SITE TYPOLOGY AND ARCHITECTURE
Genevieve N. Head
As long as archeologists have worked in the American Southwest they have sought to reduce the variability of the sites and materials they found in order to more easily make sense of past human behavior. The purpose of this chapter is to categorize the numerous descriptive feature types used by the Pecos survey into a small number of site types that reflect where people lived, farmed, and pursued other subsistence activities. This typology is used in other chapters of this volume to estimate prehistoric population in the valley and to assess change through time in material acquisition, trade, and land use. A brief review of how previous researchers in the Pecos area categorized sites is followed by a consideration of the effects of recent vegetation clearing on the nature of surface remains in the park. This is followed by descriptions of the site components and features recorded by the survey. The survey sites are then grouped according to the presence/absence and character of nonportable features. These groups are placed into three functional site types and expectations for the kinds of activities that occurred at these types are given. The density of artifacts and the number of artifact classes on sites in each of these types are compared in order to examine the artifactual parameters of this classification defined by nonportable features. Additional discussions of the artifactual characteristics of these functional site types are given in Powell's chapter on ceramics (Chapter 8) and Kilby and Cunningham's chapter on lithics (Chapter 9). Descriptions of the survey sites follow, ordered by site type. The last section of this chapter briefly assesses how the formation of Pecos Pueblo affected subsistence and social life in the Upper Pecos Valley by considering two specific feature groupsagricultural features and kivas.
Site Typology in the Upper Pecos Valley
Adolph Bandelier (1881) was the first scholar to work in the Upper Pecos Valley, and his observations from 1880 are characteristically astute and accurate. He did not provide a standard ordering of the sites he recorded however. This may be explained by the fact that Bandelier was only in the area for a very few days (making his extensive measurements and observations all that more impressive) and that Pecos was his very first archeological work in the Southwest.
In 1904 Edgar Hewett categorized the sites in the Upper Pecos Valley into three classes based on size and morphology (Hewett 1904). Pecos Pueblo is the only member of his first class, Class I. His second class includes "several ruins of smaller communal houses" (1904:432) generally with 200-300 rooms, built around a plaza, often with two stories. Also included in Class II are smaller similarly built structures of 10-50 rooms without a plaza. He also noted that Class II sites are generally older than Pecos Pueblo. Hewett's Class III sites consist of rock shelters "of a very primitive type." Although he describes these as numerous in the Pecos area, subsequent investigations, including this survey, have found relatively few rock shelters.
In his 1924 "Introduction to Southwestern Archaeology," A. V. Kidder provides a classification of sites for the Rio Grande area in general (Kidder 1924:84-87). This classification is based primarily on ceramic change, which is not surprising given Kidder's research interests in ceramic chronology. After mentioning an early "pre-Pueblo type" of small settlements that were at that time not yet described, Kidder identifies a stage of single story "small houses" with 3-50 rooms accompanied by black-on-white and corrugated pottery. These are followed by much larger pueblos built around plazas containing kivas. These structures are often at least two stories high. Kidder offers Rowe Pueblo and the early structure under Pecos Pueblo as examples of this stage of architecture. These structures are accompanied by black-on-white ceramics and by early red ware with white exterior decorationprobably White Mountain Red Ware. Kidder's scheme moves to glaze ware, and he notes that with the advent of glaze paint the "ruins become larger, more compact in floor plan, and fewer in number" (1924:86). With the use of red paint on glaze ware came the concentration of settlement in the Rio Grande area into a few large pueblos, such as Pecos Pueblo.
These classificatory schemes focus on the larger "communal houses" of the valley. Kidder classifies sites with three rooms but puts them in the same category as those with up to 50 rooms. These classifications also contain a unilinear evolutionary perspective of complexity, with the smaller houses always assumed to precede the larger. In the 1960s and 1970s, Southwestern archeologists became interested in the smaller sites and settlement systems in general (see articles in Ward 1978). With the shift in Americanist archeology toward more concern for systemic processes, archeologists began to pay more attention to small nonhabitation structures and sites without any architecture at all. Chronological refinements also demonstrated that the less complex sites found throughout the region were not necessarily earlier than the large pueblos but often played different roles in a contemporaneous settlement system.
In the Upper Pecos Valley work done by Larry Nordby for the National Park Service exemplifies this change in interests. In a research design developed for the survey of the original Pecos National Monument, Nordby creates a site typology that divides multiroom pueblos into three groupings based on number of rooms and morphology, similar to those of Hewett and Kidder. Nordby's typology also includes circular rings possibly representing tipis or wickiups, water control structures, pueblo shrines, pithouses and field houses. This typology was apparently preliminary in nature, since it accompanied notes for the research design for the survey (notes on file, National Park Service, Santa Fe), and the reports produced later do not mention it at all. It demonstrated a greater interest in subsistence and land use than is found in the earlier typologies and did not assume that smaller sites necessarily preceded larger ones.
Survey work done near Rowe Pueblo (Cordell 1998) by the University of New Mexico (UNM) field school was performed with the purpose of determining the kinds of sites surrounding Rowe Pueblo, particularly agricultural fields and lithic procurement areas. The types of sites defined by the Rowe survey include lithic and ceramic scatters, agricultural features, lithic quarry/workshops, and three separate structure classes divided by number of roomssingle room, two to five rooms, and greater than five rooms (Cordell 1998:203, Table 26). These groupings are similar to some of the component types identified by the Pecos survey.
The Effects of Mechanical Vegetation Removal on Site Interpretation
One issue that must be considered before any discussion of surface remains at Pecos National Historical Park is the impact of the widespread mechanical removal of piñon and juniper trees. After the livestock boom of the late nineteenth century and the overgrazing and fire suppression that occurred in its wake, grasslands in the Western United States were rapidly encroached by woody shrubs and trees (Bahre 1991, 1995:256; Dick-Peddie 1993:88). During the 1950s and 1960s federal land management agencies encouraged ranchers throughout the Southwest to use mechanical means to remove trees, such as piñon and juniper, decreasing woodlands and increasing grasslands for cattle forage (Bahre 1995:247).
Ninety-five percent of the land that is presently the Pecos Unit of Pecos National Historical Park was formerly part of the Forked Lightning Ranch. Bulldozers were used on the Forked Lightning to clear trees from large portions of the gently rolling plains of the Pecos Valley on the west side of the river (Gilbert Ortiz, former foreman Forked Lightning Ranch, personal communication 1995; Bernie Quintana, machine operator, personal communication 2000). This work was done with bulldozers alone, not by chaining.1 After uprooting, the cleared trees (and any soil or brush brought up or dragged along with them) were pushed into arroyo heads or into large piles. Many of these piles, especially those in open areas, were subsequently burned.
The fact of this clearing was known when the Pecos survey was begun, though its extent and effect were not well understood. Two things were done in site recording to acknowledge and track vegetation removal. First, the survey recording system used several categories for structures and features designed to encapsulate uncertain function. The category of "possible structure" was used to capture ambiguous features "that suggest a structure location (rubble, debris, and/or mound areas) but the nature, extent, and condition of the remains make this a questionable assignment" (Appendix A, Glossary). Figure 5.1 shows a possible structure at the edge of a bulldozed area. The category "unknown structure" was used to designate structures of indeterminate size or nature. The entire feature group "unknown rock features" can record features whose functions were not obvious due to disturbance. None of these recording categories were used exclusively for features disturbed by vegetation removal however. Second, vegetation removal is included as a type of site impact and was recorded on each site where its effects were observed. Survey crews used the presence of brush or dirt piles and obvious scrape marks from bulldozer buckets or treads as indicators of mechanical vegetation removal.
Information collected by the survey will be used to examine the effects of bulldozing on the surface character of archeological sites by comparing sites where vegetation removal was recorded as an impact to sites where it was not recorded. The null hypothesis is that there are no statistically significant differences between sites impacted by vegetation removal and those that were not. This examination is by necessity limited to information collected by the survey, not all of which was intended to directly capture the effects of mechanical clearing. The following is far from an exhaustive study of the effects of heavy machinery on cultural resources in Pecos National Historical Park.
Common sense leads one to conclude that just driving a caterpillar tractor across the artifact scatters and concentrations of rock that characterize surface sites at Pecos would practically obliterate them. Figure 5.2 shows an example of extreme bulldozer damage to a structure. Most archeologists would shudder to imagine what might happen to sites when the buckets of those tractors were used to dislodge and uproot trees. It might thus be expected (1) that there are no or very few sites of any sort within cleared areas of the park. Previous studies of the effects of vegetation clearing have measured the horizontal displacement of surface artifacts and show an average amount of movement as great as 52 cm (20 inches) (Wildesen 1982:56-59). Scatters on sites with removal impacts are therefore (2) expected to be more extensive than scatters on nonimpacted sites because the blade and treads of a bulldozer would spread soil and artifacts. Very little has been reported regarding the effects of clearing on nonportable surface features. Bulldozing would disrupt rock alignments and possibly bury portions of surface features, making them more difficult to specifically characterize. Thus it is expected (3) that there are more features in the "unknown rock features" group recorded on impacted sites. Bulldozing would knock down or consolidate structural walls and truncate structural mounds, making them unrecognizable. It is therefore expected (4) that there are fewer impacted sites with structures and (5) that any structures on impacted sites will be "possible structures" or "unknown structures."
