ECOLOGICAL COMPLEMENTARITY IN COCHABAMBA: VARIATION AND POLITICAL BASIS IN PREHISTORIC INTERREGIONAL INTERACTION. Presented at the 62nd annual meeting of the Society for American Archaeology, Nashville, April 1997.

To abstract ......... To papers list ......... To Bookshelf

 Today I will address the implications of recent research in Cochabamba for the understanding of the Middle Horizon in the south central Andes. Two problems are developed in light of the Cochabamba data: the interregional interaction models used to assess political economy strategies of the Tiwanaku polity; and what is currently known about the interaction between the Tiwanaku polity core area and local regional populations in the south-central Andes.

First, several systematic analyses using comparative models to assess political interaction have been conducted in the region. Regional survey data aimed at addressing interaction between Middle Horizon large-scale polities in the Upper Moquegua Valley and between external and local groups in the Lower Moquegua Valley are an example. Hence, Tiwanaku studies in the periphery are following and improving paths set in the study of other polities in the Andes. These studies show that analyses of regional databases attain a higher significance when interpreted in light of a set of interregional interaction models.

Second, we know a great deal about the character of Tiwanaku interaction in the south-central Andes. We know enough to allow scholars to suggest flexibility of the interaction patterns adopted by the Tiwanaku "state," and interpret it as adaptations molded to the core's interests in the region. However, only recently are local conditions and local populations part of the equation in a relationship that was conceived as one-way affair. Many models had assumed that it is Tiwanaku populations that move, and that presence or absence of Tiwanaku-style materials are evidence of migration of altiplano populations. To address interaction patterns in its full reach evidence of local pre-Tiwanaku occupations is accounted for in regional surveys as an important component defining the strategy following the interaction process. Such data has shown that Tiwanaku interaction in certain regions is not as clear cut as assumed.

Political economy analyses of the Tiwanaku polity have often been guided by the expectation to find archaeological scenarios of the verticality strategy across space. Verticality, a variant of ecological complementarity, is a key model to consider in interregional interaction studies in the Andes. Reflecting its importance, a full session at last year's meetings explored the character and time-depth of this strategy in the Moquegua Valley. Stanish commented that while Murra's verticality concept was aimed at defining an "Andean mode of production," a not so "Andean" approach to the strategy was to benefit Andean archaeology.

Upgrading how verticality is assessed by analyzing it at par with other political economy models is an important step in that direction: it will dissociate the strategy from its colonial context, and test it other areas than the steep and narrow valleys on the Andean western slopes. Recent contributions (Stanish and Van Buren, in particular) are careful about questioning, for example, the legal ties with the homeland, or what consumers profited from colonial production, both central to the operation of verticality.

Variation in Andean ecological settings is key to view the deployment of political economy strategies. The Cochabamba Valleys have greater agricultural potential than the western Andes valleys, where agricultural soils are limited in size. Prime agricultural land in Cochabamba is vast. With no circumscription of agricultural lands, strategies adopted by external populations, by local populations, and in the coexistence between both, may vary. No Andean setting has such availability of good agricultural soils with small elevation differences as Cochabamba. Therefore any Middle Horizon interaction strategy in Cochabamba, if qualified as verticality, would represent a new archaeological variant of the already ethnohistorically known large-scale archipelagos.


Understanding of Tiwanaku's political economy has made important progress in the last decade. Extensive excavations in different sectors of the Tiwanaku core site, and surveys in hinterland and peripheral regions in the south-central Andes have provided data to start presenting alternative models for Tiwanaku's political, economic and social organization. In the core area the current hypothesis of a central bureaucracy for Tiwanaku, supported by arguments for complex labor organization involved in vast drainage works, is now challenged by the model of "nested hierarchies" for the social and political organization of Tiwanaku's society.

The concept of smaller political entities making the Tiwanaku "state" in the Tiwanaku Valley is not new. Browman viewed the Tiwanaku polity as a federation of smaller polities or ayllus. However, Albarracin presents the first attempt to test it with settlement distribution data and, as in the bureaucracy model, on assumptions of labor organization --irrigation works could be handled at the communal level of secondary sites, following Ericksson and Graffam. He suggested that ayllu-like communities were the active political and economic entities in the organization that revolved around, and were not politically controlled by, the site of Tiwanaku.

The model of nested hierarchies has important implications for the understanding of episodes of interregional interaction that produced the widespread distribution of Tiwanaku-style artifacts in the south-central Andes. Variation in the political economy strategies of the Tiwanaku polities is supported by differences in the composition and distribution of the material evidence in Moquegua and San Pedro (and perhaps Azapa), and in land use and settlement patterns on the western shore of the Titicaca Basin and in Cochabamba. It is fair to suggest that differences in regional strategies may be the result of ayllu-level decisions and not centralized decisions from the core site, hence the important distinctions between the cases we know for Tiwanaku interaction.


