Archaeobotanical Remains

by Katharine D. Rainey and Sandra Jezik


To understand prehistoric plant use at Woods Canyon Pueblo, Crow Canyon researchers collected and analyzed two types of samples: macrofossil and flotation. Macrofossil samples are pieces of plant material large enough to be seen with the unaided eye, and excavators retrieve them by hand as they dig and screen sediment in the field. Macrofossil specimens constitute a subjective sample of the contents of a given deposit, because collection depends on what an individual excavator decides to gather (not every piece of plant material seen in the course of excavation is collected). The macrofossil samples considered in this study consist predominantly of wood charcoal.

Flotation samples are samples of sediment taken from specific archaeological contexts. Because they are collected and processed systematically, they provide the most important and reliable data about prehistoric plant use at a site. Most flotation samples consist of a standard volume of sediment collected from contexts where archaeobotanical remains are expected to be plentiful—for example, primary refuse, secondary refuse, and roof-fall deposits. The samples are poured into a bucket of water, which is agitated to separate the organic botanical material from the sediment. The plant remains that float to the top (hence the term "flotation") are then examined under a microscope to identify the different plants present. Flotation is useful in recovering small seeds and other remains that ordinarily would not be captured in a standard screen used during excavation. Although flotation samples provide an unbiased sample and therefore better—and more comparable—data, it is the combined macrofossil and flotation data sets that provide the best understanding of how the inhabitants of a prehistoric site used plant resources.


Macrofossil Samples

The first step in analyzing the macrofossil samples from Woods Canyon Pueblo was to create a subsample by randomly choosing 20 pieces of wood charcoal from each sample. If fewer than 20 pieces were present, all were examined. If more than 20 pieces were present, once the first set of 20 had been analyzed, any additional pieces that were morphologically distinctive to the naked eye were examined. The pieces were snapped in order to expose a fresh transverse (cross) section, which was then examined under a dissecting binocular light microscope at magnifications of 10 to 45X. Specimens were identified only to the level of genus; identifying items to the species level would have required examining radial and tangential sections, which we were unable to do because of time constraints and limited microscope capability. We identified taxa using a modern wood charcoal comparative collection backed by voucher specimens in the University of Arizona herbarium.

Flotation Samples

Sediment samples for flotation were processed by the Crow Canyon laboratory staff and were usually a standard size of 1 liter. If more than 1 liter of sediment was collected in the field, only 1 liter was processed, and the remainder was set aside for future research. If less than 1 liter was collected, the entire sample was processed and the volume was recorded.

The samples were poured into a bucket of water, and the water was stirred to free the organic materials. The mostly nonorganic material that settled to the bottom of the bucket constituted the "heavy fraction," which was collected, allowed to dry, and curated. Carbonized (burned) remains that floated to the surface were poured into a fine screen, .355-mm mesh, to be captured as the "light fraction." The light fraction was allowed to dry for several days and then was poured through a series of geologic sieves. This process separated the light fraction into 4.75-mm, 2.80-mm, 1.40-mm, .71-mm, and .25-mm portions, which were then individually bagged and labeled. (Although the size of the original mesh used to capture the light fraction is .355 mm, smaller particles that adhere to larger particles while wet can detach as the residue dries and be caught in the .25-mm screen used during dry screening. Plant remains also continue to break into smaller pieces whenever a sample is handled.)

The light fractions were sorted in two steps. The first step involved subsampling the wood charcoal. We examined 20 pieces of wood charcoal from every sample whenever possible. The pieces were randomly selected from the 4.75-mm portion first, because the larger size allowed for more confident identification. If it was necessary, we chose individual pieces from the 2.80-mm portion to round out the 20 pieces. After the subsample of 20 pieces was analyzed, any morphologically distinctive pieces were examined.

The second step of the procedure was to analyze the rest of the light fraction (the 0.25-mm sieve size was not sorted, because it was assumed to contain broken pieces of plant remains from the larger-sized sieves). The 4.75- and 2.80-mm portions were completely sorted for seeds. A "species area curve" was used to subsample the 1.40- and .71-mm portions. This approach maximizes the number of taxa recorded while minimizing the volume of sample sorted (Adams 1993*1:196). Because the interpretative emphasis of this study was on the presence or absence of taxa, our goal was to identify the maximum number of taxa present in a sample, rather than to record the total number of items identifiable to those taxa. The 1.40- and .71-mm portions were sorted in increments of 0.90 ml. Each unit of 0.90 ml was measured with a graduated cylinder. Three successive 0.90-ml units were sorted. If no new taxa were observed the third time, no more units were sorted for that sieve size. If new taxa were observed, then another 0.90-ml portion was sorted; this process continued until no new taxa were found.

We analyzed the light fraction using a dissecting binocular light microscope at magnifications ranging from 10 to 45X. Specimens were identified using a modern seed comparative collection backed by University of Arizona herbarium voucher specimens. Reference texts (such as Martin and Barkley [1961*1]) were also used. If none of these approaches was successful, the specimen was measured and described as "unknown."

The Sample

Of the 160 flotation and 451 macrofossil samples collected during the three seasons of excavation at Woods Canyon Pueblo, 58 flotation and 73 macrofossil samples were analyzed (36 percent and 16 percent, respectively, of the total samples within each category). In keeping with Crow Canyon's sampling strategies at other sites, the samples selected for analysis were collected primarily from secondary refuse (for example, middens), primary refuse (for example, ash in thermal features), and mixed deposits (Table 1).1 In addition, we selected samples collected from naturally redeposited secondary refuse because in situ secondary refuse (midden) deposits were not preserved on the steep slopes of the site. Roof-fall deposits and "unspecified" cultural deposits are also represented. The distribution of samples by type of deposit is presented in Table 2.

A minimum of 34 plant taxa were present in the analyzed flotation and macrofossil samples (Table 3). Specimens were identified to the most specific taxonomic category possible. In most cases, this was the genus level. Occasionally specimens could be identified only to the level of family, and in rare cases, specimens were identified to species. Because of the potential overlap between the listed families and genera, more species may be present than are actually indicated; for example, the category Pinus-type may include both Pinus edulis and Pinus ponderosa.