A first step in getting a handle on the extent of clearing activities within Pecos National Historical Park is approximating how much of the park area is affected. Figure 5.3 is a map of the Pecos Unit of the park showing cleared areas, sites where vegetation removal was recorded as an impact, and those where it was not.2 In reference to the first expectation outlined above, that there would be no archeological sites in cleared areas, it is apparent from Figure 5.3 that the impact of clearing was quite variable. Though many of the cleared areas have fewer archeological sites than uncleared areas, they are by no means devoid of sites. Some cleared areas are even ringed by archeological sites, most of which are impacted by vegetation removal.
It is estimated that 25 percent of the present Pecos Unit of the park was treated by vegetation removal. These areas still are readily visible as clearings, though after 30 years without continuing treatment some are almost entirely woodland once again. Vegetation removal was recorded as an impact on 176 sites, 28 percent of all sites recorded.
To address the second expectation discussed above, that artifact scatters on affected sites would be larger than those on unaffected sites Table 5.1 presents descriptive statistics for the area of artifact scatters on the two classes of sites. The survey recorded 837 artifact scatters. After removing outliers (scatters over 30,000 m2, n=5) and those with missing values (n=3), 829 scatters were used in this analysis. The mean area for affected scatters is 2,669.03 m2 while unaffected scatters have a mean area of 1,160.98 m2. A t-test of the null hypothesis that the mean areas of these two groups of scatters are from the same population shows a significant difference between their means (t=5.1527; df=827; alpha=.05; p=.0000).3 This indicates that scatters on sites affected by vegetation removal are indeed more extensive than scatters on sites that were not affected.
Unfortunately specific impacts were not recorded at the feature level so the impact of vegetation removal to individual features cannot be separately quantified. It can be assumed, however, that if a site is affected by vegetation removal, all features on that site are affected.
Table 5.2 contains information to examine the third expectation outlined above, that ambiguous features would be more common on affected sites. There are 277 features recorded on sites with vegetation removal, 27 percent of the total 1,032 features recorded for all sites. Table 5.2 compares the count and proportions of different categories of features on affected and unaffected sites. A chi square analysis was performed to test the null hypothesis that there is no relationship between vegetation removal impacts and the types of features recorded on sites. This test resulted in a chi-square value of 76.772, (n= 1032, df=8) giving a p value of .001. With an alpha of .05 these results require rejection of the null hypothesis, demonstrating that there is a statistically significant relationship between vegetation removal and feature type.
To examine the fourth expectation, that there would be fewer structures on affected sites, a contingency table comparing presence of vegetation removal impacts and structures was created and is shown in Table 5.3. A chi square test of the null hypothesis that presence of vegetation removal and presence of structures are independent leads to rejection of the null hypothesis, indicating a statistically significant relationship between the presence on sites of vegetation removal impacts and structures (alpha=.05; chi-square=3.9287, df=1, p=.04). The high p value indicates this is a fairly weak relationship, but it is significant nonetheless. Sites affected by vegetation removal have slightly fewer structures than would be expected if these variables were independent.
It appears, however, that vegetation removal does not reduce the recognition of structural sites. Consideration of Tables 5.2 and 5.3 shows that though structures are not as frequent as expected on sites impacted by vegetation removal, they make up a greater proportion of the features on sites with vegetation removal than they do on sites without. It might be, however, that clearing activities created concentrations of stone and artifacts that looked more like structures than any other feature type.
The fifth expectation was that there are more structures classified as "possible" or "unknown" on affected sites. Possible or unknown structures together (n=34) make up 37 percent of the structures on affected sites, while they constitute 33 percent (n=60) of structures on unaffected sites. Though not statistically significant, this indicates that clearing makes it slightly more difficult to specifically characterize structures. Relevant to this point, the severely disturbed structure shown in Figure 5.2 was recognized and recorded by the field crew as a small structure.
From this discussion it is apparent that the range rejuvenation activities of the 1950s and 1960s had definite impacts upon the surface cultural resources of Pecos National Historical Park. Though the bulldozers did not obliterate all sites in their paths, they certainly affected many of them. Scatter sizes on affected sites are larger than those on unaffected sites, indicating significant horizontal displacement of artifacts. More features on impacted sites were recorded as "unknown rock features." Structures were less common on impacted sites, and when they were recorded, they were more likely to have been categorized as "possible" or "unknown."
As stated in the introductory chapter, it is difficult to assess the implications these findings have for the project results. Certainly artifact densities on cleared sites will be lower than they would have been pretreatment, and the number of features in unknown categories are more numerous than they might be in surveys of areas with no clearing. The project design used these unknown and possible categories specifically to guarantee that the definite categories (such as "pueblo" and "small structure") are used to record only those features that can be clearly and unambiguously described. Beyond these precautions, little else is done to account for the affects of vegetation clearing in the analyses that follow.
Components and Features
The Pecos survey recorded 629 individual sites and 678 separate components. As described in Chapter 3 of this volume a component was defined as "an intra-site grouping of temporally and/or culturally related remains.... Multi-component sites were recorded when it was possible to attribute the site remains to two or more distinct occupations or cultural periods" (Chapter 3, pp. 49-50). The site component types used in the field were designed to be general yet descriptive and easily identify the dominant feature on a site. Twenty-nine different component types were listed in the site form manual (Appendix A). Eight of these were specifically for Euro-American remains (Table 5.4) while 21 (including "Other") were more general and used in most instances for Native American and prehistoric features, though some could be used for Euro-American remains as well (Table 5.5; see also Table 10.1). Two component types defined in the site manual (Kiva and Campsite) were not used in the field. The Euro-American component types account for 35 of the 678 components recorded (5 percent), while the remaining 643 components (95 percent) fall into the more general component types. The category "Other" was used for unusual or unanticipated component types. Table 5.6 lists the descriptions of components recorded as "Other."
Components are made up of nonportable features and scatters. The survey recorded 1,032 nonportable features and 836 artifact scatters. At the beginning of the project, feature types were organized according to functional groupings. Table 5.7 shows the type and number of nonportable features recorded in the field, organized by these groupings. Definitions of specific feature types can be found in the glossary for the site form manual (Appendix A). While some feature types appear on Table 5.7 they do not appear in the lists of component types (Tables 5.4 and 5.5) because component types were intended to capture only the dominant feature, not every feature, within the component. The three most common types of features are rock concentrations (n=192), grinding areas (n=110), and small structures (n= 109). An example of each is shown in Figures 5.4, 5.5, and 5.6, respectively.
Nonportable features, including structures and nonstructural features, were made using a wide variety of fabrics and construction techniques. Crews recorded several construction techniques and fabric materials for each structure or feature but always listed the predominant technique or material first. Tables 5.8 and 5.9 show the predominant stone fabric types used in structures and construction techniques, while Tables 5.10 and 5.11 show the predominant stone construction techniques and fabrics used with stone in structures and features, respectively. Tables 5.12 and 5.13 show the predominant nonstone fabric types at structures and features, and Tables 5.14 and 5.15 show the predominant construction techniques used with nonstone materials.
Table 5.16 shows the type and number of artifact scatters recorded in the field. Descriptions of the different scatter types can be found in the site form manual (Appendix A).
Classification of the Survey Sites
The component and feature types discussed above provided a means for crews to efficiently and accurately describe and categorize the phenomena they encountered in the field. They are less useful for analysis however, because they supply few indications of the kinds of activities we think were performed on the sites or how the sites functioned within the settlement system. In order to provide analysts with site groupings that more closely represented function two steps were taken. The prehistoric/Native American Pecos survey sites were first classified according to the presence or absence of architecture or other nonportable features and by the type and variety of features present on the site. (The historic/Euro-American sites are discussed separately in Chapter 10). The groupings thus produced were then placed into one of three mutually exclusive site types, discussed below.
Categories for the Pecos Survey Sites
Nonportable features were chosen as the foundation for this classification because they are basic and often highly visible elements of the surface archeological record. Their relatively permanent nature makes them less subject to influence by postdepositional collecting than artifacts. This is a great advantage in an area like the Upper Pecos Valley, where humans have used and reused the landscape intensively for at least 700 years. Because of their portability, artifacts are more often subject to collection than architectural elements. As an example, metates are not common in surface assemblages on any kind of site (see Chapter 9). Though metates are generally less portable than their smaller partners, manos, they apparently are attractive for uses other than grinding. The broad hearth before the fireplace in the great room of the Forked Lightning Ranch house, built in 1926, is made entirely of slab metates. There is no record of the origins of these items, but it can be assumed they came from sites within the park. Other kinds of artifacts may have been functionally transformed in similar ways, without such prominent evidence of their displacement.