In Cochabamba a new avenue for analysis has been set with four models of interregional interaction, all of which stem from ecological complementarity needs of prehistoric groups: direct control, prestige-good economy, verticality and status quo. I assumed that these models could be assessed by different patterns of regional spatial organization. I used therefore systematic data on settlement location and land use recorded at a regional scale. In contrast, pottery and architecture have been used in exploring interaction patterns -and verticality-, and biological data in defining migration patterns, in other regions of the south-central Andes. If pottery data had been used as prime evidence in Cochabamba this study would have had little object.

The use of evidence on land use and settlement location requires refinement in the recordation and analysis of data. For instance, number of sites by period is not a relevant variable when comparing two periods. In contrast, estimation of site size, total occupation size and occupation by period are required for the analysis.

The survey of the Capinota and Mizque sections of Cochabamba was aimed at analyzing the effects of interregional interaction between local populations and the highland Tiwanaku polity. Interaction in the Middle Horizon was assessed diachronically vs. the previous Early Intermediate Period, and regionally by comparing its effects in those two ecologically distinct sections of Cochabamba. The spatial analysis was aimed at assessing the role of good agricultural lands in Middle Horizon settlement, following Kolata's suggestion of a direct colonization and control of Cochabamba in a vertical mode by the Tiwanaku polity. By this assumption, the richest agricultural area, Mizque, even though being farther away from the altiplano was expected to have major shifts in settlement location and size, reflecting strategies for intensification of maize production.

The Status quo model is proposed as the outcome of interregional interaction between the Tiwanaku polity and local populations in Capinota and Mizque. Middle Horizon settlement does not focus on agricultural intensification in any of the two areas. First, there is no difference in the estimated occupation size between them in this period. Second, there is no preference for settling the best soils within either area. Third, Middle Horizon occupation maintains the occupation size and location strategies of the Early Intermediate Period in both areas and does not expand occupation size n a fashion that could have been correlated with a strategy that maximized agricultural production. The hypothesis of a Tiwanaku polity aiming at a larger agricultural production cannot be suggested. (In comparison, in the previous Early Intermediate Period, local populations in Mizque had a larger occupation than Capinota, showing a strategy of concentrating settlement on better soil valleys.)

In Capinota, three new settlement locations occur in the Middle Horizon in addition to reoccupation of Early Intermediate Period sites. Even if there is no significant differences in occupation size between both periods, the deployment of new sites is an important factor to differentiate it from Mizque's case. Nonetheless, this shift is far from sufficient to argue for direct incorporation of the Capinota region into the Tiwanaku polity. The scale of the settlement reorganization --that has no impact on the size of the occupation-- has no comparison with other cases in the Central Andes.

The scale of the Middle Horizon occupation for Cochabamba, systematically recorded in surveys, and judgementally known by other accounts, clearly exceeds the spatial scale of the traditional verticality strategy, a variant that was not expected in our research. However, the large-scale archipelago variant could not be documented either based on the settlement and land use data. Tiwanaku style material occupation, highly dominant in this period, is not clustered intensively on the most productive agricultural soils in either area, and there is no synchronic coexistence of local styles with Tiwanaku style materials in single or coeval settlements.


The large agricultural capacity of the Cochabamba region called for a thorough analysis with four models of interaction -which did not include the nested hierarchies model. This model will be added for future research when the expectations for its archaeological testing and recognition in the field can be figured out.

In political terms, the suggested status quo model implies that a dense presence of highland Tiwanaku populations is highly improbable. This research was relevant by assuming that the dominance of Tiwanaku-style materials is not an indicator of any interaction strategy, much less a direct control policy by Tiwanaku. Rather, dominance of those materials responds to rules to stylistic adoptions and shifts that are not fully understood. I think that a process of adoption of Tiwanaku style pottery by local populations is an option that cannot be discarded.

Cochabamba has areas with higher productivity than the ones studied. I predict that the Central Valley -focus of Inka agricultural exploitation- and the Valle Alto -the area of Bennett's Arani- would produce a relatively larger occupation in good agricultural soils that in Capinota and Mizque. This scenario would still suggest a model of status quo, where local populations control agricultural strategies, if Middle Horizon occupation mimics previous patterns. Clearly, occurrence of spatially and temporal differences in patterns of land use and settlement location in richest areas would suggest differential interaction in the Cochabamba region. Such a scenario was expected by choosing areas with different agricultural potential at the start of the research. This scenario occurs with the Inka occupation in Cochabamba which densely concentrates in the Central Valley, whereas Capinota -40 km south of the Inka zone- has a very scanty occupation. Further systematic regional research in Cochabamba is expected to produce differential patterns for the Middle Horizon and complement data from Mizque and Capinota. At that point, great synthesizers for the Andes will be able to present a more solid picture of highland-lowlands interaction in the Middle Horizon with data from the richest region in the south-central Andes.

To abstract ......... To papers list ......... To Bookshelf

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.