During analysis, one of three levels of confidence is indicated for each identification: "absolute," "-type," or "cf." As an example, an "absolute" identification to the level of species "implies that all the species of a given genus in the surrounding area have been examined and the specimen under the microscope could easily disappear in a population of seeds of the named species" (Bohrer and Adams 1976*1:1). The label "-type" is used when a well-preserved specimen has morphological characteristics that are identical to those of the named species, but not all of the species of that genus that occur in the local environment have been examined (Bohrer and Adams 1976*1:1). The label "type" is included only in the tables of this report and should be assumed to apply in the text. The abbreviation "cf." is applied to specimens that are difficult to recognize because of poor preservation or other problems (Bohrer and Adams 1976*1:1). These three labels are not limited to identifications made at the level of species; they can be adapted for use with any taxonomic rank.

When a specimen is listed with two genus names, such as "Amelanchier/Peraphyllum" or "Prunus/Rosa," it means that the specimen could be a member of either of the two genera, and a more precise determination is not possible. The order in which the names are given does not indicate which genus is most likely; the names are simply listed alphabetically. Some wood charcoal taxa may be listed in the database with both the genus and species listed. For the purposes of this study, members of the same genus were combined because it was impractical to examine the necessary radial and tangential sections. Thus, the genus level was considered to be the most accurate identification. In this way we strove to record as much information about a specimen as possible, while reducing the chances for erroneous identifications.

Both charred and uncharred archaeobotanical remains were recovered from Woods Canyon Pueblo (Table 4). Charring is generally considered to probably be the result of prehistoric cultural activities. Unburned remains are more likely to be modern or to have been introduced through noncultural processes—for example, rodent activity and wind—unless a good case to the contrary can be made on the basis of their archaeological context (Minnis 1981*1:147). Because we can make no such argument for the uncharred remains from Woods Canyon Pueblo, they are excluded from consideration in this chapter. Partly charred specimens were included with the charred specimens for the purposes of this analysis.

More-detailed information relating to our analyses and subsequent interpretations can be found in two on-line publications. For complete descriptions of the criteria used to identify each plant taxon and part identified in the Woods Canyon archaeobotanical assemblage, refer to the Plant Identification Criteria. For sources of the ethnographic information presented in this chapter (for example, possible uses of plants), refer to the Ethnographic Uses of Plants.


The data derived from the analysis of archaeobotanical remains allow us to at least partly reconstruct plant use at Woods Canyon Pueblo. In the following sections of this chapter, we examine evidence of plant use for food, fuel, and construction, and we compare the various sections of the site in an attempt to discern patterning in the archaeobotanical assemblages (refer to "Architecture and Site Layout" for a discussion of the subdivision of the site into four sections for purposes of comparative analysis). Additional discussions address a variety of issues related to the physical environment and how the inhabitants of Woods Canyon Pueblo lived on the land, including the proximity of agricultural fields to the pueblo, resource depletion, food stress, and seasonality. Finally, we present a basic reconstruction of environmental conditions as they may have existed when the village was occupied.

Food Plants

The archaeobotanical remains from Woods Canyon Pueblo demonstrate that the inhabitants were corn agriculturalists (Table 5). The parts of the corn (Zea mays) plant found most commonly were cupules and kernels, although cobs, embryos, and stalks were also recovered. The ubiquity of corn parts (found in 36 of 131 total analyzed samples) in the Woods Canyon Pueblo archaeobotanical record suggests that this plant was a major source of food. The cupules and cob fragments are probably the remains of cobs that were burned for fuel after the corn was shelled.

Although it is likely that the people of Woods Canyon Pueblo also ate squash (Cucurbita) and beans (Phaseolus), as did inhabitants of other ancient Pueblo sites in the region, no remains of these plants were identified in the flotation or macrofossil samples that were analyzed. It is possible that squash and beans were not part of the inhabitants' diet, but it is more likely that the remains are missing because of poor preservation and/or the methods used to prepare these plants for consumption—they are very fragile and might have been prepared in ways (for example, boiling) that would not have led to accidental charring and discard. It is also possible that bean and squash remains are present in the many samples that we did not examine; more samples were not analyzed than were analyzed.

A variety of wild resources were available to the people of Woods Canyon Pueblo for food (Table 5). Remains of virtually all these potential food plants were identified in the flotation samples. Cheno-am (Chenopodium/Amaranthus) seeds were the most common, occurring in 16 samples. Goosefoot (Chenopodium) and pigweed (Amaranthus) grow in disturbed habitats such as are found in fields and along paths; therefore, they were probably abundant in the vicinity of the site. The inhabitants of Woods Canyon Pueblo might have collected only the seeds, or the seeds might have been transported incidentally with whole plants, which can be eaten as greens. Since the vegetative parts of cheno-ams and purslane (Portulaca) are usually quite bitter by the time their seeds are mature, the presence of seeds in the Woods Canyon assemblage probably indicates that it was the seeds, not the greens, that were intentionally collected as food.

Groundcherry (Physalis) seeds were also quite common in the archaeobotanical record at Woods Canyon Pueblo, being found in 12 flotation samples. Groundcherry is another weedy species that grows in disturbed environments. People probably brought the fruits, which mature in late summer and early fall, into the pueblo, and the seeds somehow were charred and became part of the archaeobotanical record.

The presence of burned pine cone fragments in four samples is indirect evidence of the consumption of pinyon pine (Pinus edulis) nuts by the inhabitants of the pueblo. Pinyon pine grows on the site today. Since the nut-crop yield varies from year to year, this resource probably would not have been dependable, but in good years, it would have provided abundant food. Purslane (Portulaca) seeds could have been found in many of the same places as the cheno-am and groundcherry plants, as it, too, thrives in disturbed habitats. Other seeds occurred in lower numbers but could have been eaten by the people of Woods Canyon Pueblo nevertheless.

