Settlement Patterns in the McElmo Dome Study Area
Our purpose in this chapter is to provide the demographic and social context for the development of Sand Canyon Pueblo (Site 5MT765) in the mid–A.D. 1200s. We examine Pueblo Indian settlement patterns from A.D. 600 to 1280 on the McElmo Dome, an area of southwestern Colorado that includes portions of the McElmo drainage system, which is a major northern tributary of the San Juan River (Figure 1). More specifically, our study area is bounded by McElmo Creek on the south, by Yellow Jacket Creek on the north, and by Hartman Draw on the east. The west boundary is along section lines about 3 km west of West Rock Canyon. Sand Canyon Pueblo is located in the approximate center of the study area.
For two reasons, this is an ideal area in which to study the dynamics of ancestral Pueblo Indian settlement patterns. First, archaeologists at the Crow Canyon Archaeological Center have conducted several major research projects in this area since 1983, including the Sand Canyon Archaeological Project (Lipe 1992*1; Varien 1999*7; Varien and Wilshusen 2002*1), the Shields Pueblo excavations (Duff and Ryan 1999*1, 2000*1, 2001*1), and most recently, the Goodman Point Archaeological Project (Coffey and Kuckelman 2006*1; Hovezak et al. 2004*1; Kuckelman and Coffey 2007*1; Kuckelman et al. 2004*1). Second, numerous archaeological surveys have been conducted in the study area, including large block surveys (e.g., Adler 1990*1; Hovezak et al. 2003*1; O'Neil 1998*1), linear-transect surveys, and smaller quadrat and transect surveys. This study is the first attempt to synthesize and analyze the results of these surveys.
In this chapter, we present an inventory of all recorded sites in our study area, classify them as to their general functional categories, and examine their distribution through time and across space. Using these data, we reconstruct patterns of population growth and decline and the clustering of settlements and public buildings in communities. We also look more closely at the population history of specific communities in the study area to identify some of the social processes underlying the formation of large sites that we interpret as community centers. As such, this study provides a broad context for Crow Canyon's long-term research in the Sand Canyon locality.
The Study Area
The study area (Figure 1) is part of the McElmo drainage unit of the northern San Juan archaeological region (Lipe 1992*3:2). We call the area considered in this analysis the "McElmo Dome study area," because its boundaries approximate the extent of the geologic feature named the McElmo Dome (Ekren and Houser 1965*1:51–52). The McElmo Dome is an asymmetric structural dome that comprises a central dome and satellite anticlines (Ekren and Houser 1965*1:51–52). The approximate center of the dome is the head of Sand Canyon, and the dome and its anticlines extend about 33 km east-west and 17 km north-south.
Our study area also closely approximates the boundaries of the Sand Canyon locality (Lipe 1992*3:2), although our boundaries are farther east and west. Willey and Phillips (1958*1) defined a locality as a spatial unit larger than a site and smaller than a region. The Sand Canyon locality was defined by Lipe and Bradley (1986*1; see also Lipe 1992*1) as the sustaining area for two of the largest ancestral Pueblo communities in southwestern Colorado. It was defined as an area of about 200 km2 centered on Sand Canyon and Goodman Point pueblos (Lipe 1992*3:2), which were large villages during the thirteenth century A.D. (Adler 1990*1; Adler and Varien 1994*1; Coffey and Kuckelman 2006*1; Hovezak et al. 2004*1; Kuckelman et al. 2004*1; Ortman and Bradley 2002*1; Wilshusen 1991*1), and both were the central places of larger communities that included smaller sites (Varien 1999*1).
The Sand Canyon locality is bounded to the south by McElmo Creek and to the north by Yellow Jacket Creek. These boundaries have also been used to define the northern and southern boundaries of the McElmo Dome study area. To the east and west, the boundaries of the Sand Canyon locality were defined by an arc with a 7.5-km radius, with the western arc being drawn from Sand Canyon Pueblo and the eastern arc from Goodman Point Pueblo (Lipe 1992*3:2). These arcs made the west fork of Rock Canyon the approximate location of the west boundary and Alkali Canyon the approximate location of the east boundary.
