Main page

Tiwanaku research page

 Andean Research




This research studied changes in land use and settlement patterns in Cochabamba, in the Bolivian south central Andes, in the transition from the Early Intermediate Period (ca. A.D. 200-500) to the Middle Horizon, a period marked by the ubiquity of Tiwanaku style materials. It has been stated that in this latter period Cochabamba is colonized by the Tiwanaku polity for agricultural purposes. This observation is tested here by hypothesizing a denser Middle Horizon occupation in areas with higher agricultural production in Cochabamba. Survey data is evaluated against land use and settlement patterns expected from four models of interregional interaction. The results show that there is no change in land use strategies and only minor shifts occur in settlement distribution during the Middle Horizon. This data leads me to suggest that indigenous agricultural strategies remained dominant in structuring settlement and subsistence strategies during the Middle Horizon, despite interaction with the Tiwanaku polity.


La interacción regional de la entidad política Tiwanaku durante el Horizonte Medio (500-1000 d.C.) es esencial para la comprensión de la expansión territorial de estados prehispánicos en los Andes. En esta investigación se ha estudiado cambios en los patrones de asentamiento y el uso de tierras en Cochabamba, en los Andes centro-sur de Bolivia, en la transición del Periodo Intermedio Temprano (200-500 d.C.) al Horizonte Medio, un periodo caracterizado por la presencia de cerámica de estilo Tiwanaku. Se ha afirmado que en este último periodo Cochabamba es colonizada por la entidad política Tiwanaku para explotar recursos agrícolas de zonas mesotérmicas. Esta observación es analizada en este artículo con una hipótesis en la que se espera que una ocupación humana más densa durante el Horizonte Medio ocurra en areas con mayor productividad agrícola de Cochabamba. La información de la prospección es evaluada frente a los patrones de asentamiento y de uso de tierras que se espera de cada uno de los cuatro modelos de interacción política propuestos: subordinación política, verticalidad, economía de bienes de prestigio, e independencia. La ausencia de cambios en las estrategias de uso de tierras y variaciones menores en los patrones de asentamiento del Horizonte Medio no pueden ser sugeridos como expresión de expansión territorial de Tiwanaku; mas aun, interacción con Tiwanaku no genera ningún cambio que se asemeje a estrategias de intensificación agrícola. Se postula el modelo de independecia de las poblaciones locales para entender este proceso de interacción. Ello implica que procesos de organización locales fueron más importantes en estructurar estrategias de asentamiento y explotación agrícola que la influencia externa del altiplano. Este estudio sugiere un nuevo tipo de interacción regional que amplia nuestros conocimientos de las relaciones territoriales de la entidad política Tiwanaku y de sociedades expansionistas de los Andes.

Ver Investigaciones en Cochabamba: Asentamientos Humanos y Uso de Tierras en los Valles de Cochabamba, Bolivia para otra versión de esta investigación.

Introduction to the research

The Tiwanaku polity is considered a dominant political entity in the south central Andes, that evolved in the southern Lake Titicaca basin starting ca. A.D. 400-500 (Kolata 1993a; Ponce Sanginés 1972). Its "dominance" is revealed by a broad regional distribution of Tiwanaku style pottery and other artifacts. As part of this pattern, the Cochabamba region has recorded a very dense concentration of Tiwanaku style pottery.

The Cochabamba region, about 600 km southeast of Lake Titicaca, has been suspected to be a hinterland exploited and dominated by the Tiwanaku polity (Kolata 1993a, 1993b). However, this has been assumed rather than archaeologically tested. Cochabamba is the most productive region in the eastern slopes of the Andes, and as such was a prime area for agricultural exploitation in prehistory, as confirmed by the region's later Inka occupation (Wachtel 1982). Because of its agricultural features, this region is ideal to analyze strategies deployed by expansive prehistoric polities both Andean and World-wide to interact with populations in other territories and extract resources from them (Algaze 1993; Sinopoli 1994).

Two issues are evaluated in this study: (1) the spatial shifts in settlement location and land use preferences consequence of the relationships of the Cochabamba populations with the Tiwanaku polity; and (2) the role of pre-Middle Horizon patterns in shaping Middle Horizon agricultural strategies. The archaeological evidence used to explore these issues is settlement location and land use patterns. This evidence will produce a sequence of shifts in relationships between human occupation and agricultural strategies. This is the data used to interpret Tiwanaku-Cochabamba interaction in the context of four suggested models of interregional interaction.

Tiwanaku style pottery distribution is viewed by some researchers as proof of the presence of highland populations in Cochabamba (Kolata 1993a:269; Owen 1994; Ponce Sanginés 1972). Despite this latter view, it is useful to explore other models that include causality of local populations as well as highland presence in the analysis. At the same time, the use of alternative lines of archeological evidence such as the spatial evidence explored in this research, can reveal patterns that contrast with patterns shown by pottery alone (cf. Flannery 1972). Therefore, a pots-equal-political economy axiom joins the long debated pots-equal-people assumption.

Tiwanaku in Cochabamba

The presence of Tiwanaku style ceramics in Cochabamba is long known (Bennett 1936; Byrne 1984; Rydén 1954, 1959). Kolata (1993a:269) writes that Cochabamba was "the focus of intense Tiwanaku colonization," and that, "Tiwanaku directly colonized and subsequently controlled key economic resources in lower-lying regions, such as the Cochabamba Valley." (1992:80). This latter interpretation is tested in this study by analyzing its land use and settlement distribution implications. Browman, who has widely written on the political economy of the Tiwanaku polity, suggests the demise of Tiwanaku-Cochabamba relationships coinciding with the collapse of the Wari polity to the north (ca.800-900; Browman 1980:109). Cochabamba had been integrated into a Tiwanaku "federation", but detaches to develop its own resources, with no Tiwanaku traders as middlemen. Tiwanaku then adopts the Archipelago extraction strategy in Moquegua to compensate loss of rich Cochabamba (Browman 1978:332; 1985). However, Browman (1997: 233) will suggest all along the sequence, Tiwanaku interaction with Cochabamba will be made through trade and a prestige goo economy, and not with colonization or conquest strategies. The rich pre-Tiwanaku social developments in Cochabamba would have been a factor in the type of strategies implemented by highland populations. Bermann (1997:108) also suggests an early trade link with Cochabamba, that "may have been rooted in interaction between local elites and Tiwanaku rulers and manifested in local adoption of Tiwanaku-style goods".