Remains in Primary Refuse

The best information on the use of plants for food at Woods Canyon Pueblo comes from two sources: primary refuse from hearths and secondary refuse from middens. Primary refuse, which was found at the place where it was generated, gives us a picture of the last few cooking episodes in a hearth and thus provides direct evidence of what the inhabitants ate. Flotation samples from four hearths in three kivas (Structures 2-S, 6-S, and 8-S) yielded evidence suggestive of food preparation at the site (Table 6). Unfortunately, one hearth (Feature 8) in Structure 2-S and the hearth in Structure 6-S contained no reproductive plant parts. The charred cheno-am seeds found in the hearth in Structure 8-S (Feature 1) may be the remains of a batch of seeds that were accidentally burned in the process of being parched. Groundcherry seeds also were among the food remains found in this hearth. In ethnographic studies, groundcherry fruit is usually described as being boiled, which generally does not lead to accidental burning, but the contents of the cook pot could have bubbled over into the fire, allowing some seeds to become charred. It is also possible that the Woods Canyon inhabitants roasted the fruits.

A seed belonging to the Solanaceae family was found in one of the hearths in Structure 2-S (Feature 3). Given the prevalence of this genus in other flotation samples from the site, it is probably another groundcherry seed; however, other members of this family also grow on the surrounding landscape today. Corn in various forms was found in both hearths that yielded reproductive plant parts. The kernels could have been roasted or parched over the fire, but the cupules are probably remnants of cobs used for tinder and fuel.

Remains in Secondary Refuse

Flotation samples from secondary refuse in middens complement the samples of primary refuse from hearths. Secondary refuse is trash, including ash from hearths, that accumulates over a period of time; it provides indirect evidence of what foods were consumed, by revealing the things that were discarded. Middens in each area of the site were tested, and 40 flotation samples were collected from in situ secondary refuse. Many more taxa were present in these samples (Table 7) than were present in the primary refuse samples (Table 6).

Remnants of pine cones were found in four samples. Historically, green pinyon cones have been heated over fire or among coals to release the nuts, a process that can result in charring. Alternatively, the cones could have been used for tinder or fuel.

Charred goosefoot (Chenopodium) and purslane (Portulaca) seeds were also found. These seeds could have entered the archaeobotanical record in a variety of ways. They could have been charred in the process of being parched, or they could have been unintentionally fried with the greens and discarded into the fire. As stated earlier, however, their presence is most likely the result of a charring accident during parching.

Ricegrass (Stipa hymenoides) caryopses were found in two flotation samples. The grain of this late spring-early summer plant has a very tough outer coat, which must be removed to expose the edible interior. In the ethnographic literature (refer to the Ethnographic Uses of Plants), the most commonly recorded technique for accomplishing this is to light the grass over a fire, then catch the inner grains as they fall out. Hedgehog cactus (Echinocereus) seeds were found in two flotation samples. Because a common method of removing spines from cactus fruit is to burn them off, these seeds probably represent a roasting accident. There were other charred seeds that might have been leftovers from food processing and consumption, and they are listed in Table 7.

Corn (Zea mays), groundcherry (Physalis), and cheno-ams were the most common taxa identified in secondary refuse deposits, mirroring their prevalence in the primary refuse samples. These three foods are thus most likely to have been dietary staples for the residents of Woods Canyon Pueblo, although additional food resources also appear to have been used on occasion.

Fuel Sources

At least 20 wood taxa were present as charcoal in the flotation and macrofossil samples (Table 8). Overall, juniper (Juniperus) and pine (Pinus) were the most common woods, followed by sagebrush (Artemisia) and mountain mahogany (Cercocarpus). Juniper and pine are the dominant woods on the landscape today, suggesting that a pinyon-juniper forest was also present during the occupation of Woods Canyon Pueblo. Sagebrush is very common in fallow fields. Mountain mahogany is usually associated with pinyon-juniper woodlands and is common today in the area of the site. Cottonwood/willow (Populus/Salix) charcoal is less common in the assemblage than these taxa, but still present. Cottonwood and willow grow in riparian environments and are found in the bottom of Woods Canyon. Excavators found other charred wood taxa that are also common on the modern landscape in upland and lowland habitats (Table 8).

One way to study fuel use is to examine wood charcoal found in kiva hearths. Plant remains from four hearths in three kivas (Structures 2-S, 6-S, 8-S) were analyzed for this study (Table 9). These samples were collected from primary refuse and thus represent the final use of the hearths. Juniper and pine were the most common taxa, appearing in samples from all four hearths. The large limbs of these trees and their prevalence in the area would have made them the best choices for fuel.

Sagebrush was the next-most-common taxon identified in the hearth samples (present in three of four hearths). Ethnographically, sagebrush has been considered a second-choice fuel. It is, however, one of the first shrubs to grow back in old-field succession, which means that it would have become more prevalent as more and more of the pinyon-juniper forest was cleared for agricultural fields. Charcoal from serviceberry/peraphyllum (Amelanchier/Peraphyllum), mountain mahogany, and ephedra (Ephedra) were found in two hearths; these and the other woods all appear on the landscape today, but they might have been less common when Woods Canyon Pueblo was inhabited. Alternatively, these shrubs might have been used less often for fuelwood because of their other desirable qualities—serviceberry, for example, yields an edible fruit, which might have made this plant a more important food resource than fuel source.

Another way to study fuel use is to examine wood charcoal cleaned from hearths and discarded in the middens. The charcoal from the secondary refuse in the middens (Table 10) presented much the same picture as did the charcoal from primary refuse in hearths. Juniper and pine are the dominant taxa, followed distantly by sagebrush and mountain mahogany. These four taxa appeared in middens across all four sections of the site.

Three additional taxa were found in the midden samples—saltbush (Atriplex), bitterbrush (Purshia), and rose (Rosa). These taxa were relatively rare. They probably were found only in the middens because more midden samples were analyzed and because middens were subject to longer periods of deposition than were hearths. The remains of these shrubs were still very rare in the secondary refuse deposits, occurring in only one or two samples, which suggests that they were used infrequently as fuelwood.

