Historical Ecology in the Mesa Verde Region: Results from the Village Ecodynamics Project

In a recent paper in American Antiquity, van derLeeuw and Redman (2002) call for archaeol-ogy to situate itself at the center of socio-natural

studies, by which they mean the study of long-terminteractions among components of ecosystems thatinclude human populations (see also Fisher and

Feinman 2005; Redman 1999). This is a timelyand important challenge for which archaeology iswell suited. It is also a line of research that archae-ologists in the American Southwest have beenaddressing, in one form or another, ever since A.E. Douglass discovered that the history of south-

HISTORICAL ECOLOGY IN THE MESA VERDE REGION:RESULTS FROM THE VILLAGE ECODYNAMICS PROJECT

Mark D. Varien, Scott G. Ortman, Timothy A. Kohler, Donna M. Glowacki, and C. David Johnson

Using the occupation histories of 3,176 habitation sites, new estimates of maize-agriculture productivity, and an analysisof over 1,700 construction timbers, we examine the historical ecology of Pueblo peoples during their seven-century occu-pation (A.D. 600–1300) of a densely settled portion of the Mesa Verde archaeological region. We identify two cycles of pop-ulation growth and decline, the earlier and smaller peaking in the late-A.D. 800s, the later and larger in the mid-A.D. 1200s.We also identify several episodes of immigration. Formation of aggregated settlements, which we term community centers,is positively correlated with increasing population and the time elapsed in each settlement cycle, and it persists during peri-ods of regional population decline, but it does not correlate with climatic variation averaged over periods. Architecturaland land-use practices depleted pinyon-juniper woodlands during the first cycle, but more stable field systems and greaterrecycling of construction timber resulted in more sustainable management of wood resources during the second cycle, despitemuch higher population densities. Our estimates for maize production are lower than previous estimates, especially for theA.D. 1200s, when population reached its peak in the study area. Even so, considerable potential agricultural productionremained unused in the decades that immediately preceded the complete depopulation of our study area.

En este trabajo examinamos la ecología histórica de las gentes Pueblo a través de siete siglos de ocupación (600-1300 d.C.)en una densa porción asentada en la región arqueológica de Mesa Verde. Usamos como datos las historias ocupacionales de3,176 sitios habitacionales, unos cálculos nuevos de la productividad agrícola del maíz, y un análisis de más de 1,700 mues-tras de madera de construcción. Identificamos dos ciclos de crecimiento y disminución poblacional, el más temprano y pequeñocon su apogeo hacia finales de los 800 d.C., y el más tardío y grande con su máximo auge hacia los 1200 d.C. También iden-tificamos varios episodios de inmigración. La formación de los asentamientos agregados, los cuales denominamos como cen-tros comunitarios, está correlacionada positivamente con el incremento de la población y el tiempo asociado con cada ciclodel asentamiento, y persiste durante los periodos de disminución de la población regional, pero no se correlaciona con lavariación climática promediada a través de los periodos. Durante el primer ciclo, las necesidades de materiales constructivosy del uso del suelo dieron lugar a la reducción de los recursos en los bosques de piñón y enebro. Sin embargo, y a pesar deque las densidades de población fueron más grandes, durante el segundo ciclo se manejaron los recursos de la madera demanera más sustentable a través de sistemas de cultivo más estables, y del reciclamiento de la madera de construcción. Nue-stros cálculos de la producción de maíz son más bajos que los previamente calculados, especialmente para los 1200s d.C.,cuando la población alcanzó su máximo desarrollo en el área de estudio. Aún así, una considerable proporción del potencialde la producción agrícola no se empleó en las décadas que precedieron inmediatamente el despoblamiento total del área deestudio.

Mark D. Varien and Scott G. Ortman ■ Crow Canyon Archaeological Center, 23390 County Road K, Cortez, CO 81321.970-565-8975 ([email protected], [email protected])Timothy A. Kohler ■ Department of Anthropology, Washington State University, Pullman, WA 99164-4910 & Santa FeInstitute. ([email protected])Donna M. Glowacki ■ School of Human Evolution and Social Change, Arizona State University, Tempe, AZ 85287-2402([email protected])C. David Johnson ■ Department of Anthropology, Washington State University, Pullman, WA 99164-4910. ([email protected])

American Antiquity, 72(2), 2007, pp. 273–299Copyright ©2007 by the Society for American Archaeology

273

AQ72(2) varien 4/6/07 9:16 AM Page 273

274 AMERICAN ANTIQUITY [Vol. 72, No. 2, 2007

western climate could be read from tree-rings (Dou-glass 1929).

Research on human ecology in the Southwestis extensive and often implicit, and we could nothope to accomplish our goals for this paper if wewere to review it properly here. However, based onour knowledge of the literature that focuses explic-itly on human-environment relations we find it pro-ductive to characterize recent studies as followingone of three courses of a braided stream. The firstand most venerable focuses on the effects of long-and short-term climatic fluctuations for agricul-tural success and population movement (Berry1982; Dean 1985, 1988; Dean et al. 1985; Euler etal. 1979; Jett 1964; Lipe 1970; Petersen 1988). Asecond, more recent course models agriculturalproductivity on specific landscapes by integratingclimate information with geomorphology, soils,and likely agricultural strategies (Hill 1998; Huck-leberry and Billman 1998; Kohler et al. 2000; Stoneand Downum 1999; Tuggle et al. 1984; Van West1994). The third and perhaps least-developedcourse considers anthropogenic changes in pastsouthwestern environments and the impact of thesechanges for humans (Diamond 2005:Chpt. 4;Kohler 1992, 2004; Kohler and Matthews 1988).

In this paper we weave these branches togetherto consider impacts of climatic variation, agricul-tural productivity, and anthropogenic change simul-taneously using data generated by the VillageEcodynamics Project (Kohler et al. 2007). We pre-sent results of several studies. First is an assessmentof the occupation histories of the 3,176 habitationsites in our project study area. Second is a newreconstruction of potential maize productivity forthe project area. This in turn makes possible ananalysis of settlement dynamics using the VillageProject archaeological site database incorporatingcomparisons with the productivity data. Finally,we analyze long-term use of one particularresource, wood. Together these studies make it pos-sible for us to view the historical ecology of ances-tral Pueblo1 (Anasazi) peoples from severaldifferent perspectives.

In the previous paper, Ortman and others (thisissue) describe how we reconstructed the occupa-tional history of all recorded habitation sites in theVillage Project study area. In this article, we buildon these results to describe and interpret the set-tlement history of this area, which included two

cycles of population growth and decline, and twonearly coincident cycles of aggregation into vil-lages.

We also summarize the methods used to gener-ate new, annual maize-agriculture paleoproductiv-ity estimates that extend and refine the well-knownestimates produced by Van West (1994). These newestimates are the first to incorporate annual varia-tion in temperature in addition to precipitation.They provide a detailed retrodiction of localclimate-driven variation in agricultural potential, animportant factor that Pueblo farmers had to con-sider when selecting areas to settle and farm.

Finally, we analyze a database of wood samplesfrom sites in the project study area for which yearof harvest could be determined through tree-ringanalysis. This dataset includes the year in whicheach dated timber was harvested and the species oftree. Nearly all of these come from structural con-texts, and we assume they represent timbers har-vested for construction. We also examine thenumber of rings in the specimens when both thepith and the outside ring is present, which gives theage of the section of the tree that was sampled andanalyzed by analysts at the Laboratory of Tree-Ring Research.2 We use these data to address howthe Pueblo populations affected study-area wood-lands through centuries of harvesting timbers.

We begin this new synthesis of settlement his-tory in the central Mesa Verde region by describ-ing these three datasets and the methods used toanalyze them. Then we present our results and con-sider their implications for the historical ecologyof Pueblo peoples in the northern Southwest.

Site Data

The site database for our study contains informa-tion on 8,948 sites. Most were recorded during themany archaeological surveys conducted within theproject study area in the context of problem-oriented research or cultural resource managementwork. These surveys cover approximately 15 per-cent of the study area and include large, block sur-veys (which were used in many of the settlementanalyses that follow), transect surveys, and a vari-ety of smaller surveys (Figure 1). Although resur-vey would undoubtedly discover additional siteswithin certain of these blocks, we will assume herethat all habitation sites within the surveyed areas

AQ72(2) varien 4/6/07 9:16 AM Page 274

have been recorded and that only a small portionof the total has been recorded outside these sur-veyed areas.

In Table 1 we present the general time periodsand primary functions inferred for all sites in theVillage Project database. The project study areahas been used by humans from Paleoindian to his-toric times, but 97 percent of all sites for whichperiod of occupation could be inferred date fromthe occupation of the study area by Pueblo peoples,between about A.D. 600 and 1280. Most sites forwhich a primary function has been recorded arehabitation sites. We used the presence of a trashmidden and one or more pit structure depressionsas evidence that a given settlement was a year-round residence for at least one household; theanalysis of features and assemblages at excavatedsites with these characteristics supports this inter-pretation (Varien 1999a; Varien, ed. 1999). A totalof 3,176 sites in the database possess these featuresand are accordingly interpreted as single habita-

tions, multiple habitations, or community centers.The occupational histories of these settlements,reconstructed using the methods outlined by Ort-man and others (this issue), suggest that some ofthese habitations were occupied for more than oneof the 14 modeling periods into which we have sub-divided our sequence.

Based on recent research in our study area (Caterand Chenault 1988; Lightfoot 1994; Lipe 1989;Ortman 1998; Varien, ed. 1999) we infer that eachpit structure was the central building used by a sin-gle household consisting of individuals related bylineage or marriage, two to three generations deep.About two-thirds of the habitation sites in the data-base contain only one pit structure, and thus rep-resent farmsteads occupied by a single household.Approximately 850 sites contain evidence ofbetween 2–8 pit structures and are termed multiplehabitations. Finally, the largest sites are called com-munity centers. For the purpose of these analyses,community centers are defined as settlements with

Varien et al.] HISTORICAL ECOLOGY IN THE MESA VERDE REGION 275

Figure 1. Digital elevation model showing the study area boundary and the location of archaeological surveys.

AQ72(2) varien 4/6/07 9:16 AM Page 275

nine or more pit structures, 50 or more total struc-tures, or sites with public architecture.

We interpret these large sites as community cen-ters because their inhabitants could not have allbeen lineally related, they often contain publicarchitecture such as a great kiva or plaza, and theyare typically the largest site within a cluster of sites(Adler and Varien 1994; Lipe and Varien 1999:345;Ortman and Varien 2007; Varien et al. 1996). Thesecenters generally have longer histories than mostof the smaller habitations in the region (Varien1999:202–207; Varien and Ortman 2005). Finally,excavations at these centers indicate that they werelocations of social, economic, and political activi-ties not occurring at smaller habitations (Adler1994; Bradley 1988, 1993, 1996; Driver 1996; Lipe2002; Muir 1999; Muir and Driver 2002; Ortmanand Bradley 2002; Potter 1997, 2000; Potter andOrtman 2004).

As a result of 15 years’effort to document largesettlements in the greater Mesa Verde region (seeVarien et al. 1996; Varien 1999a), we have identi-fied 106 community centers in the Village Projectstudy area. As we compiled the database we eval-uated existing records for these centers and con-ducted new fieldwork at 59 poorly documentedcenters. A team led by Glowacki mapped thesesites and obtained size data (site area, number of

pit structures, size of the roomblock area, and theestimated number of rooms), compiled an inven-tory of cultural features, and analyzed the surfacepottery to collect new chronological information.As a result, we believe that we have a relativelycomplete dataset of all community center sites inthe Village Project study area, regardless of whetheror not the center occurs within a survey block asdefined in Figure 1. This assumption will play animportant role in our efforts to model populationdynamics below.

Paleoproductivity Estimates

Macrobotanical and stable isotope data suggest thatin the northern Southwest maize was the single-most important subsistence resource by at leastA.D. 600, and perhaps even earlier in the Basket-maker sequence (Kantner 2004:60–67; Matson1991:90–101; Matson and Chisholm 1991). VanWest’s (1994) well-known estimates for maize pro-duction in our project study area provided the start-ing point for modeling paleoproductivity in theVillage Project. Following Van West, we use thePalmer Drought Severity Index (PDSI) as a basisfor our maize productivity calculations. The PDSIis a relative measure of soil moisture developed byPalmer (1965); Burns (1983) was the first to note


Table 1. Primary Temporal and Functional Classification of Sites in the Village Project Database.