In defining the area for this study, we have extended the east and west boundaries of the Sand Canyon locality to include two large block surveys that have been completed in recent years: the Indian Camp Ranch survey (Fetterman and Honeycutt 1994*1) on the east and the Canyons of the Ancients National Monument survey (Hovezak et al. 2003*1) on the west. The east boundary of the McElmo Dome study area extends to the central portion of Hartman Draw. The west boundary runs north-south along section lines about 3 km west of the West Fork of Rock Canyon. This larger area includes about 300 km2. For an introduction to the natural environment of the study area, see Adams and Petersen (1999*1).
The Settlement Database
Most known archaeological sites in the study area were recorded during one of the many surveys that have been conducted as part of either problem-oriented research or cultural resource management work. These surveys include small quadrat surveys, linear-transect surveys, and large block surveys (Figure 2); a few sites were recorded individually. A total of 2,242 sites have been documented in the study area (Figure 3). Excavations have been conducted at numerous sites on the McElmo Dome, including Sand Canyon Pueblo and 14 other sites excavated by the Crow Canyon Archaeological Center (Kuckelman 2000*1; Varien 1999*7). Approximately 73 km2, or about one-fourth of the study area, has been covered by block surveys. The major block surveys that have been conducted in the study area are listed in Table 1. Eighty-four percent of the sites that have been documented in the study area (1,877 sites) were recorded during these surveys.
All sites in the study area were analyzed using methods developed for a larger project funded by the National Science Foundation (Kohler et al. 2007*1). These methods are explained in detail by Ortman et al. (2007*1) and are summarized briefly here. We used these methods to determine: (1) the function(s) of every site; (2) the period(s) of use of every habitation, field house, isolated public building, and pottery kiln; and (3) the number of households that occupied each habitation site during each period of use.
Site function was assessed from information on state site forms, information in the database of the Colorado Office of Archaeology and Historic preservation, and information on forms for sites that were not recorded on state forms or that have not yet been added to the state register. The information considered in determining site function includes the presence or absence of architectural features, the types of artifacts present, and the interpretations made by the archaeologists who recorded the sites. Using these data, we identified sites as habitations, field houses, public architecture, or one of a variety of types of limited-activity sites. Table 2 lists the sites by site type for the subset of sites recorded in block surveys, and it gives the total for all sites that have been identified in the study area.
Limited-activity sites—numbering 930 sites, or 48 percent of the total sites where function could be inferred (1,933)—are the most common type of site. As seen in Figure 4, the most common limited-activity sites are artifact scatters, kilns, and field houses. Limited-activity sites that occur less frequently include storage sites, rock art panels, water-control features, stone quarries, ceremonial sites, reservoirs, and clay quarries. Numerous limited-activity sites could not be assigned to one of these categories because of a lack of data. Figure 5 shows the locations and types of limited-activity sites in the study area. Habitations are the next most common site type. There is a total of 887 habitation sites, including single habitations (640) and multiple habitations (247). Habitations are 46 percent of the sites for which function could be inferred.
Periods of Occupation and Populations of Habitation Sites
The first step in analyzing periods of use for habitations, field houses, isolated public architecture, and kilns was to compile a calibration data set of about 80 excavated and well-dated habitation sites in the central Mesa Verde region. These calibration data were used to determine how the styles of architecture and pottery changed over time. Eighteen architectural attributes and 24 pottery types were examined. The calibration data were used to specify the probability that each architectural attribute and each pottery type dated from each of 14 time, or modeling, periods (Table 3). As seen in Table 3, these 14 periods are of unequal length. These probability distributions were combined with the sample data from each site to create distributions that specify the relative probability of use of each surveyed site during each time period on the basis of decorated pottery, undecorated pottery, and architectural attributes. We also converted both the estimated date of use recorded by surveyors for each site and the tree-ring dates that were available for particular sites into probability distributions that we could combine with those generated from the pottery and architecture data. Finally, we determined the occupational history of the neighborhood around each site by using the subset of sites within a 7-km radius of each site in question for which decorated pottery tallies of a sufficient sample size were available. We then averaged all available lines of evidence to estimate for each site the relative probability of use during each of the 14 time periods in our study (Ortman et al. 2007*1).