The chronological sequences used here are based on data produced by other researchers in several stratigraphic cuts from the Central Valley and Mizque (Anderson and Céspedes 1994; Brockington et al. 1987). Therefore, only the Capinota sequence is extrapolated here from the nearby Central Valley. Data from the Santivañez Valley mimics the sequence from the Central Valley (Vetters, pers. comm.), suggesting that the sequence adopted for Capinota is reliable. The sequences were used to define temporal occupation of the recorded settlements in five broad temporal units, based on the surface material.

Research objectives

Kolata (1993a, 1993b) stated that Tiwanaku colonization of Cochabamba was aimed at acquiring agricultural resources such as maize. Therefore, one factor for differences in the density of Middle Horizon settlement occupation within Cochabamba would be agricultural potential. If this is true, the location of Middle Horizon occupation, bearing Tiwanaku style materials, should correlate with areas of higher agricultural productivity in the region. In addition, within each area where occupation of this period is found, settlements should correlate with soils of higher agricultural potential. At this latter smaller spatial scale, there should be, if not a one-to-one correlation of best soils and individual settlement location, at least a close spatial correlation between both variables. Thus, settlement near good soils is as important as settlement on good soils (Hastorf 1993:140; Stanish 1994). However, site location patterns at any given period is also measured in comparison to a previous period. In this case, Middle Horizon settlement distribution on or near good soils would be less significant, and would fail to show agricultural intensification, if it repeats location patterns (i.e., no site shifts) or occupation size (i.e., no occupation size growth) recorded in the Early Intermediate Period. Shifts in agricultural patterns have been documented in the Juli-Pomata region where Stanish (1994:table 1) recorded agricultural intensification in the Middle Horizon by settlement shifts, and increase in occupation both on and close to the most productive soils.

Following the hypothesis, agricultural potential is tested as one factor for shaping Middle Horizon occupation in the region. However, other factors, such as political strategies, pre-existing patterns of local development, or distance from the altiplano, are known to shape spatial distribution of settlements and overall occupation of a region. In effect, regional data on settlement sequences (Hastorf 1993; Schreiber 1992) or data allowing comparison of strategies by different large-scale polities (Browman 1970; D'Altroy 1992; Earle et al. 1987; Hastorf 1993) show that agricultural factors alone are less important to define strategies to occupy a region. For example, in the agriculturally rich Mantaro Valley, the Middle Horizon Wari occupation (Browman 1970) and the later Inka occupation (D'Altroy 1992; Hastorf 1993) occur in mutually exclusive zones of the valley. In the Inka case, productivity of the Yanamarca Valley seems to be a deciding factor for concentrating its occupation, given that Late Intermediate Period settlements occur in the whole valley (Browman 1970:map 11; Hastorf 1993). In contrast, the first phase of Middle Horizon settlement concentrates in the southern end of the valley where is located the densest Early Intermediate Period 3 occupation (Browman 1970:map 8). In a contrary case, despite wide-scale Middle Horizon agricultural intensification in the Carhuarazo Valley, Inka occupation is far less important (Schreiber 1992).

Bearing this in mind, the purpose of this research is to explore specifically patterns related to agricultural factors. The correlates for the four models proposed below, which represent different political and economic motives, emphasize archaeological evidence of settlement location and land use.

Areas of research

A survey program was set in the Capinota Valley and the Mizque Valley in the Cochabamba region to test the hypothesis of Middle Horizon focusing settlement on the most productive lands. These areas were chosen for their differences in ecological conditions and in agricultural potential, their size, by which the survey areas could encompass a significant portion of the drainage, the documented surface evidence for Early Intermediate Period and Middle Horizon occupation, and their accessibility for study because of their non-urban setting and low population density.

The Mizque Valley has a higher agricultural potential than Capinota. Both valleys have similar precipitation and temperature patterns, with 600 mm per year and 17° annual average, respectively. However, they differ in the amount of good alluvial soils available for agriculture, in the quality of soils within the piedmont zone, and, more importantly, in the availability of water year-round (e.g., note five rivers flowing into Mizque's main valley. In sum, Mizque has a larger total of all three variables. Given these ecological conditions, a higher density of Middle Horizon occupation is expected in the Mizque area, despite Mizque's greater distance from the altiplano.

For the analysis of land use, the survey areas were divided in three zones of good, medium and poor soils for agricultural use. These are characterized as "water-table" and flooding land, marginal flooding and canal-irrigable land, and dry farming to grazing lands, respectively (Nicholas 1989). The three zones were collapsed from FAO's seven-category division, which is based principally on water availability, and other factors such as salinity, acidity, drainage, and slope (CIDRE 1987, 1988). Therefore, current soil productivity data is used to provide a relative measure of prehistoric soil quality (Feinman and Nicholas 1990; Hastorf 1993; Kirkby 1973; Nicholas 1989). If used to produce relative estimates with error ranges this data is very useful for assessing prehistoric productivity. Additionally, the spatial analysis explored site location patterns with respect to topography. Both areas were divided, then, based on elevation and vegetation features, into alluvial plain, piedmont, and mountain zones.

Models for interregional interaction

Settlement and land use patterns are tested against four models of interregional interaction. The first model is political subordination, and implies direct control of the region by highland populations (D'Altroy 1992; Hassig 1985). The second is exploitation under a verticality system on a large scale, or archipelago (Mujica 1985; Murra 1972, 1985a; Salomon 1985). The third model is one of a prestige-good economy that generates an increase in the economic complexity of local groups (Schortmann 1989; Helms 1979). Finally, the last model is one of independence, in which very little change occurs in the agricultural strategies of the Middle Horizon with respect to the Early Intermediate Period.

These models are not all new in the Andes, and have been proposed for the interaction of other large-scale polities with provincial populations (D'Altroy 1992; Owen 1994; Schreiber 1987, 1992). Another important contribution is the "nested hierarchy" model to reassesses political "centralization" of Tiwanaku's core area rather as a set of local independent polities in the core (Albarracín-Jordan 1992; McAndrews et al. 1997). This model is not included in this research, since no data was produced to build the settlement hierarchy needed. However, it has provoking implications for a more detailed understanding of regional interaction by the Tiwanaku "state", or rather, for interaction managed by the local political units of the core area.

Only recently are local conditions and local populations part of the equation in a relationship that was conceived of as a one-way affair (Bermann 1993, 1994; Costin and Earle 1989; D'Altroy and Earle 1985; Graffam 1992; Hastorf 1991, 1993; Janusek 1994; Morris 1972; Schreiber 1992). This perspective allows the possibility that local developments could have played a primary role in regional organization, and could have been, in part, responsible for the outcome of the process of interaction (Bermann 1994). Thus, avoiding interpretation of recorded patterns solely in function of the politics of the dominant polity. In this research, the first two models consider a causality effect emanating from interaction with the Tiwanaku polity, and two of the models proposed, prestige-good and independence models, emphasize patterns that could have been generated by a degree of local organization.