Archaic Ancestral through Pueblo

Paleo- Basket- (A.D. Numic / Not Primary Function indian maker II 600-1280) Navajo Historic Unknown recorded Total

Not recorded 4 48 740 9 27 1 1642 2471Indeterminate 4 174 4 24 17 223Isolated Find 3 9 1 18 9 40Artifact scatter 1 56 546 3 1 154 66 827Ceremonial site 15 2 2 19Clay quarry 2 1 3Field house 1 618 1 20 640Kiln 259 6 3 268Limited activity 1 10 565 5 8 73 200 862Reservoir 14 2 16Rock art 15 2 6 13 36Stone quarry 1 12 10 36 59Storage 1 107 8 116Water control 45 4 11 9 69Single habitation 8 1991 5 75 2079Multiple habitation 3 852 1 16 872Indeterminate habitation 4 227 11 242Community center 106 106Total 6 139 6297 19 52 313 2122 8948

AQ72(2) varien 4/6/07 9:16 AM Page 276

its utility for archaeology. Our studies show thatPDSI values, calculated from historic weatherrecords, are reasonably well correlated with recordsof maize-and-bean production in our study areafrom 1931–1960. Moreover, we can estimate thePDSI values from tree-ring data relatively accu-rately and thus project them into the prehispanicperiod.

Our study modifies Van West’s approach by (1)including portions of the study area for which soilsmaps did not exist when she did her work, (2) giv-ing temperature variability a more explicit role indeveloping production estimates, (3) modifyingthese estimates to represent harvest rates more sim-ilar to those of maize grown prehispanically in ourarea, (4) reducing estimates of production based onsoils that are evaluated as unsuitable for hand plant-ing (Ramsey 2003), and (5) extending the recon-structions back to A.D. 600.

This work was accomplished in a series of steps;here we give a general overview with additionaldetails deferred to other publications. The first stepwas to use the available tree-ring data to developproxies for temperature and identify local cold orshort summers. We eventually settled on two high-elevation bristlecone pine sequences, one fromAlmagre Mountain about 350 km east-northeast of

the project area (Graybill 1984), and one from SanFrancisco Peaks about 335 km southwest of the pro-ject area (Figure 2). We used the scores on the firstprincipal component extracted from both sequences(Table 2) as an independent variable in a regres-sion formula retrodicting productivity, as explainedbelow. These scores are positively related to aver-age temperatures for local weather stations duringthe summer months, with the correlations strongestand most significant for late summer (e.g., in Sep-tember, when the r at Mesa Verde is .49 [p < .0001],at Yellowjacket, .53 [p = .01], and .31 in Cortez [p= .02]). Although these correlations are not terri-bly strong, their only purpose is to assure us thatthese sequences are sensitive to local late summertemperatures. We also produced a rather similarreconstruction in which the Almagre sequence byitself provided the temperature proxy, but that willnot be discussed further here.

In the second step we assessed differences insoils in the study area and defined 14 groups of soilsthat are similar with respect to their productivity.Third, using instrumented data from four weatherstations, each representing a different elevationalband, we produced PDSI reconstructions for these14 groups of soils for the years between 1931 and1960. This produced 56 (4 x 14) PDSI sequences


Figure 2. Location of the study area in Southwest Colorado, relative to two tree-ring sequences used to construct a tem-perature proxy.

AQ72(2) varien 4/6/07 9:16 AM Page 277

for the study area. Fourth, we determined that eachof these 56 PDSI sequences had a significant rela-tionship with the Mesa Verde Douglas-fir indexedseries, with values of r2 ranging from .32 – .67(mean = .54, s = .11; all with p > F < .001 and allbut three with p > F < .0001). Fifth, using theseregression relationships we retrodicted PDSI val-ues to A.D. 600, using the Mesa Verde series as theindependent variable for each of the 56 combina-tions of soil/elevation.

In the sixth step, we produced a weighted aver-age of the PDSI values for those soils producingmaize and beans between 1930 and 1960 in Mon-tezuma County, which we call the “bean soils,”below. The seventh step regressed those valuesagainst the historic maize and bean production inMontezuma County, and against the two tempera-ture proxies, introducing a third independent vari-able, year, which would allow us to assess and holdconstant the “technology trend” (Burns 1983:70)that this represented. This produced the followingresults for the first principal component (PC1) ofthe Almagre and San Francisco series:

We standardized all independent variables prior toregression, so the partial slopes associated witheach may be considered beta weights. Therefore,the single-most important variable affecting maizeyields during the calibration period was “year,” avariable standing in for the technology trend

(increasing use of fertilizer, mechanized equip-ment, more productive varieties, and so on). Fol-lowing that, the PDSI, affected mostly byprecipitation, and the PC1 score, affected mostlyby temperature, are approximately equal in impor-tance.

Because these maize production estimates applyto the average productivity of the soils farmed his-torically (the bean soils), the eighth step was toadjust the productivity for each soil class in thestudy area relative to the bean soils. We did this bycalculating the ratio of the average productivity forsoils in that class to the average productivity of thebean soils, which was gleaned from the normal-yeartotal dry weight production published in the Cortez-area soil survey (Ramsey 2003:Table 7).

The ninth step takes into account the fact thatthe farming practices of Pueblo peoples in the studyarea likely produced lower yields than thoseobtained using historic seed varieties and plantingpractices. We used historical data from Zuni andHopi and ethnoagricultural experiments byMuenchrath et al. (2002),Adams et al. (1999), andother sources, to adjust the yields. Using thesesources, we determined that 500 kg/ha was a rea-sonable estimate for the mean yield in our best soils(i.e., the bean soils). We then renormed our pro-duction figures so that the mean production in thebean-field soils is 500 kg/ha by multiplying theyields from step 8 by a factor of .68.

The tenth step further reduced maize productionon soils reported as unsuited for hand planting. Thedetailed soil descriptions in the soil surveys pro-vide “major management factors” that indicate the


Table 2. Principal Components Analysis Used to Generate Temperature Proxy for Paleoproductivity Reconstruction.

Ring-Width Approx. Years Index Series Elevation Used (Abbreviation) Reference Species (m) (A.D.)

San Francisco Peaks (SFP) Salzer (2000) and bristlecone pine 3535 560–1983personal communication

Almagre Mountain B (ALM519) Graybill (1984) bristlecone pine 3536 560–1983

Eigenvalues of the Correlation Matrixn Eigenvalue Proportion Cumulative

1 1.233 0.617 0.6172 0.767 0.383 1.000

Eigenvectors Prin1 Prin2ALM519 0.707 0.707SFP 0.707 -.707

AQ72(2) varien 4/6/07 9:16 AM Page 278

suitability of each soil complex for a variety ofuses. In addition, and somewhat independently ofthese general land suitability classes, for each soilthe surveys report hand-planting suitability restric-tion codes ranging from 0 to 1.0. (A value of 0means no restrictions; 1 indicates the soil to becompletely unsuitable, even for hand planting.) Forthose soils in our study area reported as “unsuit-able” in the “cropland suitability” field, we multi-ply the yields from step 9 by the inverse of thehand-planting restriction value. This furtherreduces the yields for soil complexes that in ourstudy area for the most part represent off-mesa,steep, stony soils. In a few places there are also bogsor highly alkaline soils that have their potentialyields reduced through this correction. Altogetherthese soils represent 53.2 percent (by area) of the1817-km2 study area.

The eleventh step applied a “cold correction”factor. This factor disallowed any maize produc-tion above 7,900 feet (2,400 m) and progressivelyreduced production in colder-than-average years in

the elevational band between 7,054 and 7,900 feet.3

Discounting production in this elevational bandwas accomplished using a formula that takes intoaccount both the elevation and how cold it was,using the tree-ring-based temperature proxy.

Figure 3 displays the average potential yields peryear in our study area that result from this process.These estimates are not affected by any fallow fac-tor, possible land degradation, or improvements inagricultural productivity over time due to innova-tions in farming practices or the evolution of moreproductive maize varieties through human selec-tion. These estimates are spatial and provide ayearly estimate for each 4-ha plot of land in thestudy area, but for Figure 3 we have averaged thosevalues across the study area to provide a measureof average annual agricultural potential. Of course,we can also estimate the potential productivity ofland adjacent to known sites that was likely farmedwhen those sites were occupied.

We emphasize that what we have produced isonly a model for paleoproductivity. Our claim, in


Figure 3. Average study-area potential maize yields in kg/ha per year A.D., unsmoothed (light gray), and spline smoothed(black); mean annual yield of 254 kg/ha shown for comparison.

AQ72(2) varien 4/6/07 9:16 AM Page 279

constructing the model this way, is that soil mois-ture, length of growing season, elevation, and localsoil characteristics are the most important factorsaffecting production. Future refinements couldattempt to take into account other variables of prob-able local importance, including patterns of cold-air drainage and aspect, but we have been unableto locate data that would allow us to calibrate theimportance of these factors. The final strength ofthe fit we obtain with the historical data on cropyields (adjusted R2 = .58) compares favorably withthat obtained by Van West in her widely cited andaccepted reconstruction (unadjusted R2 = .49; VanWest 1994:101–102) and is incomparably betterthan the traditional archaeological technique ofestimating maize production by inspecting a graphof tree-ring departures through time (which implic-itly, and surely incorrectly, assumes a perfect cor-relation). The essence of our approach—and alsothat of Van West—is to try to understand whateffects these tree-ring departures have on maizeproduction on a specific landscape. Therefore, ourreconstruction does not necessarily apply to neigh-boring regions.

We reconstruct noteworthy periods of lowpotential production in the late 600s, the middle700s, the late 800s and early 900s, around 1000,around 1100, from about 1130–1150, in the early1200s, and in the late 1200s. Our reconstruction dif-fers from Van West’s most dramatically in the late1160s and early 1170s, when we estimate relativelyhigher potential production, and in the early 1200s,a cold period when we estimate lower potentialproduction. Van West’s estimates are relativelyhigher throughout than these new estimates.

Tree-Ring Database

The tree-ring database for the central Mesa Verderegion was compiled by archaeologists at the CrowCanyon Archaeological Center from records at theLaboratory of Tree-Ring Research at the Univer-sity of Arizona. The database includes 1,784 cut-ting dates from 91 sites in the study area.4 The vastmajority of these samples with cutting dates werecollected from burned deposits near the floors ofcollapsed structures, and therefore we assume theyrepresent timbers harvested for use in construction.Because large timbers are difficult to cut and shapewhen dry, we also infer the date recorded for most

cutting dates not only represents the year that treeor limb died, but also the year in which the tree orlimb was harvested for eventual use in construc-tion. The species of tree recorded for each date alsoprovides information on species selection andavailability. We therefore believe our database pro-vides a record of a specific cultural practicethrough time.

Sampling issues, however, prevent us fromviewing this sample as representative of the rate ofconstruction-timber harvesting through time. Theavailable sample is biased in favor of periods thatarchaeologists have studied most intensively andcontexts where preservation is most likely. Thisincludes periods when open sites were burned,when cliff dwellings were built, and when there wasa large amount of construction generally. The sam-ple is biased against periods in which large num-bers of recycled timbers were used (Bradley 1993;Varien 1999a; Varien and Kuckelman 1999). Forthese reasons we do not take the numbers of cut-ting dates through time as a proxy for population,and consider them to offer only a general indica-tion of construction activity. Because we believecutting dates indicate the year when timbers wereharvested for construction, we do believe our sam-ple conclusively identifies years when construc-tion occurred and characterizes patterns in thespecies of timbers harvested. The age of the par-ticular sample can also be determined when bothpith and cutting dates are present. These data areinterpreted cautiously to examine changing pat-terns in wood procurement.

Momentary and Total Population Estimates

In this section we reconstruct the overall popula-tion history of the study area. We begin by calcu-lating the momentary population, in households, ofsites in our database for each of our 14 modelingperiods. Then we use average momentary house-hold estimates for sites in the database to estimatethe total momentary population for the entire studyarea.

Ortman and others (this issue) describe the prob-ability density analysis5 we used to reconstruct theoccupational histories of habitation sites and thusestimate the total number of households that occu-pied these sites at some point during each period.Here, we use these estimates of total households to


AQ72(2) varien 4/6/07 9:16 AM Page 280

calculate the average momentary population ofhouseholds in our sample. Momentary householdestimates take into consideration the use-lives ofhouses in small sites and community centers vis-à-vis the lengths of our modeling periods. Duringthe final few centuries of our sequence the averageuse-life of a house in both small sites and commu-nity centers was equal to or greater than the lengthof our modeling periods, so for these periods theuse-lives of houses in small sites and centers areeffectively subsumed by the probability densityanalysis (Ortman et al. 2007) and have no effect onmomentary household calculations. For earlierperiods, however, the average use-life of a housewas shorter than the length of the modeling periodwithin which our analysis suggests a house wasoccupied, so for these periods we need to take intoaccount the variable use-lives of houses in smallsites and centers, and the length of the corre-sponding modeling period.

House use-life estimates for early small siteswere calculated by measuring the accumulation ofcooking pottery, following the methods developedby Varien and others (Varien 1999a; Varien andMills 1997; Varien and Ortman 2005). Excavationsand the analysis of pottery assemblages indicatethat houses in community centers were used forlonger periods than houses in small sites (Kohlerand Blinman 1987; Ortman et al. 2000), so wedeveloped a new method for estimating the use-lifeof houses in early community centers. This methoduses the probability density analysis and point esti-mates for the total accumulation of cooking pot-tery at two early community centers excavatedusing stratified random samples: Grass Mesa(Kohler 1988) and Rio Vista villages (Wilshusen1986). To calculate the average use-life of housesin these two centers we divided the total householdsinferred for all periods from the probability den-sity analysis by the total household-years of occu-pation, based on point estimates of the total cookingpottery accumulation and Varien’s (1999a:107)accumulation rate per household year.