Architecture and other nonportable features also can be assumed to reflect human activity on the landscape at a general level and so are an appropriate basis for the present classification. The basic research questions presented in the introduction to this volume are concerned with broad changes and adaptations at Pecos. These are best addressed with a fairly general site classification, rather than attempting to approximate specific uses of particular areas through a more fine-grained classification. Grouping sites by their nonportable features alone also allows the analyses of artifacts later in this report without the danger of circularity that would occur if artifact variables were also used to create the site categories.
There are certainly drawbacks to using features to classify surface remains. What is now present on the surface probably represents only those constructions built of permanent materials, primarily stone. Features created of easily eroded or biodegradable materials, such as brush or jacal, will often be overlooked or difficult to distinguish because they leave few lasting remains. It is also necessary to clarify what kinds of phenomena are meant here by the use of the term "feature." For the purposes of this chapter, this term includes human modifications of the landscape, such as terraces, petroglyphs, and grinding slicks as well as architectural structures that would have contained roofed space. Features differ from artifacts in that they are nonportable. Their permanence gave them the potential to influence patterns of land use and land tenure as they became elements of the Pecos landscape. The presence of a certain kind of permanent feature, a grinding slick for instance, could increase the likelihood that grinding activities would continue to occur in that location. The presence of a relatively labor-intensive feature, such as a one to two room structure, also indicates an investment of time and labor. The structure had the potential not only to shelter but to indicate ownership as well (Head and Snead 1992; Kohler 1989).
Figure 5.7 shows the decision tree for categorizing the survey sites. Sites were first divided based on the presence or absence of nonportable features. Sites with no features were subdivided by the number of scatter types present on the site. Sites with features were split between those with structures and those with only nonstructural features. Structural sites were divided by the size of the largest structure as measured by the estimated number of rooms: 1-5 rooms for small structures, 6-99 rooms for pueblos, and 100 rooms or more for communal pueblos. Examination of the distribution of estimated room numbers suggests that the break between pueblos and small structures is a reasonable division (Figure 5.8). The category "Other structure" was used to capture structures that did not fit in these groups. It includes features such as shrines, tipi rings, and unknown structures.
The decision to divide sites by the presence/absence of structures was based on the assumption that structures serve as permanent or semipermanent shelter and so attract repeated use and overnight stays with concomitant domestic activities. The number of rooms was deemed relevant because it is assumed to reflect the number of people using a site. Structural sites were further subdivided by those that were accompanied by other features and those that were not. Significantly, there are no communal pueblos or pueblos not accompanied by other features.
Nonstructural feature sites were divided by whether they contained multiple types of features or a single type of feature. This distinction was made on the assumption that the number of different feature types, rather than the number of features alone, will most accurately reflect the variety of activities that took place at a site. A site with 20 terraces was most likely the location of a single kind of agricultural activity, albeit an intensive practice of that activity. A site with three terraces, a check dam, a grinding slick, and a fire-cracked rock concentration was very likely the location of a greater variety of agricultural and food processing activities. The ten subtypes derived by this method are shown in the first column of Table 5.17.
Functional Types for the Pecos Survey Sites
While descriptive of the kind of features at sites, these categories do not go much further than the component types used in the field to provide information about the kinds of behaviors or activities that took place at the sites. To answer the questions put forth in the research design, namely what were the affects of aggregation on life in the Upper Pecos Valley, we needed to have some idea of where very general activities occurred. We wanted to know where most people lived, so reasonably accurate estimates of population could be obtained. We wanted to know where agricultural and horticultural activities took place so it could be determined how Pecos Pueblo was funded and maintained. And we also wanted to know where less intensive uses of the landscape, such as gathering of raw materials and wild resources, were performed to get a better idea of their role in the support of the pueblo.
Based on these needs a tripartite division of the survey sites was created. This division is based on a model of the general nature of Ancestral Puebloan land use (Kane 1983; Sebastian 1983). In this model, most Puebloan groups had primary habitations which were the places where most domestic activities took place and which gave shelter throughout the year to families, households, and communities (Kane 1983:23; Sebastian 1983:407). Supportive activities were performed at nonhabitation sites that were used regularly and often intensively during certain seasons (Kane 1983:21; Sebastian 1983:411). In addition, specific resource procurement sites exploited particular resources for very short periods of time (Kane 1983:22; Sebastian 1983:412).
Three site categories were defined: habitation sites, designating primary domiciles; seasonal sites, designating sites used intensively on a seasonal basis; and special-use sites, designating sites used infrequently throughout the year. The ten architectural site groups discussed earlier were classified into these categories (see Figure 5.7 and Table 5.17).
In addition to feature characteristics, there are certain expectations for the kinds of activities at each of these site types that have concomitant expectations for the kinds of artifacts and the locations for each type of site. These expectations are laid out briefly here and are explored further in later chapters of this report.
Communal pueblos and pueblos were placed in the habitation category, since the great number of rooms they contain is assumed to be for the purpose of housing large populations. Habitations are expected to be the location of most domestic activities for the majority of the populationsleeping, eating, food processing, storage, tool maintenance, and tool manufacture. As the domestic foci of communities, habitations are expected to have dense and diverse artifact scatters, the result of discard by large numbers of people throughout the year, over a number of years. Habitations are also expected to have kivas, plazas, or other socially integrative features. Most trade activities are expected to have taken place at habitations, so sites in this type should have the majority of trade goods found in each period.
Small structures and other structures (with and without features) as well as nonstructural sites with multiple feature types were placed in the seasonal category. The structures likely provided roofed and contained space for protection from the elements during short overnight stays for both people and goods, while the multiple types of features suggests a variety of activities took place there. Seasonal sites are expected to be the location of most agricultural activities. We predict that they are located near resources or arable land. They are expected to have served some short-term domestic functions and so may have evidence of small-scale food processing, storage, and tool maintenance. However, the size and complexity of the scatters should be smaller than those at habitations, since seasonal sites are expected to have been used by smaller groups for shorter periods of time. Some seasonal sites might show evidence of activities not related to subsistence, primarily through the specific kinds of structure(s) or feature(s) at the site.
Sites with only a single feature type and those comprised of scatters without structures or features were placed in the special-use category under the assumption that the lack of permanent structures implies that these sites were used infrequently or on a very limited basis. If they were revisited, subsequent visits were made because of the location of the site, not necessarily because of the features there. Special-use sites should be located near wild resources. They should not show signs of storage, because resources harvested from these sites would be moved to other locations for long-term keeping. Artifact scatters at special-use sites should be less dense and complex than those at either habitations or seasonal sites and may show evidence of tool maintenance and/or very small scale food processing. As with seasonal sites, there is the possibility that some special-use sites also were not subsistence oriented at all but were the focus of ritual activities. Because of the inclusion of scatter-only sites in the special-use sites category, there is also the great possibility that many of these sites are not Puebloan but are either earlier Archaic or contemporary Plains nomad sites.
General Artifactual Parameters of the Site Types
As mentioned above, it is undesirable to use artifact parameters to create the site types since that would create circularity in later discussions of ceramics and lithics at the site types. Artifacts are important indicators of site function, however, and should not be neglected in any classification of archeological sites. In order to determine very general artifactual parameters for the site types, statistical tests were run on the number of artifact types and the density of artifacts at sites in each category. A series of three tests was performed on the mean number of artifact types on sites, and another on the mean density of artifacts. The null hypothesis for both series of tests is that the three site types are samples from the same population of sites and so will show no significant difference in the number of artifact types or the density of artifacts. A finding of a significant difference would require rejection of the null hypothesis, indicating that the site categories were distinctly different.
Because this method required three comparisons (habitation to seasonal, habitation to special-use, seasonal to special-use), a Bonferroni adjustment (Sokal and Rohlf 1995:240) was made by dividing the test alpha (.05) by the number of comparisons (3). This gave an alpha of .017. The Bonferroni adjustment is made to control for the greater likelihood of rejecting the null hypothesis even when it is true (a Type I error) that occurs with multiple comparisons. A further adjustment had to be made to the data. The frequency distributions of the number of artifact types and artifact densities were not normal, both being highly right skewed. A log transformation was performed on both distributions, which brought them closer to normality, allowing the use of parametric statistical tests.
Table 5.18 shows the results of the t-test comparisons of the logged artifact type data. The null hypothesis could be rejected in all three cases, indicating significant differences between the number of artifact types at the three site types. In general, the artifact assemblages at habitation sites have more artifact types, making them more complex than assemblages at the other site categories. As can be seen in Table 5.18, the log of the number of artifact types is larger at habitations than at either seasonal or special-use sites. The logs translate into actual numbers of a mean of 66 artifact types at habitations, a mean of 13 artifact types at seasonal sites, and a mean of eight artifact types at special-use sites.