Each model has specific correlates for the following variables: (1) regional distribution of settlements (i.e., to define significant differences in the settlement occupation size between both areas in the Early Intermediate Period and Middle Horizon); (2) shifts in settlement location between both periods in each area; (3) shifts in settlement occupation size within each area; and (4) settlement-soil relationships (i.e., establishing preferences for settling certain soil or topographic zones within each survey area).

Modeling settlement and land use correlates attempts to produce clear-cut differences for patterns of each model. However, data must be evaluated against the set of correlates for the four variables of each model, and not against single variables. There are similarities in the correlates for a single variable between two models. For example, a same correlate of no shifts in settlement location within a settled area for both the prestige-good economy model and the independence model. However, in the former model growth in occupation size would be generated by a few settlements, reflecting a trend towards complexity of economic organization.

A single model might be appropriate for "explaining" land use and settlement patterns for both survey areas. Alternatively, as originally predicted, if Tiwanaku relationships in Mizque were greater than in Capinota, or of a special character, explanation might require different models for each survey area. Models can be neatly defined on paper. However, there is a fine line between them depending on the quality of the archaeological evidence obtained, and variation in the features of political organization. D'Altroy (1992) has indicated a broad range of variation in political organization oscillating between the opposing direct control and indirect control strategies. Yet, building models are a useful tool for producing comparative studies in archaeology.

Political subordination model

The political subordination model expects incorporation of Cochabamba populations into the Tiwanaku realm through direct political and territorial control. Mechanisms for territorial incorporation in the Andes range from strategies of military coercion to reciprocity forms (Morris 1985), or most certainly, in large-scale territorial states, a combination of these latter two strategies.

A key feature in territorial control is the implementation of a strategy of high resource extraction (e.g., intensification of agricultural production; D'Altroy 1987:6; Hassig 1985). This should be accompanied by implementing the core's administration features, detectable in architecture, into the subjugated area (Goldstein 1989; 1993; Schreiber 1992: 28). This model should result in a sharp reorganization of settlements to facilitate resource extraction, and to break up, or use, local political structures (Hastorf 1993; Schreiber 1992). Although settlement will also be shaped by factors other than the state's concerns of extraction, it is expected that direct control would be followed by evidence of increased surplus mobilization, through the spread of agricultural technologies or the movement of populations to areas with higher potential for surplus extraction.

Archaeological correlates for this model are: (1) shifts in land use and settlement patterns to concentrate occupation on a restricted but very -or most- productive land in the region; (2) a reorganization in site distribution in the settled area, possibly with identification of sites bearing features of Tiwanaku style public architecture (e.g., sunken temples); (3) a significant growth in the size of occupation with respect to the previous period in the settled area; and (4) new sites within the area are not necessarily on the best soils, since the model implies a total control of the area.

Judging by archaeological cases recorded in the Andes, such an extreme form of political interaction will be readily visible archaeologically. An example is the Wari conquest of the Carhuarazo Valley, and the accompanying settlement shifts in the Willka phase, when much of the valley was terraced (Schreiber 1992:260). The state-directed emphasis on maize production led to the movement of villages from higher elevations to lower elevations and the kichwa zone. Similar disjunctions in settlement could be described for the Inka conquest of the Mantaro Valley and Cochabamba (D'Altroy 1992; Hastorf 1991; Wachtel 1982).

Verticality Model

Vertical exploitation is a key model in studies of interregional interaction in Andean prehistory (Conrad and Rice 1989; Dillehay 1979; Hastings 1987; Raymond 1992; Stanish 1989, 1992; Van Buren 1996). The strategy of verticality has been documented ethnohistorically in highland populations as a mechanism for obtaining lowland resources (Murra 1972). This subsistence strategy consists in the "simultaneous control by a single ethnic group of several geographically dispersed ecological tiers" (Murra 1985a:3). This model is defined at different spatial and political scales (Murra 1972, 1985b), as well as different economic features (Salomon 1985).

The regional scale of the interaction process studied here requires consideration of the archipelago strategy, instead of the other variants of single-valley scale verticality (Murra 1972). There are three important differences in the present regional analysis of verticality for the Cochabamba region. First, verticality is evaluated at a regional scale combined with three other models and not alone. A regional approach to verticality was made in the Upper Moquegua Valley for the Late Intermediate Period (Stanish 1989, 1992). However, for the Middle Moquegua Valley, Middle Horizon interpretation of vertical organization was based only on data from the sites of Omo and Chen Chen (Goldstein 1989). Second, this research explores land use, occupation size shifts, and spatial distribution of settlements as alternative lines of evidence to pottery and architecture (Marcus and Silva 1988; Stanish 1989). The evidence used here explores the essence of the concept of verticality by addressing agricultural productivity, crucial to the functioning of the strategy. Finally, the high agricultural capacity and less land circumscription make of Cochabamba a drastically more productive ecological setting than the regions where ethnohistoric cases for archipelagos were made (Murra 1972). Therefore, different settlement strategies adopted by foreign colonial populations in Cochabamba should make a new case study for the variability in archipelago verticality.

Stanish (1992:43) distinguishes several scales of archipelago colonies. The largest scale is one in which different regions --even small individual valleys within a larger drainage such as Cochabamba-- are controlled by different ethnic groups, leading to a pattern of differences among settlement systems. At a smaller scale, ethnically distinct colonies will be visible in a defined area. And at the smallest scale, "multiethnicity [...] within an individual settlement," would be characterized by distinct and discrete barrios (Stanish 1992:44). Architectural styles and pottery assemblages were used by Stanish to define distinct settlements in a region or discrete barrios in a settlement. In this research, a new consistent set of correlates to identify "ethnic" heterogeneity (Stanish 1992:45), in the form of agricultural strategies, is made with land use, occupation size shifts, and spatial distribution of settlements.

The vertical archipelago strategy was expected to generate a distribution of productive enclaves, especially in, but not limited to, Mizque, the area with the highest agricultural potential. The spatial patterns for expected in this situation are: (1) occupation by new populations of several areas, expecting a larger occupation size in richer regions; (2) no shifts in the location of local settlements, but new locations chosen by new foreign settlements. If shared, settlements would stay in the location of local sites; (3) growth in occupation size resulting from the introduction of new foreign sites, whereas size of local sites remains constant. There is an overall growth in occupation size in both areas; and (4) new foreign sites should be concentrated on or near the most productive soils, with or without local sites on the same soils.