With these use-life estimates in hand, we obtainmomentary household estimates by dividing themean house use-life for a given site type and periodby the length of that period, and multiply the resultby the corresponding total household estimate forsites of that type and period in the site database.These calculations are presented in Table 3. Again,

note that total and momentary household estimatesdiffer for the years between A.D. 600 and 1100because the length of our modeling periods in thisinterval exceed our estimates of house use-life.From A.D. 1100 to 1280, however, house use-lifewas equal to or greater than the length of the peri-ods, so the total and momentary population esti-mates are the same.

The average momentary household estimatesfor habitations in the database form the basis forestimating the total momentary population for theentire study area. Estimating population fromarchaeological evidence is notoriously difficult(Powell 1988), and a variety of different methodscan be used, each of which produces differentresults. We estimated total population using threemethods, the results of which are compared inFigure 4. Method 3 produces middle-range esti-mates that we believe are most reasonable, so wepresent the details of these calculations in Table4.6 Under our preferred method we make twoassumptions: (1) that the small-site momentaryhousehold density for the block-surveyed area isrepresentative of the entire study area, and (2) wehave a 100 percent sample of community centers.We first determine the proportion of the study areacovered by block survey, and the momentaryhousehold density of small sites in this surveyedarea. Then, we multiply the small-site momentaryhousehold density in the surveyed areas by theinverse of the sampling fraction to get totalmomentary household estimates for small sites.We then add the total momentary households incenters and the estimates for small sites to esti-mate the total momentary households in the studyarea. Then, we convert total momentary house-holds into estimates of the momentary total per-sons who resided in the study area during eachperiod by multiplying the total momentary house-hold figure for each period by six, based on Light-foot’s (1994) ethnographic review of householdsizes in historic Pueblo groups. Finally, we usethese total population figures to calculate the pop-ulation density of humans per square kilometer ofpotentially arable land in the study area. Theseestimates indicate that population peaked duringthe A.D. 1225–1260 period, when approximately19,500 persons, nearly 11 persons per square kilo-meter of land below 2,400 m elevation, lived inthe Village Project study area.


AQ72(2) varien 4/6/07 9:16 AM Page 281


Tabl

e 3.

Cal

cula

tion

of M

omen

tary

Pop

ulat

ions

of

Pueb

lo H

abita

tion

Site

s in

the

Vill

age

Proj

ect S

ite D

atab

ase.

Per

iod

Occ

upat

ion

Spa

n (y

ears

)S

pan

/ P

erio

d L

engt

hTo

tal

Hou

seho

lds

Mom

enta

ry H

ouse

hold

s

Smal

l C

omm

unity

Sm

all

Com

mun

ity

Smal

l C

omm

unity

Sm

all

Com

mun

ity

Beg

in (

A.D

.)E

nd (

A.D

.)L

engt

h (y

ears

)si

tesa

cent

ersb

site

sce

nter

ssi

tes

cent

ers

site

sce

nter

sTo

tal

600

725

125

828

0.06

40.

224

1567

1010

0.3

2.2

103

725

800

7513

280.

173

0.37

354

510

594

.539

.213

480

084

040

1828

0.45

00.

700

436

9719

6.2

67.9

264

840

880

4018

280.

450

0.70

048

314

921

7.4

104.

332

288

092

040

1828

0.45

00.

700

158

154

71.1

107.

817

992

098

060

1828

0.30

00.

467

240

6172

.028

.510

098

010

2040

1828

0.45

00.

700

362

3816

2.9

26.6

190

1020

1060

4021

400.

525

1.00

032

635

171.

235

.020

610

6011

0040

2140

0.52

51.

000

721

269

378.

526

9.0

648

1100

1140

4040

401.

000

1.00

059

130

959

1.0

309.

090

011

4011

8040

4040

1.00

01.

000

579

394

579.

039

4.0

973

1180

1225

4545

451.

000

1.00

048

759

848

7.0

598.

010

8512

2512

6035

3535

1.00

01.

000

763

967

763.

096

7.0

1730

1260

1280

2020

201.

000

1.00

034

385

034

3.0

850.

011

93a S

mal

l-si

te o

ccup

atio

n sp

ans

afte

r V

arie

n an

d O

rtm

an (

2005

:138

–140

).b C

omm

unity

-cen

ter

occu

patio

n sp

ans

for

mod

elin

g pe

riod

s pr

ior

to A

.D. 1

020

base

d on

the

ratio

of

tota

l hou

seho

ld y

ears

fro

m th

e B

ayes

ian

anal

ysis

and

the

accu

mul

atio

n of

coo

king

pot

-te

ry a

t Gra

ss M

esa

(Koh

ler

1988

) an

d R

io V

ista

(W

ilshu

sen

1986

) vi

llage

s.

AQ72(2) varien 4/6/07 9:16 AM Page 282

Settlement History

In this section, we use the total momentary popu-lation estimates to examine population dynamicsand settlement ecology in the project study area.The results of these analyses are presented in Fig-ure 5 as a series of line charts.

Population Dynamics

Figure 5A presents total momentary populationestimates for the study area. This population recon-struction shows that occupation of the regionoccurred in two general cycles. The early cyclespans A.D. 600 to 920 and the late cycle 920 to1280. Population density became much higher dur-ing the late cycle than at any point in the earliercycle, and peaked in the mid-A.D. 1200s.

Figure 5B translates the total momentary pop-ulation estimates into population growth rates foreach period. The dashed lines on this chart refer-ence positive or negative annual growth rates of .7percent (an intrinsic rate of natural increase of .007).Based on Cowgill’s (1975) work, we interpretpoints above and below the dashed lines as periodswhen growth or decline may have included eitherimmigration or emigration.7 Using this guideline,settlement in the region included four probableperiods of immigration and two of emigration (Fig-ure 5B). Since habitation sites prior to A.D. 600 are

nearly absent in the study area (Table 1), there wasactually a fifth episode of immigration not shownon Figure 5B, when the study area was first colo-nized by Pueblo farmers around A.D. 600.

This reconstruction presents a dynamic pictureof regional population history that contrastsmarkedly with the view of gradual populationgrowth and culture change over the seven centurieswhen Pueblo groups occupied the area (e.g., Rohn1989). The probability that people moved into andout of the study area so often makes it likely thatinhabitants of the region came from different areasand had distinct histories (Glowacki 2006;Wilshusen and Ortman 1999). As discussed below,many important events in the Pueblo Indian historyof the Mesa Verde region occurred in the contextof these dramatic swings in population.

Population Aggregation

Figure 5C examines settlement aggregation andthe development of community centers. It uses datafrom block surveys within the study area, wherewe have a 100 percent sample of both centers andsmaller habitation sites, to calculate the portion ofhouseholds living in larger centers during eachperiod. This chart shows that the formation of cen-ters also occurred in two cycles, with peak aggre-gation at the end of each cycle, during the A.D.880–920 and 1260–1280 periods. Perhaps the most


0

500

1000

1500

2000

2500

3000

3500

4000

4500

600 700 800 900 1000 1100 1200 1300

Initial year of modeling period (A.D.)

Po

pu

lati

on

(h

ou

seh

old

s)

Method 1: % in centers Method 2: elevation Method 3: small site density + centers

Figure 4. Total population estimates for the Village Project study area. Method 1 and Method 2 are discussed in note 6,Method 3 is discussed in the text and presented in Table 4.

AQ72(2) varien 4/6/07 9:16 AM Page 283

interesting and unexpected outcome of this analy-sis is that aggregation peaked during periods ofpopulation decline at the end of each cycle, and notduring the period of peak population in each cycle.We examine this pattern in more depth in our sum-mary discussion.

Figure 5D examines the formation of centers,summarizing the number of new centers foundedduring each period, and the total number of cen-ters occupied in each period. From these data wecan infer that centers tended to remain occupied forseveral periods after their initial founding. The per-sistence of centers over time in our reconstructioncontrasts with views suggesting that villages in thenorthern Southwest were ephemeral and unstable(Adams 1989, 1991; Lekson and Cameron 1995;Schlanger and Wilshusen 1993).

Community Centers and Agricultural Potential

Figure 5E integrates our new potential paleopro-ductivity estimates into our settlement patternreconstruction. The solid line shows the meanpotential productivity of the two-kilometer catch-ments around community centers founded in eachperiod, during that period. We chose a two-kilometer radius as the size of a community centercatchment based on both ethnographic research,which suggests this radius approximates the areawithin which community members would have

interacted most regularly and farmed most inten-sively (Varien 1999a:153–155; Varien et al.2000:51–52), and also archaeological research,which demonstrates that clusters of residential set-tlements, centered on villages and/or public archi-tecture, tend to have a radius of two kilometers inour study area (Adler and Varien 1994; Ortman andVarien 2007). The thin line in Figure 5E shows themean paleoproductivity of the study area overallduring each period.8 This chart thus incorporatesboth spatial and temporal variability in agriculturalpotential in a catchment analysis.

For most of the early demographic cycle theagricultural potential of community-center catch-ments was well above the mean. This is not sur-prising, as one would expect the earliest villages ina sparsely populated landscape to emerge in highlyproductive catchments. There is an interlude of lowpopulation density at the beginning of the secondsettlement cycle, between A.D. 920 and 1060. Dur-ing this interlude the agricultural potential of catch-ments surrounding new centers was about averagefor the study area overall. Why these new centerswere not founded in the high-potential catchmentsof the early-cycle centers is puzzling, and invitesspeculation that the catchments on which early-cycle centers were founded had been so degradedby the end of the cycle that they were unattractiveplaces to live and farm for a period of time (Kohler


Table 4. Best Estimate for Total Population in the Village Project Study Area.

Period Database momentary households Study area momentary households

Small sites All Begin in block community Small Total Population(A.D.) End (A.D.) surveys centers sitesa Total personsb Densityc

600 725 40.4 2.2 302.2 304.4 1826 1.03725 800 38.3 39.2 286.6 325.8 1955 1.10800 840 102.6 67.9 767.7 835.6 5013 2.82840 880 123.8 104.3 925.9 1030.2 6181 3.48880 920 35.1 107.8 262.6 370.4 2223 1.25920 980 34.8 28.4 260.4 288.8 1733 0.98980 1020 83.7 26.6 626.2 652.8 3917 2.211020 1060 85.1 35.0 636.3 671.3 4028 2.271060 1100 149.1 269.0 1115.6 1384.6 8307 4.681100 1140 218.0 309.0 1631.1 1940.1 11641 6.551140 1180 225.0 394.0 1683.5 2077.5 12465 7.021180 1225 231.0 598.0 1728.4 2326.4 13958 7.861225 1260 303.0 967.0 2267.1 3234.1 19404 10.931260 1280 123.0 850.0 920.3 1770.3 10622 5.98aBlock survey small sites / proportion of study area surveyed.bTotal households x 6, after Lightfoot (1984).cTotal persons / 1776 km2 of land below 2400 m (7900 ft.) in study area.

AQ72(2) varien 4/6/07 9:16 AM Page 284


0.0

3.0

6.0

9.0

F.

0

5000

10000

15000

20000

A.

-40

-20

0

20

40

B.

0

0.2

0.4

0.6

0.8

C.

0

20

40

60

D.

New centers Occupied centers

0

50

100

150

200

600 700 800 900 1000 1100 1200 1300

G.

175

275

375

E.

New center catchments Study area

Figure 5. Population and settlement dynamics in the Village Project study area. Notes: Points in all graphs represent theinitial date of a modeling period. A) Total population (individuals); B) Rate of population change (per 1000 per year,dashed lines mark limits of in situ growth/decline); C) Proportion of block-survey households in centers; D) Number ofnewly-constructed and occupied centers; E) Mean productivity (kg maize/ha); F) Tree-ring cutting dates/years in period;G) Mean and standard deviation of construction timber age.

AQ72(2) varien 4/6/07 9:16 AM Page 285

and Matthews 1988). This possibility reminds usthat our paleoproductivity reconstruction providesestimates of agricultural potential that are unaf-fected by historically contingent human impacts,either positive or negative. This is clearly an areawhere subsequent phases of this research couldfocus.

There was a burst of new center constructionassociated with immigration during the late A.D.1000s, the early years of the late cycle. These cen-ters were once again located on better-than-averagecatchments at the time they were founded. But aspopulation increased over the course of the latecycle, the productivity of new community centercatchments fell below the study area mean. Thiscannot be due to abandonment of centers foundedearlier in this cycle, for example due to local soildegradation, because many centers founded on rel-atively good catchments early in the late cycle werestill occupied even as new centers were built inless-optimal catchments over time.

