of emergent inequality between houses may have been a marked reduction in the public preparation and consumption of food; roots may have been processed or at least consumed on an intramural basis. We agree with Lepofsky and Peacock (2004) that more research is required to explore these ideas, but also suggest that a combination of factors may have acted to reduce incidences of root harvest and cooking in the Mid-Fraser area during early Period IIc.

On the Columbia Plateau, salmon remained a critical resource during Period IIc, although mammal predation was also important. In some contexts, such as Avey's Orchard and the Chief Joseph Dam sites, salmon were more intensively exploited. Elsewhere, at sites such as Ferry Canyon “A”, Alpowai, and 45PO137 [Calispel Valley (Andrefsky et al., 2000)] large mammals were emphasized. Upper Columbia villages are generally thought to have emphasized medium to large mammals to a greater degree during this period than Plateau groups elsewhere did (Goodale et al., 2004). As Goodale et al.(2004: Fig. 3.3) and Thoms (1989: Fig. 46) illustrate, camas use in portions of the Columbia Plateau peaked during the early portion of Period IIc and subsequently declined. This decline parallels the climatic trend towards drier conditions in late Period IIc, with associated reductions in the extent of wet meadows.


The lithic technology of the Complex Collectors in the Mid-Fraser has been thoroughly analyzed and described for Period IIc (Hayden, 1997a, 2000c; Hayden et al., 1996b; Prentiss, 1993, 2000, 2001; Spafford, 1991, 2000). The winter village model of technological organization is expressed particularly clearly during this period [which contributed the assemblages on which the model was based (Prentiss, 1993)], despite the fact that biface production becomes increasingly prominent at Keatley Creek (Godin, 2004), Ferry Canyon “A” (Di Scipio, 1997), and Wildcat Canyon (Dumond and Minor, 1983) in contexts containing evidence for increased hunting.

Projectile point style appears to split between the north and south (Fig. 16). Canadian Plateau groups used a small side-notched arrow point called the “Kamloops point” (Rousseau, 2004a: Fig. 1.7). The side-notched arrow point was also adopted on the Columbia Plateau as Plateau side-notched, but many groups also retained small corner-notched varieties (Andrefsky, 2004). The Dalles area differed from the rest of the region as “pin-stemmed” types became dominant (Ames et al., 1998). Frequencies of end scrapers, which were presumably used as hide processing tools, increase in contexts where hunting was prominent. Groundstone and bone/antler tools remain relatively common and morphologically similar to those of Period IIb.
Fig. 16.

Artifacts of Period IIc: (a, c, and d) Kamloops side-notched projectile points; (b) chipped stone drill; (e) small thin-stemmed projectile point; (f) small corner-notched projectile point; (g) pestle; (h) nephrite adze; (i) digging stick handle; (j) sandstone saw (Keatley Creek site); (k) copper bead; (l) steatite bead; (m) polished steatite pipe fragment with hole drilled for use as a pendant (Keatley Creek site); (n) harpoon (leister) tip; (o) toggling harpoon valve; (p) bone gaming piece; (q) net sinker; (r) hand maul.