Highland groups would have sought to establish colonies in the areas of highest agricultural potential. Naturally, this might have been difficult under certain local conditions. The "multiethnic" settlements documented for some archipelago systems represent one cooperative solution for sharing access to desired lands. A pattern in which sites with Middle Horizon were independent and limited to lesser quality soils might indicate that highland populations were unable to set up colonies on the richest soils.

In the few cases where a Tiwanaku colonial strategy has been reconstructed, the inference has been that Tiwanaku colonists were, in fact, able to gain residence on the best lands. In the Moquegua Valley, the large Tiwanaku site of Omo and subsidiary communities are overlooking the best agricultural lands in the valley (Goldstein 1989:238). Little previous local settlement is signaled for the same area and there is, up to now, limited evidence of other Tiwanaku sites in the surrounding areas. A similar argument has been advanced for the Azapa Valley, where Tiwanaku colonies represented a marked shift in settlement location. "Es claro," wrote Mujica et al. (1983:103), "que la población Tiwanaku inicia la explotación de microzonas anteriormente no utilizadas por los pobladores locales como especialmente las partes medias de los valles." This is not the case in Cochabamba. There is local human occupation settled in both areas during the Early Intermediate Period.

Prestige-good economy model

Increasing political complexity in peripheral societies has been observed to be a common consequence of interaction with more complex polities. The increase in complexity takes the form of the emergence of an elite stratum or, if this stratum already exists, further empowerment of elites through an increase of status through the operation of a prestige-good economy (D'Altroy 1987; Gledhill 1988; McGuire 1989; Paynter 1981). This model would imply Middle Horizon interaction with highland populations (Browman 1984b).

In prestige-good economies, non-local items are critical to establishing, maintaining, or enhancing power relationships within a population (Costin and Earle 1989; Helms 1979; Oakland 1993; Paynter 1981; Schortmann 1989). Such goods are typically exotic materials, products of craft specialization, and may bear esoteric decoration. This is a model, then, that relies principally on the identification of "exotic" goods amid local or standard artifact assemblages. Because of the dominance and widespread distribution of Tiwanaku style materials shadowing local styles, there is nothing exotic to this assemblage in the region (cf. Alconini 1993). Granted, pottery is the most common material documented until now bearing Tiwanaku style in the region. However, Oakland (1985) analyzed textile samples from Cochabamba and suggested their highland origin.

This model explains, based on several artifacts types present, but pottery, the interaction between Tiwanaku and San Pedro de Atacama (Browman 1984b; Nuñez and Dillehay 1979; Oakland 1992). This process is then possible between Tiwanaku and Cochabamba. Furthermore, it could be identified with spatial data given that in many complex societies, the power and privilege of elites is based on domination of economic processes, including agricultural production (Earle 1991). A set of correlates with the spatial evidence emphasized in this research is presented for this model.

This economic organization could be identified through the following features of land use and settlement patterns: (1) occupation of several areas with little attention to soil productivity; (2) no shifts in settlement location; (3) growth in occupation size of local settlements occurs in a few settlements. Overall, we might expect growth in settled areas, although a less pronounced growth that the one implied by foundation of sites or site enlargement in "multiethnic" settlements of the archipelago model; and (4) sites are not necessarily located on the most productive soils since control of each area would be achieved by paramount sites (which in this research are not identified in a hierarchy of settlements). The overall site distribution would follow previous Early Intermediate Period distribution.

Independence model

The final model proposed is, simply put, that the nature of interaction between Tiwanaku and the Cochabamba populations may not have been such as to lead to shifts in subsistence strategies or settlement patterns, shifts that were aimed, in the Middle Horizon, at intensifying agricultural production . It is very possible that this "non-result" may be manifesting more complex and fine-grained relationships between Andean communities, such as interaction managed by the smaller political unities of the nested hierarchies model (Albarracín-Jordan 1996; cf. Mayer 1985; Salomon 1985). The term "independence" is chosen in contrast to an autochthonous model proposed by Saignes (1986) for ethnohistorically-documented relationships between economic groups and the organization of vertical space among communities.

In this model, Middle Horizon settlement generated little changes in land use strategies or shifts in settlement location. The archaeological correlates of this model are: (1) occupation of several areas, mostly the same as in the Early Intermediate Period, and with no particular correlation with the most productive zones; (2) no shifts in settlement distribution, as site location follows previous patterns; (3) no significant growth in site occupation size in comparison to the Early Intermediate Period; and (4) no change in soil location or in soil preference, if this latter pattern occurs in the previous period Early Intermediate Period.

Research design

Regional settlement survey is recognized as a powerful tool for investigating how humans settled, exploited, and managed a region (Blanton et al. 1981, 1982; Drennan et al. 1985, 1991; Johnson 1977; Kowalewski 1990; Parsons 1972; Paynter 1983; Wilson 1988). Survey data was critical for evaluating the archaeological correlates described for the models in the previous section. This research adopted a random sample survey strategy (Nance 1983; Read 1986; Whalen 1990). A sampling strategy permitted a coverage of a large area and a record of comparable data sets from two survey areas. A comparative analysis of two areas was needed to define differences in human occupation depending on agricultural potential in the Middle Horizon in Cochabamba. Therefore, patterns of settlement hierarchy for these models (Johnson 1981), obtained with so-called "full-coverage" surveys (Kowalewski 1990), are not evaluated here.

Two survey areas of 200 km² were set in Capinota and in Mizque. A survey area this size encompassed in both valleys all three type of soil (i.e., good, medium and poor soils), and all three topographic zones (i.e., alluvial, piedmont, and mountain zones). Despite the general expectations for settlement location (e.g., sites are likely to occur along water sources like rivers or seasonally watered quebradas) the survey area included by design lands outside the alluvial zone to provide a complete picture of regional settlement distribution.

A grid oriented north-south divided each survey area into a 250x250 m grid or 6.25 ha. square units. The size of the survey quadrat was defined after a preliminary assessment of average site size in the survey areas. Every type of site was recorded, such as occupation sites, cemeteries, isolated burial sites, and agricultural works associated to a site. Size of settlement was defined by surface artifact scatter and architectural and construction features. Total size of all sites found in the quadrat sample was recorded even if a site exceeded the arbitrary boundaries of the quadrat. However, only the area of the fraction of a site within quadrat boundaries (i.e., .35 ha that corresponds to one-third of the site within the quadrat) is used for the aggregate figure of occupation size by period, and, to estimate total occupation size by period in each survey area (Nance 1983). Occupation size in hectares by period is more meaningful than a simple count of sites for our purpose of comparing settlement sequences. The sum of site fractions within quadrats produces is the statistical number of sites, usually smaller than the actual number of settlements recorded.