Artifact densities (used as a proxy for total counts, which were not available for the many sites that were sampled) are not as clear cut, but still show some significant differences. In t-tests of logged artifact densities (still using the Bonferroni adjustment), habitations were significantly different from both seasonal sites and special-use sites (Table 5.19). This indicates that there are real differences in the density and complexity of assemblages between habitations and other site types, reflecting the larger size of habitations and the wide variety of tasks performed there. Seasonal and special-use sites are not significantly different from one another in terms of artifact density, however. This indicates that seasonal sites, while roughly of the same density as special-use sites, have more different types of artifacts, suggesting more complex assemblages.
Nonstructural Feature Groups and Site Types
Another means to explore the nature of these site types is to examine the occurrence of different nonstructural features at each site type. It will be remembered that the type of feature (other than presence/absence and size of structures) was not used to create the site types, though the number of different feature types was important. In the survey recording scheme, nonstructural features were grouped into five general categories based on assumed use and/or material/fabric: architectural features; fire-related features; unknown/other rock features; agricultural features; and bedrock features. Structures are not included in this discussion, since they were used to create the site types. Circulation features (trails and roads) are omitted since none were recorded that were prehistoric. Unknown and other features (n=4) are also omitted from this discussion (Table 5.20 shows specific feature types in each group; see Appendix A for descriptions of these types).
The null hypothesis that the type of site and the kinds of nonstructural features found on sites are independent was tested using a chi-square test, with an alpha of .05. Table 5.21 shows the contingency table of site type and feature group. The results of this test (sample size=571 features, df=8, chi square=25.448, p=.001) lead to rejection of the null hypothesis and recognition of a definite association between site type and feature group (Table 5.21). Examination of Table 5.21, which shows the observed frequency, expected frequency, and difference of observed from expected for each cell, reveals that seasonal sites have a higher than expected number of bedrock features, while special-use sites have a higher than expected frequency of rock features. The association of feature groups with site types provides independent information which, when combined with the artifactual parameters of the site types, suggests that there are real differences between these three site types.
Change in Site Types through Time
Table 5.22 shows the number and proportion of components in each functional site type for each period (because each component could potentially occur in multiple periods, summing the numbers and proportions within columns is not of interest). The overall structure of the site types changes little over time. Special-use sites generally make up a little less than two-thirds of the sites in each period, seasonal sites make up a little more than one-third, and habitations make up a small fraction of the sites in each period. This suggests generally stable land use and settlement patterns during the ceramic periods. The biggest change can be seen in the great decrease in the number of habitations from Period 3 (A.D. 1325-1450) to 4 (A.D. 1450-1575), indicating the aggregation of population at Pecos Pueblo.
Site Types across Space and through Time
Figures 5.9-5.13 show the spatial distribution of the three different site types throughout the Pecos Unit of the park for Periods 2 through 6 (A.D. 1200-1700; there were so few sites in Periods 0 and 1 that they are not included here). Visual inspection of these maps reveals several shifts through time. During Period 2 (A.D. 1200-1325), before the building of the Pecos Pueblo Quadrangle, most sites cluster around Forked Lightning Pueblo (PECO 226, LA 672), PECO 23 (LA 14124), PECO 30 (LA 14113), PECO 215 (LA 118811) and Glorieta Creek (Figure 5.9). The area of Pecos Pueblo is completely empty at this time.
The increase in the number and density of sites in Period 3 (A.D. 1325-1450) is quite apparent from Figure 5.10, with Pecos Pueblo (PECO 228, LA 625) and Loma Lothrop (PECO 227, LA 277) surrounded by both seasonal and special-use sites. A cluster of sites is still evident in the vicinity of Forked Lightning Pueblo. The southern portion of the study area and the east side of the Pecos River are dominated by special-use sites, and a cluster of sites is beginning to form at the northern boundary of the park, west of and above the Pecos River, northeast of Pecos Pueblo.
Period 4 (A.D. 1450-1575; Figure 5.11) appears to see increased clustering of activity to the north of Pecos Pueblo and intensification of the cluster to the northeast, above the river, and a slight increase of seasonal sites in the southern portion of the study area, along the Pecos River and to the east of the river.
Period 5 (A.D. 1575-1700; Figure 5.12) sees the apparent breakup of the cluster of sites around Pecos Pueblo and more seasonal activity along the lowlands of the Pecos River, with more special-use sites in the higher elevation areas on the east side of the river. The cluster of sites above the river to the northeast of Pecos Pueblo persists. A severe decline in the number of sites is clearly apparent during Period 6 (post A.D. 1700; Figure 5.13). Chapter 6 looks at the effects of topography, other sites, and local soil productivity on Native American site placement in more detail, using the functional site types developed here. Euro-American sites and artifacts are discussed in Chapter 10.
This portion of this chapter has described the components and features recorded by the Pecos survey and placed them into three general functional types based on their architectural characteristics. A major limitation of this method is that it does not use artifacts to approximate the activities that occurred at sites. This is also its greatest advantage, however, since it allows use of site types in the artifact chapters without fear of circular arguments. It also makes use of salient and easily recognizable information (nonportable features) in a way that can be readily reproduced. Placing components into three site categories reduces the variability of the component and feature types derived from field recording into groups that have associated activities and expectations for other data suites such as settlement, ceramics, and lithics. Comparison of two gross artifact assemblage parametersartifact density and the number of artifact typesand general feature groups between site types demonstrates that these types are separable from one another using independent information. The following section of this chapter provides descriptions of individual habitation sites and more detailed information on the nature of features and artifact scatters on seasonal and special-use sites.
Description of the Survey Sites
The next section of this chapter provides descriptive information about the Native American sites recorded by the Pecos survey (Euro-American sites and artifacts are discussed in Chapter 10). As mentioned earlier, the large number of sites recorded does not allow for individual description of all sites. The discussion is organized by site type. Individual site descriptions are provided for habitation sites (communal pueblos and pueblos) since those are few in number and will be used for estimating population later in this volume. Seasonal and special-use sites are described through tables primarily, with commentary on certain aspects of these sites in the text. Table 5.20 summarizes the different feature types found at sites in each category with features.
Communal Pueblos (100 rooms or more)
The survey recorded six communal pueblos at four sites. PECO 228 (LA 625), Pecos Pueblo (Figure 5.14), consists of three presently visible communal pueblosthe Quadrangle (also referred to as North Pueblo), South Pueblo, and the West Mound.4 There is also a structure known as 700 House, so named in Kidder's notes because it occurred along the 700-foot line of his excavation grid (J. C. White, personal communication 2001), immediately south of the Quadrangle. The Quadrangle is estimated to have had 893 rooms (see Table 7.1) (Room estimates for the Quadrangle are discussed in detail in Chapter 7). From survey data, South Pueblo was estimated to contain 150 rooms; the West Mound was estimated to have 100 rooms; and 700 House was estimated to have 18 rooms. During excavation Kidder discovered structures underlying the Quadrangle and its trash deposits on the mesilla (1958:55-62). These have been referred to as Black-on-white House. He attributed these structures to Black-on-white and Glaze I and II times and noted that they were constructed without any plan, but just by the addition of rooms as needed. There was no evidence of these structures when Kidder began his excavations, nor can they be seen on the surface today.
The Pecos Pueblo structures (Quadrangle, South Pueblo, West Mound, and 700 house) are enclosed by a low rock wall that outlines the top of the mesilla, separating the pueblo complex from the outside world (Kidder 1958:113-118). Kidder estimated this wall was nowhere more than 3.5 feet (1 m) in height (1958:113), and given its great length it would have been ineffective as a physical barrier against attackers. It more likely served to clearly define Pecos Pueblo for both its residents and visitors. Castañeda, who documented Coronado's expedition of the 1540s, notes that visitors to Pecos from the Plains were not allowed to spend the night within the pueblo wall (Winship 1896:524). The date of construction of this wall is unclear, though Kidder notes it must have been after Glaze I times (A.D. 1315 to 1425).
The structures within this wall are all constructed of shaped and unshaped sandstone, and the Quadrangle may incorporate some adobe construction. The room blocks of the Quadrangle form a single enclosed plaza, and a second open plaza is formed by extensions of room blocks on the south end of the Quadrangle (Kidder 1958:Figures 20 and 21). A third plaza might be considered to be formed in the open area between the West Mound and South Pueblo.
Kidder (1958:64) shows 27 kivas within the enclosing wall both inside and outside the plaza of the Quadrangle and within or attached to the Quadrangle room blocks. Three of these kivas predate the construction of the Quadrangle (Kidder 1958:144). Other than these, Kidder did not associate any kivas with the other structures.5. Five of Kidder's 27 kivas were noted as visible by the 1990s survey field crew.
The Pecos Pueblo complex, as recorded by this survey, is surrounded by three separate middens, one on the east slope of the mesilla, below the east room block of the Quadrangle and South Pueblo; a second on the west slope of the mesilla, below what Kidder refers to as the "West Terrace," west of the west room block of the Quadrangle; a third midden was defined on the fairly shallow slope west of the West Mound. In addition two mixed artifact scatters were recorded with this complex. One was south and west of the Quadrangle, between it and the West Mound. This scatter occurs below an area of high sheet wash that has been denuded of vegetation. It is a mixed scatter of lithics and ceramics. The second scatter was defined within the area of the enclosing wall.