However, other factors may explain the declin-ing potential of new center catchments over thecourse of the late cycle. First, new centers foundedrelatively early in this cycle continued to occupytheir high-productivity catchments, so these werenot available to subsequent communities. In addi-tion, the population density of the study areareached unprecedented heights during the final cen-tury of occupation, and it may well be that thisforced people to build new centers on less-productive catchments. It is also possible that agri-cultural potential came to have a lower priority,relative to other factors, in choosing center loca-tions. For example, centers founded early in the latecycle, such as Yellow Jacket and Albert Porter pueb-los, were located on and adjacent to deep soils thatwere ideal for farming, whereas centers foundedlater in the cycle, such as Sand Canyon and WoodsCanyon pueblos, were in canyon settings close todomestic water, building stone, and constructiontimber, but further from the best agricultural soils(Varien et al. 1996). It may be that farming theselocalities for centuries had resulted in land tenuresystems that were widely acknowledged such thatpeople no longer had to claim fields by livingdirectly on these lands (Adler 1996; Varien 1999a).So as the late cycle progressed, people who had ini-tially lived near fields and walked to water increas-ingly came to live near their water, stone, and timber

resources and walked to their fields. One can alsocharacterize these new center locations as beingmore defensible than earlier locations (e.g., Kuck-elman, ed. 2000), and conflict and warfare havebeen documented during this period (Kuckelman2002; Kuckelman et al. 2000, 2002; Lightfoot andKuckelman 2001). However, safety at the latestcenters could have been found in numbers as wellas in the physiographic setting.

Construction-Timber Procurement

Figure 5F presents the number of tree-ring cuttingdates recovered from sites in the study area duringeach period, standardized according to the lengthof the period. There are at least a few cutting datesfor every period, demonstrating that the study areawas occupied virtually continuously for 700 years.In addition, the number of cutting dates per yearduring the three periods from A.D. 880 to 1020 isconsistently low. This is the period of low popula-tion density at the start of the second demographiccycle.

In other respects, however, the corpus of tree-ring dates does not provide a reliable record of pop-ulation dynamics; the r2 value for the regressionbetween numbers of tree-ring cutting dates andpopulation by period is .008. The current distribu-tion of dates appears to represent a complex com-bination of ancient practices—includingconstruction activity, recycling of timbers, andburning—and modern archaeological practices,such as differential intensity of sampling for data-ble timbers and the research foci of major excava-tion projects. As an example, tree-ring cutting datesare especially abundant for A.D. 840–880 becauseexcavations by the Dolores Archaeological Pro-gram focused on community centers of this period,and many sites constructed during this periodburned at abandonment. Likewise, cutting datesare especially abundant for the A.D. 1200s becauseseveral major projects have focused on this century,and many sites constructed during this period werealso burned at abandonment.

Although we do not consider it productive tomake specific comparisons between cutting datesand our demographic reconstruction, these data canbe used for other purposes. Figure 5G summarizesthe age distribution of the archaeological samplesthat have both pith and cutting dates. These data


AQ72(2) varien 4/6/07 9:16 AM Page 286

must be interpreted with caution because many fac-tors affect the age of the sample, including the por-tion of the tree that was harvested and the portionof the sample that was analyzed. But these data indi-cate that the mean age of the archaeological sam-ples, and their standard deviation, decreasedgradually over the course of the early settlementcycle. The local woodlands were relatively unaf-fected by human impacts when occupation of thestudy area began about A.D. 600, due to limiteduse by humans prior to that time. During subsequentuse of the area, however, Pueblo agriculturalistscleared forested areas for fields, collected wood forfuel, and harvested suitable timbers for supportposts, walls, and roofs of their houses. The fact thatthe mean age of timbers and the standard deviationof timber ages decreased during the first occupa-tional cycle may indicate that the use of woodlandssurrounding sites created stands of younger, andmore even-aged, trees. Forests may have recoveredsomewhat during the period of low population atthe onset of the second cycle, as reflected by theincreasing age of archaeological samples during thecourse of the A.D. 900s. During the late cycle, itappears that the mean age of utilized timbers onceagain gradually declined, but the standard devia-tion of these ages did not decline as it had duringthe early cycle. This suggests that the late-cyclepopulation, despite its larger size, did not impactthe local forests as dramatically as the early cyclepopulation.

This interpretation of tree-ring sample age dis-tributions is supported by the archaeobotanicalanalysis of wood used for fuel. Kohler andMatthews (1988) demonstrate a pervasive shiftfrom high-quality fuel woods such as pinyon andjuniper to low-quality fuel woods like Populus androsaceous shrubs among the Dolores Pueblo I vil-lages. By contrast, Adams and Bowyer (2002:134,141–142) demonstrate relative stability in use ofjuniper and pinyon for fuels throughout the thir-teenth century in our study area. Although theirarchaeobotanical evidence indicates some land-scape disturbance by the end of the thirteenth cen-tury, the presence of small juniper and pinyon plantparts such as twigs, needles, and bark suggests thatintact pinyon-juniper woodland was never absentfrom the landscape or at a great distance from sites(Adams and Bowyer 2002:141). Perhaps, then,thirteenth-century communities did not impact their

local woodlands to the same extent as did ninth-century communities in the Dolores River valley.

Figure 6 summarizes proportions of ponderosapine, juniper, and pinyon pine among cutting datesfor each period. These are the three species usedmost commonly in construction. Although Pon-derosa pines produce large, straight timbers idealfor support posts and primary beams, this speciesis rare in the study area, occurring primarily at ele-vations that are too high for maize agriculture, suchas the upper slopes of Ute Mountain and the uplandsadjacent to the Dolores River. Most Ponderosa tim-bers date from A.D. 600 and 880 and come fromsites in the Dolores River valley, where they werelikely used as posts and beams in pit structures. Dur-ing the late cycle the Dolores River valley waslargely unoccupied, and pit structure roofs weresupported by masonry walls instead of posts, sothere was little incentive to procure Ponderosa tim-bers from a distance. The one exception to this ruleappears to have occurred during the era of Chacoaninfluence when Ponderosa timbers were used toroof great houses. All 32 Ponderosa cutting datesfrom the late cycle come from great houses exhibit-ing Bonito-style architecture, including Ida Jean,Wallace Ruin, Escalante Ruin, and Lowry Ruin. Ahallmark of Bonito-style architecture in the SanJuan basin is roof construction using Ponderosapine vigas, often imported from a substantial dis-tance (Betancourt et al. 1986; Durand 1999;Reynolds et al. 2005; Windes and Ford1996:303–306). Thus the restriction of Ponderosatimbers to great houses in the second cycle proba-bly reflects the energy expended in construction ofthese structures and not depletion of local Pon-derosa forests.

Juniper occurs throughout the study area, andwas the most commonly used species overall.Pinyon also occurs throughout the study area butwas little used after A.D. 880. Juniper was alwayspreferred over pinyon for construction, perhapsbecause it is stronger and more rot resistant. So thedecline in pinyon over time may be due in part tothe increasing use-life of houses in the late cycle,which led to an even stronger preference for juniper.The decline in juniper over the course of the earlycycle therefore may indicate that land-use prac-tices led to a shortage of suitable junipers within areasonable distance of settlements. In contrast,nearly all dated construction timbers were of


AQ72(2) varien 4/6/07 9:16 AM Page 287

juniper throughout the late cycle, indicating eitherthat land use did not deplete local woodlands to thepoint that builders were forced to use the less-desirable species, or that traveling farther to obtaindesirable construction timber was an accepted costof building more durably. In sum, both sample ageand species distributions suggest that the largerpopulations of the late cycle degraded local wood-lands to a lesser degree than did the smaller popu-lations of the early cycle. We will consider theimportance of this for the study of human-environment relations in our concluding discus-sion.

Implications for Pueblo History

The reconstruction presented in this paper con-tributes to our understanding of the long history ofPueblo peoples in the Mesa Verde region. First, itprovides new population estimates for a denselyoccupied portion of the northern Southwest. Oneestimate that is intrinsically interesting, especiallyto Pueblo people we work with, is that cumulativelyabout 197,000 people lived in the study area dur-ing the seven centuries between A.D. 600 and1300.9 A second figure that is especially useful forcomparative studies is our estimate of the peak pop-ulation of the Village Project study area. Based onthe methods presented in this paper and in the pre-vious paper by Ortman and others (2007), we esti-mate that approximately 19,500 persons, or 11persons per square kilometer of land below 2400m elevation, inhabited the study area during the

mid-thirteenth century. This new estimate is withinthe range (15,000 to 30,000) suggested by Rohn(1983:176; 1989:166) for the entire Mesa Verderegion, but is higher than estimates produced byDuff and Wilshusen (2000:173–184); they calcu-lated 23,000 people as a maximum estimate for theportion of the Mesa Verde region in southwesternColorado, but argued that a peak population ofbetween about 6,000 and 11,000 people is morelikely. Our figure is also higher than the estimateof about 12,700 people produced by Wilshusen(2002) in a follow-up study. Our peak populationestimate for the Village Project study area coversonly a portion of the larger areas considered inthese studies, and thus suggests a still higherregional population. We nevertheless prefer ournew estimate because it is based on a more thor-ough analysis of a larger body of data, building upfrom the original site forms, and profiting fromyears of excavation to calibrate the surface potteryand architectural data.

A second important feature of our reconstruc-tion is the dynamic population history it suggests.In contrast to traditional views of population con-tinuity, in situ growth, and gradual culture changein the Mesa Verde region, our work suggests up tofive episodes of immigration and two periods ofemigration, each of which is tied to importantevents in Pueblo Indian history. In this sense ourreconstruction offers support for the contention thatpopulation movements played a central role in cul-tural transitions throughout Pueblo history (e.g.,Berry 1982; Wilshusen and Van Dyke 2006).


0.0%

25.0%

50.0%

75.0%

100.0%

600-725

725-800

800-840

840-880

880-920

920-980

980-1020

1020-1060

1060-1100

1100-1140

1140-1180

1180-1225

1225-1260

1260-1280

Modeling period (A.D.)

Juniper Ponderosa Pinyon

Figure 6. Species distribution of harvested timbers through time (N = 1,784).

AQ72(2) varien 4/6/07 9:16 AM Page 288

The first episode of immigration in our studyarea occurred when it was initially settled by a sub-stantial population of year-round residents, in theA.D. 600–725 period. We know little about the rea-sons for this migration or the source of the migrants;these remain important problems for futureresearch. The second episode occurred as the well-documented villages of the Dolores River valleyformed in the northeast corner of our study area(Kane 1986, 1989; Kohler 1992; Orcutt et al. 1990;Potter 1997; Schlanger 1988). Schlanger(1988:787) was the first to argue for large-scalemigration into the Dolores River valley during themid-ninth century to which Wilshusen and Ortman(1999) added the suggestion that these villageswere created by people with distinct cultural andhistorical backgrounds, with one source in theupper reaches of the San Juan drainage to the eastand another in southeastern Utah to the west. Ourreconstruction, in agreement with Wilshusen andOrtman (1999:380, 389), suggests that these groupsmoved into the study area primarily during the A.D.800–840 period and then coalesced into the largevillages of the Dolores River valley in the subse-quent period.

This period of village formation was followedby the first large-scale emigration from the studyarea between A.D. 880 and 920. It now appears thatmany of these migrants went south into north-western New Mexico (Wilshusen and Wilson 1995)and the San Juan geologic basin (Judge 1989:216;Windes 2005). Several recent studies have arguedthis movement from north to south contributed tothe initial development of the Chacoan regionalsystem (Varien 2001:53; Varien et al. 2007;Wilshusen and Van Dyke 2006; Windes 2005).Despite a third episode of possible immigration inthe late tenth century, population levels remainedlow until the fourth episode of immigration in A.D.1060 to 1100. This is a surprising and significantfinding because construction of Bonito-style greathouses in the central Mesa Verde region began inearnest around A.D. 1080, which our reconstruc-tion suggests was during a period of immigration.This fact needs to be taken into account in futurediscussions of Chaco influence in the Mesa Verderegion (also see Cameron 2005).

A fifth period of population growth and possi-ble immigration occurred during the A.D. 1225 to1260 period, when the peak population of the entire

sequence was reached. This increase in populationdensity has been largely unrecognized in previousstudies, and it is associated with other importantchanges in settlement patterns (Glowacki 2006).First, both the number of community centers andthe proportion of people who lived in centersincreased. Second, these new centers were built incanyon settings that had less-productive catchmentsthan established centers in the uplands, but did con-tain domestic water sources, especially springs,and ready access to stone for masonry buildings(Fetterman and Honeycutt 1987) and timber forconstruction and fuel.