Social Organization

Social organization, as measured in two dimensions peaks in complexity among populations of the Mid-Fraser. First, Hayden and colleagues (Hayden, 1997a,b; Hayden and Spafford, 1993; Lepofsky et al., 1996; Prentiss, 1993, 2000; Spafford, 1993) demonstrate that co-residential corporate groups lived in large housepits. The Housepit 7 floor at Keatley Creek has evidence for five to seven domestic units marked by redundant groupings of hearths, cache pits, and clusters of bedding (needles), food debris, and lithic artifacts. A corporate group of perhaps three to four families is reflected at the somewhat smaller Housepit 3 of the same site. While similar behavior is also presumed to have existed during Period IIb at this site and others nearby, the complete excavation of the floors in Housepits 7 and 3 provides the best evidence for its signature. The second dimension of complexity is status organization. Variability in food remains, as outlined earlier under Period IIb, indicates that house size may have been a marker of rank within the community. Analyses of the horizontal distribution of subsistence debris on the floor of Housepit 7 have also been used by Hayden's group (Hayden, 1997a; Lepofsky et al., 1996) as evidence for ranking between cohabiting families. For example, they suggest that clusters of salmon bones on the east side of the house reflect the need for low status families to save these materials for soup. Deer bones are more ubiquitous across the floor, but Hayden (1997a) speculates that members of the household's highest-ranking families acquired marrow from the limb bones, but left lower ranking family members with the more intensive task of bone grease extraction to lower ranking families. Other data suggest variations in household member access to prestige trade goods. Henry and Hayden (2000) even posit that projectile point distributions could mark status distinctions. Assuming that older technologies carry less prestige, they point out that dart points are found on the (poorer) east side of Housepit 7, while arrow points are mostly found on the west side. Although there are alternative explanations for each of Hayden's assertions, he does present enough data to suggest there may have been status differentiation between families and perhaps individuals within Housepit 7. Subsistence, feature, and artifact data support an argument for inter-household status differences at Keatley Creek. Schulting's (1995) assessment of burial data also supports this position, though sample sizes are very small.

A similar pattern of household organization may have existed in the big villages of the Upper Columbia (particularly Slocan area). Mohs (1982) recognized floor distributions at Vallican suggestive of co-residential multi-family groups. Household-based status differentiation has not been tested for this area.

Apart from accumulated prestige goods in burials, middens, and on some small house floors (Chatters, 2003; Hayden and Schulting, 1997), status differentiation is also difficult to recognize on the Mid- or Lower Columbia. We do recognize indicators of complex co-residential groups in the Mid-Columbia however, as exemplified in the Avey's Orchard site.


Period IIc is recognized for some changes in the population histories of Classic and Complex Collectors of the Plateau. Classic Collector populations appear to have covaried with the availability of moisture on the Plateau. Columbia Plateau groups were successful as long as conditions favored access to salmon and wet-meadow geophytes (e.g. camas). Increasingly wet conditions between ca. 1400 and 1100 cal. b.p. had provided that bounty, allowing populations to peak early in Period IIc. Warm and dry conditions of the peak Little Climatic Anomaly (1100–700 cal. b.p.), however, are likely to have reduced the availability of salmon as well as camas and other roots in this region. Dry conditions appear to have been problematic for southern Classic Collectors, whose populations appear to have become less stable later in this period before dropping near the Period IId boundary.

In contrast, northern Classic Collector populations generally rose. Some Complex Collectors of the Mid-Fraser and Upper Columbia also appear to have benefited from drier conditions, perhaps controlling the most productive hunting and fishing locales using the larger and more-powerful groups from villages such as Keatley Creek. They also benefited from increasing game and geophyte abundances as aridity opened forest canopies and expanded the steppe. By 700 cal. BP, however, these groups too had fallen apart. While local disasters could have contributed to this process, it is also likely that local abandonments were a reflection of regional processes, perhaps including a collapse of salmon populations in the eastern Pacific (Finney et al., 2002; Tunnicliffe et al., 2001).

Period IId: Final Collectors (650–250 cal. B.P.)


Finney et al.(2002) provide evidence that the productivity of eastern Pacific fisheries pattern is inverse to that of the North Pacific (Gulf of Alaska). While salmon were probably abundant along the Northwestern Coast between 2000 and 1000 cal. b.p., their populations apparently declined to far below average levels soon after this date. Meanwhile, salmon populations boomed in the Gulf of Alaska after 800 cal. b.p. As the droughts of the Little Climatic Anomaly were replaced by Little Ice Age moisture and cooler temperatures, camas patches probably expanded in the south, while geophytes declined on the Canadian Plateau. Population proxies (Figs. 3 and 4) suggest a general decline in human numbers throughout the Columbia Plateau and high but fluctuating numbers on the Canadian Plateau during the final period. These changes may have initially been linked to the unpredictability of salmon and geophyte resources, but were later due also to the influx of European diseases. Changes in the frequencies of radiocarbon dates from the Canadian Plateau during Period IId could also be at least partially an artifact of sampling and the imprecision of the radiocarbon method for dating the prehistoric and early contact periods.