It was realized at the outset that this survey could contribute little further knowledge of the main structures of Pecos Pueblo because of the extensive excavation and recording done by previous researchers (see Chapter 2). One contribution this survey could make, however, was the discovery and description of minor non-structural features in the immediate environs of the pueblo structures. The area within the scatters was searched for features. Twenty-one features were discovered and recorded in this manner, including the enclosing wall. Eight of these features appear to relate to excavations, State Monument, or National Park Service activities (one possible structure, six rock concentrations, and one coursed stone wall). The rest of these features are bedrock grinding surfaces, including cupules and flatter grinding areas, one petroglyph (with a cupule) and a bed rock mortar. These features occur in bedrock outcrops on and below the mesilla rim and on the boulders and slabs that have eroded from the mesilla and ring its edge.
The Pecos Pueblo complex was occupied during ceramic Periods 3 through 5 defined for this project (A.D. 1325-1700). Tree-ring dates suggest construction episodes in the 1430s (see Chapter 4 and Appendix E) and historical documents indicate the final residents of the pueblo left in 1838 (Bandelier 1881; Hewett 1904; Kidder 1958).
PECO 710 (LA 251), known as Arrowhead Pueblo (Figures 5.15 and 5.16) consists of one communal pueblo and one pueblo. This site is located on a ridge near Glorieta Pass in the Pigeon's Ranch Unit of the park. The communal pueblo has an estimated 105 rooms and was partially excavated in the 1930s and 1940s by Texas Technical College under the direction of William Holden (Holden 1955; Pearce 1937) (see Chapter 2, this volume). Arrowhead Pueblo is constructed of coursed masonry using shaped and unshaped sandstone with some adobe.
Though it is irregular in shape, the four room blocks of the pueblo form a rough square around a plaza with a single opening to the east. An additional room block is attached to the northwest corner of the main structure. Two kivas are evident in the plaza, one built against the west wall of the plaza (Figure 5.16) and one in the northwest corner of the plaza. This latter was subsequently broken into rooms (Holden 1955:107).
The pueblo adjacent to the communal pueblo is evident from a mound with several rock alignments. It is estimated to have contained only eight rooms, based on the size of the mound. Field crews noted that this mound might have been an earlier structure that was partially dismantled for the construction of the larger pueblo. It was apparently built of shaped and unshaped sandstone, probably coursed masonry.
A series of three terraces that may actually constitute a grid garden was recorded near the pueblo, as were two ground cupules. Other features on the site such as hearths, trails, a scratched petroglyph, and a benchmark, are associated with excavation and historic use of the communal pueblo as a tourist attraction. Field crews recorded a mixed ceramic and lithic artifact scatter on the mesa top surrounding Arrowhead Pueblo and another on the slope below the structure.
According to the chronology defined for this project, Arrowhead Pueblo was occupied during Periods 2 and 3 (A.D. 1200-1450). These dates reflect the early end date for Santa Fe Black-on white used in the chronology construction, however, as well as the removal of glaze ware ceramics from the surface by visitors to the site (Chapter 4). Excavation data and tree-ring dates define a later occupation, running from A.D. 1370 to 1450 (Holden 1955; see also Appendix E, this volume).
The two other sites with communal pueblos are Forked Lightning Pueblo and Loma Lothrop. Both were partially excavated early in the century by Kidder or by individuals associated with his project (see Chapter 2). Both lie in the main Pecos Unit of the park, on the west side of Glorieta CreekForked Lightning to the south and west of Pecos Pueblo, Loma Lothrop to the north and west (see Figure 1.2).
Forked Lightning Pueblo (PECO 226, LA 672) (Figure 5.17) lies in a low area at the juncture of Glorieta Creek (also called Arroyo del Pueblo) and a tributary drainage (called by Kidder "North Arroyo," 1958:Figure 2). Glorieta Creek has eroded an unknown number of rooms from the eastern portion of the structure. Forked Lightning Pueblo was constructed of coursed adobe, with some walls and floors of stone. Little is visible of the structure at the present time (Figure 5.18). Survey crews estimated 150 rooms, though Kidder comments that he never successfully delimited the full extent of the structure (1958:8-9). He shows 140 rooms in the East Pueblo (1958:6, Figure 2).
The shape of Forked Lightning is irregular, reflecting, as Kidder noted, the growth of the pueblo through the agglomeration of rooms and room blocks. The room blocks excavated by Kidder appear to have formed a plaza with at least one opening to the west, though he mentions that the overall form may have included several plazas (Kidder 1958:9). Kidder identified seven kivas, two of which were subsurface, five of which were above ground (1958:29). None of these are presently visible on the surface, though the crew recorded a depression that might be a kiva. As recorded on survey, Forked Lightning Pueblo was accompanied by a variety of nonportable features including: a two-room small structure; three concentrations of burned adobe that may be from excavation or the burning of a portion of the pueblo noted by Kidder (1958:43); seven rock alignments, some of which may be walls within the pueblo; six rock concentrations; and an upright slab. The survey crew also recorded a second component on the site consisting of features associated with Kidder's field camp and laboratory, including the frame of one of his field vehicles. This is most likely "Pecos Black" the less well-known of Kidder's infamous field vehicles. "Old Blue" apparently made it back to the East Coast (Elliott 1995:42-43; Judd 1968:94) (Ann Rasor, personal communication 2001).
The occupation of Forked Lightning Pueblo falls into ceramic Periods 2 and 3 (A.D. 1200-1450) though tree-ring and ceramic evidence suggest a slightly earlier start date of A.D. 1175 (see Chapter 4 and Appendix E). Forked Lightning was probably empty by the time of the construction of the Pecos Pueblo Quadrangle.
Loma Lothrop (PECO 227, LA 277) (Figure 5.19) sits on a knoll above and immediately west of Glorieta Creek. It appears to have been constructed of hand-molded adobe, possibly with unshaped sandstone and cobble footings (Metzger 1990b:5-3). Today the pueblo is evidenced by numerous rock alignments and clear room outlines (Figure 5.20). It is in the shape of an E with room blocks forming two small plazas. Crews estimated 150 rooms in the pueblo. No kivas were visibly associated with the structure, nor found during the excavations of the 1920s (Kidder 1958:47). The pueblo was associated with one possible structure, a concentration of fire-cracked rock, a terrace with upright slabs, a rock alignment, and seven probable agricultural features made of linear cobble scatters parallel to the slope below the pueblo. A single scatter of lithics and ceramics was recorded on the site. Loma Lothrop was occupied during ceramic Period 3 (A.D. 1325-1450).
Pueblos (6-99 rooms)
Eight pueblos were recorded by the survey at eight different sites. One pueblo has been described in conjunction with the Pecos Pueblo complex (700 House) and one in conjunction with Arrowhead Pueblo. The remaining six were found at three named pueblos (Long House, Dick's Ruin, and Shin'po) and three unnamed sites (PECO 23, 30, and 215).
Although two excavated rooms are obvious at Shin'po (PECO 307, LA 267) (Figures 5.21 and 5.22), there is no record of it ever having received scholarly archeological attention. The site sits on a knoll about 50 m above a north-south drainage area west of the Pecos River. Although it is not dissected, this low area generally flows toward the south and joins the arroyo that runs past Rowe Pueblo, approximately 3.2 km (2 mi) to the south. From this knoll, the structure of Shin'po has a clear view of a knoll above the river to the west and the two access points to the river on either side of it. It also has views south to Rowe Pueblo and the bottom end of the Upper Pecos Valley, northwest to Pecos Pueblo, and beyond to the pass at Glorieta.
Shin'po contains approximately 65 rooms (Figure 5.23), but very few artifacts were found in association with the structureno more than 30 sherds were recovered across the site. The slopes below the structure were also examined, and though a few artifact scatters were found at the base of the slope, nothing suggests the kind of midden that might be expected from a site of this size had it been occupied for any long period of time. A small dense lithic scatter located immediately north of the pueblo was accompanied by ashy soil and some burned sandstone and may relate to later, nonoccupational use of the area (Head 1998) (1995 field notes on file at Pecos National Historical Park). A second lithic scatter north of the structure is composed almost entirely of small obsidian flakes. The structure appears to have been built of masonry. No kiva was visible. Jemez oral history (Levine et al. 1995) suggests that Shin'po was used mostly as a large field house or temporary farming location but not as a full time habitation.
Long House (PECO 402, LA 118967) (Figure 5.24) is a single room block of an estimated 40 rooms, apparently two rooms wide, in an open flat area west of a deep bend in the Pecos River (Figure 5.25). Long House also had very few artifacts, though this might be partially explained by its close proximity to the Forked Lightning Ranch House about 0.4 km (0.25 mi) to the southwest. The structure was primarily of masonry construction, using both sandstone and unshaped limestone. Limestone is the predominant bedrock in areas near the Pecos River. This pueblo is accompanied by a small structure and was apparently occupied during Periods 3 and 4 (A.D. 1325-1575).