These changes in settlement pattern, unparal-leled in their scope and swiftness, anticipate thelargest change of all: the complete depopulation ofthe region during the late thirteenth century. Duffand Wilshusen (2000) argued that emigration fromthe region was a long-term process that began inthe late A.D. 1100s or the early 1200s, and theynote that robust intrinsic growth could increase theoverall population of an area even in the context oflimited emigration. But in a separate study,Wilshusen (2002:118–119) argued that populationlevels remained high until the middle-to-late 1200s.Although it may not apply to the entirety of thelarger areas studied by Duff and Wilshusen, ourreconstruction supports elements of both models.It suggests that emigration from the Village Pro-ject study area began at least two decades prior tofinal depopulation, but it also suggests that morethan 10,000 individuals remained in our study areaafter A.D. 1260. These people either emigrated ordied over the following two decades. The latesttree-ring cutting-date from the Mesa Verde regionin our database is A.D. 1280, and it is likely thatthe entire region was devoid of Pueblo people bya few years later.

Implications for Historical Ecology

In this final section we discuss several findings ingreater depth and consider their implications for thestudy of historical ecology. The first finding of gen-eral importance is that one cannot account for thehistory of aggregated settlements in our study areawithout considering both ecological and social fac-tors. This conclusion is supported by several linesof evidence. First, Figure 5D shows that commu-nity centers were relatively resilient in our study


AQ72(2) varien 4/6/07 9:16 AM Page 289

area and tended to persist for several successiveperiods after their initial founding. Second, Figure5B shows that the proportion of population livingin centers grew over the course of each settlementcycle, reaching a peak during the phase of emi-gration that occurred at the end of each cycle. Thisindicates that most of the initial emigrants at theend of each settlement cycle moved out of smallsites rather than centers.10 Evidently, centers, espe-cially the largest centers, exerted a stronger holdon their inhabitants than did isolated farmsteads.This might be explainable through the differing“sunk costs” in these two settlement types (Janssenet al. 2003); it could be due to the threat level onthe landscape (Haas and Creamer 1996; Kuckel-man 2002; Kuckelman et al. 2002; LeBlanc 1999;Lightfoot and Kuckelman 2001; Wilcox and Haas1994); or perhaps due to some other factor thatneeds to be adduced. In any case, this is an inter-esting topic for future research.

Figure 7 presents a series of scatterplots thatillustrate our general point in greater detail. Each

of these plots contains one or two series of 14points, and each of these points characterizes oneof our 14 modeling periods using two summaryvariables generated from our settlement or paleo-productivity reconstructions. Regression lines andcorrelations are also given for each series. Figure7A illustrates that there is practically zero correla-tion between long-term climatic fluctuations, asmeasured by mean potential productivity of thestudy area across the years in each period, and theproportion of households that lived in communitycenters during that period. This evidence suggeststhat agriculture was not as marginal in our studyarea as is often assumed. Instead, these data indi-cate that the landscape was productive enough ona per hectare basis that people could live in villagesduring any of these periods. These data also indi-cate that aggregation was not particularly respon-sive to climatic variation on an annual basis, at leastas averaged within periods.

Figure 7B shows that the number of occupiedcenters was proportional to the overall population


A.

R2 = 0.000

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

230 240 250 260 270

Mean Kg maize / ha for modeling period

Pro

po

rtio

n o

f h

ou

seh

old

s in

cen

ter

B.

R2 = 0.833

R2 = 0.0363

0

10

20

30

40

50

60

70

80

0 5000 10000 15000 20000

Study area population

Number of centers% households in centers

D.

R2 = 0.7236

R2 = 0.873

0

20

40

60

80

100

120

140

160

0 5000 10000 15000 20000

Study area populationi

Mean center (households)

Largest center (households

C.

R2 = 0.6224

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0 50 100 150 200 250 300 350

Elapsed time in settlement cycle (year)

Pro

po

rtio

n o

f h

ou

seh

old

s in

cen

ter

Early CycleLate Cycle

Figure 7. Correlations among aggregation, population, climate, and history in the Village Project study area. Note: In allscatterplots, each point represents a modeling period, characterized according to two summary measures: A) Proportionof population in centers vs. agricultural potential; B) Percent of population in centers vs. total population; C) Proportionof population in centers vs. elapsed time in settlement cycle; D) Community center size (mean and maximum households)vs. total population.

AQ72(2) varien 4/6/07 9:16 AM Page 290

of the study area, but that the proportion of popu-lation actually living in these centers was notclosely related to these same population levels. Thisis because aggregation peaked during the initialperiod of population decline at the end of eachcycle, as mentioned previously.

Figure 7C shows that the proportion of house-holds living in centers during a given period cor-relates more strongly with the elapsed time in therelevant settlement cycle than with either averagemaize productivity (7A) or overall population den-sity (7B). This scatterplot compares the percent ofhouseholds living in community centers with thetotal years elapsed in each occupational cycle; thefirst cycle begins at A.D. 600, the second at A.D.920. The number of years elapsed is the cumula-tive years passed in a given occupational cycle atthe end of each period within that cycle. So therewas something about the passage of time that pro-moted population aggregation on this landscape.Based on initial results of Village Project model-ing efforts (Reynolds et al. 2003), we suspect thatsocial factors, such as the growth of exchange net-works with “hub-nodes” at community centers, areimportant to this pattern.

Figure 7D, however, suggests that ecologicaland sociopolitical factors limited the growth of cen-ters inhabited by subsistence-farming householdson this landscape. As population grew, people builtnew centers (Figure 7B), but the size of the largestcenters (Figure 7D)—primarily Grass Mesa Village(Kohler 1988) during the early cycle and YellowJacket Pueblo (Kuckelman 2003) during the latecycle—did not keep pace with rising populationdensity. As can be seen in 7D, the size of the largestcenters did not increase in direct proportion withthe total study area population. One factor that mayhave limited center growth was the increasing dis-tance new arrivals had to travel to fields, whichwould encourage the establishment of new centersover the continued growth of existing centers.Sociopolitical factors may also have played a lim-iting role. Cross-culturally, there appears to be amajor threshold of sociopolitical organization atroughly 500 members in the “largest organizationalunit” (Johnson 1982). Johnson and Earle(1987:314) consider this a transition from “localgroups” (roughly, autonomous villages and com-munities) to larger polities of various sorts. The rel-ative difficulty of crossing this threshold with

available models of sociopolitical organization maycontribute to the flattening out of the best-fit curve(Figure 7D) as it begins to approach 500.

A second general conclusion of our study is thatalthough increasing population strongly influencedthe size of aggregates, it did not necessarily lead toincreased environmental impacts. An intriguingfinding from our analysis of construction timber useis that both the minimum-age distributions of har-vested timbers and species-distributions of cuttingdates through time suggest that the early-cycle pop-ulation did far more damage to the woodlands sur-rounding settlements than did the late-cyclepopulation, despite being only one-third its size.This interpretation is supported by archaeobotani-cal analysis of fuel use. This is striking evidencethat anthropogenic impacts to ecosystems need notincrease in lockstep with population density, anencouraging finding for our contemporary world.

We can think of two factors that may be respon-sible for the lowered impact of the larger late-cyclepopulation on local woodlands. First, changes inarchitecture may have made it easier to conserveand recycle construction timbers. During the earlycycle both roofs and walls of houses were built ofwood and earth, and a variety of lines of evidence(Ahlstrom 1985; Cameron 1990; McIntosh 1974;Schlanger 1987; Varien 1999b, 2006) suggests thatconstruction timbers in contact with the groundwere used only for 8–12 years, a length of time thatcorresponds to the average use-life of houses basedon pottery accumulations. In contrast, during thelate cycle load-bearing walls were constructed ofstone masonry, so wood was used almost exclu-sively for roofing. The roof timbers in late-cyclehouses did not contact the ground and were lessprone to rot and decay; as a result, these timberslasted much longer (Ahlstrom 1985:642; Varien1999b). Thus, a higher proportion of timbers weresalvageable when an old house was demolishedand a new one built during the late cycle than dur-ing the early cycle. Both factors, longer use life andincreased recycling, would have lowered thedemand for construction timber.

A comparison of recycling rates at two sites inour database with similar occupation spans, and atwhich complete houses were excavated, supportsthis view. The Duckfoot site was occupied betweenA.D. 850–880 at the end of the early cycle; only16 percent of the 215 cutting dates recovered from


AQ72(2) varien 4/6/07 9:16 AM Page 291

it date prior to A.D. 850, the approximate year ofinitial construction (Lightfoot 1992, 1994). In con-trast, 68 percent of the 275 cutting dates from SandCanyon Pueblo, a late-cycle village occupiedbetween A.D. 1260–1280, date prior to the con-struction date of the earliest dated house (Ortmanand Bradley 2002:48–53). In fact, 30 percent ofthese dates are prior to A.D. 1225, and stem-and-leaf plots of the dates from specific roofs exhibitmultiple date clusters that Bradley (1993) inter-prets as groups of recycled timbers from severalprevious generations of houses. Excavations atsmall family farmsteads surrounding Sand CanyonPueblo (Varien, ed. 1999) suggest that many ofthese recycled timbers were originally cut down tobuild earlier houses in the area. This dramaticincrease in recycling may have compensated for thelarger population of the late cycle and mitigatedimpacts of construction on local woodlands.

A second factor that may have lowered theimpact of the late-cycle population on local wood-lands is changing land-use practices. During theearly cycle houses were short-lived and populationdensity was low, indicating that households relo-cated often and had ample opportunity to claimpreviously unfarmed land (Varien 1999a, 2006).The easiest way for Pueblo farmers to clear newland for agriculture was to burn it (Kohler 1992;Matson 1991), but in the process large numbers oftrees that could provide construction timbers wouldbe destroyed. Thus, swidden agriculture, even bya relatively small population, could have had a sig-nificant negative impact on the slow-growingforests of this high-desert environment over threecenturies. In contrast, during the late cycle popu-lation density was high and houses came to be occu-pied for multiple generations, and farmland waslikely inherited from one generation to the next(Varien 1999a; Varien and Ortman 2005). Thischange from swidden to more settled, intensiveagriculture may also have moderated the impact ofthis larger population on the woodlands surround-ing settlements.

This scenario appears broadly consistent withfindings from cultural ecology. Although severalstudies have argued that swidden agriculture pro-duces a stable mosaic forest in tropical environ-ments (Robbins 2004:33), Netting (1993) has foundthat intensive, settled, smallholder agriculture isthe dominant, and apparently more sustainable,

form in drier or more temperate environments.These findings suggest that a key variable affect-ing the relative sustainability of swidden versussettled agriculture is the rate of forest successionand regrowth: somewhat obviously, forests thatregenerate slowly are more prone to degradationfrom swidden agriculture than are forests thatregenerate quickly. Data from the Village Projectsupport this view in suggesting that, over a com-parable length of time and in the same environment,settled, intensive agriculture by a large populationimpacted local woodlands less extensively than didswidden agriculture by a smaller population.Regardless of whether recycling, agricultural strate-gies, or both are responsible, the patterns weobserve in construction timber age and species dis-tributions support the conclusion that resource-usepractices have the ability to moderate the per per-son impact of humans on their sustaining environ-ments. Fisher (2005) has documented this samephenomenon in the Tarascan region of PostclassicMesoamerica.

Our third general conclusion is that the notionof carrying capacity cannot account for the collapseof our settlement system. The peak population ofthe Village Project study area was reached in themid-thirteenth century, when it was inhabited by11 persons per square kilometer of arable land.Within a generation of reaching this peak popula-tion, the settlement system had collapsed and theregion was totally depopulated. Despite the popu-lation increase, we do not believe this collapse canbe explained as a population exceeding the absolutecarrying capacity of its local environment.

Our paleoproductivity reconstruction suggeststhat, absent significant anthropogenic impacts, thestudy area was amenable to additional intensifica-tion at the end of the late cycle. The mean annualproductivity of all hectares in the study area dur-ing the thirteenth century was 235 kg maize perhectare, with a standard deviation of 41 kg maizeper hectare. If our paleoproductivity reconstruc-tion is roughly accurate, it indicates that about 35percent of the land below 2400 m in the study areacould produce a two-year supply of the daily caloricneeds of the mid-thirteenth-century population dur-ing a year that was one standard deviation belowthe long-term mean for annual productivity (nutri-tional needs and caloric yield of cornmeal providedby the US Department of Agriculture).


AQ72(2) varien 4/6/07 9:16 AM Page 292

This rough calculation assumes that people canmake use of large areas of low production that con-tribute to the mean production, and it ignores thelimiting effect of runs of bad years. Nevertheless,these figures do suggest that total potential agri-cultural productivity did not directly limit this pop-ulation and highlight the fact that local populationsin the thirteenth century did not use much appar-ently productive land in the eastern third of the pro-ject area (see also Kohler 2000). Pueblo farmers inour study area thus could have put more land intoproduction as one means of addressing shortfalls,unless this option was impossible for security rea-sons; or they could have intensified production bycapturing more of the available moisture (e.g.,through more use of check dams and terraces), orby extending the growing season (e.g., through theuse of rock mulches). This finding focuses ourattention on the need to better understand why pop-ulations did not use all the apparently productiveland available: was use limited by more severe anddurable anthropogenic impacts to the study areathan we currently recognize? Did conflict and war-fare, which has been documented during the thir-teenth century (Kuckelman, ed. 2000, 2002;Kuckelman et al. 2002), restrict settlement to spe-cific areas? Was it the relative indefensibility of therather flat “eastern marches” of the project areathat prevented its use?