Complex Collectors of the Mid-Fraser disappeared for approximately 300 years (ca. 800–500 cal. b.p.) before returning to some villages in smaller numbers during the final centuries of Period IId (Hayden and Ryder, 1992; Prentiss et al., 2003). The Bridge River site was occupied by households of various sizes during Period IId. House sizes vary from large, at approximately 18 m diameter, to medium and small at 5–13 m with smaller houses tending to be the rule. Villages that appear included no more than 15 contemporaneously occupied housepits between 500 and 250 cal. b.p (Prentiss et al., 2004, 2005b). These houses generally have thin floors and moderate quantities of cultural materials. Some houses such as Housepit 20 have numerous stratified floors (Prentiss et al., 2004). There has been insufficient excavation to characterize rim accumulations, but they appear to be relatively thin. Overall, Period IId housepits appear to have been occupied frequently, though possibly not for the extended periods typical of the same village during Period IIb.

The Keatley Creek site was also reoccupied after about 500 cal. b.p. The nature and scale of occupation are not clear however. Occupation did occur within the village core during this period, but may not have consisted of more than temporary camps in and adjacent to abandoned housepits (Alexander, 1989). Recent analysis of midden materials and stratigraphy in Housepit 7 suggests, however, that some housepits were reoccupied during the late period by people following a subsistence strategy and technological model similar to that of pre-800 cal. b.p. (Jendresen, 2004). Even if some housepits were reoccupied during the final prehistoric and early contact periods, it does not appear that households were as large or complex as they had been in earlier times. Despite the smaller-scale use of the site, Hayden and Adams (2004) document a pattern of housepit occupation on the peripheries of the village that they think reflects the activities of shaman and secret societies. These houses have thin floor and rim middens, but often contain unique and interesting collections of artifacts (e.g. a large crescent-shaped biface, basketry, gaming pieces, etc.) and faunal remains. While this is a problem requiring substantially more research, it does raise the interesting possibility that the ruins of some ancient villages may have taken on new roles as places for conducting special ceremonial activities.

Complex Collectors may have persisted in portions of the Mid- and Upper Columbia. Chatters (1986) documents a large house floor at Wells Reservoir, indicative of a multi-family group, although incomplete excavation of the house floor leaves uncertainty as to whether the dwelling housed multiple domestic units. The Vallican site (Mohs, 1982) contains late-dating houses and earthworks similar to those of Period IIc. A child burial with extensive grave goods also occurs at Vallican during Period IId. Similar to Vallican, the Ilthkoyape site (Chance and Chance, 1982) contains occupational patterns much like those of the prior period. Substantially more research is necessary to more fully document Complex Collector residential patterns of this area.

Complex Collectors appear to have moved into The Dalles area from the vicinity of the mouth of Columbia after about 550 cal. b.p. (Ames et al., 1998). This incursion is marked by the sudden appearance of house depressions reflecting the rectangular plank houses typical of Chinookan speakers, who also lived downstream (Butler, 1960; Caldwell, 1956). Above-ground ossuaries, which appear on islands during this period, also indicate a Chinookan presence. Although the archaeological record of this expansion has been insufficiently studied, the cultural pattern is relatively well understood ethnographically (French and French, 1998).