Using a three-step process, we estimated the peak population of each habitation site. First, we estimated the number of pit structures present at each site by counting pit structure depressions, and, for sites where depressions were not recorded by surveyors, by estimating the number of pit structures on the basis of roomblock area and total site area. Second, on the basis of the total pit structure estimate and the probability distributions for excavated sites, we developed a multiple regression equation that predicts the proportion of total pit structures occupied during the period of peak occupation for which peak populations are known. Third, we assumed that each occupied pit structure represented a single household.
The final step in our analysis was to integrate the peak population estimate with the probability distribution to estimate the number of households living at each habitation site during each period in which it was occupied. For habitation sites with peak populations of one household, we assumed that one household lived there during the single most probable period of occupation, and inferred that an additional household lived there during periods corresponding to secondary modes in the probability distribution for that site. For sites with peak populations of more than one household, we again used a multistep process. First, as we did for single habitations, we assigned the peak population to the most probable period. Second, on the basis of the probability distributions and peak populations of excavated sites for which occupational histories are known, we developed a multiple regression equation that predicts the probability threshold above which occupation of a given site should be inferred. Third, on the basis of the ratio of probabilities between the peak period and the chosen nonpeak period, we assigned additional households to each period for which there was a probability of occupation above this threshold. Fourth, on the basis of the ratio of probabilities between the peak period and the period corresponding to the secondary mode, we assigned additional households to periods corresponding to secondary modes in the probability distribution.
Figure 6 illustrates the total number of households that we infer inhabited the recorded sites in the McElmo Dome study area during each of our 14 modeling periods. It is important to note that these are not the momentary populations of these sites, but rather are estimates of the total number of households that occupied each site during each modeling period. To convert these total household estimates to momentary estimates, we must account for variation in both the lengths of our modeling periods and the use lives of houses through time (see Varien and Ortman 2005*1). Figure 7 presents these "momentized" population estimates. These estimates clearly show that ancestral Pueblo occupation of the study area occurred in two cycles characterized by population growth and decline, the first dating between A.D. 600 and 920 and the second between A.D. 920 and 1280.
In this section, we use momentary population estimates generated from the block-survey data to estimate the total number of individuals living in the McElmo Dome study area during each of the 14 modeling periods. To move from momentary resident households at known sites to estimates of the total momentary population of the study area, we make several assumptions. First, we assume that the surveyed area of the McElmo Dome is representative of the entire study area. Whether or not this is actually the case is a complex problem that is beyond the scope of this chapter. Second, we assume that all the large sites containing nine or more pit structures in our study area (see discussion of community centers, below) have been recorded. The surveyed portion of the McElmo Dome is approximately 24 percent of the total study area, so to estimate the momentary population of small sites for each modeling period, we multiplied the total momentary population of small sites by the inverse of the sample proportion. Then we added the momentary populations of the known large sites to this figure to estimate the total momentary population in households. Finally, on the basis of Lightfoot's (1994*1) review of ethnographic data on Pueblo household sizes, we multiplied this household figure by six persons per household to calculate rough estimates of the number of persons who lived in the study area during each modeling period.
Figure 8 compares the total momentary population estimates for the entire study area with the momentary population estimates of known sites. Using this method, we estimate that total momentary population in the study area was around 500 people during the first modeling period, A.D. 600–725. Population increased to a peak of nearly 1,500 people by about A.D. 880, the end of the first cycle of occupation. During the subsequent A.D. 880–920 period, the number of people in the study area dropped dramatically to a low of around 200 people. During the second demographic cycle, population increased steadily to a peak of more than 3,000 people between A.D. 1180 and 1260. During the peak of the second cycle, the population density of the study area was approximately 10 persons per square kilometer.