One interesting flotation sample from Nonstructure 7.6-N contained saltbush, rabbitbrush (Chrysothamnus), and lemonade berry (Rhus) charcoal—three of the four traditional Hopi kiva fuels (only greasewood [Sarcobatus] was missing from the mix). This flotation sample was collected from secondary refuse, so the origin of the charcoal is uncertain, although it likely represents discarded fuelwood.

Construction Materials

Wood charcoal from Woods Canyon Pueblo can also help us identify construction materials used at the site. The small sample from construction contexts (N = 8) limits the conclusions that can be made, however. One flotation sample and seven macrofossil samples were analyzed from roof-fall and wall-fall deposits in Structures 1-S, 2-S, 3-S, 5-S, 6-S, and 7-S (Table 11). The samples from Structures 1-S, 2-S, and 6-S did not contain any cultural materials. Juniper charcoal and a corn cob were found in the wall-fall and roof-fall debris of Structure 7-S. The roof of Structure 7-S is believed to have burned, so the charcoal might be a piece of a juniper roof beam. The corn could have been sitting on the roof or hanging from the rafters when the roof burned and collapsed, or it could have been incorporated into the roofing materials during construction.

The two macrofossil samples collected from roof-fall deposits in Structure 5-S contained juniper, pine, and sagebrush charcoal. The juniper and pine might be remnants of burned roof beams. The sagebrush could have been used as secondary roofing material. The flotation sample from Structure 3-S was collected from a layer of ash in the roof fall of Structure 3-S; it yielded juniper wood, pine bark scales, and a cheno-am seed. If the ash is the remains of burned roofing materials, then the juniper wood might be from a roof timber, and the pine bark scales from earlier pine beams. Alternatively, the ash could have been dumped as secondary refuse after the roof collapsed, or it could have rested on top of the roof and collapsed with it. If the deposit was secondary refuse, then the plant remains recovered could be from a hearth that was cleaned out. Cheno-ams are a weedy species, so the seed found in the sample might have blown or fallen onto the roof and become incorporated into the roof-fall deposits when the roof burned.

In summary, although our conclusions are limited by the small sample size, it seems that the wood of choice for roof timbers was juniper. Juniper is the predominant wood in the tree-ring samples from the village, along with some pinyon pine. Sagebrush appears to have been used as roofing material as well.

Intrasite Comparisons

In this section, we compare the plant remains found in the four spatially discrete sections of Woods Canyon Pueblo: the upper west side, the canyon bottom, the east talus slope, and the rim complex (see "Architecture and Site Layout" and Database Map 330). Comparisons of plant use among these four areas were central to the questions posed in "Research Objectives and Methods." In this chapter, we present the results of spatial and temporal analyses of the archaeobotanical assemblage in an attempt to (1) shed light on functional differences, if any, between the four sections of the site and (2) discern changes in plant use through time. Together, these analyses provide a picture of plant use at Woods Canyon Pueblo.

Plant remains from several different types of deposits were included in these studies (Table 2). The main data are derived from analysis of samples from primary and secondary refuse. A third important source of information is naturally redeposited secondary refuse—that is, trash that has been moved from its original location by nonhuman forces such as wind, water, or gravity. Although this material has been redeposited, it almost certainly is in the same section of the site where it was originally discarded.

Other types of deposits represented in the archaeobotanical assemblage from Woods Canyon Pueblo include mixed refuse (an assemblage containing a combination of the main refuse types) and cultural deposits that are "not further specified" (deposits known to be the result of prehistoric human activity, but for which we cannot determine the exact type). In our analysis, we look at these types of deposits collectively as well as separately. Grouping the samples from different types of deposits allows us to increase the sample size; considering them separately provides insights into the different types of activities that might be represented. We begin by looking at how sample size and preservation affect analytic results and interpretations.

Methodological Concerns

Sample Size and Diversity

Sample size can affect analytic results in two ways: a large sample is expected to have a greater diversity of remains, while a smaller sample may not include a representative sample of all the plants that were used at a site. There are two types of sample size that concern us here: flotation sample size and the size of the overall site sample. "Flotation sample size" refers to the amount of sediment processed for each sample. "Overall-site sample size" refers to the number of flotation samples taken from each section of the site, as well as the total number collected from the entire site. Because the Woods Canyon Pueblo flotation samples were a uniform 1.0 liter in volume, we can eliminate flotation sample size as a potential source of bias in this study.

We cannot, however, discount the possible effect that the overall-site sample size (that is, the number of samples from each section and from the entire site) might have had on our results. Because there were a limited number of appropriate primary and secondary refuse contexts in each section of the site, we selected an unequal number of flotation samples from each for analysis (see Table 12). Twice as many samples from the canyon bottom were analyzed (N = 27) as were analyzed from the upper west side (N = 12) and east talus slope (N = 13), and four times as many samples were analyzed for the canyon bottom as for the rim complex (N = 6). The small number of analyzed samples from the rim can be attributed to the small number of proveniences that were test-excavated in that section of the site. Clearly, the variation in number of analyzed samples from the different site sections creates the potential for overall-site sample size bias.

In an attempt to compensate for potential bias, we applied the statistical program DIVERS (Kintigh 1998*1) in our examination of (1) the distribution of plant remains among the four sections of the site, (2) the distribution of plant remains among the various types of deposits, and (3) the distribution of plant remains in the in situ secondary refuse assemblages. The DIVERS program is well suited to such analyses, because it compares the actual assemblage with simulated assemblages that are created with respect to probability distribution of the actual assemblage. Since sample size may have affected the quantity and types of archaeobotanical remains observed, we wanted some way to make spatial and temporal comparisons that would not be biased by the effects of sample size. The DIVERS program allowed us to determine whether the diversity of assemblages was greater than what we would have expected given their sample size (Kintigh 1984*1:44).

There are two assumptions behind the DIVERS approach to simulating assemblages (Kintigh 1984*1:45). The first is that "for a given artifact typology and cultural situation there is an underlying frequency distribution" of items in the classification system, determined through culture and societal norms (Kintigh 1984*1:45). The second assumption is that the assemblage was created by choosing randomly from the potential pool of elements. That is, a component's ultimate presence in the assemblage is influenced by the probabilities based in the societal norms (Kintigh 1984*1:45). Through these assumptions, the model provides a randomly created grounds for comparison. That, in turn, allows us to draw conclusions about the extant data set (Kintigh 1984*1:45).