With rare exceptions, Classic Collectors of Period IId are generally found in small residential settlements. Typical small housepit sites are found at Three Springs Bar (Daugherty et al., 1967), Ferry Canyon “B” (Di Scipio, 1997), Chief Joseph Dam (Campbell, 1985), Wildcat Canyon (Dumond and Minor, 1983), Curr (Carlson, 1980), Harper Ranch (Wilson, 1980), and Tryon Creek (Hackenberger and Thompson, 1995). Although the Miller site (Fig. 17), located on Strawberry Island in the Lower Snake River contains approximately 150 housepits, the pattern of occupation does not appear to be similar to the large Mid-Fraser villages (Schalk, 1983b). From the limited excavations undertaken at this site, houses are small, floors are not internally complex, and there is little evidence for inter-household variability in prestige markers. There are a few larger houses at this site and at other large villages, such as 45BN53 (Osborn, 1957) that could have served as gathering places for communal activities, dances, or ceremonies for dispersed corporate groups. A general lack of domestic debris in the floors of these structures strongly indicates they served a special purpose. Large villages on islands such as the Miller site may also have been positioned for defense (Chatters, 2004). Columbia Plateau sites (including Miller) typically contain thin floors and evidence for multiple reoccupations; rim accumulations have not been studied adequately. Thicker floors and somewhat denser artifact accumulations are found at Curr, Harper Ranch, and Tryon Creek.
Fig. 17.

Map of the Miller site, a large Period IId village (adapted from Schalk, 1983b).


The densest clusters of housepits still occur at the optimal fishing places near mouths of rivers (confluences of the Snake and Columbia and the North and South Thompson) and at narrows (The Dalles, Mid-Fraser rapids, Kettle Falls). It is not surprising, therefore, that salmon are dominant in faunal assemblages from most of these contexts (Hayden and Adams, 2004; Orcholl, 2004; Prentiss et al., 2003; Schalk, 1980, 1983b). Salmon are also important in most other residential contexts (Campbell, 1985; Chatters, 1986). Ferry Canyon “B” is mammal dominated, however (Di Scipio, 1997), as are most sites in Hells Canyon (Hackenberger and Thompson, 1995; Lyman, 1991)

Goodale et al.(2004) and Thoms (1989) demonstrate that a decline in the frequency of camas processing occurred in the Upper Columbia area during Period IId. Since conditions were probably good for camas, we suggest that this can only be a consequence of declining human population, perhaps associated with changes in the variability of salmon, but also with the impacts of both pre-Contact (e.g. Rousseau, 2004a) and post-Contact diseases. Lepofsky and Peacock (2004) provide provocative data that indicate geophyte intensification on the Canadian Plateau early in Period IId (ca. 500–800 cal. b.p.). This was followed by a near-cessation of geophyte exploitation until near the end of this period. They note that while roasting ovens became smaller, their, frequency increased dramatically (ca. 800–500 cal. b.p.). Lepofsky and Peacock (see also Kuijt and Prentiss, 2004) suggest that this change could be associated with a decline in the sizes of residential groups after the abandonment of the Mid-Fraser villages. New evidence from the Bridge River site suggest a resurgence in large-oven cooking after ca. 400 cal. b.p. (Prentiss et al., 2004). It is not clear, however, that the Bridge River pits are associated with roots; many may be meat and/or fish roasting ovens (Dietz, 2004). This overall pattern of an early peak in root processing followed by a collapse makes sense if we assume that patches expanded during dry conditions of the Little Climatic Anomaly, but subsequently contracted as moisture increased during the Little Ice Age. We suspect that a decline in resource availability is a better explanation than population decline, as it is clear that many Mid-Fraser villages were reoccupied after 500 cal. b.p. and it is precisely at this point that geophyte roasting disappears in both lowland and upland settings.

There is indirect evidence in the eastern periphery of the Plateau that some peoples attempted to stay the decline in open meadows by controlling the forests. Increases in the proportion of large charcoal particles, which are indicative of local burning, during a period when catastrophic fires were relatively rare, has been interpreted by Chatters and Leavell (1995) as evidence for controlled burning after the end of the Little Climatic Anomaly.