Dick's Ruin (PECO 434, LA 276) (Figures 5.26 and 5.27) sits in a low area immediately above the Pecos River. It is roughly a C-shaped pueblo with an additional room block attached to the western side (closed side of the C). A low wall between the two ends of the C blocked off the opening of the plaza. The approximately 50 rooms of this structure were constructed of unshaped limestone and sandstone. The structure may have been two stories in some places. As in other pueblos, stones from Dick's may have been used in the creation of later structures, since there is a Hispanic site very close by. No kiva was noted from surface examination, but one corner kiva was found during excavations of the site during the 1920s (Kidder 1958:47). A possible tipi ring with a separate scatter was located just south of the pueblo (Figure 5.28). Dick's was occupied during ceramic Periods 2 and 3 (A. D. 1200-1450), though the dominance of black-on-white ware and the very small number of early glaze ware sherds suggests an end date closer to 1425.
The three unnamed pueblos are all in close proximity to Pecos Pueblo and Forked Lightning Pueblo. They are quite enigmatic but were recorded as pueblos in 1976 or 1992 by Nordby, and after field examination, the 1990s crews confirmed these attributions.
PECO 23 (LA 14124) (Figure 5.29) is the most enigmatic of the three unnamed sites, consisting of concentrations of burned adobe and a small rubble mound with a vegetation anomaly/depression that may be a kiva. It is in association with four petroglyph panels and numerous oval grinding areas on a bedrock outcrop immediately above a currently active spring. The possible kiva is in close association to the grinding areas and petroglyphs. The pueblo structure was estimated to have contained 10 rooms based on the extent of the adobe and rubble scatter.
PECO 30 (LA 14113) (Figures 5.30 and 5.31) consists of mounds and rock scatters with some rock alignments. It is immediately south of Square Ruin (PECO 35, LA 14114), a Spanish-era enclosure (Nordby and Creutz 1993b) (James Ivey, personal communication 1997). The field crew noted that stone from the structure at PECO 30 was probably used in the later construction of Square Ruin. This probable stone robbing made it difficult to interpret the exact shape and size of the structure. Crews estimated 45 rooms based on the extent of the mound and rubble. The structure was predominantly masonry, with some adobe.
At PECO 215 (LA 118811) (Figure 5.32) the pueblo structure is most easily recognized by a masonry corner exposed in a recent bulldozer cut (Figure 5.33). This corner is visible about 0.3 m (1 foot) below present ground surface and is at the west end of the site. On the surface to the east of this exposed corner are found rock alignments that are in line with the southern wall of this corner and suggest the extent of the pueblo, estimated to have contained 16 rooms. This pueblo structure is accompanied by an unknown structure that shows a slight depression (possibly a kiva), a tight concentration of rock, and a small structure evidenced by two alignments of upright cobbles that form a corner. An ash stain containing burned adobe and slag is also in immediate proximity to the pueblo.
These three pueblos, though all difficult to detect from the surface, suggest an occupation of the Upper Pecos Valley that may have predated Forked Lightning Pueblo. In a consideration of occupation probabilities based on ceramic types, these pueblos appear to date slightly earlier than Forked Lightning (see Chapter 4). Adobe construction and the borrowing of rocks from these sites to build later structures would explain their ambiguous appearance.
Seasonal sites are defined as those with structures of six or fewer rooms, with or without features, and sites with no structures but with multiple nonstructural feature types present. There are 206 prehistoric sites in this site type. The kinds of nonstructural features found at seasonal sites are presented in Table 5.20. The types of structural features on seasonal sites are presented in Table 5.23. In Table 5.23, the category "Unknown structure" represents features that crews could determine were certainly structures but could not determine their exact nature due to disturbance from bulldozing, trees and over growth, or vandalism.
The most common types of features at seasonal sites are rock concentrations and petroglyphs/pictographs. In Table 5.21, a cross-tabulation of site type and feature group, the greatest difference between expected and observed frequencies is in the association of seasonal sites with bedrock features. Table 5.20, showing feature types at site types, indicates a high frequency of petroglyphs/pictographs and grinding areas on seasonal sites. Eighty-two percent of all petroglyphs/pictographs and 61 percent of all grinding areas are found on seasonal sites.
Table 5.24 shows the kinds of artifact scatters associated with seasonal sites. Fifteen seasonal sites had no artifact scatters at all. Lithic dominant lithic and ceramic scatters (where lithics account for more than 80 percent of the artifacts) were the most frequent, followed closely by mixed lithic and ceramic scatters (where neither artifact type makes up more than 80 percent of the assemblage). Lithic only scatters were also fairly common.
Special-use sites are defined as those with only a single feature type or with no features whatsoever. There are 394 sites in this category. Two-hundred ninety-nine, or 76 percent, are artifact scatters with no features. The types of features on special-use sites are shown in Table 5.20. Rock concentration/scatters are the single most frequent kind of feature at special-use sites, followed by grinding areas, rock alignments, and petroglyphs/pictographs. The kinds of artifact scatters on special-use sites, both with and without features, are shown in Table 5.25. Lithic dominant artifact scatters are the most common type of scatter on special-use sites, followed by mixed ceramic and lithic, then lithic only scatters. A chi-square test of the null hypothesis that the variables in Table 5.25 are independent did not lead to rejection of the null hypothesis, indicating no significant relationship between the presence/absence of nonportable features and the kind of scatters on special-use sites.
The greatest apparent difference between scatters on special-use and seasonal sites is that lithic only scatters are much more common on special-use sites than on seasonal sites28 percent and 14 percent respectively. Otherwise, the proportions of the different scatters at the two site types are roughly equivalent (see Tables 5.24 and 5.25). This may reflect the presence of preceramic scatters among the special-use sites.
Subsistence and Social Change through Architecture
The purpose of the third section of this chapter is to examine the changes that took place pre-historically in the Upper Pecos Valley as seen through architectural data. Two categories of architecture will be considered specifically because of their inherent interest and their suitability for addressing concerns set out in the introduction of this volume, namely, how Pecos Pueblo differed from its predecessors and how it supported itself. Agricultural features are of interest because they can inform upon the methods and strategies used by the residents of the Upper Pecos Valley to grow crops and how aggregation and trade affected these strategies. Kivas are of interest because they can give us clues to how the Pecos people were organized and how that social organization was changed or maintained throughout the life of the pueblo.
A population the size of Pecos must have relied on its own food production. Spielmann's simulation study for the protohistoric period indicates that in most years Pecos should have been able to supply corn for itself and at least three months' worth for trade (Spielmann 1991b). Bone chemistry analysis suggests that until the late 1600s, corn made up at least 75 percent of the diet at Pecos (Spielmann et al. 1990). Agriculture is thus assumed to have been a crucial activity to the residents of the Upper Pecos Valley.
Before the founding of Pecos Pueblo, agriculture is expected to have been performed primarily at the household level and to have required only minimal improvements for the retention of soil and water. Most agricultural activities should have taken place near habitations (this specific expectation is addressed in the Settlement chapterChapter 6) and should have required little in the way of shelter or field houses.
With the foundation of Pecos Pueblo and the increase in population and population density, we expect that there should be an increase in the intensification of agriculture. Though the household remains the basic unit of the pueblo economy (Ford 1972; Snow 1981), agricultural intensification may have involved the addition of community-level activities (Adler 1994; Snead 1995). Intensification should appear as increased investment in agriculture, evidenced by an increase in the number of agricultural features and the creation of more complex or labor-intensive features. The number of agricultural features is expected to increase throughout the occupation of Pecos Pueblo to supply the population there, as well as to fund Pecos' trade relationships.
These expectations are tested by examining the number and type of agricultural features recorded by the Pecos survey. The number and types of agricultural features should increase from ceramic Period 2 to Period 3 (A.D. 1200-1450) (formation) and remain high through ceramic Period 6 (post A. D. 1700) (maintenance). There should be more agricultural features in general in these periods than in earlier ones, and there should be more features that would require more than a single household to create and maintain, such as reservoirs or irrigation systems. There should also be more small structures, reflecting an increase in agricultural investment, as well as exploitation of areas farther from the habitation, necessitating greater shelter at field locations.
The survey recorded a total of 38 agricultural features. Table 5.26 shows the different kinds of agricultural features recorded and the types of sites on which they occur. Check dams were defined by the survey as "linear arrangements of stone (one to a few courses high) built across drainages for the purpose of slowing runoff and perhaps capturing water and soil" (Pecos Survey Site Manual Glossary, Appendix A). Grid gardens were defined as "rectilinear arrangement of rocks, often contiguous, forming a grid-like pattern; built to help retain soil and water in garden plot areas; typically single course alignments." Terraces were defined as "rock alignments or walls (one to a few stones high) built perpendicular to the slope direction ("on contour"), to control surface runoff and down-slope movement of soils." Reservoir/catchments were defined as "natural and/or modified (excavation, damming) catchments or basins used as water collection areas."