The peak population density on the McElmo Dome during the A.D. 1200s approached the levels that preceded the emergence of regional polities and hereditary ranking in formative Mesoamerica. For example, the density on the dome was comparable to the population density of the Valley of Mexico between 900 and 650 B.C., as documented by Sanders and his colleagues (Sanders et al. 1979*1:217). This was about 600 years after the initial colonization of the valley by farmers and immediately before the appearance of the first regional polities (Sanders et al. 1979*1). Likewise, the population density of the Valley of Oaxaca increased to 15 persons per square kilometer of prime agricultural land between 1150 and 850 B.C., about 750 years after the earliest agricultural villages formed and during the period in which hereditary-rank society emerged (Marcus and Flannery 1996*1:106). Population densities comparable to those in Mesoamerica were reached on the McElmo Dome a similar length of time after initial colonization by farmers who lived in villages and made pottery. However, instead of hereditary ranking, regional polities, and (eventually) archaic state polities emerging, the Mesa Verde region was completely depopulated. Recent studies on the origins of archaic state polities (Trigger 2003*1; Yoffee 2005*1) note that these polities emerged in a variety of environments, including strongly seasonal, high-desert environments similar to that on the McElmo Dome. Why Pueblo society in the northern Southwest did not develop the same way is an interesting question for further research.
It is clear that the entire ancestral Pueblo population emigrated out of the McElmo Dome study area and the larger San Juan drainage system by the early A.D. 1280s, as evidenced by the complete absence of post–A.D. 1281 tree-ring cutting dates in a regional sample of more than 4,600 cutting dates from more than 350 sites (Kohler et al. 2007*1). It is intriguing that momentary populations of the sites in our sample and our estimates of the total momentary population in the entire study area diverge during the final modeling period, which is the final two decades of Pueblo occupation in the northern San Juan region. Our estimates indicate that the total estimated population of the McElmo Dome began to decline during these two decades, but the momentary population of sites in the block surveys increased markedly during this same interval. Our methods and assumptions may be to blame for this apparent contradiction, and thus the shapes of either the known population curve or the total population curve, or both, could be incorrect. But it is possible that the overall population of the McElmo Dome declined during the final two decades of occupation, even as population density grew in the surveyed portions of the study area.
Assuming that the shape of our total population history is reasonable, we can use the change in population levels through time to investigate patterns of immigration to and emigration from the McElmo Dome. Figure 9 presents our total population history along with a population growth- rate curve. The dashed lines provide an upper and lower boundary for the range in growth rates that can be plausibly accounted for by in situ births and deaths. These boundaries were calculated on the basis of ethnographic literature (Cowgill 1975*1). Growth rates lying outside this range likely result from people moving into or out of the study area. The growth-rate curve indicates that people moved onto the McElmo Dome during the middle A.D. 800s, the late A.D. 900s, and possibly the middle A.D. 1000s, and that people moved out of the study area during the late A.D. 800s and late A.D. 1200s. Note also that it is plausible to conclude that there was modest in situ growth, and a relatively even ratio of population movement into and out of the McElmo Dome, from the middle A.D. 1000s through the middle A.D. 1200s.
Settlement Patterns Through Time
Figure 10 illustrates the locations of all sites in the study area that date from the late Archaic and Basketmaker II periods (1000 B.C.–A.D. 600). These sites were occupied sometime during the 1,600 years immediately before the initial modeling period. This figure shows that there are few late Archaic and Basketmaker II sites. In addition, most of these are limited-activity sites rather than habitation sites. The few habitation sites present were occupied for only a short period of time and probably were not year-round residences. Finally, the length of the late Archaic and Basketmaker II periods combined is more than twice the entire span of time encompassed by the later ancestral Pueblo occupation of the study area. These points indicate that the population density on the McElmo Dome was very low between 1000 B.C. and A.D. 600. The distribution of sites is interesting—most sites dating from this time period are located in the western half of the study area, including a concentration of sites around the reliable spring near the head of Sand Canyon.