The winter village pattern of lithic technological organization (e.g., Prentiss, 2000) persists through Period IId. Nearly all major residential sites are dominated by a core/flake pattern of lithic tool production emphasizing flake tools and to a much lesser degree, bifaces and portable specialized tools like end scrapers and drills. The late reoccupation of the Bridge River site still contains abundant examples of ground slate. Hayden and Adams (2004) document cases where biface production debris is concentrated in housepits thought to be associated with secret societies or shamans. High rates of biface production are also recognized at Ferry Canyon “B” in the Lower Columbia area (Di Scipio, 1997). Groundstone and bone tools change little from the preceding period. Projectile point styles are also similar to Period IIc with one exception. Multi-notch Kamloops type arrow points (Fig. 18) appear during Period IId on the Canadian Plateau (Richards and Rousseau, 1987; Rousseau, 2004a). European trade goods (glass beads, metal tools, etc.) begin to appear in many contexts during the final decades of this period (e.g. Chatters, 2003; Dumond and Minor, 1983; Mohs, 1982).
Fig. 18.

Artifacts of Period IId: (a) Kamloops Side-notched projectile point; (b) Kamloops multi-notched projectile point; (c) small corner-notched projectile point; (d) small un-notched projectile point; (e) ground slate scraper (Bridge River site); (f) net sinker; (g) steatite pipe; (h) large curvilinear biface (Keatley Creek site); (i) chipped stone drill; (j) copper bead; (k) steatite bead; (l) dentalium shell bead.

Social Organization

Social organization has not been adequately studied in Period IId villages. Variability among house sizes and frequencies of prestige artifacts (stone beads, pendants, pipes, etc.) at the Bridge River site, however, suggests complex forms of labor organization and status differentiation (Prentiss et al., 2004). Hayden and Adams' (2004) assessment of small houses on the margins of the Keatley Creek village as special places also suggests complex forms of social organization. A similar situation exists in the Upper Columbia where earthworks, medium to large houses, and at least one lavish burial indicate complexity in social status and the organization of labor. Finally, expansion of Chinookan plank houses into The Dalles area likely brought with it socioeconomic complexity. Hayden and Schulting (1997) document large numbers of prestige artifacts in the Mid-Fraser and Lower Columbia areas. They note that during ethnographic times there was a well defined exchange relationship between these areas, as if populations at equivalent levels of political organization were communicating across a gulf of less-hierarchically-organized folk.

The large, apparently special purpose houses seen near the mouth of the Snake River may be a strong indication of continued egalitarianism in that region. Settings where a large proportion of the adult occupants of a village could congregate may be evidence for consensus-based decision making, or at least free debate among village dwellers. Alternatively, however, large dwellings may have functioned as meeting places of dispersed corporate groups or for winter ceremonies, which were a significant aspect of inter-village communication and commerce during the ethnographic period (e.g. Walker, 1998).

Mortuary assemblages from the same region may contradict the idea of egalitarianism. Cemetery lots that date stylistically to the latest prehistoric and early protohistoric show a high degree of disparity in number and quality of grave goods (Combs, 1963; Crabtree, 1957), indicating that certain individuals held higher status than others. Occasionally, graves of children have been found associated with large numbers of dentalium or glass beads, suggesting status may have been ascribed to some degree.


Curiously, the latest period is, in many ways, the least understood of all late prehistoric periods on the Plateau. We know that populations generally declined throughout the region. At times these declines were precipitous, as in the Mid-Fraser area. The last expansion of Complex Collectors probably occurred in the Lower Columbia after 500 cal. b.p. perhaps brought by the Chinookan groups recognized in the Contact period. Some Mid-Fraser villages were reoccupied, but generally on smaller scales than 500–1000 years earlier. Despite fluctuating population dynamics in Periods IIc and IId, the basic collector strategy remained firmly in place with settlement, subsistence, and technological strategies similar to those seen in the previous period.