Another point which bears on this discussion isthat the population density of the project study areanear the end of our late cycle approached levels thatpreceded the emergence of regional polities andhereditary ranking in formative Mesoamerica. Forexample, it was similar to the population densityof the Valley of Mexico between 900–650 B.C.,about 600 years after initial colonization of the val-ley by farmers, and immediately prior to the appear-ance of the first regional polities (Sanders et al.1979:217). Likewise, the population density of theValley of Oaxaca increased to 15 persons per squarekilometer of prime agricultural land between1150–850 B.C., about 750 years after the earliestagricultural settlements, and during the period inwhich hereditary rank society emerged (Marcusand Flannery 1996:106). Similar population den-sities were reached in the Village Project study areaabout the same length of time after its initial colo-nization by people who lived in year-round habi-tations, made pottery, and grew maize as a dietary

staple. But in this case—instead of the emergenceof hereditary ranking, regional polities, and ulti-mately archaic state polities—we have regionaldepopulation and the disintegration of the settle-ment system. Perhaps the population density of themid-thirteenth century did exceed the carryingcapacity for a society comprised of politicallyautonomous villages, but since additional intensi-fication and increased production on this landscapeappears to have been possible at the time of the finaldepopulation, we must ask why Pueblo peoples inour region did not follow the path of societies inMesoamerica when confronted with similar popu-lation pressures.

Our final conclusion for the study of historicalecology is that issues of scale loom large. Figure 8illustrates our reconstruction of agricultural pro-ductivity for each year of the thirteenth century rel-ative to mean productivity for the A.D. 600–1300period. This graph shows that the 1200s wereplagued by drought and cooler-than-normal tem-peratures, which resulted in poor harvests. Therewere 69 years when agricultural potential wasbelow the long-term average and only 31 years thatwere above average. Yet this was a period whenpopulation grew, including growth that may havebeen due to immigration. This finding cautions usto recognize that climatic fluctuations occurred overlarge areas on an annual basis, and variation in rain-fall and especially temperature would have affectedthe relative agricultural potential of lands beyondour study area in complex ways. The best lands dur-ing a warm and dry period would have been dif-ferent than the best lands when it was cool and dry,warm and wet, or cool and wet. The amount of pop-ulation movement into and out of our study areaover time (Figure 5B) suggests that the scale of theeffective environment that Pueblo peopleresponded to was much larger than our project studyarea. This is an important point that one must con-sider in studying historical ecology.

Although our paleoproductivity reconstructiondeals with a much smaller area than the effectiveenvironment that Pueblo people responded to inlocational decision-making, we do not think thisundermines the utility of our endeavor. We havelearned a great deal about human-environment rela-tionships and Pueblo Indian history in the centralMesa Verde region through the Village Project,despite modeling the environment only within our


AQ72(2) varien 4/6/07 9:16 AM Page 293

study area. But to tackle the larger problems of thehistorical ecology of Pueblo peoples, we will even-tually need to work at larger scales. We believe ourstudy illustrates that the methods needed to accom-plish this effort are at hand, and applying thesemethods to a larger region would produce excitingand substantive results.

Acknowledgments. This work is supported by the NationalScience Foundation (BCS-0119981), the ColoradoHistorical Society (2004-01-056), and the Crow CanyonArchaeological Center. We thank many colleagues includingJeff Dean (Laboratory of Tree-ring Research, University ofArizona) who advised on the revisions to the maize paleo-productivity we summarize here; Doug Ramsey (NaturalResource Conservation Service, Cortez) for endless adviceon updating soils data; and Matt Salzer (Laboratory of Tree-ring Research, University of Arizona), who allowed us to usehis bristlecone pine ring-width index series from the SanFrancisco Peaks. We are grateful to William D. Lipe(Anthropology, WSU) and Carla Van West (SRI Foundation)who provided thorough and valuable comments on an earlydraft of this manuscript, and we thank three anonymousreviewers for their helpful comments. Finally, we thank all ofour colleagues on the Village Ecodynamics Project and at theCrow Canyon Archaeological Center. We thank Ian

Robertson and Oralia Cabrera for translating the abstractinto Spanish. Any errors are the responsibility of the authors.

References Cited

Adams, E. Charles1989 Changing Form and Function in Western Pueblo Cer-

emonial Architecture from A.D. 1000 to A.D. 1500. In TheArchitecture of Social Integration in Prehistoric Pueblos,edited by W. D. Lipe and M. Hegmon, pp. 155–160. Occa-sional Papers No. 1. Crow Canyon Archaeological Cen-ter, Cortez, Colorado.

1991 The Origin and Development of the Pueblo Katsina Cult.University of Arizona Press, Tucson.

Adams, Karen R., and Vandy E. Bowyer2002 Sustainable Landscape:Thirteenth-Century Food and

Fuel Use in the Sand Canyon Locality. In Seeking the Cen-ter Place: Archaeology and Ancient Communities in theMesa Verde Region, edited by Mark D. Varien and RichardH. Wilshusen, pp. 123–142. University of Utah Press, SaltLake City.

Adams, Karen R., Deborah A. Muenchrath, and Dylan M.Schwindt

1999 Moisture Effects on the Morphology of Ears, Cobsand Kernels of a South-Western U.S. Maize (Zea Mays L.)Cultivar, and Implications for the Interpretation of Archae-ological Maize. Journal of Archaeological Science26:483–496.


Figure 8. Potential agricultural productivity for each year of the thirteenth-century A.D. plotted against the long-term(A.D. 600–1300) mean productivity.

AQ72(2) varien 4/6/07 9:16 AM Page 294

Adler, Michael A.1994 Population Aggregation and the Anasazi Social Land-

scape: A View from the Four Corners. In The AncientSouthwestern Community: Models and Methods for theStudy of Prehistoric Social Organization, edited by W. H.Wills and R. D. Leonard, pp. 85–101. University of NewMexico Press, Albuquerque.

1996 Land Tenure, Archaeology, and the Ancestral PuebloSocial Landscape. Journal of Anthropological Archaeol-ogy 15:337–371.

Adler, Michael A., and Mark D. Varien1994 The Changing Face of the Community in the Mesa

Verde Region A.D. 1000–1300. In Proceedings of theAnasazi Symposium, 1991, edited by A. Hutchinson andJ. E. Smith, pp. 83–97. Mesa Verde Museum Association,Mesa Verde, Colorado.

Ahlstrom, Richard V. N.1985 The Interpretation of Archaeological Tree-Ring Dates.

Unpublished Ph.D. dissertation, Department of Anthro-pology, University of Arizona, Tucson.

Berry, Michael S.1982 Time, Space, and Transition in Anasazi Prehistory.

University of Utah Press, Salt Lake City.Betancourt, Julio L., Jeffrey S. Dean, and Herbert M. Hull

1986 Prehistoric Long-Distance Transport of ConstructionBeams, Chaco Canyon, New Mexico. American Antiquity51:370–375.

Bradley, Bruce A.1988 Wallace Ruin Interim Report. Southwestern Lore

54(2):8–33.1993 Planning, Growth, and Functional Differentiation at

a Prehistoric Pueblo: A Case Study from SW Colorado.Journal of Field Archaeology 20:23–42.

1996 Pitchers to Mugs: Chacoan Revival at Sand CanyonPueblo. Kiva 61:241–255.

Burns, Barney T.1983 Simulated Anasazi Storage Behavior Using Crop

Yields Reconstructed from Tree Rings: A.D. 652–1968.Unpublished Ph.D. dissertation, Department of Anthro-pology, University of Arizona, Tucson.

Cameron, Catherine M.2005 Exploring Archaeological Cultures in the Northern

Southwest: What Were Chaco and Mesa Verde? Kiva70(3):227–253.

Cater, John D., and Mark L. Chenault1988 Kiva Use Reinterpreted. Southwestern Lore

54(3):19–32.Cowgill, George

1975 On Causes and Consequences of Ancient and Mod-ern Population Changes. American Anthropologist77:505–525.

Dean, Jeffrey S.1985 Review of Time, Space, and Transition in Anasazi

Prehistory, by M. S. Berry. American Antiquity50:704–705.

1988 A Model of Anasazi Behavioral Adaptation. In TheAnasazi in a Changing Environment, edited by George J.Gummerman, pp. 25–44. Cambridge University Press,Cambridge.

Dean, Jeffrey S., Robert C. Euler, George J. Gumerman, FredPlog, Richard H. Hevly, and Thor N. V. Karlstrom

1985 Human Behavior, Demography, and Paleoenviron-ment on the Colorado Plateaus. American Antiquity50:537–554.

Diamond, Jared2005 Collapse: How Societies Choose to Fall or Succeed.

Viking Press, New York.

Douglass, Andrew E.1929 The Secret of the Southwest Solved by Talkative Tree

Rings. National Geographic 56:737–770.Driver, Jonathan C.

1996 Social Complexity and Hunting Systems in South-western Colorado. In Debating Complexity: Proceedingsof the Twenty–Sixth Annual Chacmool Conference, editedby D. A. Meyer, P. C. Dawson, and D. T. Hanna, pp.364–374. Archaeological Association of the University ofCalgary, Calgary, Alberta.

Duff, Andrew I., and Richard H. Wilshusen2000 Prehistoric Population Dynamics in the Northern San

Juan Region, A.D. 950–1300. Kiva 66:167–190.Durand, Stephen, R., Phillip H. Shelley, Ronald C. Antweiler,

and Howard E. Taylor1999 Trees, Chemistry, and Prehistory in the American

Southwest. Journal of Archaeological Science26:185–203.

Euler, Robert C., George Gumerman,Thor N. V. Karlstrom, Jef-frey Dean, and Richard Hevly

1979 The Colorado Plateaus: Cultural Dynamics and Pale-oenvironment. Science 205:1089–1101.

Fetterman, Jerry, and Linda Honeycutt1987 The Mockingbird Mesa Survey, Southwestern Col-

orado. Cultural Resource Series, No. 22. Bureau of LandManagement, Denver.

Fisher, Christopher T.2005 Demographic and Landscape Change in the Lake

Patzcuaro Basin, Mexico:Abandoning the Garden. Amer-ican Anthropologist 107:87–95.

Fisher, Christopher T., and Gary M. Feinman2005 Introduction to “Landscapes over Time.” American

Anthropologist 107:62–69.Fritz, Harold C., David G. Smith, Carl A. Budelsky, and John

W. Cardis1965 The Variability of Ring Characteristics within Trees

as Shown by a Reanalysis of four Ponderosa Pine. Tree-Ring Bulletin 27(1–2):3–18.

Glowacki, Donna M.2006 The Social Landscape of Depopulation: the Northern

San Juan, A.D. 1150–1300. Unpublished Ph.D. Disserta-tion. School of Human Evolution and Social Change,Ari-zona State University, Tempe.

Graybill, Donald A.1984 Almagre Mountain B. International Tree-Ring Data

Bank. Electronic document, http://www.ncdc.noaa.gov/paleo/ftp-treering.htm, accessed 8/2002.

Haas, Jonathan, and Winifred Creamer1996 The Role of Warfare in the Pueblo III Period. In The

Prehistoric Pueblo World, A.D. 1150–1350, edited by M.A. Adler, pp. 205–213. University of Arizona Press, Tuc-son.

Hill, J. Brett1998 Ecological Variability and Agricultural Specialization

among the Protohistoric Pueblos of New Mexico. Journalof Field Archaeology 25:275–294.

Honeycutt, Linda1995 Dryland Gardening in Southwest Colorado: Past and

Present. In Soil, Water, Biology, and Belief in Prehistoricand Traditional Southwestern Agriculture, edited by W.Wolcott Toll, pp. 369–373. New Mexico ArchaeologicalCouncil Special Publication 2. Albuquerque.

Huckleberry, Gary A., and Brian R. Billman1998 Floodwater Farming, Discontinuous Ephemeral

Streams, and Puebloan Abandonment in SouthwesternColorado. American Antiquity 63:595–616.

Janssen, Marco A., Timothy A. Kohler, and Marten Scheffer


AQ72(2) varien 4/6/07 9:16 AM Page 295

2003 Sunk-Cost Effects and Vulnerability to Collapse inAncient Societies. Current Anthropology 44:722–728.

Jett, Stephen C.1964 Pueblo Indian Migrations: An Evaluation of the Pos-

sible Physical and Cultural Determinants. American Antiq-uity 29:281–300.

Johnson. Allen W., and Timothy Earle1987 The Evolution of Human Societies: From Foraging

Group to Agrarian State. Stanford University Press, Stan-ford, California.