The number of quadrats to be surveyed in each area was calculated based on an preliminary assessment of three sites per km², with an estimated standard deviation 1.5 sites, and a desired error range of not more than .5 sites at the 99% confidence level. This produced a figure of 60 sites needed in the sample. Therefore, with an mean of .1875 site per survey quadrat (60 sites divided by 3200 quadrats in a survey area), 318.5 survey quadrats, or 320 by rounding up, needed to be surveyed. Oddly enough, this made a total area of 20 km² surveyed, or 10.0%, of the survey area. This figure of 10% surveyed stands in contrast to the same proportion often taken, with no clear reasoning, as a "good" sample to survey.

Settlement preferences are determined by comparing percentages of human occupation in each soil and topographic zone (i.e., 12% of the alluvial zone is occupied), rather than using the distribution of human occupation in, say, each soil zone (i.e., 27% of the occupation is on good soils). The percentage of occupation is calculated by dividing the estimate of total occupation size in each zone by the total size of a zone in the survey area (and multiplying by 100). This relative figure is needed because of the different size each soil and topographic zone have in each survey area (i.e., a 23% and 45% occupation of the alluvial plain zone in Capinota and Mizque, respectively, is a meaningful difference only if this topographic zone has the same size in both areas. Since it does not, I rather compare a 2% and 3% of the same topographic zone occupied in each survey area, indicating very similar use of the alluvial zone). This procedure is also used for comparisons of occupation within each survey area (e.g., comparing percentage of occupation of alluvial plain, piedmont, and mountain). A 95% confidence level error range is attached to the estimates of occupation size by period, and to the percentage of occupation of a single soil or topographic zone to compare the patterns between and within survey areas.

Preference for a particular zone would be determined by (1) a larger percentage of occupation in one particular zone, and (2) a 95% confidence level error range for that same zone that did not overlap with the lower occupation percentage means of the two zones. Comparison of means and error ranges for a single variable in a temporal sequence is useful to measure significant or "abrupt", or non-significant differences through time. Concluding abrupt or subtle changes or continuity from the data at any confidence level is, however, a decision of the researcher (cf. Bermann 1997).

@The field survey produced different results than expected. Mainly, the expected statistical total of 60 sites to be recorded was not attained; in other words, I overestimated the figures used for setting the number of quadrats to survey, and a larger sample would have been needed. Therefore, statistical estimates on the collected data are made at a 95% confidence level, instead of the 99% level set at the start. For example, 49 sites were recorded in the Capinota survey area, which correspond, adding up site fractions within quadrats, to a total of 35.04 sites. For the Mizque survey area, a total of 40 sites, made 18.22 sites for the statistical analysis, mostly because site sizes in this area were much larger, and hence a smaller fraction of the site was recorded in a quadrat.

Despite this shortcoming, the survey data has provided systematic and comparable evidence to understand the spatial shifts in settlement location and land use during the Middle Horizon in two areas of the Cochabamba region. The data in the following section describes the recorded settlement data, and its relationship to soil and topography factors.

The Capinota sequence

The survey recorded 13 settlements occupied in the Early Intermediate Period in Capinota, and a statistical number of 6.98 sites (see Research strategy), with an occupation size of 17.7 ha. The sample obtained yields an estimate of 157.7±95.4 ha of total occupation size for the period in the survey area. This period in Capinota is marked by settlements with the local Tupuraya and Mojocoya pottery styles. Tupuraya materials dominate in 97% of the settlement size recorded for this period. The densest occupation for this period is at sites 10, 16, 43, 44 and 48 (Figure), which total 9.27 ha.

Settlement in this period often occurs in multi-component sites. Site 10, Sites 16 (with Tupuraya material ranging from 6% to 48% in the collections) and 44 (ranging from 9% to 38%), and the mainly Tiwanaku burial site 42 (with Tupuraya percentages from 4% to 14%), are sites with both Tupuraya and Tiwanaku style materials. In other words, two of the largest Tupuraya-bearing sites, sites 16 and 44, will have Middle Horizon settlement. Tiwanaku style materials, however, will not be limited to "Tupuraya" sites and will occur exclusively at two new sites in the following period.

Early Intermediate Period occupation is concentrated on the least productive or poor soils (61% of the occupation size), followed by occupation of medium and good soils (21.5% and 17% of the total occupation size, respectively; Table 1). Comparison of the percentages of occupation of each zone reveals no significant difference, at the 95% confidence level, in the occupation of the three soil zones (Graph). Note that the error range for medium soils overlaps the mean percentage for the two other soil zones. Therefore, preference cannot be suggested for the occupation of any of the three soil zones in the Early Intermediate Period.

More than half of Early Intermediate Period settlement, or 54% of the total occupation size, is in the piedmont zone. The occupation size of the alluvial zone is very similar to the mountain zone occupation, with 22% and 24%, respectively (Table 2). Comparison between the percentage occupied in each topographic zone indicates no difference, at the 95% confidence level, between the occupation of the three zones (Graph). In effect, the error range of the occupation percentage of the alluvial zone overlaps the percentage of piedmont and mountain zones. No preference can be suggested, therefore, for any topographic zone in the settlement of the Early Intermediate Period.

In the Capinota survey area, unlike the Mizque survey area, the piedmont zone is dominated by poor soils. Therefore, settlement in the piedmont here entailed occupation of the least productive soils. Yet, neither the bulk of the settlement occupation size in the piedmont zone or on poor soils reflects a settlement preference in the Early Intermediate Period.

Twelve Middle Horizon settlements were recorded in Capinota, totaling a statistical number of 7.9 sites, with an occupation size of 17.3 ha. The total occupation size for the period is estimated at 159.9±99.7 ha. Occupation of the Middle Horizon is marked by Tiwanaku style pottery, and the local Omereque, and Gray Ware style assemblages ( Figure; Rydén 1959). Omereque and Gray Ware materials occur in the same sites as the Tiwanaku style; only the latter material is found alone. Human occupation bearing Tiwanaku style material was recorded in all 12 Middle Horizon sites, where this style dominates with more than 90% of the pottery collected. Low percentages of Omereque style material were recorded at three sites (with an average of 3.5%), and Gray Ware material is found at four sites ranging from 10 to 20%. These two styles co-occur with Tiwanaku style material in sites with burial evidence.

Tiwanaku style materials are concentrated at seven principal sites (7, 10, 11, 16, 40, 42 and 44; Figure) representing 74% of the Middle Horizon occupation size. Site 11, a meseta-like site overlooking a river bed, and Site 7, an isolated small burial site, are the only two sites with exclusive Tiwanaku style occupation. Site 10, a low mound on the alluvial plain resembling highland Tiwanaku mounds, had Early Intermediate Period occupation.