Figure 11 is a series of images illustrating the locations of habitation sites in the McElmo Dome that were occupied between A.D. 600 and 1280, during each of our 14 modeling periods. The total number of households per site is illustrated in these images. The large number of sites occupied during the first modeling period of 125 years (A.D. 600–725), compared with the small number of sites dating from the previous 1,600 years, indicates that a significant number of ancestral Pueblo people moved into the McElmo Dome study area during this span of time. However, it is important to remember that these images illustrate the total, not the momentary, number of households at these sites, and that this span is much longer than residences were typically occupied during that time. Only a small proportion of the sites plotted were occupied at any given point in time during this period.
There are fewer total inhabited sites that date from the subsequent period (A.D. 725–800), although the momentary population increased because this period is shorter, the use life of houses was longer, and more sites that contained multiple households date from this period. There are still fewer sites that date from the subsequent A.D. 800–840 period, in part because this period is only 40 years in duration (momentary population increased again during this period). Between A.D. 725 and 880 the greatest population densities, and almost all of the large sites, were located in the eastern portion of the McElmo Dome study area.
There was a noticeable decline in occupation during the A.D. 880–920 period. The few sites occupied during the early A.D. 900s included a few larger sites in the eastern portion of the study area. Population remained low throughout the A.D. 900s, but there was an interesting shift in settlement during this century as the western portion of the McElmo Dome replaced the eastern portion as the area of highest population density. Population continued to increase in the mid-eleventh century, and, over the next two centuries, the eastern half of the study area was almost completely depopulated, even as people clustered ever more densely in the western half. There was also a trend toward increasing aggregation of population into fewer, larger settlements during this period, which culminated during the final two decades of Pueblo occupation of the region.
Communities in the McElmo Dome
In 1983, archaeologists at Crow Canyon began studying the social organization of ancient Pueblo communities in the central Mesa Verde region (Adams 1983*2, 1984*2; Adler and Varien 1994*1; Lipe 1992*3; Lipe and Bradley 1986*1; Varien 1999*1, 1999*7; Varien et al. 1996*1; Varien and Wilshusen 2002*1). In these studies, a community is viewed as a group of people who live in proximity to one another within a geographically limited area, who have face-to-face interaction on a regular basis, and who share access to resources in their local sustaining area (Adler 1990*1; Lipe 1992*3:3; Varien 1999*1:4, 19–23). Previous survey work suggested that each community consisted of a distinct settlement cluster with a densely settled core surrounded by an area of more dispersed settlement. The densely settled core has come to be known as the "community center" (Lipe and Ortman 2000*1; Varien 1999*1; Varien et al. 2000*1). The occupation spans of community centers varied (Ortman et al. 2000*1), but most of the longest-occupied sites in the central Mesa Verde region were community centers. In addition, there is evidence that numerous social, economic, ritual, and political activities took place in community centers and not in smaller habitations (Adler 1994*1; Bradley 1988*1, 1993*1, 1996*1; Driver 1996*1; Lipe 2002*1; Muir 1999*2; Muir and Driver 2002*2; Ortman and Bradley 2002*1; Potter 1997*1, 2000*1; Potter and Ortman 2004*1).
In the next section, we extend previous analyses of communities in the central Mesa Verde region using our database of all known sites in the McElmo Dome, our newly refined demographic estimates for these sites, and a variety of spatial analysis methods. We first examine the locations and spacing of community centers and then examine the clustering of smaller settlements around these centers through time. Finally, on the basis of our spatial analyses, we assign habitation sites to specific communities and use these associations to examine the demographic histories of the Goodman Point, Lower Sand Canyon, and Upper Sand Canyon communities.