Johnson, Gregory A.1982 Organizational Structure and Scalar Stress. In Theory

and Explanation in Archaeology: the Southampton Con-ference, edited by Colin Renfrew, Michael J. Rowlands,and Barbara Abbott Segraves, pp. 389–421. AcademicPress, New York.

Judge, W. James1989 Chaco Canyon-San Juan Basin. In Dynamics of South-

west Prehistory, edited by L. S. Cordell and G. J. Gumer-man, pp. 209–261. Smithsonian Institution Press,Washington, D.C.

Kane, Allen E.1986 Prehistory of the Dolores River Valley. In Dolores

Archaeological Program: Final Synthetic Report, editedby D. A. Breternitz, C. K. Robinson, and G. T. Gross, pp.353–435. Bureau of Reclamation, Engineering andResearch Center, Denver.

1989 Did the Sheep Look Up? Sociopolitical Complexityin Ninth Century Dolores Society. In The SociopoliticalStructure of Prehistoric Southwestern Societies, edited byS. Upham, K. G. Lightfoot, and R. A. Jewett, pp. 307–361.Westview Press, Boulder, Colorado.

Kantner, John2004 Ancient Puebloan Southwest, Cambridge University

Press, Cambridge.Kohler, Timothy A.

1988 The Probability Sample at Grass Mesa Village. InDolores Archaeological Program: Anasazi Communitiesat Dolores: Grass Mesa Village, edited by W. D. Lipe, J.N. Morris, and T. A. Kohler, pp. 51–74. Bureau of Recla-mation, Engineering and Research Center, Denver.

1992 Prehistoric Human Impact on the Environment in theUpland North American Southwest. Population and Envi-ronment: A Journal of Interdisciplinary Studies13:255–268.

2000 The Final 400 Years of Prehispanic Agricultural Soci-ety in the Mesa Verde Region. Kiva 66:191–204.

2004 Pre-Hispanic Human Impact on Upland North Amer-ican Southwestern Environments: Evolutionary Ecologi-cal Perspectives. In The Archaeology of Global Change:The Impact of Humans on Their Environment, edited byCharles L. Redman, Steven R. James, Paul R. Fish, and J.Daniel Rogers, pp. 224–242. Smithsonian Books, Wash-ington, D. C.

Kohler, Timothy A., and Eric Blinman1987 Solving Mixture Problems in Archaeology: Analysis

of Ceramic Materials for Dating and Demographic Recon-struction. Journal of Anthropological Archaeology 6:1–28.

Kohler, Timothy A., James Kresl, Carla Van West, Eric Carr,and Richard H. Wilshusen

2000 Be There Then: A Modeling Approach to SettlementDeterminants and Spatial Efficiency among Late Ances-tral Pueblo Populations of the Mesa Verde Region, U.S.Southwest. In Dynamics in Human and Primate Societies:Agent-based Modeling of Social and Spatial Processes,edited by Timothy A. Kohler and George J. Gumerman,pp. 145–205. Santa Fe Institute Studies in the Sciences of

Complexity. Oxford University Press, New York.Kohler, Timothy A., and Meredith H. Matthews

1988 Long-Term Anasazi Land Use and Forest Reduction:A Case Study from Southwest Colorado. American Antiq-uity 53:537–564.

Kohler, Timothy A., C. David Johnson, Mark D. Varien, ScottG. Ortman, Robert G. Reynolds, Ziad Kobti, Jason Cowan,Kenneth Kolm, Schaun Smith, and Lorene Yap

2007 Settlement Ecodynamics in the Prehispanic CentralMesa Verde Region. In Modeling Socionatural Systems,edited by Timothy A. Kohler, and Sander van der Leeuw,pp. 61–104. SAR Press, Santa Fe.

Kuckelman, Kristin A.2002 Thirteenth-Century Warfare in the Central Mesa Verde

Region. In Seeking the Center Place: Archaeology andAncient Communities in the Mesa Verde Region, edited byM. D. Varien and R. H. Wilshusen, pp. 233–253. Univer-sity of Utah Press, Salt Lake City.

Kuckelman, Kristin A. (editor)2000 The Archaeology of Castle Rock Pueblo:A Thirteenth-

Century Village in Southwestern Colorado. Electronic doc-ument, http://www.crowcanyon.org/castlerock, accessedSeptember 22, 2005.

2003 The Archaeology of Yellow Jacket Pueblo: Excavationsat a Large Community Center in Southwestern Colorado.Electronic document, http://www.crowcanyon.org/castle-rock, accessed September 22, 2005.

Kuckelman, Kristin A., Ricky R. Lightfoot, and Debra L. Mar-tin

2000 Changing Patterns of Violence in the Northern SanJuan Region. Kiva 66:147–165.

2002 The Bioarchaeology and Taphonomy of Violence atCastle Rock and Sand Canyon Pueblos, Southwestern Col-orado. American Antiquity 67:486–513.

LeBlanc, Steven A.1999 Prehistoric Warfare in the American Southwest. Uni-

versity of Utah Press.Lekson, Stephen H., and Catherine M. Cameron

1995 The Abandonment of Chaco Canyon, the Mesa VerdeMigrations, and the Reorganization of the Pueblo World.Journal of Anthropological Archaeology 14:184–202.

Lightfoot, Ricky R.1992 Architecture and Tree-Ring Dating at the Duckfoot

Site in Southwestern Colorado. Kiva 57:213–236.1994 The Duckfoot Site, Volume 2: Archaeology of the

House and Household. Occasional Papers No. 4. CrowCanyon Archaeological Center, Cortez, Colorado.

Lightfoot, Ricky R., and Kristin A. Kuckelman2001 A Case of Warfare in the Mesa Verde Region. In

Deadly Landscapes: Case Studies in Prehistoric South-western Warfare, edited by G. E. Rice and S. A. LeBlanc,pp. 51–64. University of Utah Press, Salt Lake City.

Lipe, William D.1970 Anasazi Communities in the Red Rock Plateau, South-

eastern Utah. In Reconstructing Prehistoric Pueblo Soci-eties, edited by W. A. Longacre, pp. 84–139. University ofNew Mexico Press, Albuquerque.

1989 Social Scale of Mesa Verde Anasazi Kivas. In TheArchitecture of Social Integration in Prehistoric Pueblos,edited by W. D. Lipe and M. Hegmon, pp. 53–71. Occa-sional Papers No. 1. Crow Canyon Archaeological Cen-ter, Cortez, Colorado.

2002 Social Power in the Central Mesa Verde Region,A.D.1150–1290. In Seeking the Center Place: Archaeologyand Ancient Communities in the Mesa Verde Region, editedby M. D. Varien and R. H. Wilshusen, pp. 203–232. Uni-versity of Utah Press, Salt Lake City.


AQ72(2) varien 4/6/07 9:16 AM Page 296

Lipe, William D., and Mark D. Varien1999 Pueblo III (A.D. 1150–1300). In Colorado Prehistory:

A Context for the Southern Colorado River Basin, editedby W. D. Lipe, M. D. Varien, and R. H. Wilshusen, pp.290–352. Colorado Council of Professional Archaeolo-gists, Denver.

Marcus, Joyce, and Kent V. Flannery1996 Zapotec Civilization. Thames and Hudson, New York.

Matson, R. G.1991 The Origins of Southwestern Agriculture. University

of Arizona Press, Tucson.Matson, R. G., and Brian Chisholm

1991 Basketmaker II Subsistence: Carbon Isotopes andOther Dietary Indicators from Cedar Mesa, Utah. Ameri-can Antiquity 56:444–459.

McIntosh, Roderick J.1974 Archaeology and Mud Wall Decay in a West African

Village. World Archaeology 6:154–171.Muir, Robert J.

1999 Zooarchaeology of Sand Canyon Pueblo, Colorado.Unpublished Ph.D. dissertation, Department of Archaeol-ogy, Simon Fraser University, Burnaby, British Columbia.

Muir, Robert J., and Jonathan C. Driver2002 Scale of Analysis and Zooarchaeological Interpreta-

tion: Pueblo III Faunal Variation in the Northern San JuanRegion. Journal of Anthropological Archaeology21:165–199.

Muenchrath, Deborah A., Maya Kuratomi, Jonathan A. Sandor,and Jeffrey A. Homburg

2002 Observational Study of Maize Production Systems ofZuni Farmers in Semiarid New Mexico, Journal of Eth-nobiology 22(1):1–33.

Netting, Robert McC.1993 Smallholders, Householders: Farm Families and The

Ecology of Intensive, Sustainable Agriculture. StanfordUniversity Press, Stanford, California.

Orcutt, Janet D., Eric Blinman, and Timothy A. Kohler1990 Explanations of Population Aggregation in the Mesa

Verde Region Prior to A.D. 900. In Perspectives on South-western Prehistory, edited by P. E. Minnis and C. L. Red-man, pp. 196–212. Westview Press, Boulder, Colorado.

Ortman, Scott G.1998 Corn Grinding and Community Organization in the

Pueblo Southwest, A.D. 1150–1550. In Migration andReorganization: The Pueblo IV Period in the AmericanSouthwest, edited by K. A. Spielmann, pp. 165–192.Anthropological Research Papers No. 51. Arizona StateUniversity, Tempe.

Ortman, Scott G., and Bruce A. Bradley2002 Sand Canyon Pueblo: The Container in the Center. In

Seeking the Center Place: Archaeology and Ancient Com-munities in the Mesa Verde Region, edited by M. D. Varienand R. H. Wilshusen, pp. 41–78. University of Utah Press,Salt Lake City.

Ortman, Scott G., Donna M. Glowacki, Melissa J. Churchill,and Kristin A. Kuckelman

2000 Pattern and Variation in Northern San Juan Village His-tories. Kiva 66:123–146.

Ortman, Scott G., and Mark D. Varien2007 McElmo Dome-Sand Canyon Locality Settlement

Patterns. In The Archaeology of Sand Canyon Pueblo:Intensive Excavations at a Late-Thirteenth-Century Villagein Southwestern Colorado, edited by Kristin A. Kuckel-man Electronic document, http://www.crowcanyon.org, inpress.

Ortman, Scott G., Mark D. Varien, and T. Lee Gripp2007 Empirical Bayesian Methods for Archaeological Sur-

vey Data: An Application from the Mesa Verde Region.American Antiquity 72:241–272.

Palmer, Wayne C.1965 Meteorological Drought. Research Paper No. 45. U.S.

Department of Commerce, Office of Climatology, U.S.Weather Bureau, Washington, D.C.

Petersen, Kenneth Lee1988 Climate and the Dolores River Anasazi: A Paleoen-

vironmental Reconstruction from a 10,000-Year PollenRecord, La Plata Mountains, Southwestern Colorado.Anthropological Papers No. 113. University of Utah, SaltLake City.

Potter, James M.1997 Communal Ritual and Faunal Remains: An Example

from the Dolores Anasazi. Journal of Field Archaeology24:353–364.

2000 Pots, Parties, and Politics: Communal Feasting in theAmerican Southwest. American Antiquity 65:471–492.

Potter, James M., and Scott G. Ortman2004 Community and Cuisine in the Prehispanic American

Southwest. In Identity, Feasting, and the Archaeology ofthe Greater Southwest, edited by Barbara J. Mills, pp.173–191. University Press of Colorado, Boulder.

Powell, Shirley1988 Anasazi Demographic Patterns and Organizational

Responses: Assumptions and Interpretive Difficulties. InThe Anasazi in a Changing Environment, edited by G. J.Gumerman, pp. 168–191. Cambridge University Press,Cambridge.

Ramsey, Douglas K.2003 Soil Survey of Cortez Area, Colorado, Parts of Dolores

and Montezuma County. United States Department ofAgriculture, Natural Resources Conservation Service; inCooperation with United States Department of the Inte-rior, Bureau of Land Management and National Park Ser-vice; and the Colorado Agricultural Experiment Station.Natural Resources Conservation Service, Washington,D.C.

Redman, Charles L.1999 Human Impacts on Ancient Environments. University

of Arizona Press, Tucson.Reynolds, Amanda C., Julio L. Betancourt, Jay Quade, P.

Jonathan Patchett, Jeffrey S. Dean, and John Stein2005 87Sr/86Sr Sourcing of Ponderosa Pine Used in Anasazi

Great House Construction at Chaco Canyon, New Mex-ico. Journal of Archaeological Science 32:1061–1075.

Reynolds, Robert G., Timothy A. Kohler, and Ziad Kobti2003 The Effects of Generalized Reciprocal Exchange on

the Resilience of Social Networks: An Example from thePrehispanic Mesa Verde Region. Computational and Math-ematical Organization Theory 9:227–254.

Richerson, Peter J., Robert Boyd, and Robert L. Bettinger2001 Was Agriculture Impossible during the Pleistocene but

Mandatory during the Holocene? A Climate ChangeHypothesis. American Antiquity 66:387–411.

Robbins, Paul2004 Political Ecology: A Critical Introduction. Blackwell

Publishing, Malden, Massachusetts.Rohn, Arthur H.