Tiwanaku style remains are highly dominant at site 42, a large cemetery, ranging from 27% to 75% of the total assemblage. Most of the rest of the collections is made of Gray Ware and large Sauces-type urns (Ibarra Grasso 1965) used for burial. Similarly, at site 44, which had an important Tupuraya style occupation, Tiwanaku style materials make up 33% to 85% of the collections. Sites 11, 16, and 44 are the largest domestic settlements with Tiwanaku style materials.

Middle Horizon settlement is split between poor soils, with 77.5%, and medium soils, with 17.75% of the total 17.3 ha of occupation. The additional .83 ha on good soils represents Site 10 (Table 1). The differences between the percentage of occupation for the estimated totals of each soil zone are not significant at the 95% confidence level (Graph). Therefore, no preference for settling any soil zone is indicated.

The five principal Tiwanaku occupations are distributed in three groups: Site 10 is located in the good soils zone, Site 42 is located in the lower piedmont zone in medium soils, and finally, Sites 7, 11, 44 and 16 are located on the poor soils zone. Sites of this latter group are adjacent to the alluvial plain and residents would have had easy access to richer soils. The advantage is clear here of being near good soils, except for Site 11 where access to good soils meant crossing the Tapacari River. However, this is a pattern of site location that occurred already in the Early Intermediate Period.

Most Middle Horizon occupation is in the piedmont zone, with 88% of the total 17.3 ha of occupation (Table 2). Comparison of the occupation percentage estimates reveals a significant difference, at the 95% confidence level, in occupation of the piedmont zone ( Graph). Thus, a preference for the piedmont is documented for the Middle Horizon, where are located four of the five principal sites of this period. Only Site 10, the platform mound site, is located on the alluvial plain. All the piedmont locations are adjacent to the alluvial plain zone, except for Site 11.

In sum, the land use and settlement patterns in the Capinota survey area failed to meet our expectations of finding a preference for settlement on the best soils in the Middle Horizon. And in the transition to the Middle Horizon, there is almost no change in the settlement occupation size, despite two new Middle Horizon sites, as was expected if residents were maximizing maize exploitation. This might suggest that Early Intermediate Period and, more importantly, Middle Horizon inhabitants were uninterested in expanding productive capacity. The preference for settling the piedmont zone in the Middle Horizon, which follows a large but not preferential piedmont occupation in the previous period, is not tantamount to settlement on the best soils. Hence, the pattern of the Middle Horizon does not represent a drastic shift with respect to the Early Intermediate Period. But location near good soils would not have precluded ready access to good agricultural plots. The strategy of settling in piedmont, is on the other hand, a way to avoid extensive settlement flooding. Even supposing that the observed location of settlements near good soils in Capinota validates one of the expected relationships between settlement and soils, comparison of other variables between the Early Intermediate Period and the Middle Horizon (e.g., no growth in occupation size) indicate another scenario. These results will be discussed after describing the Mizque sequence.

Mizque Sequence

A total of 14 sites where recorded for the Early Intermediate Period in Mizque, and a statistical number of 5.15 sites with an occupation size of 53.8 ha. The total estimated occupation size based on the sample is 487.2±324.4 ha for this period. The Early Intermediate Period pottery assemblage is dominated by the Tupuraya style, followed by Sauces, Gray Ware, and Mojocoya styles (Ibarra Grasso 1965; Walter 1966). Tupuraya pottery is found in 62% of the occupation size recorded for the period. These styles, however, co-occur in the same settlements. For example, Grey Ware pottery is usually found with Tupuraya material, Tupuraya and Sauces co-occur in five sites, and Mojocoya is found in two sites with Sauces and Tupuraya material (Figure).

In Mizque, settlement on good soils is only slightly larger that settlement on poor soils; occupation on both zones is 82% of the total settlement size (Table 3). The percentages of occupation of each soil zone show no significant differences at the 95% confidence level in this period (Graph). Therefore, the data show no preference for settlement on any soil zone. The occupation size for good soils is heavily influenced by the size of sites 13 and 14, which make up roughly 40% of the occupation on this soil zone. Settlements in less productive soil zones show evidence of agricultural intensification, such as site 34, located on medium soils of the piedmont zone.

Settlement on the piedmont zone amounts to 44.7 ha, or 83% of the total occupation, whereas occupation of the alluvial and mountain zones is lower, with 15% and 2% of the total occupation, respectively (Table 4). Comparing the percentages of occupation of these zones reveals a significantly larger occupation of the piedmont zone (at the 95% confidence level) suggesting a preference for settlement in that topographic zone (Graph).

Preference for settling the piedmont zone implies in Mizque settling a good portion of good soil plots. The most common setting for settlements during this period was in the lower piedmont zone on the south bank of the Mizque River. A proportion of 97% of Tupuraya occupation (and Grey Ware occupation), and all but one Sauces occupation are in this setting. Therefore, settlements in this period are settling the most productive soils of the valley.

The survey recorded ten settlements for the Middle Horizon, producing a statistical number of 4.24 sites, and 40.9 ha in total occupation size. An estimated occupation size of 365.9±259.1 ha has been calculated from the sample. Middle Horizon occupation is marked by Tiwanaku style, Omereque style and, to a lesser extent, Gray Ware style pottery. Pottery of these three styles have been documented as co-occurring in burials but not in domestic contexts (Rydén 1959; Walter 1966).

Tiwanaku style pottery represents about 60% of the total occupation, followed by Gray Ware assemblage with 29% of the occupation, and 11% for the Omereque assemblage. Tiwanaku style materials were found at nine sites, but only in Sites 13 and 30 do they surpass a proportion of 20% of the assemblage ( Figure; adjacent Site 26 is the Late Period Lakatambo site). Omereque materials occur at seven sites, including Sites 13 and 30, and have their highest proportion of 18% only in these latter two sites. Grey Ware style pottery was recorded at three sites, including 13 and 30, averaging 13% of the assemblage.

All the Middle Horizon sites recorded are multi-component sites. There is not a site which had exclusive, or new, occupation with Tiwanaku style pottery. Rather, Tiwanaku style remains were found at sites which also had local Early Intermediate styles, Sauces, Tupuraya and Grey ware, and the Omereque style. Sites 13 and 30 are the only ones with all four pottery styles, along with the highest proportions of Tiwanaku style material in Mizque. The densest concentration of Middle Horizon occupation occurs at sites 13, 15, and 30, which make up 57% of the occupation size for the period.