This review of the late prehistoric Plateau archaeological record has implications for improving our understanding of several critical topics. We focus our discussion on three areas: subsistence and demographics among Classic Collectors, emergence and dispersal of Complex Collectors, and the development of social inequality.
Fig. 19.

Overview of population and resource trends on the Plateau. Parts (A)–(C) provide summaries of more detailed graphs in Figs. 3 and 4. Part (D) is a summary of variation in marine productivity in the Eastern Pacific based on Tunnicliffe et al.(2001) and Finney et al., (2002). General high productivity in fish and diatom populations is correlated with increased numbers of spawning salmon. Salmon productivity in the eastern Pacific tends to pattern inversely to that of the Gulf of Alaska (Chavez et al., 2003; Finney et al., 2002). Part (E) is a summary of upland geophyte intensification based on Lepofsky and Peacock (2004). Part (F) is a summary of camas intensification based primarily on Thoms (1989).

The Classic Collector strategy persisted on the Plateau since about 3600 cal. b.p. Despite stability at the strategic level, we recognize a substantial amount of variation in practice and implementation of the strategy at the tactical level (e.g. Chatters and Prentiss, 2005). Columbia Plateau Classic Collectors varied substantially in preferred foods through time and across space. Some groups were more mammal dependent, while others clearly favored fish. We are not clear on variability in the use of geophytes and, consequently, must rely on the record of the Calispel Valley as a proxy measure.

Examining the record at the most broad level, we see patterns of rising and falling human populations (as indicated by our radiocarbon proxy) roughly correlated with indicators of geophyte intensification and above-average predicted marine productivity (Fig. 19). These data suggest that the ability to sustain relatively high population densities was predicated on sources of high-quality protein and, perhaps even more fundamentally, carbohydrates from camas, a wet-meadow adapted geophyte. This conclusion is further warranted by the data on Classic Collectors of the Canadian Plateau. There, human populations appear to have risen under warmer and drier conditions that favored better geophyte production (Fig. 19). Conversely, they declined under higher moisture regimes that would have eliminated many geophyte meadows.

Our study has implications for developing an understanding of the emergence of more socioeconomically complex forms of collecting. We argued that our “Complex Collectors” reflect not a simple case of small groups developing complex social organizations featuring inequality, but, rather a more fundamental change in socioeconomic structure. We argue that it was a broad strategy that incorporated new demographic arrangements (large multi-family households) permitting control of resources and intensive simultaneous harvesting of diverse, simultaneously available resources on a scale that was impossible for Classic Collectors. It may have been a particularly good adaptation to highly abundant but clustered resources, such as might occur in mesic environments near major river narrows where spawning salmon would readily provide for basic subsistence and support a diversification of a local economic system.

If we view complex collecting as an economic strategy, then we recognize that it may have required special conditions to emerge (Prentiss and Chatters, 2003a). In other words, selective conditions and normal hunter–gatherer cultural transmission mechanisms (e.g. McDonald, 1999) generally would not have permitted such a radical behavioral pattern to emerge in most places. Optimal conditions may have existed in lower portions of some large river systems—especially the Lower Fraser Valley. Lepofsky et al.(2005) have presented data to indicate that Complex Collectors were there by 2500 cal. b.p. They suggest that highly clustered salmon in the Fraser River and a very patchy adjacent alpine environment would have offered big rewards for those who could effectively harvest and store the clustered resources and consequently expand alliance networks for increased economic opportunity. While Lepofsky and her coworkers favor self-interested elites as stimulators of that process, we suggest that this is not necessary. Rather, a rapid but relatively minimal niche reorganization (e.g. Chatters and Prentiss, 2005; Prentiss and Chatters, 2003a) could have offered such benefit to the community at large that strategy adjustments could have come quickly within small, optimally positioned groups. Big payoffs would have promoted growth and budding-off of the participating community and the rapid transmission of these ideas; in other words; higher fitness of the new socioeconomic strategy. In fact, current data suggest that, once developed, the Complex Collector strategy spread very rapidly along the Central Coast.