1983 Budding Urban Settlements in the Northern San Juan.In Proceedings of the Anasazi Symposium, 1981, editedby J. E. Smith, pp. 175–180. Mesa Verde Museum Asso-ciation, Mesa Verde National Park, Colorado.

1989 Northern San Juan Prehistory. In Dynamics of South-west Prehistory, edited by L. S. Cordell and G. J. Gumer-man, pp. 149–177. Smithsonian Institution Press,Washington, D.C.


AQ72(2) varien 4/6/07 9:16 AM Page 297

Salzer, Matthew W.2000 Dendroclimatology in the San Francisco Peaks Region

of Northern Arizona, USA. Ph.D. Dissertation, Depart-ment of Geosciences, University of Arizona. UMI Dis-sertation Services, Ann Arbor, Michigan.

Sanders, William T., Jeffrey R. Parsons, and Robert S. Santley1979 The Basin of Mexico: Ecological Processes in the

Evolution of a Civilization. Academic Press, New York.Schlanger, Sarah H.

1987 Population Measurement, Size, and Change, A.D.600–1175. In Dolores Archaeological Program: Support-ing Studies: Settlement and Environment, edited by K. L.Petersen and J. D. Orcutt, pp. 568–613. Bureau of Recla-mation, Engineering and Research Center, Denver.

1988 Patterns of Population Movement and Long-TermPopulation Growth in Southwestern Colorado. AmericanAntiquity 53:773–793.

Schlanger, Sarah H., and Richard H. Wishusen1993 Local Abandonments and Regional Conditions in the

North American Southwest. In Abandonment of Settle-ments and Regions: Ethnoarchaeological and Archaeo-logical Approaches, edited by C. M. Cameron and S. A.Tomka, pp. 85–98. Cambridge University Press, Cam-bridge.

Stone, Glenn Davis, and Christian E. Downum1999 Non-Boserupian Ecology and Agricultural Risk: Eth-

nic Politics and Land Control in the Arid Southwest. Amer-ican Anthropologist 101(1):113–128.

Tuggle, H. D., J. J. Reid, and R. C. Cole1984 Fourteenth Century Mogollon Agriculture in the

Grasshopper Region. In Prehistoric Agricultural Strate-gies in the Southwest, edited by S. F. Fish and P. R. Fish,pp. 101–110. Anthropological Research Papers 33. ArizonaState University, Tempe.

van der Leeuw, Sander E., and Charles L. Redman2002 Placing Archaeology at the Center of Socio-Natural

Studies. American Antiquity 67:597–605.Van West, Carla, R.

1994 Modeling Prehistoric Agricultural Productivity inSouthwestern Colorado: A GIS Approach. Reports ofInvestigations No. 67. Department of Anthropology,Wash-ington State University, Pullman, and Crow CanyonArchaeological Center, Cortez, Colorado.

Varien, Mark D.1999a Sedentism and Mobility in a Social Landscape: Mesa

Verde and Beyond. University of Arizona Press, Tucson.1999b Regional Context:Architecture, Settlement Patterns,

and Abandonment. In The Sand Canyon ArchaeologicalProject: Site Testing, edited by M. D. Varien, p. Chapter21. Electronic document, http://www.crowcanyon.org/sitetesting, accessed 04/25/06.

2001 We Have Learned A Lot, But We Still Have More toLearn. In Chaco Society and Polity: Papers from the 1999Conference, edited by Linda S. Cordell, W. James Judge,and June-el Piper, pp. 47–61, New Mexico Archaeologi-cal Council Special Publication 4, Albuquerque.

2006 Occupation Span and Organization of ResidentialActivities: A Case Study from the Mesa Verde Region. InAncient Households of the Americas: ConceptualizingWhat Households Do, edited by J. G. Douglass and N.Gonlin, University of Colorado Press, Boulder, in press.

Varien, Mark D. (editor)1999 The Sand Canyon Archaeological Project: Site Test-

ing [HTML Title], edited by M. D. Varien, Chapter 22. Elec-tronic document, http://www.crowcanyon.org/sitetesting,accessed March 8, 2006.

Varien, Mark D., and Kristin A. Kuckelman

1999 Summary and Conclusions. In The Sand CanyonArchaeological Project: Site Testing [HTML Title], editedby M. D. Varien, Chapter 22. Electronic document,http://www.crowcanyon.org/sitetesting, accessed March8, 2006.

Varien, Mark D., William D. Lipe, Michael A. Adler, Ian M.Thompson, and Bruce A. Bradley

1996 Southwestern Colorado and Southeastern Utah Set-tlement Patterns: A.D. 1100 to 1300. In The PrehistoricPueblo World, A.D. 1150–1350, edited by M. A. Adler, pp.86–113. University of Arizona Press, Tucson.

Varien, Mark D., and Barbara J. Mills1997 Accumulations Research: Problems and Prospects for

Estimating Site Occupation Span. Journal of Archaeolog-ical Method and Theory 4:141–191.

Varien, Mark D., Carla R. Van West, and G. Stuart Patterson2000 Competition, Cooperation and Conflict: Agricultural

Production and Community Catchments in the CentralMesa Verde Region. Kiva 66:45–65.

Varien, Mark D., and Scott G. Ortman2005 Accumulations Research in the Southwest United

States: Middle Range Theory for Big-Picture Problems.World Archaeology 37(1):132–155.

Varien, Mark D., Scott G. Ortman, Susan M. Ryan, and KristinA. Kuckelman

2007 Population Dynamics among Salmon’s NorthernNeighbors in the Central Mesa Verde Region. In SalmonRuins: Chacoan Outlier and Thirteenth Century Puebloin Middle San Juan Region, edited by Paul F. Reed, Uni-versity of Utah Press, Salt Lake City, in press.

Wilcox, David R., and Jonathan Haas1994 The Scream of the Butterfly: Competition and Con-

flict in the Prehistoric Southwest. In Themes in SouthwestPrehistory, edited by G. J. Gumerman, pp. 211–238.School of American Research Press, Santa Fe.

Wilshusen, Richard H.1986 Excavations at Rio Vista Village (Site 5MT2182), a

Multicomponent Pueblo I Village. In Dolores Archaeo-logical Program: Anasazi Communities at Dolores: Mid-dle Canyon Area, edited by A. E. Kane and C. K. Robinson,pp. 210–658. Bureau of Reclamation, Engineering andResearch Center, Denver.

2002 Estimating Population in the Central Mesa VerdeRegion. In Seeking the Center Place: Archaeology andAncient Communities in the Mesa Verde Region, edited byM. D. Varien and R. H. Wilshusen, pp. 101–120. Univer-sity of Utah Press, Salt Lake City.

Wilshusen, Richard H., and Scott G. Ortman1999 Rethinking the Pueblo I Period in the San Juan

Drainage:Aggregation, Migration, and Cultural Diversity,Kiva 64:369–399.

Wilshusen, Richard H., and Ruth M. Van Dyke2006 Chaco’s Beginnings. In The Archaeology of Chaco

Canyon: An Eleventh Century Pueblo Regional Center,edited by Stephen H. Lekson, pp. 211–259. School ofAmerican Research Press, Santa Fe.

Wilshusen, Richard H., and C. Dean Wilson1995 Reformatting the Social Landscape in the Late Pueblo

I-Early Pueblo II Period: The Cedar Hill Data in RegionalContext. In The Cedar Hill Special Treatment Project:Late Pueblo I, Early Navajo, and Historic Occupations inNorthwestern New Mexico, compiled by Richard H.Wilshusen, pp. 43–80. La Plata Archaeological Consul-tants, Research Papers No. 1.

Windes, Thomas C.2005 Early Puebloan Occupations in the Chaco Region:

Excavations and Survey of Basketmaker III and Pueblo I


AQ72(2) varien 4/6/07 9:16 AM Page 298

Sites, Chaco Canyon, New Mexico, Volume I. Anthropol-ogy Projects, Cultural Resources Management, Inter-mountain Region Support Office, National Park Service,Department of the Interior, Santa Fe, New Mexico.

Windes, Thomas C., and Dabney Ford1996 The Chaco Wood Project: The Chronometric Reap-

praisal of Pueblo Bonito. American Antiquity 61:295–310.

Notes

1. We use the term “Pueblo” to refer to the AmericanIndian groups that occupied our study area during the timeperiod considered by this study (A.D. 600 to 1300). In usingthis term, we acknowledge the historical continuity betweenthe peoples who occupied our study area and the modernPueblo people who live today in New Mexico and Arizona.Although many different terms have been used to describe thepeople who inhabited our study area during the period ofinterest, our consultations with modern Pueblo groups haveestablished their preference for this term.

2. The number of years between the pith and the last ringproduced by a specimen does not yield the age of the treeunless the precise point where the meristem meets the roots issampled, something that is impossible to determine witharchaeological samples and something that is difficult todetermine even when sampling living trees (see Fritz et al.1965). The number of years between the pith and the last ringproduced by a specimen therefore gives a minimum age ofthe tree and the true age of the portion of the sample that wasanalyzed by the tree-ring laboratory.

3. These correction factors were derived in a semi-quantitative fashion using data on production from the high-elevation San Juan Basin Branch Station at Hesperus,Colorado, and observations on local dryland farming byexperienced local gardeners (and archaeologists, e.g.,Honeycutt 1995).

4. The laboratory of tree-ring research has developed anumber of suffix codes to distinguish various factors affectingthe interpretation of a date relative to the year of death of thetree from which the sample was taken. In this study we con-sider all dates except those associated with “vv” or “++” suf-fixes as cutting dates. The former case indicates that anunknown number of rings is missing from the outermost ringof the sample to the last growth ring of the tree, and the lattercase that the date of the outermost ring was estimated bycounting from the last datable ring. This latter situation isproblematic because certain species, especially juniper (themost common species used for construction in our studyarea), often put on extra rings or skip rings depending on soilmoisture conditions.

5. For the remainder of this paper we will use the termprobability density analysis as a gloss for the procedures pre-sented in Ortman et al. (2007). For those who do not wish toread the companion piece, our method formalizes and makesexplicit the reasoning process followed by fieldworkers whenthey record new sites on survey. It uses the known popula-tions, periods of occupation, pottery assemblages, and archi-tectural characteristics of excavated sites to quantify therelationship between surface evidence and occupational his-tory. These calibration data are then used as prior knowledge

in a Bayesian statistical framework to model the occupationalhistories of all recorded sites in the study area. There are ofcourse many additional details (see Ortman et al. 2007).

6. The Method 1 and Method 2 estimates were generatedin the following ways. For Method 1, we assume a 100 per-cent sample of community centers for the entire study area, a100 percent sample of habitation sites in areas covered byblock survey, and that the proportion of momentary house-holds in small sites within block surveys is representative ofthe missing proportion of households surrounding centersthat lie outside block survey areas. We calculate the momen-tary household estimates for centers and the proportion ofmomentary households in centers versus small sites in theareas covered by block survey, and then multiply the momen-tary household estimate for community centers by the inverseof this proportion to obtain the total momentary householdsfor the entire study area. Method 1 probably underestimatestotal momentary households because the proportion of house-holds in centers varies across block-survey areas, and severalof the largest surveys surround clusters of large communitycenters. The unsurveyed portion of the study area may there-fore have a higher proportion of small sites relative to centersthan is suggested by the block survey sample.

For Method 2, we divide the study area into four spatialquadrants and four elevational zones to create 16 discretestrata. We then determine the portion of block survey withineach stratum (the sampling fraction), the momentary house-hold density in block survey areas in each stratum, and mul-tiply the population density by the inverse of the samplingfraction to estimate the population for each stratum. Finally,we sum the stratum estimates to estimate the total populationfor the study area. Method 2 produces the highest estimates,and may overestimate total momentary households somewhatbecause it assumes that the block survey sample in each stra-tum is representative of the population density of the entirestratum. This approach neglects the distribution of other crit-ical resources, such as fresh water from springs, and does notreflect the known distribution of community centers verywell.

7. We recognize that this is a very conservative estimate.Richerson et al. (2001:396–397), for example, consider .01 (1percent per year) to be a conservative estimate for the obtain-able r (the intrinsic rate of natural increase).

8. Note that neither set of lines tracks year-to-year vari-ability; period means are joined by straight lines.

9. This was calculated by taking the total households insmall sites in block surveys / surveyed portion, and addingthis to the total households in all community centers to gettotal households by period in the study area. These estimateswere multiplied by 6 to get the total persons by period. Thisfigure was multiplied by 50/modeling period length (accountfor lifespan of a person and length of modeling period, as wedo for calculating momentary households). The results foreach period were summed to get a grand total.

10. It also appears that population decreases in smallercenters before it decreases in larger centers (Glowacki 2006).

Received October 28, 2005; Revised June 28, 2006;Accepted June 28, 2006.


AQ72(2) varien 4/6/07 9:16 AM Page 299

AQ72(2) varien 4/6/07 9:16 AM Page 300

Historical Ecology in the Mesa Verde Region: Results from the Village Ecodynamics Project

Documents