The Woods Canyon Pueblo data do not violate the first assumption, but it is possible that they violate the second. This is not a great concern, however, since we would expect that the archaeobotanical assemblage at Woods Canyon Pueblo was not created randomly. The purpose of a simulation is to create a confidence interval through which we can determine whether the actual assemblage deviates from expected ranges, and if so, by how much.

The DIVERS program employs a Monte Carlo–style method (Kintigh 1998*1:51) that creates a set of simulated assemblages by drawing items independently and at random, according to the probabilities found in the combined frequency distribution of the individual assemblages. From the given assemblage, a sample is chosen randomly with the same sample size as the actual assemblage; if the actual sample has 25 elements, then the simulated assemblage will have 25 elements as well. Finally, the approach creates a large number of randomly simulated assemblages; from these, it computes an expected range of diversity for the samples (Kintigh 1984*1:47). Therefore, using DIVERS analysis, we can attempt to compensate for the small sample size from certain sections of Woods Canyon Pueblo.

The first step in examining the four sections of the site for diversity of plant remains was to compare their assemblages with the randomly assorted simulated assemblages (Table 12, Figure 1). The results of the simulation show that the numbers and ubiquities of plant taxa in the upper west side and east talus slope are very similar to the distributions expected if the plant taxa were randomly apportioned among the samples—that is, they fall within the 90 percent confidence interval. The rim complex and the canyon bottom differ from the other two sections in that they have more plant taxa than we would expect if the archaeobotanical remains were apportioned randomly. This greater richness might be the result of sample size, but it could also be related to preservation or prehistoric human behavior.

With 34 different plant taxa and parts, the in situ secondary refuse assemblages have the greatest number of taxa of all deposit types (Table 13). Because 40 of the 58 Woods Canyon Pueblo flotation samples are from in situ secondary refuse, these samples have the greatest effect on the analyzed assemblage as a whole (Table 2, Figure 2). When we consider only the archaeobotanical remains from in situ secondary refuse, the pattern of greater richness for the rim complex and canyon bottom is less striking (Table 14). The richness of the rim-complex assemblage is at the upper limit of the 90 percent confidence interval, whereas the richness of the canyon-bottom assemblage is close to the upper limit but still below the boundary (Figure 3).

Why do the samples from the canyon bottom and rim complex have more plant taxa than expected? More flotation samples were analyzed from the canyon bottom than from the other sections, which would lead us to expect to find a greater number of plant taxa there. The diversity analysis theoretically corrects for this bias, however, which leads us to infer that the canyon bottom has even more taxa present than its large sample size would predict. Thus, larger sample size alone does not seem to account for the greater-than-expected richness in the canyon bottom. Instead, preservation or behavioral factors could be playing a role. The rim complex also has a greater-than-expected richness of plant taxa and parts. This is unexpected, because the rim complex has the fewest analyzed flotation samples (six). To understand the canyon-bottom and rim-complex assemblages, we must turn to alternative explanations such as differential preservation and behavioral differences.


Factors that influence the preservation of archaeobotanical materials include temperature variation and extremes, sediment conditions such as pH and moisture content, and the depth of sediment covering the plant remains. Exposure to the elements probably played a large role in the preservation of plant remains at Woods Canyon Pueblo. The rim complex, situated high above the canyon bottom, is susceptible to considerable wind and water erosion, as well as to temperature extremes. In addition, it has much thinner midden deposits, which would afford less protection to the plant materials in those deposits. We would expect preservation to be poorer there. The canyon bottom, on the other hand, has deeper middens, which would afford buried plant remains better protection from the elements. The middens in the canyon bottom are visibly ashier, lending support to the idea that organic materials in this part of the site were better preserved. The archaeobotanical assemblages from both the rim complex and the canyon bottom, however, are characterized by greater-than-expected richness. Thus, although differential preservation may have had some effect on the Woods Canyon archaeobotanical assemblages, this factor alone cannot explain all of the observed patterns.

Spatial Analysis

Spatial analyses of the archaeobotanical remains from Woods Canyon Pueblo were conducted in an effort to determine whether the four sections of the site—the upper west side, the canyon bottom, the east talus slope, and the rim complex—had different functions. First, to maximize sample size, we evaluated the data from all analyzed flotation samples from all contexts. Then, in an attempt to identify specific activities that might have taken place in the different sections, we focused on flotation samples from in situ secondary refuse and on macrofossil and flotation samples from primary refuse in kiva hearths. Within each group, we looked at reproductive plant parts for insights into food use and at wood charcoal for evidence of fuel use. Comparisons were made on the basis of the ubiquity of each plant taxon and part. Data patterns were tested using the DIVERS program, which was run at least twice for each set. If the two trials produced different results, the program was run a third time; if the results of the first two trials agreed, no further trials were run. The results of these tests are presented below.

All Contexts

When the flotation data from all contexts at Woods Canyon Pueblo are considered, the four site sections are basically similar in terms of taxa and parts present; generally, the same taxa appear in all four areas (Table 12). Yet, as the diversity analysis demonstrated, the rim complex and canyon bottom have a greater number of taxa than would be expected when compared with the simulated assemblages (Figure 1). Apparently, some factor is exerting an influence on the samples from the rim and canyon bottom that is not a factor in the samples from the other two sections of the site.

When only reproductive taxa are considered, the four sections of the site seem generally similar in terms of food use—corn (Zea mays), cheno-ams, and groundcherry (Physalis) were the most commonly recovered food plants in the analyzed samples (Table 15). However, the canyon bottom has the greatest diversity, with 17 seeds and other parts indicative of food use, and the east talus slope has the fewest food taxa, with six. Rarer seeds, such as juniper (Juniperus), chokecherry/rose (Prunus/Rosa), and ricegrass (Stipa hymenoides), occurred only in the canyon bottom.