Middle Horizon occupation declines from 53.8 ha in the Early Intermediate Period to 40.9 ha. The number of settlements occupied is also reduced from 14 to 12. However, these differences, compared with the estimated total occupation size for each area, are not significant at a 95% confidence level (Graph). What is more relevant in this transition is a trend toward settlement aggregation into sites 13 and 30. This aggregation is not limited to the occupation of Tiwanaku style ceramics, although this style is dominant in the collections; the densest occupation of the two local pottery groups is also at those two sites.

As in the Early Intermediate Period, the bulk of Middle Horizon occupation was on the most productive soils, with 44.7% of the occupation, or 18.3 ha, on good soils ( Table 3). Occupation of medium and poor soils constituted 24.8 and 30.5% of the remaining occupation size, respectively. However, the percentage of occupation of good soils zone is not significantly different, at the 95% confidence level, from occupation in the other two soil zones (Graph). But, as occurred during the previous period, the settlement focus in the piedmont zone led to an occupation of good soil plots.

In fact, a very high 98.5% of the Middle Horizon occupation size of our sample is in the piedmont zone, while the remaining 1.5% is located in the alluvial zone ( Table 4). No sites were found in the mountain zone. Comparison of the percentages of occupation shows that the difference in the higher proportion of occupation in the piedmont zone is significant at the 95% confidence level. There is, then, a preference for occupying the piedmont zone. This pattern of topographic distribution seems to be a continuation of a trend initiated in the Early Intermediate Period. Site 13, Condadillo, one of the two most important Middle Horizon sites, is on good soil plots in the lower piedmont zone. Settlement at Site 30 is located on poor soils and in the piedmont zone, but close to the alluvial plain. Two other sites, sites 14 and 15, with a smaller occupation of this period (with less than 15% of Middle Horizon pottery) are also located on lower piedmont spots close to the river bed ( Figure).

In sum, a significant spatial evidence in Mizque is the preference for settling the piedmont topographic zone in the Early Intermediate Period and in the Middle Horizon. Occupation by soil zones did not reveal any preference patterns. The conditions of the Mizque Valley with large portions of good soils in the piedmont zone, imply that, indirectly, the preference for that zone in Middle Horizon could be suggested as a pattern for settling good soil zones. But this pattern does not start in the Middle Horizon, and is rather a continuation of a settlement strategy of the Early Intermediate Period. This Middle Horizon pattern would have been more relevant if, in addition to following similar location patterns, there had been significant growth in occupation size (e.g., there is a decrease in occupation size) or foundation of new settlements. Unlike the patterns in Capinota, soil productivity would have been a factor more important than topography, when a fairly productive piedmont zone is settled.

The pottery distribution patterns documented in both survey areas show that Tiwanaku style materials are largely found on sites originally established during the Early Intermediate Period. Therefore, there is little evidence in the sample to suggest foundation of new and exclusive Middle Horizon sites with Tiwanaku style material. Site 11 in Capinota is the only exception. In addition, there is no exclusive occupations bearing local pottery in either survey area during the Middle Horizon. In fact, there is little pottery of the local Omereque style in Capinota, except at Site 42, the Tiwanaku burial site.

A significant observation on both spatial and pottery patterns in Mizque is settlement aggregation showing occupation with Tiwanaku style and local style materials at the same two sites. Neither of these two style groups occur alone at any site. A dominant Tiwanaku style assemblage coexists with Omereque and Grey Ware pottery. Not surprisingly, this observation confirms the overwhelming presence of the Tiwanaku style in Cochabamba's Middle Horizon. The two "multiethnic" sites in Mizque show a pattern of population aggregation that would be expected in the archipelago verticality model, although it contrasts with the spatial correlates for the same model. However, the dominant presence of Tiwanaku style pottery vs. local styles is not a common expectation of this model. Mizque's case shows a Tiwanaku style that is not fully dominant in the area.


The spatial data on land use and settlement distribution obtained in the Capinota and Mizque sub-valleys of Cochabamba suggest that Middle Horizon spatial organization can be best explained by the independence model. As expected, this choice is not as clear cut as desired. Land use and settlement data is hard to match with patterns of spatial distribution of pottery produced in this research. Although, the spatial patterns used here do not match the expectations for the vertical archipelago or prestige-good model, pottery patterns alone might reveal some affinities to these interaction models.

The spatial data offers a panorama where a process of interaction emanating from the Tiwanaku polity was much less important in the spatial organization of the region than previously thought. Little external impact could be suggested on Middle Horizon spatial organization because local Early Intermediate Period settlement location and land use patterns endure during the Middle Horizon. These Middle Horizon patterns do not reflect agricultural intensification in either survey area, an implication of the hypothesis of the direct exploitation of Cochabamba resources by the Tiwanaku polity. The patterns do, however, reflect a settlement distribution that favors exploitation of the best soils in Mizque in the Middle Horizon. But again, that pattern was already present in the region and hence does not represent a significant shift in agricultural strategy.

The results obtained here are only the start in our understanding of the political organization of the Cochabamba region in the Middle Horizon. In fact, the results of the spatial analysis refer only to patterns of agricultural exploitation. These results allow me to strongly suggest that Middle Horizon interaction with Cochabamba did not generate an increase of agricultural production at the request of the Tiwanaku core, nor was settlement focusing on the richest area studied here. A comparison of the estimated occupation sizes calculated for each period shows no difference in the size of Middle Horizon settlement between the Capinota and Mizque survey areas. Mizque had a significantly larger occupation size only during the Early Intermediate Period. During the Middle Horizon, for which a settlement concentration on most productive areas was hypothesized, the advantages of the Mizque survey area did not lead to a significantly larger occupation size than in Capinota. Therefore, a difference in agricultural potential in the two regions analyzed here is not adequate to explain the distribution of Middle Horizon occupation.

Similarly, room for variation in the modeled correlates must remain open for future analysis. Indeed, the use of the proposed interaction models based only on spatial evidence requires caution. The evidence examined seems to suggest that Tiwanaku colonies, direct political control, or local development of elite economies linked with highland economic networks, were less likely outcomes of the interaction between Cochabamba and the Tiwanaku polity. However, two observations made on other than spatial evidence should be pursued: the "multiethnic" settlements in Mizque and the foundation of a new Middle Horizon habitation settlement in Capinota.

In addition, in the methodological arena, the validity of spatial shifts and occupation growth to interpret transition in spatial patterns and political organization models should earn further testing. It is possible, but highly unlikely based on other case studies, and regardless of the type of political organization emanating from the highlands towards Cochabamba, that no changes were produced in the spatial organization of the studied areas. In other words, it would be unusual in Andean studies that the organization seen in Capinota, with one single small new site, little growth in occupation size, and no shifts in location patterns, could be reflecting a political subordination organization. More substantial changes were expected in spatial organization deriving from the political subordination model. On the other hand, even if independent ayllus were heading the interaction process we would still expect to see some changes in settlement and land use patterns.