Given the chronology of developments on the Coast, it seems logical that the Complex Collector strategy offered significant enough payoffs for expanding human groups to bring it with them as they moved inland. Conversely, cultural transmission, perhaps tied to interactions associated with alliance networks, may have permitted the strategy to spread with little actual immigration. The earliest unequivocal Complex Collectors on the Plateau occur in the Mid-Fraser area at dates shortly after 2000 cal. b.p. Given this area's geographic proximity and similarity in resource configurations to the place where this strategy developed, the emergence of Complex Collectors on the Plateau was not an independent development. Given similarities in household structure and a variety of technologies (including ground slate), it is likely that the strategy was first brought into the area by groups who originated nearer to the coast (or at least Lower Fraser Valley).

The Complex Collector strategy appears in the Lower Columbia River at least by ca. 2000 cal. b.p., but it is not clear whether it spread inland into The Dalles area before the ca. 500 cal. b.p. upriver migration of Chinookan peoples. The earlier presence of prestige items in this area may have resulted from the “trade fair” that accompanied seasonal conclaves of many groups who assembled at such places as Celilo Falls and the Long Narrows to take advantage of tremendous concentrations of salmon. Peak resource conditions at ca. 1200 cal. b.p. may have favored exploration of the Mid-Columbia by Complex Collectors from the Fraser System or experimentation with the strategy by resident groups. Whatever happened earlier, it is clear that, by 800 cal. b.p., Upper Columbia groups practiced some form of complex collecting. Following Ames et al.(1998), we suggest that this could have been the consequence of another population expansion into these areas or a byproduct of good resource conditions and cultural transmission. The Upper Columbia case is very intriguing but understudied.

The best data concerning the emergence of social inequality come from the Mid-Fraser area. Current data indicate that institutionalized inequality (e.g. Wiessner, 2002) that includes hereditary ranking and competitive feasting complexes (Hayden's (1995) “entrepreneurs”) was a development that was not originally associated with complex collecting as a socioeconomic strategy. The Keatley Creek site appears to have had persistent inequality in house size, but artifact indicators of status differences only appear after 1200–1300 cal. b.p., when salmon resources may have declined while others, such as geophytes, improved. Changes in salmon abundance may have had severe impacts on such other local groups as those of Bridge River, leaving villages like Keatley Creek in a good economic and political position. One implication is that institionalized inequality only developed when the resource base was reconfigured (e.g. Arnold, 1993, 2001) such that only a few localities could support large groups. With large numbers of disenfranchised people in the area, it was no stretch for the already-large households at villages like Keatley Creek and Bell to assert greater degrees of control over resources and perhaps other, more intangible aspects of culture (e.g. origin myths, crest animals, and hereditary status positions). They could have leveraged this position to attract larger kin groups and clients and thus enhance their socioeconomic position.


Plateau archaeology is entering a dynamic period. Even as late as the early-1980s, Plateau archaeologists were primarily focused on developing culture history. As we have described, one byproduct of this was the development of a myriad of cultural chronologies for different subareas within the Plateau (Andrefsky, 2004; Rousseau, 2004a). Despite this impediment, a number of important archaeological studies conducted during the late-1970s and 1980s, including Ames and Marshal (1980), Campbell (1985), Chatters (1986, 1989a), Galm and Masten (1985), Hayden et al.(1985), Lohse and Sammons-Lohse (1986), Nelson (1973), and Schalk (1981), had the impact of steering the study of Plateau archaeology in new directions. During the 1990s and early-2000s, Plateau archaeologists published important works on subsistence intensification and the emergence of social complexity (Butler and Campbell, 2004; Chatters, 1995a,b; Hayden, 1994, 1997a, 1998). We can now identify four particularly important research themes in Plateau archaeology as we move into the 21st century: intensification, social complexity, Darwinian evolutionary processes, and indigenous archaeology.