These differences may be due to preservation or human behavior, but we cannot rule out the effects of sample size. The diversity analysis of reproductive taxa shows that the assortment of taxa is within the expected range (Table 15, Figure 4). The notable variety of foods in the canyon bottom could be the result of longer occupation in this part of the site—because this section was occupied first, we would expect to see a greater variety of foods accumulate over time in the archaeobotanical record. Also, the deeper middens would have promoted preservation of plant refuse. Thus, although the assemblage of reproductive plant parts from the canyon bottom is not unexpectedly diverse for its sample size, it could still reflect better preservation and longer occupation. The range of food remains recovered could also indicate that the canyon bottom was an area of the site where foods were routinely prepared.

The small number of food-plant taxa found in the rim complex warrants discussion. The two most common food plants at the site—corn and cheno-ams—are present in the samples from the rim complex, yet few other seeds were found there. If the rim complex was a public space used for ceremonial activities or feasting, we might not expect to find much evidence of food preparation. Rather, food might have been prepared at residences and brought to the public space, with the refuse being discarded at the residence locations. The middens in the rim complex are very shallow, which suggests that refuse was not routinely discarded there. Because the rim complex is so exposed and the middens are so shallow, we would also expect poor preservation compared with the preservation of plant remains in other parts of the village. Yet the number of food taxa present, although small, is not unusually small for the size of the sample.

The archaeobotanical remains from the upper west side and east talus slope do not show any patterns out of the ordinary with regard to food use. Although these sections of the site have fewer food taxa than were found in the canyon-bottom samples, many of the same plants are represented. The presence of food taxa is consistent with the interpretation of the upper west side and east talus slope as areas of the village where food was prepared.

Just as food use was similar in the four sections of the site, so, too, was fuelwood use. When the taxa identified in the wood charcoal assemblage from the 58 flotation samples were tabulated, the four areas had almost identical plant lists (Table 16). The most common fuel types in all areas are juniper (Juniperus), pine (Pinus), and sagebrush (Artemisia). One reason for the apparent similarity in fuel use among the four sections is that the environment offered all people the same choices of wood for food preparation, heat, and light.

When fuelwood diversity is examined by site section, both the rim complex and the canyon bottom have a greater-than-expected richness of fuel remains (Figure 5), which in turn contributes to the higher overall richness of these two sections discussed above. The deeper midden deposits in the canyon bottom might have promoted the preservation of less-common woody taxa. The increased diversity could also be partly the result of more people inhabiting this part of the pueblo for a longer period of time. In the case of the rim complex, preservation is not believed to have contributed to the richness of fuelwood taxa, because preservation there was very poor. Instead, it is possible that human behavior—possibly associated with public functions—contributed to the observed diversity.

In Situ Secondary Refuse

Forty flotation samples from in situ secondary refuse were included in this analysis (Table 14). Because secondary refuse is discarded trash, the analysis of reproductive plant parts and wood charcoal remains in samples collected from these contexts provides indirect evidence of food and fuel use, respectively. Patterns discerned in the assemblage of reproductive plant parts in the samples from in situ secondary refuse are similar to those seen when all contexts are considered together. Unlike the occurrence of reproductive plant parts in all contexts, however, the occurrences of assorted reproductive parts in the in situ secondary refuse are all within the ranges predicted by the diversity analysis. A similar situation obtains for the wood charcoal assemblage, that is, the assemblage from the in situ secondary refuse presents patterns that are similar to those documented for all contexts combined. And, unlike the distribution of woody taxa when all contexts are considered together, the distribution of woody taxa in secondary refuse was within the expected range.

Primary Refuse in Kiva Hearths

To complete our intrasite comparison, we examined eight samples from primary refuse contexts, which provide direct evidence of food and fuelwood use. Seven of these samples were flotation samples from hearths; one was a macrofossil sample from a nonstructure surface. The samples were taken from the rim complex, canyon bottom, and upper west side; no good primary refuse contexts were excavated on the east talus slope (Table 17). Overall, the three areas were similar in terms of the taxa identified in primary refuse, and they all had the expected number of taxa for their sample sizes. There is, however, one interesting departure from the sample profile: the samples from primary refuse in the rim complex contained wood charcoal only. The absence of reproductive plant parts in the kiva hearth in the rim complex suggests that no foods were prepared in the last fires burned in this structure. This inference is consistent with the earlier interpretation that the rim complex might have been an area to which people brought foods prepared elsewhere in the village.

Temporal Analysis

We turn now to an examination of temporal patterning in the Woods Canyon Pueblo archaeobotanical assemblage. Specifically, we want to understand whether the inhabitants used plants in different ways over time. Temporal comparisons among the four sections of the site are difficult to make, because chronological data are limited; nonetheless, an attempt was made to identify early and late components on the basis of tree-ring, pottery, architectural, stratigraphic, abandonment, and structure-location data (see "Chronology").

It is thought that the canyon bottom was the first inhabited section of the site, with occupation there beginning in the mid–A.D. 1100s. This area is also believed to have been occupied longer than the other areas, although it may not have been in use during the final years of occupation. Occupation in the upper areas of the site—that is, the rim complex, the upper west side, and the east talus slope—started later, sometime in the 1200s, but there appears to have been some overlap between the occupation of these areas and that of the canyon bottom.

For the purposes of this study, we used two temporal groupings—early and late—but we defined them differently for different analyses. For the first set of temporal comparisons, we examined early and late sections of the site: the canyon bottom is "early," and the upper west side, rim complex, and east talus slope are "late" (refer to "Chronology" for a discussion of the assignment of site sections to early and late periods). Fifty-eight flotation samples were included in these comparisons (Table 18). In the second set of temporal groupings, we used a much smaller subset of specific dated contexts—all of them nonstructures—that had been assigned to the early and late phases of occupation on the basis of pottery types and stratigraphy. In addition, Structure 7-S was assigned to the late group because a tree-ring sample collected from it yielded a cutting date of A.D. 1257. In all, six flotation samples and 18 macrofossil samples from specific dated contexts were used in the second set of comparisons (Table 18). We chose to group the flotation and macrofossil samples together for this analysis in order to create a larger set of specifically dated contexts. Our interpretations are based on the ubiquity of the plant taxa for each time period.