In Capinota, Site 11 is a new Middle Horizon and Site 7 is an isolated Tiwanaku burial site. However, Capinota sites hardly generated any growth in Middle Horizon occupation size. Therefore, this change essentially represents a shift in settlement distribution rather than a demographic expansion. Foundation of one habitation site of 2.5 ha is in itself far from sufficient to argue that interaction Capinota could resulted in a political subordination system in the area. More substantial new sites producing a larger increase in occupation size were expected. Indeed, the scale of settlement growth recorded in Capinota is not remotely as dramatic as witnessed in other cases of imperial conquest: Moquegua by the Tiwanaku polity (Goldstein 1989, 1993); Carhuarazo by the Wari polity (Schreiber 1987, 1992); Cochabamba (Wachtel 1982), and the Mantaro Valley by the Inka empire (D'Altroy 1992; Hastorf 1993).

It has been signaled that some regional Tiwanaku settlement sequences do not show important shifts between periods, as for example in the Tiwanaku Valley after the Formative Period (Albarracín-Jordan 1992; Mathews 1992). That is, settlement patterns in the Tiwanaku Period maintain previous patterns, showing a low key decision making system emanating from the site of Tiwanaku to its immediate hinterland. Rather, a tradition of small-scale independent polities would have been maintained with endurance of basic land use and settlement patterns. In this case, similar site location patterns would be functioning in a loose political organization of the Formative Period and in a "centralized" or cult-unified organization of the Middle Horizon (Bermann 1997). Future analysis should show if this same type of subtle transition might have been occurring in Cochabamba under the independence model, where political organization could change with no detectable shifts in land use or settlement patterns. For instance, Middle Horizon occupation maintains previous levels of agricultural production, but political organization has changed from the Early Intermediate Period.

Analysis of other archaeological evidence can provide additional leads to this methodological problem. Even if pottery patterns were not set up for analysis --where patterns favor the political subordination model the evidence obtained in the two "multiethnic" settlements in the Mizque survey area, and smaller similar cases in Capinota, cannot be ignored. They represent an example of "ethnic" coexistence of artifacts in shared settlements that is theoretically characteristic of the vertical archipelago or the prestige-good model. A difference with the vertical archipelago model is that the proportions of Tiwanaku style materials are much higher that what would be expected from colonies settling amid local populations, in which the former style would be outnumbered by much denser local styles. This pattern could match with the prestige-good model, if the Tiwanaku style pottery is not the exotic good, but rather textiles, wooden artifacts or metals bearing that style, or in the unlikely scenario that the "local" styles are the exotic products given their small percentage in the assemblage.

In addition, the spatial evidence is showing little focus on agricultural intensification in the form deployed by an archipelago strategy. First, it was expected that distribution of settlement in both areas would not be different in a region with little circumscription as Cochabamba. Then, preference for settling the piedmont in Mizque, given the soil composition of the region, can be interpreted as a preference for settling on the best soils in the valley. Then, as expected within a survey area, Middle Horizon occupation favors the best agricultural spots in the richest survey area. However, Middle Horizon settlement follows a trend already established in the Early Intermediate Period and, in addition, does not produce any evidence of intensification of agriculture as expected. Similarly, the preference for piedmont settlement in the Capinota survey area during the Middle Horizon represents an enlargement of piedmont settlement in the previous period, but in Capinota there are no soil advantages by settling this zone.

The decrease in occupation size in Mizque is another reason why the preference Middle Horizon settlement on good soils in Mizque can hardly be suggested as evidence of intensification of agricultural exploitation. In the decrease from 53.8 ha to 40.9 ha occupation of good soils is reduced and occupation of medium soils is slightly increased. Human occupation size in Capinota has a slight decrease from 17.7 ha in the Early Intermediate Period to 17.3 ha in the Middle Horizon. In this transition, settlement moves from good soils to poor soils. This occupation relocation is an indicator of agriculture emphasis, as good soils are cleared. However, the change is not significant as the estimate for occupation of poor soils in the Middle Horizon is almost identical to the one in the Early Intermediate Period ( Table 1). The lack of growth in occupation size in Mizque or Capinota cannot be a result of limited resources in the valley. In both areas, Late Period occupation size will have very significant growth with respect to the Middle Horizon (Higueras 1996). An explanation for Mizque's decrease in occupation in this latter period could be that distance for the altiplano is combined with a pattern of settling in areas more productive than Mizque itself.

Concluding this research by suggesting the independence model could be a disappointing "non-result" in the context of many paradigms popular in Andean archaeology. However, finding that the appearance and distribution of Tiwanaku style materials was not concomitant with changes in settlement and land use (at least as I am approaching them) would open a door to more meaningful and powerful explanatory approaches in the future, when all evidence available could be explained by a finer model of political organization. As a first step in understanding the long term evolutionary trajectory of Cochabamba populations, this research does not support the important role ascribed to the Tiwanaku polity. Choosing the independence model calls into question not only Kolata's hypothesis, but other models that see interaction with Tiwanaku as a prime-mover in causing cultural change. One of the obvious implications of this result is that the distribution of Tiwanaku style materials, which is a very significant phenomenon, and by extension, the distribution of any rather loose pottery style, are, alone, a poor basis from which to reconstruct issues of regional interaction and societal development, in comparison, say, to settlement and land use patterns.


The 1993-94 phase of the Proyecto Arqueológico Expansión Tiwanaku en Cochabamba was co-directed by Mr. Ricardo Céspedes, researcher at the Museo Arqueológico de la Universidad Mayor de San Simón, Cochabamba, and the author. A Postdoctoral fellowship at the American Museum of Natural History, New York, allowed me to prepare this paper. Financial support for the research was provided by the Tinker Foundation, through the Center for Latin American Studies of the University of Pittsburgh, the Wenner-Gren Foundation for Anthropological Research (#5628), the National Science Foundation (SBR-9312906), and the Organization of American States. I wish to thank Marc Bermann, Robert D. Drennan, and James B. Richardson III, members of my dissertation committee, the Howard Heinz Foundation, and the staff of the Cochabamba museum for their support. Five anonymous reviewers provided crucial comments to this paper. Charmaine Steinberg provided useful editorial advice. Any inadequacies in this paper are, however, of my own responsibility.
top of page

  Main page

Tiwanaku research page

 Andean Research



Tiwanaku and Andean Archaeology
Copyright © 1998-01, Alvaro Higueras. Derechos Reservados/All rights reserved.
Please send comments on content and presentation to
URL of this document:
Revised: 29 May 1999