In this study, we consider any site containing at least nine pit structures, 50 total structures, or some form of public architecture (Flannery 1972*2:38–39, 1976*1:334–335) to be part of a community center. By this definition, there are 15 community center sites in the McElmo Dome study area. The locations of these sites and associated public architecture are shown in Figure 12. We have also drawn a circular buffer with a 2-km radius around each community center; ethnographic research suggests that this radius approximates the area within which community members would have interacted most regularly and farmed most intensively (Varien 1999*1:153–155; Varien et al. 2000*1:51–52). Some community centers consist of one site. Other community centers, especially those that lasted for centuries, comprise several sites. For example, the residences and public buildings clustered around the heads of Goodman and Sand canyons must have each composed a community center that lasted more than two centuries in each location. Figure 12 also illustrates that community centers were relatively evenly spaced across the McElmo Dome and that there was a low degree of overlap in the 2-km catchments around each center.
Figure 13 presents the peak populations of community centers, the general time period in which the peak population was reached at each center, and the 2-km catchments around each center. The color of each dot corresponds to the period in the Pecos Classification in which the peak population was reached, and the size of each dot reflects the peak population size in households during the peak period. Three important points are apparent from this map. First, community centers that peaked during the Basketmaker III, Pueblo I, and Pueblo II periods (A.D. 600–1140) are located throughout the eastern and central two-thirds of the study area, whereas the centers that peaked during the Pueblo III period (A.D. 1140–1280) are all located in the central one-third of the study area. This temporal pattern mirrors that of the distribution of all habitation sites. Second, no community centers dating from any period are known for the western one-third of the McElmo Dome, even though numerous small habitations are present in this area and additional community centers are located beyond the western edge of the study area. Third, there is relatively little overlap in 2-km catchments even between centers that reached their peak populations during different time periods. This suggests that the community locations were stable through time, even if their populations were not always clustered in central villages. It also suggests that community locations were established early in the Pueblo sequence, and were fixed in space and extent through time, although the fortunes of communities waxed and waned.
Clustering of Population Around Centers
The fact that community center spacing appears to have been somewhat independent of time suggests that communities may have existed in these catchments even during periods when a distinct center was lacking or when the center had yet to reach its peak population. Figure 14 analyzes the degree to which community populations, including households in smaller habitation sites as well as community center sites, clustered within the catchments that we defined using community center locations. The series of 14 images in Figure 14 presents the distribution of households in all habitation sites during each modeling period. In addition, contours representing the density of households in a 2-km radius around each 50-x-50-m pixel are illustrated, with scaling and shading as indicated in the legend. These contours were generated using the kernel density estimate (KDE) spatial analysis routine implemented in ArcGIS, Version 8.0. For an early discussion of KDE as an analysis method, see Baxter et al. (1997*1), and for a recent archaeological application using a geographical information system, see McMahon (2004*1).
One must take the distribution of surveyed areas into account in interpreting the KDE results because isolated sites recorded outside of block-survey areas tend to elevate local densities. Even after taking this into account, there remains a striking correspondence between kernel density estimates of total site populations during each modeling period and the 2-km catchments defined by community center sites. Especially notable is the degree to which, for nearly every modeling period, there is a falloff in density that corresponds to the areas where the 2-km catchments around the Goodman Point and Upper Sand Canyon communities overlap, despite the fact that this area has been thoroughly surveyed. This pattern is apparent even during the initial modeling period, A.D. 600–725, when there were no community centers in the Goodman Point or Upper Sand Canyon communities, but there were clusters of dispersed, single-household residences in both catchment areas.
We draw several conclusions from these results. First, the fact that populations formed clusters that were centered on locations of villages and public buildings supports our identification of villages and public buildings as community centers. Second, the fact that population clusters corresponded in spatial extent to the 2-km catchments drawn around community centers supports the use of 2-km catchments as a proxy for the spatial extent of what Lipe (1992*3:3) calls "first-order" communities and what Mahoney et al. (2000*1:77) term "residential communities." Third, the fact that populations clustered even during periods in which few villages or public buildings were being constructed suggests that evidence of population clustering can be used to define communities in the absence of central sites. Fourth, the existence of population clusters in a given location over several consecutive modeling periods indicates that communities persisted in particular places for longer periods than are suggested by the occupation spans of community centers. It appears likely that both ecological factors—such as proximity to springs, farmland, and canyon resources—and social factors—such as the distances to existing communities—were involved in decisions as to where new communities would be established. But once this social landscape was established, community locations stayed fixed in space through time.