Food and Fuel Use Over Time

A comparison of all charred plant parts between the early and late sections of the site reveals general similarities in assemblage composition and plant ubiquity (Table 19). The same pattern is seen in a comparison of the smaller set of specific contexts dated on the basis of pottery types and stratigraphy, plus one tree-ring cutting date (Tables 20 and 21). These results suggest that people gathered many of the same foods and fuels through time.

When all taxa for the early and late sections of the site are examined using the DIVERS program, however, the early section of the site appears to have a greater richness of taxa than would be expected for the number of samples examined (Figure 6). The early occupants of Woods Canyon Pueblo may have had access to a greater variety of plants, or they may have preferred a more diverse set of resources to satisfy their needs, than did later occupants. Because the canyon bottom, which was occupied early in the site's history, has deposits that are very well preserved, we cannot completely rule out preservation as a contributing factor. No DIVERS analysis was attempted on the smaller dataset composed of specific contexts, because the sample size was too small.

Not included in the foregoing discussion are insights pertaining to changes in specific taxa over time, although Popper (1988*1:61) warns that it can be risky to compare ubiquities of taxa. Among six food resources present through time, the only major change is that cheno-am seeds decrease notably in ubiquity in the later sections of the site (Table 19), suggesting less access to, or harvesting of, this wild food resource. An increase in the recovery of sagebrush (Artemisia) charcoal in the late sections of the site may indicate that more agricultural land was returning to fallow during this time. A drop in the ubiquity of pine (Pinus) charcoal relative to juniper (Juniperus) through time (Table 19) may partly relate to the ability of juniper to regenerate more quickly than pine, which would have led to a greater proportion of juniper on the landscape late in the Pueblo occupation of the area (Kohler 1992*2:263).

Proximity of Agricultural Fields

The archaeobotanical record for Woods Canyon Pueblo was examined for evidence that might allow us to determine the proximity of agricultural fields to the site. If the fields were close by, it would have been easier to transport whole ears of corn to the pueblo for processing, which would have resulted in the deposition of cobs, stalks, and cupules at the site. If the fields were distant, it is more likely that corn would have been shelled in the field to save transportation costs, since the volume of shelled kernels is half the volume of the original complete ears (Thornton 1984*1:267 [1845]). In that case, we would expect to find only corn kernels at the site. Corn cobs, stalks, and cupules were found in both flotation and macrofossil samples from Woods Canyon Pueblo, indicating that the agricultural fields were probably located close to the village.

It is possible that Nonstructure 1-N, in the canyon bottom, was an agricultural garden, although the evidence is ambiguous. Pieces of corn plants, such as cupules, kernels, and stalk segments, do occur in the flotation and macrofossil samples from the canyon bottom, but they also occur in samples from the other areas of the site (Table 22).

Resource Depletion and Food Stress

Resource depletion and food stress are related, though not identical, concepts. Resource depletion can lead to food stress, and vice versa. Resource depletion occurs when people use up plants that were formerly abundant, so that the plants become less common. Food stress occurs when there is not enough to eat, or when there are not enough nutritious foods to eat. It is thought that resource depletion and food stress might have occurred in the Mesa Verde region before the major emigrations from the area (Kohler and Matthews 1988*1:559; Stiger 1979*1:142).

If resource depletion occurred at Woods Canyon Pueblo, we would expect previously used taxa either to stop appearing or to be found in fewer contexts in the later period. Both juniper and pine, however, are still present in the late assemblages (Table 19). These woods seem to have provided most of the fuel and construction wood throughout the occupation of the village. Structures 10-S and 11-S, at the base of the cliff east of the main drainage, date to the post–A.D. 1280s and have juniper roof beams. Structure 7-S has a construction date of A.D. 1257, which places it later in the occupation of the pueblo. In this kiva, pieces of juniper and pine wood were found on Surface 2, suggesting that suitable juniper and pine timbers were still available during the later period of occupation.

There is a slight increase in the prevalence of sagebrush in later contexts; however, even then it does not seem to have been a major fuel source. The ethnographic literature suggests that sagebrush is a second-choice fuel because of its unpleasant odor when burned. It is common in fallow fields and would have become more available as land was cleared for agriculture. Although cottonwood and willow occur in samples from early contexts, there is less of these types in the later samples, which may mean that the people of the village had cleared out the riparian areas for fields. In summary, some resource depletion may have occurred at Woods Canyon Pueblo, but it does not seem to have placed severe limits on the inhabitants' choice of fuel.

There are a variety of coping mechanisms for dealing with food stress. People can eat more of the foods that they normally use to supplement poor harvests, or they can shift to foods they would normally ignore. They also can alter their methods of food preparation so that foods yield either more energy or more bulk (Adams and Bowyer 2002*1). Adams and Bowyer (2002*1) focus on the relative presence of higher-cost or less-desirable foods. "Higher-cost" foods are those that require considerable processing to be edible, or can be acquired only from great distances. Two examples of higher-cost foods are cactus fruits, with their numerous spines, and lemonade berry, with its sour fruit and large seeds.

"Less-desirable" foods are more difficult to categorize because they are deemed less desirable largely on the basis of personal preference. Higher-cost foods can also be less-desirable foods, because people may decide that eating the plant is not worth the costs involved. Less-desirable foods also might have a bad odor, be less nutritious, or have an unpleasant taste. An example of such a food is wolfberry (Lycium), the fruit of which, unless picked at just the right stage of ripeness, is very bitter.

If food stress occurred at Woods Canyon Pueblo, we would expect to find more higher-cost or less-desirable foods, as compared with weedy species, such as cheno-ams, groundcherry, and purslane. Weedy plants produce abundant seeds that are easy to gather; moreover, they grow in disturbed areas (such as agricultural fields) and might even be encouraged to grow in such locations. If environmental conditions were to fluctuate and the weather were to change, the agricultural crops and the associated encouraged weeds might fail, leaving only the hardier higher-cost or less-desirable resources to eat.