The last two feature descriptions in Table 5.26 require some explanation. The seven field borders are the linear alignments on the slopes below the pueblo of Loma Lothrop (PECO 227) (Figure 5.19). The field crew did not feel they fit the description for either terraces or grid gardens but that their arrangement suggested more the definition of a field area. The two rock piles both occur on the same site (PECO 334) along with a small structure, an unknown structure, and two check dams. The crew noted that both rock piles appeared to be for use as water or soil control and hence could be considered agricultural.
Over 73 percent of these agricultural features occur on seasonal or special-use sites, and well over half (58 percent) occur on seasonal sites. This lends verity to the expectation that agricultural activities would take place predominantly at seasonal sites. These 38 agricultural features occur on a total of 24 sites. Twenty agricultural features (over 52 percent) occur on a site with another agricultural feature.
Table 5.27 shows the type and number of agricultural features in each period. There are three features on sites that could not be dated with ceramics. Two are grid gardens; one is a terrace. Many agricultural features appear on sites dated to multiple periods. It is assumed for this discussion that an agricultural feature was initially created in the first period in which it appears. This is considered the time of initial investment. The feature is then assumed to have been used with little cost in the subsequent periods to which it was assigned. The number of new features in each period is shown in the right hand column of Table 5.27 as an indication of investment in agriculture in each period.
Agricultural features first appear in Period 2 (A.D. 1200-1325), with 13 new features and four feature types. The number of features doubles in Period 3 (A.D. 1325-1450), and the number of types increases with the addition of a reservoir northeast of Pecos Pueblo and a complex of field borders on the slopes below Loma Lothrop. Grid gardens are also added. Period 3 appears to be the time of greatest agricultural investment, with 19 new agricultural features (73 percent of the total for the period) created. This finding supports the expectation that aggregation will be accompanied by more agricultural features, more types of agricultural features, and the creation of larger, more complex features.
The number of agricultural features declines in Period 4 (A.D. 1450-1575), however. Only three (17 percent) new agricultural features are created in this time period, and only 18 appear to have been used at all. This represents a decline of 69 percent from the number of agricultural features used in ceramic Period 3 (A.D. 1325-1450). The number of agricultural features declines by 33 percent in Period 5 (A.D. 1575-1700), and there are no new features created. This decline continues through Period 6 (post A.D. 1700), with only one agricultural feature dated to this period, and no new features created.
It would appear from these results that the initial formation and occupation of Pecos Pueblo and the peak of population estimated from survey sites during ceramic Period 3 (A.D. 1325-1450) (see Chapter 7, Figure 7.1) was indeed accompanied by an increase in agricultural intensity. The peak of aggregation that occurred in ceramic Period 4 (A.D. 1450-1575) (see Chapter 7, Table 7.3), however, and the maintenance of Pecos Pueblo as a large aggregate up to ceramic Period 6 (post A.D. 1700) was apparently supported by reuse of preexisting agricultural features or with little agricultural activity at all. This is contrary to expectations and suggests that perhaps Pecos did, indeed, rely on interregional trade for its livelihood.
The first thing that strikes one on perusal of Tables 5.26 and 5.27, however, is that there are so few agricultural features. The thirty-eight features recorded here mean there is on average one agricultural feature for every 158 acres (64 hectares) of land surveyed at Pecos. In contrast, the Bandelier survey recorded a total of 1,156 agricultural features of types very similar to those recorded by the Pecos survey (VanZandt 1999:332-333). This gives an average of one agricultural feature for every 12 acres (5 hectares) of land surveyed on the Pajarito Plateau. If we increase the number of agricultural features at Pecos by assuming that every rock concentration and rock alignment recorded was agricultural in function, that increases the number of agricultural features to 289. That gives one agricultural feature for every 20 acres (8 hectares), still far below that for the Pajarito Plateau.
There are several explanations for this relative dearth of agricultural features at Pecos. One is previous land use within the park. The lands of the Upper Pecos Valley have been intensively used certainly since Spanish colonization (Adams and Chavez 1956; Hall 1984; Kessell 1979) and very likely before that. The lands within Bandelier National Monument historically have been heavily grazed but rarely cultivated. Much of Bandelier has been protected since the 1930s, and major portions have been designated wilderness since the 1970s, while the land at Pecos has been a productive ranch since the 1930s, and many of the archeological sites themselves have been tourist attractions. At Pecos, bulldozing performed during the 1950s and 1960s across roughly 25 percent of the lands that are now within the park may have obliterated many small features.
The number of recognizable agricultural features still seems miniscule, despite these depredations. One other explanation is that the location of much of the intensification of agricultural efforts and investments made by the Pecosenos occurred in areas outside the boundaries of Pecos National Historical Park. In a description of New Mexico missions in 1776, Fray Dominguez gives an account of the pueblo of Pecos and states that "The Indians have arable lands in all the four principal directions, but only those which lie to the north, partly east, enjoy irrigation. The rest are dependent on rain" (Adams and Chavez 1956:213). "To the north, partly east" points directly to the broad floodplain of the Pecos river in the area now occupied by the village of Pecos, outside the boundaries of Pecos National Historical Park. Through the park, the river runs in a steep canyon, with little or no floodplain on either side.
In her consideration of the subsistence system of Pecos Pueblo, Spielmann (1991b:102-104) notes that Manzano fine sandy loam soils, soils suitable for irrigation, occur only in the floodplain to the northeast of the pueblo. Evidence for capital investment in agriculture by the people of Pecos Pueblo might be found in the acequias that still water these lands (Hall 1984). Vibo and Ribera fine sandy loams, which Spielmann (1991b:102-104) notes are currently used for dry farming, can be found in the immediate vicinity of Pecos Pueblo, about 1 km (0.62 mi) due north, and on the west side of Glorieta Creek. The drop in agricultural features within the park beginning with Period 4 (A.D. 1450-1575) may indicate a shift in emphasis of agricultural activities away from dry farming by individual families in these areas to greater communal exploitation of the irrigable land to the northeast by the river.
The formation of large villages through the aggregation of human population has been called "socially difficult and economically inefficient" (Cordell 1996:230). Close proximity of many people increased the potential for interpersonal conflict and stress. The consolidation of people from several communities into a single pueblo would have created challenges for decision making and for social integrationintegration that was essential to the well being of the community and the individuals within it.
In a community where group consensus formed the basis for most decision making and where coercive measures were not used to ensure cooperation, one means of relieving the scalar-stress associated with aggregation would be to increase the size of the basal social unit, thus decreasing the number of social units involved in decision making (Adler 1989; Johnson 1982). Communitywide ritual provided a valuable means of maintaining and reinforcing social order above the level of the household, relieving scalar-stress and preventing village fragmentation (Adler 1989; Hegmon 1989). Many (though certainly not all) such rituals were held in kivas.
As community size increased through aggregation, the size of groups using kivas should also increase. As the size of groups using kivas increased, so the size of the kivas themselves should have increased. Logically, an increase in the size of base social groups is expected to shrink the number of groups in the community. If population growth or aggregation is great enough, however, or if there are cultural constraints on the minimum number of social units required for maintaining a viable community, there is the possibility that the number of basal social units will remain the same or even increase. If it can be assumed that there is a positive relationship between the number of social units and the numbers of kivas, changes in the size and numbers of kivas can reflect changes in the size and numbers of the social groups using them. With the consolidation of population into Pecos Pueblo, therefore, it is expected that there were larger kivas and most likely more kivas, as evidence of an increase in the number and size of base social groups.
Unfortunately, but not surprisingly, very few kivas were identified on survey. For this reason, most of the information on kiva numbers and sizes presented here was reconstructed from Kidder (1958) and reports of other excavations in the Upper Pecos Valley (Cordell 1998; Hayes 1974; Holden 1955). There are several pueblo sites with no excavation information and/or no surface indication of a kiva (Shin'po PECO 307; Hobson-Dressler LA 32388; Loma Lothrop PECO 227; Long House PECO 402; PECO 30). It is assumed that if the inhabitants of these pueblos used kivas, they went to those at other sites.
Table 5.28 lists all the possible kivas reported for the Upper Pecos Valley, along with the sizes and ceramic periods for use of the kivas, where these could be determined.6 Table 5.29 shows the total number of kivas estimated and the number of rooms estimated to be occupied (from the demography chapter, Table 7.2) for each site in each period. Table 5.30 shows kiva size, both average and total, for the measurable kivas from each period. Each kiva was counted in every period to which it, or the site it is on, was dated, so many appear in multiple periods.
The total number of estimated kivas in the Upper Pecos Valley increases from 16 in ceramic Period 2 (A.D. 1200-1325) to 17 in ceramic Period 3 (A.D. 1325-1450), then drops to 15 in ceramic Period 4 (A.D. 1450-1575), to nine in Period 5 (A.D. 1575-1700), and remains at nine in Period 6 (post A.D. 1700). This reflects the patterns of population growth, outlined in Chapter 7 (Demography) where population peaks in Period 3. The large drop to Periods 5 and 6 are very likely the result of population decrease at the pueblo during Spanish colonization (Kessell 1979; Levine 1999). The greatest number of kivas at a single site increases from seven at Forked Lightning Pueblo in Periods 2 and 3, to 15 at Pecos Pueblo in Period 4.