Demographic Histories of Specific Communities
In our final analysis in this chapter, we use 2-km catchments and the results of kernel density analysis to assign habitation sites to the specific community to which each belonged. Then we examine the demographic histories of these communities to determine whether population sizes in the Goodman Point, Lower Sand Canyon, and Upper Sand Canyon communities during Modeling Period 13 (A.D. 1225–1260) were large enough to have resulted in the population sizes of Goodman Point, Castle Rock, and Sand Canyon pueblos during Modeling Period 14 (A.D. 1260–1280). Figure 15 presents a kernel density analysis using the peak populations of all habitation sites, regardless of the modeling period in which the peak population was reached. The colored stars indicate the locations of Goodman Point, Castle Rock, and Sand Canyon pueblos. The dots of corresponding color are the sites that, on the basis of the kernel density map and 2-km catchments around community centers, we assigned to each of these communities. Note that, on the basis of their clustering in the lower tributaries of McElmo Creek, we have included sites from an area larger than the 2-km catchment in the Lower Sand Canyon community. Also, although our western boundary for the Upper Sand Canyon community does correspond to the 2-km catchment around Sand Canyon Pueblo and a falloff in density, the boundary somewhat arbitrarily cuts off a secondary cluster of settlements that does not have a community center in what we term the Burro Point area. Finally, note that nearly all of the areas within which we have assigned sites to a specific community have been covered by a block survey. Thus, it is likely that these sites can be used to estimate the total resident population of these communities during each of our 14 modeling periods.
In Figure 16, we present the population histories of sites assigned to the Lower Sand Canyon, Goodman Point, and Upper Sand Canyon communities. Note that we have included the secondary cluster of settlements on Burro Point with the Upper Sand Canyon sites in the bottom chart of this figure. Several points are apparent from these data. First, the two cycles of settlement apparent among all sites on the McElmo Dome are also apparent at the level of these individual communities. Second, there were enough people in the Lower Sand Canyon community cluster between A.D. 1225 and 1260 to account for the population of Castle Rock Pueblo between A.D. 1260 and 1280. Third, the Goodman Point community was the most populous community on the McElmo Dome, and it consisted of approximately 100 households, or 600 people, throughout the final century of Pueblo occupation. Fourth, the population of the Goodman Point community between A.D. 1225 and 1260 was sufficient to account for the eventual population of Goodman Point Pueblo during the final decades of Pueblo occupation. Fifth, the population of the Upper Sand Canyon community approached the population of the Goodman Point community only during our final modeling period, when Sand Canyon Pueblo was constructed. Finally, it is clear that, even when taken together, the populations of the Upper Sand Canyon community and the Burro Point area between A.D. 1225 and 1260 account for only about one-third of the population of this community after the construction of Sand Canyon Pueblo. This indicates that the population of Sand Canyon Pueblo must have included many households that immigrated into the Upper Sand Canyon community. This finding should be considered in discussions of the social milieu of the McElmo Dome study area during the final decades of Pueblo occupation.
Summary and Conclusions
This chapter has presented a discussion of the settlement patterns and population dynamics in a study area that we call the McElmo Dome. More archaeological survey has taken place in this area than in any other part of the central Mesa Verde region, with the exception of Mesa Verde National Park. The study area has been a focus of research for almost 25 years (1983 to present) by archaeologists working at the Crow Canyon Archaeological Center. This settlement pattern study synthesizes data from all recorded archaeological sites in the study area. As such, this settlement pattern study provides important contextual information for many archaeological studies, including the excavations at Sand Canyon Pueblo.