At Woods Canyon Pueblo, the most desirable weedy plants like cheno-ams, groundcherries, and purslane were still being used in large quantities in the later contexts, comparable to their use in early contexts. Hardier or higher-cost resources, such as juniper, hedgehog cactus, serviceberry, plum/rose seeds, and knotweed (Polygonum) achenes are scattered through both early and late contexts. These secondary resources seem to have been used occasionally to supplement the main wild resources (cheno-ams and groundcherries), but they do not indicate the major dietary shift we would expect if there had been food stress.


The range and diversity of archaeobotanical remains found at Woods Canyon Pueblo indicate that the site was occupied during at least a good portion of the calendar year (see Table 5). Ricegrass grains are one of the first foods available in the late spring; pigweed and goosefoot seeds are available in the summer; and it appears that the Woods Canyon inhabitants gathered groundcherry fruits and purslane seeds in the late summer and early fall. If the people of Woods Canyon Pueblo grew their corn in nearby fields, someone had to have been there to prepare and plant the gardens in the spring, weed the fields during the growing season, and then harvest the corn in the fall. From these lines of evidence, it appears that the site was occupied from at least spring into fall. Although there is no direct archaeobotanical evidence to support the inference that the pueblo was occupied during the winter months, the heavy investment in architectural facilities at the site constitutes indirect evidence for this interpretation.

It is more difficult to determine the season or seasons of emigration from the site. Since we cannot definitively say which structures were the last to be occupied at Woods Canyon Pueblo, the archaeobotanical assemblage is of little help in addressing this question.

The Past Environment

The natural environment as it existed during the occupation of the pueblo seems to have been similar to the modern natural environment. Many of the plants seen on the landscape today were available to the inhabitants of Woods Canyon. Both dry and riparian habitats were present, judging from the spectrum of archaeobotanical remains; however, differential use and preservation make it difficult to estimate the relative proportions of mesic to xeric environments in the past. The basic successional stages were all present in the past, from mature pinyon-and-juniper woodlands to old-field sagebrush and rabbitbrush shrubs, to the weedy cheno-ams, purslane, and groundcherry that grew in disturbed areas.


We analyzed 58 flotation samples and 73 macrofossil samples from Woods Canyon Pueblo for this study; most of the samples were collected from primary and secondary refuse. The results show that the people of Woods Canyon grew corn and gathered a variety of wild plants such as cheno-ams and groundcherries. Their preferred fuels appear to have been juniper, pine, and sagebrush. Pine and juniper were also used in construction.

Similar foods and fuels were used in each of the four sections of the site. Both the canyon bottom and rim complex, however, revealed some unexpected patterning. For example, a greater-than-expected number of plant taxa were documented in the samples from these two sections, a circumstance that may be attributed primarily to the greater number of wood-charcoal types present. A combination of better preservation and longer occupation might explain the pattern observed in the canyon-bottom assemblage but does not account for the same pattern in the rim complex, with its poorer preservation and shorter occupation. The recovery of few plant-food remains in rim samples, coupled with a complete lack of such remains in the hearth of the tested kiva, suggests that foods were rarely prepared in the rim complex. The greater-than-expected richness of wood-charcoal types, on the other hand, indicates that a variety of woods were used in this part of the site. The rim complex may have been a location where public activities involving the serving of food prepared elsewhere at the pueblo, as well as the burning of a greater-than-usual variety of wood, were conducted.

The inhabitants of Woods Canyon Pueblo used many of the same plants during both the early and late periods of occupation. A greater-than-expected number of total food and fuel taxa in the samples from the canyon bottom suggests that the early occupants had access to, or perhaps simply preferred, a greater variety of plants than did later occupants. We cannot, however, rule out differential preservation as a contributing factor to the observed pattern. Juniper appears to have been the primary fuelwood through time. By the later occupation, use of pine as fuel decreased, possibly because of the slow recovery of depleted pine stands, whereas the use of sagebrush for fuel increased, possibly because this shrub flourishes in fallow agricultural fields.

The people of Woods Canyon Pueblo appear to have farmed fields close to their village, as evidenced by the recovery of pieces of corn stalks and cobs in samples collected from the site. As judged by the relative proportions of corn and weedy plants to hardier, less-desirable, and higher-cost foods, significant food stress does not appear to have occurred at the pueblo. Seasonal availability of plant resources, coupled with inferences about agricultural scheduling needs, together suggest that the village was occupied at least from spring through fall; winter habitation was likely as well. Generally, the people of Woods Canyon Pueblo used many of the same plant resources that are available in the area surrounding the site today.


Much of the text of this chapter is derived from an early version of my senior honors thesis at the University of North Carolina at Chapel Hill, supervised by Margaret Scarry and Karen R. Adams. I am grateful to them and to my committee of readers, Vincas Steponaitis and Richard Yarnell. Vandy Bowyer, Donna Glowacki, Keith Kintigh, Scott Ortman, R. Lee Rainey, Virginia Rainey, Christopher Rodning, Dylan Schwindt, and Amber Van der Warker provided helpful comments and suggestions. The Crow Canyon Archaeological Center and Michael Kolb provided computing resources along the way. Finally, I would like to thank Melissa Churchill and Mark Varien, the principal investigators of the Woods Canyon project, for allowing me to work on this report and for their advice and encouragement during the course of the project.
Katharine D. Rainey

1In this chapter, we refer to several types of refuse, or trash: Primary refuse is trash that was left at the place where it was generated (Schiffer 1987*1:58). Secondary refuse is trash that was disposed of away from the place where it was created, usually into a midden (Schiffer 1987*1:58). Crow Canyon researchers distinguish between two types of secondary refuse. In situ secondary refuse is still in the place where it was originally discarded; naturally redeposited secondary refuse is trash that has moved from its original position as the result of erosional processes such as wind, water, and gravity.

References cited | To borrow, cite, or request permission

Copyright © 2002 by Crow Canyon Archaeological Center. All rights reserved.