The number of rooms per kiva shows no real trend through time (Table 5.29; Figure 5.34) but counters the expectation that there would be more rooms (as a proxy for people) for each kiva in Period 4 (A.D. 1450-1575), the peak of aggregation. This index almost doubles from Period 2 (A.D. 1200-1325) to Period 3 (A.D. 1325-1450), then drops in Period 4, increases again in Period 5 (A.D. 1575-1700) and drops precipitously in Period 6 (post A.D. 1700). The great drop in this ratio during Period 4 shows that the increase in number of kivas did not keep up with the increase in the number of occupied rooms. The rise in the ratio of rooms to kivas in Period 5 may reflect the effects of Christianity on the people of Pecos. Those who accepted Christianity may have continued to reside in the pueblo but no longer used kivas.
Kiva size changes little from Period 2 (A.D. 1200-1325) to Period 3 (A.D. 1325-1450) but increases between ceramic Periods 3 and 4 (A.D. 1325-1575), jumping from an average of 23.26 m2 to peak at 43.74 m2 (Table 5.30; Figure 5.35). Average kiva size declines to 34.07 m2 in Period 5 (A.D. 1575-1700), then drops to 30.35 m2 in Period 6 (post A.D. 1700), but never drops back down to the average sizes for Periods 2 and 3. This conflicts with the information on room to kiva ratios, which suggests that the number of people per kivas drops during Period 4 (A.D. 1450-1575) but confirms the expectation that kiva size, and hence basal social group size, increased with aggregation.
Examination of Table 5.30 and Figure 5.35 shows that Period 4 (A.D. 1450-1575) sees the introduction of one very large kiva. Kiva 12, estimated to have been built during ceramic Period 4 (Kidder 1958:199-204), covers over 134 m2almost two and a half times the size of the next largest kiva at Pecos Pueblo during the same time period. This may shed light on the low number of rooms per kiva for Period 4. Though the number of social groups seems to have dropped slightly (from 17 kivas in Period 3 [A.D. 1325-1450] to 15 in Period 4), a higher level integrative structure was introduced in Period 4, the time that aggregation peaked (Adler 1989). This suggests the existence of rituals that brought together the entire community through a single integrative facility and may have incorporated those not served by smaller kivas. Kidder positively identified Kiva 12 as a kiva, but he concluded that it was never completed or used, primarily because it was never roofed (Kidder 1958:218). Unroofed kivas or open ceremonial areas that greatly resemble kivas, are not uncommon from archeological contexts in western New Mexico, eastern Arizona, and the Mesa Verde region, however (Herr 1994:46; LeBlanc 1989:348; McGimsey 1980:171-176; Orcutt et al. 1990:200; Roberts 1939:128). Period 5 (A.D. 1575-1700) also has one kiva that is much larger than the others used in that period, though the difference is nowhere near as great as that seen for Period 4.
It would appear from this information on kiva number, size, and ratio to habitation rooms that the formation of Pecos Pueblo was not accompanied by an increase in the number of base social units as the number of kivas in the Upper Pecos Valley increases by only one from Period 2 (A.D. 1200-1325) to Period 3 (A.D. 1325-1450), then drops by two in Period 4 (A.D. 1450-1575). This suggests a consolidation of social groups, rather than a proliferation. Average kiva size does increase, however, as might be expected if more people were coming together in fewer groups. The low ratio of habitation rooms to kivas in Period 4, however, conflicts with these findings. The formation of the Quadrangle at Pecos Pueblo (late in Period 3 or early in Period 4) was also accompanied by the introduction of a kiva much larger than all others at the pueblo. This kiva may have served as a higher level integrative facility pulling together the many separate social groups that used the smaller kivas.
The pueblo was also constructed to create an enclosed plaza. Plazas can provide space for communitywide ritual, beyond that provided by kivas, and it has been suggested that enclosed plazas replace great kivas throughout the Southwest (Adams 1991). Construction of room blocks around plazas can also increase the frequency of face-to-face encounters between individuals in the community which, while they do not occur within a ritual context, can serve to reinforce integration. The presence of an enclosed plaza is not a complete departure from previous villages in the valley, however. Both Shin'po (PECO 307) and Arrowhead (PECO 710) are built to fully enclose their plazas.
Summary and Conclusions
Through description of sites, components and features this chapter has presented a general introduction to the sites recorded by the Pecos CRIS. It also organized the sites through a simple typology based on architectural characteristics. Consideration of artifact assemblage parameters and feature characteristics in each site type confirmed that basic differences exist among the types. This typology does not rely on artifact variables to classify sites and so provides an independent base for further discussion of site function in later chapters. In addition, the chapter examined subsistence and social change within the Upper Pecos Valley using two kinds of prehistoric features, agricultural facilities and kivas.
From these analyses it appears that the land-use and settlement structure of the Upper Pecos Valley was fairly stable through time, with a preponderance of special-use and seasonal sites, and very few habitation sites in any period. The major change in settlement structure was the reduction of the number of habitation sites from 10 in Period 3 (A.D. 1325-1450) to two in Period 4 (A.D. 1450-1575), with the aggregation of population at Pecos Pueblo. Consideration of agricultural features suggests that Pecos Pueblo thrived without major corporate capitol investment in agriculture, at least not in the dry-farming that was probably performed within what are now the boundaries of Pecos National Historical Park. Historic documents and information on soils suitable for irrigation suggest that the "bread-basket" for Pecos Pueblo lay in the rich floodplain of the Pecos River, in the area that is now the Village of Pecos, outside the present study area. These irrigated fields may have seen the investment in agricultural facilities that was expected to accompany the formation of the pueblo and to fund its continuation.
The consideration of integrative ritual facilities, in particular kivas, suggests that the formation of Pecos Pueblo brought with it a consolidation of social groups, as seen in the low number of kivas and the low ratio of habitation rooms to kivas. Kiva size is much larger after the formation of Pecos Pueblo than at earlier sites, however, suggesting that with greater aggregation kivas were used by more people. Pecos Pueblo also had one kiva that was much larger than all the others at the pueblo, indicating the existence of a new overarching level of integration that brought together members of the smaller kiva groups.
The image of life and change in the Upper Pecos Valley derived from the information presented in this chapter is one of basic stability, with changes in settlement and integration at key points in time. The formation of Pecos Pueblo brought about change in the organizational structure of the people of the valley, but their modes of subsistence appear to have remained fairly constant through time. The following chapters in this report make use of the typology developed and presented in this chapter in tandem with the chronology presented in Chapter 4, to give more detailed consideration to many of the issues broached here.
1. Though reportedly not employed on the Forked Lightning Ranch, chaining is a commonly used method for clearing piñon and juniper. Chaining involves dragging a heavy anchor chain between two large bulldozers or caterpillar tractors, uprooting trees in the path of the chain (deBloois et al. 1975).
2. Figure 5.3 was created using GIS layers in ArcView 3.1 and a digital orthophotograph of the park. Cleared areas within the park boundary were visually identified on the computer and their outlines screen digitized after fieldwork was completed. They had not been field checked at the time of this writing. The GIS layers and database are on file at Pecos National Historical Park and the National Park Service in Santa Fe.
4. These names are derived from Kidder 1958 and unpublished field notes. This survey did not record the Spanish mission church and convento that sit at the south end of the mesilla. For detailed information on these structures, see the recently completed Historic Structures Report (Ivey 1996).
6. The size and dates of kivas at Pecos Pueblo were reconstructed from Kidder (1958). When exact dimensions were not mentioned in the text, measurements were taken from plan view maps of individual kivas. No dimensions are given for unexcavated kivas or those for which Kidder does not give specific dimensions or map individually. Since Kidder refers to the dates of kivas by reference to the Pecos ceramic Glaze ware sequence, some rules were developed for assigning individual kivas to the ceramic periods created for this report. If Kidder noted that a kiva was used in black-on-white times, it was assigned to ceramic Period 2 (A.D. 1200-1325). If he noted that a kiva was used in Glaze I or II times it was assigned to ceramic Period 3 (A.D. 1325-1450), in Glaze III, IV, or early Glaze V it was assigned to ceramic Period 4 (A.D. 1450-1575). Kivas noted to be used in Glaze V, VI, or to have "modern painted ware" were assigned to ceramic Period 5 (A.D. 1575-1700). Those that Kidder noted to have been built historically or after the Pueblo Revolt were assigned to ceramic Period 6 (post A.D. 1700). In addition, Kidder often was able to determine the time a kiva was dug but not when it was abandoned. Likewise, there were some kivas where he was able to determine a date of abandonment but did not excavate enough to determine when the structure was built. These kivas were assigned to the single period mentioned by Kidder. If Kidder noted that a kiva was used for a very long time, it was assigned to each period covered by that interval.
Last Updated: 13-Feb-2006