This study shows that population in the study area was very low during the late Archaic and Basketmaker II time periods, an era that spanned the 1,600 years from 1000 B.C. to A.D. 600. There was a population explosion during the first modeling period examined by our settlement analysis: the A.D. 600–725 period (roughly equivalent to the Basketmaker III period of the Pecos Classification). This population increase was so dramatic that it could have resulted only from immigration into the study area during this time. The nature of this immigration remains an important problem to be addressed by future research.
Our population reconstruction shows that after the study area was settled, population growth occurred in two cycles. The first cycle spanned roughly three hundred years, from A.D. 600 to A.D. 920. Population dropped dramatically at the beginning of the tenth century, almost certainly as a result of emigration from the study area. The second cycle had relatively low population density and slow growth for the next 140 years, between A.D. 920 and 1060. There was dramatic population growth between A.D. 1060 and 1100. Population continued to increase, peaking during the middle A.D. 1200s. At that time, the population density on the McElmo Dome was comparable to that of well-studied areas of formative Mesoamerica immediately before the emergence of hereditary ranking and regional polities. On the McElmo Dome, however, this level of population density was followed by rapid depopulation, which was complete by about A.D. 1285.
Our analysis of community centers and population densities indicates that throughout the Pueblo occupation of the study area, many households formed clusters that were focused on central areas of dense settlement and public architecture. These clusters tended to be approximately 4 km in diameter and tended to be fixed in place through time. We feel that these spatial patterns provide ample evidence for the existence of community-level organization throughout the ancestral Pueblo sequence and generally support the model of community settlement patterning employed by Southwestern researchers for more than 40 years.
As part of a study funded by the National Science Foundation (NSF) and the Colorado Historical Society, Crow Canyon archaeologists have recently completed a demographic study of a larger portion of the central Mesa Verde region. We refer to this as the "Village Project," and its study area is about 1,800 km2. The south boundary of the Village Project study area is just south of McElmo Creek. The west boundary is the Colorado-Utah state line, and the east boundary is just east of Dolores, Colorado. The northern boundary is about the location of Cahone, Colorado.
In Figure 17, we compare the momentary population of known sites on the McElmo Dome to the population history of the larger Village Project study area. These data show that the general trends in the two areas are similar: population growth occurred in two cycles in both areas, and the population peak in the later cycle was much greater than in the earlier cycle. However, the peak population at the end of the first cycle was relatively higher on the McElmo Dome than in the Village Project study area. The most important difference illustrated by these data is that the population of known habitations in the Village Project area began to decline before the final period of occupation (A.D. 1260–1280), but the population on the McElmo Dome continued to increase into this same period. This indicates that population in the larger central Mesa Verde region declined for many decades before the region was completely depopulated. In contrast, population continued to increase in specific communities on the McElmo Dome right up until the final few years of regional occupation. Our analysis of community demographic histories indicates that this population growth was due to the formation of one of the largest villages in the region: Sand Canyon Pueblo.
These data show that the conditions that eventually led to regional depopulation probably developed sometime in the early to middle A.D. 1200s, when the population density of the central Mesa Verde region reached its peak. People living in the region apparently had a dramatic choice to make during the thirteenth century: Do I stay in the central Mesa Verde region, an area that has been home to Pueblo people for seven centuries? Or do I move to areas to the south? Large Pueblo villages formed in these areas to the south during the fourteenth century, and these areas remain occupied to this day.
It appears that Sand Canyon Pueblo in particular attracted people to the McElmo Dome during the late thirteenth century. This large village, in addition to Goodman Point Pueblo, Castle Rock Pueblo, and their surrounding communities, remained home to many who decided to stay on in their central Mesa Verde–region homeland. Ultimately, even these villages were depopulated, such that no Pueblo people remained in the region by A.D. 1285. This settlement study provides the centuries-long settlement context for the formation of these large, late villages. The excavations at Sand Canyon Pueblo, in turn, provide detailed insight into these final decades, and even final years and months of occupation of the region. Together these studies provide an important new perspective on the history of Pueblo people in the central Mesa Verde region.