The Plateau remains an ideal context for the study of subsistence intensification. Research into processes of salmon intensification will continue with a particular emphasis on relationships to demographics and technologies. One of the brightest emerging areas is in geophyte intensification (e.g. Lepofsky and Peacock, 2004). Researchers are exploring not only the effects of changing cultural and ecological contexts on root harvesting and processing, but they are beginning to use Plateau data to study human–plant mutualism and its implications for our understanding of the origins of agriculture in other parts of the world (Peacock, 1998).

Hayden's important studies of social processes at Keatley Creek have paved the way for future study and debate in this arena. The evolution of complex hunter–gatherer societies remains an important topic in many archaeological circles around the world (Arnold, 2001; Prentiss and Kuijt, 2004a; Price and Feinman, 1995). The archaeological record of Complex Collectors on the Plateau provides an important arena for studying variability in evolution and organization of these societies. Debates over the nature of the evolutionary process are emerging, with a focus on the Mid-Fraser area (Kuijt and Prentiss, 2004; Prentiss et al., 2003, 2005a,b). Columbia Plateau researchers need to develop similar research programs to that of Hayden in order to understand variations in social organization there as well.

Data we present in this paper show that evolutionary processes operate at multiple scales. On the Plateau, we can recognize microevolutionary processes illustrated by subtle shifts in artifact technology and style. However, we also recognize macroevolutionary change on the scale of socioeconomic strategies (Chatters and Prentiss, 2005; Prentiss and Chatters, 2003a). Substantial research is still required to sort out many details associated with the evolution of different hunter–gatherer socioeconomies across the Plateau. In addition, we should not let traditional culture area boundaries stymie our considerations. For example, it appears to us that some elements of Plateau collector organization moved on to the Great Plains around 3000 cal. b.p. In contrast, Numic speaking populations with a very different economic system impacted southern Plateau groups after 1000 cal. b.p. In this paper, we have raised the possibility that the Complex Collector socioeconomic strategy may have emerged and spread from at least one place on the Northwest Coast.

Finally, Native people have been absent for too long from archaeological research on the Plateau. A new generation of Native archaeologists working in the region with the ability to combine traditional knowledge with modern archaeological method and theory can only provide us with a richer set of tools for exploring the ancient past (e.g. Watkins, 2000). This process is beginning to happen, as facilitated by creative training programs at several universities in the region.


This paper would not have been possible without the many years of work in Plateau archaeology undertaken by our colleagues and forebears. We offer special thanks to Bill Andrefsky, Ken Ames, Michael Blake, Sarah Campbell, Cathy Carlson, Richard D. Daugherty, Tom Foor, Jerry Galm, Steve Hackenberger, Brian Hayden, Ian Kuijt, Dana Lepofsky, Martin Magne, Bob Muir, George Nicholas, Sandy Peacock, David Pokotylo, Mike Rousseau, Bill Smith, and Arnoud Stryd for their contributions and for stimulating discussions and debates over the years. The Bridge River map was drafted by Mathew Hogan. The Housepit 7 floor map was drafted by Jim Spafford. We thank Jerry Galm and one anonymous reviewer for their critical comments on the manuscript. Jerry Galm also provided us with copies of critical unpublished reports. Research conducted by Prentiss at the Bridge River and Keatley Creek sites was generously supported by the National Science Foundation, the Wenner-Gren Foundation for Anthropological Research Inc. and The University of Montana. We extend special thanks to the members of the Stl'atl'imx and Secwepemc First Nations in British Columbia for their interest and support of Canadian Plateau archaeology. Finally, we thank Angela E. Close for her patience and for providing critical commentary on this manuscript.

Copyright information

© Springer Science + Business Media, Inc. 2005