Tiwanaku Interaction and Formative Period Land Use in Cochabamba, Bolivia.

  Main page

Tiwanaku research page

 Andean Research




     Archaeologists have long viewed ecological complementarity as a pervasive and important part of the economy and political organization of the prehispanic Andean population. This notion that prehispanic peoples deliberately sought to exploit productive areas at other altitudes is largely derived from ethnohistoric accounts and ethnographic studies of modern populations (Condarco 1972; Brush 1974; Murra 1975). In the last two decades, verticality has been a popular explanation for the regional distribution of archaeological remains. In turn, inter-regional variation in artifact assemblages or stylistic preferences has often been interpreted in the context of different forms of verticality (Masuda et al. 1985; Stanish 1985, 1992). Small-scale ecological complementarity has been well-documented in late prehispanic and modern times. Not surprisingly, archaeologists have often sought to extend the verticality model to the largest political formations of the prehispanic world, arguing that polities such as the Inca empire (Dillehay 1979) and Huari empire (Isbell 1977; Raymond 1992) deliberately pursued a strategy of vertical expansion in order to exploit resources not available in the homeland.

     One of the great prehispanic polities for which a pattern of verticality has been proposed is the highland state of Tiwanaku that dominated the south-central Andes five centuries before the Inca. In discussing Tiwanaku imperial expansion, Alan Kolata (1992:81) has recently written: "There is substantial and growing evidence that Tiwanaku directly colonized and subsequently controlled key economic resources in lower-lying regions." He later adds, "the motive force underlying Tiwanaku agricultural colonization in the yungas zones was expansion of the cultivation of prestige crops, particularly maize,..." (1992:82). Despite Kolata's strong statement, the idea that the Tiwanaku-style materials found on the eastern slopes of the Andes in the warm yungas (lowlands) region of the Cochabamba Valley represent imperial verticality (rather than other forms of interaction) has been assumed rather than tested.

     The proposed research will examine the Kolata hypothesis by comparing changes in man-land relationships and productive activities in two ecologically contrasting areas of the Cochabamba Valley in which Tiwanaku-style materials are found. One area presents a moderately cold, near highland, environment. The other are is an optimal area for maize and coca production. Put most simply, if acquisition of lowland agricultural products was the motive behind Tiwanaku expansion into the Cochabamba Valley, then we would expect to see greater Tiwanaku involvement with the latter area.

     Because this investigation requires documenting pre-Tiwanaku (Formative period) settlement and land- use patterns in each area as a comparative baseline, I expect my research to provide insights into additional issues of concern to Andean prehistorians. Archaeology is experiencing a revival of interest in the effects that interaction with a state society may have on less complex societies in its periphery (Shortmann and Urban 1987; Champion 1989; Chase-Dunn and Hall 1991). The proposed research provides an opportunity to explore how interaction with Tiwanaku may have affected two Cochabamba populations. In addition, by examining settlement changes in each study area, and the types and inter-site distribution of Tiwanaku-style materials in each area, I will develop data important ultimately to determining what processes -- direct colonization, alliances with local elites, intensive trade, prestige-good exchange -- resulted in the presence of Tiwanaku-style materials in each area.


     This research takes a different approach to archaeological discernment of verticality. Traditional approaches to verticality typically focus on artifact assemblages from individual sites, present artifact comparisons to highland sites, and document the regional distribution of pottery styles (Dillehay 1979; Mujica 1985; Goldstein 1989, 1990; Raymond 1992; Stanish 1992). As Marcus and Silva (1988) point out, such approaches have not yet proven very productive in documenting or elucidating the nature of interregional interaction. In contrast to these traditional approaches, my research focuses on changes in man-land relationships and agricultural production. As the motive behind verticality is almost always ascribed to acquisition of non-highland agricultural resources (Kolata 1992), a project incorporating assessment of agricultural production speaks directly to the crux of verticality arrangements.

     Traditionally archaeologists have used the following as archaeological correlates of Tiwanaku "verticality": intrusive sites with Tiwanaku-style material (e.g. Goldstein 1989, 1990), intrasite and/or intersite variation of the pottery assemblages, including local styles; evidence of the permanent presence of altiplano populations with native local villages, forming multiethnic settlements (Murra 1975; Mujica et al. 1983:97- 101; Stanish 1989), sharing of territories (Saignes 1986), or the construction of new settlements with administrative structures in the Tiwanaku style (Goldstein 1989:20; 1993). Any of these options may appear in Cochabamba.

     My approach to Tiwanaku verticality does not presuppose a particular complementarity mechanism or form of vertical organization (e.g., colonies, trade, political alliances; see Brush 1974; Condarco 1978; Oberem 1978; Salomon 1985). Even if Tiwanaku did not install colonies in the Cochabamba, we might still expect to see changes in man-land relationships as a result of interaction with the Tiwanaku polity. Thus, I am not testing any specific model of the verticality system --such as the archipelago model-- but the simply proposition, explicitly laid out by Kolata, that Tiwanaku was interested in lowland areas for their agricultural potential. There are in fact several ways in which that interest might have been expressed: colonialism and archipelagos; trade; political alliances; and if Kolata is correct we would expect them to be focused in the maize-growing areas. This research will concentrate on analyzing potential changes in land use. In sum, in areas of differing lowland agricultural potential. If Kolata is correct, I would expect the presence of Tiwanaku- style material to be correlated spatially with the most productive lands of the region reflecting resource extraction mechanisms. Tiwanaku settlement, control of the area or intensive interaction should be reflected with a higher proportion of imported ceramics, or the presence of pottery with local clays with a consistent Tiwanaku iconography and forms, reflecting a close relationship maintained with the Tiwanaku polity (Dillehay 1979; Stanish 1989). I will compare the Cochabamba assemblages to those from other Tiwanaku sites, such as Omo, considered a Tiwanaku colony in the Moquegua Valley, and from the Azapa Valley, apparently subject to a different kind of interaction with Tiwanaku (Goldstein, pers.comm. 1992). Future work, based on this data, will be able to assess how this interest was expressed or identify the model of interregional interaction.

     This research will therefore broaden the archaeological means used to explore verticality in the prehispanic Andes by focusing on evidence of land use and soil productivity, and differential Tiwanaku interest in maize and non-maize growing areas.


     The Tiwanaku polity dominated the South-Central Andes between ca. AD 400-1200 (Kolata 1987). The capital site --Tiwanaku-- located to the south of Lake Titicaca in present day Bolivia, and at its height covered some 400 ha., with extensive monumental architecture (Bennett 1936; Ponce 1972, 1989) and a residential population of 20-40,000. The urban concentration in the Tiwanaku core area was supported by intensive agricultural production of highland tubers and grains in massive raised field systems (Kolata 1986, 1989). Survey of the Tiwanaku Valley has suggested a four-tier settlement hierarchy, a pattern indicative of state-level organization (Albarracin-Jordan and Mathews 1990). The capital site was surrounded by a set of secondary sites (e.g., Lukurmata, Pajchiri, Khonko, Wankani), each of them associated with expanses of agricultural fields. Each of these secondary sites contain lesser amounts of Tiwanaku-style public architecture, including semi-subterranean temples (Spickard 1985; Bermann 1990; Goldstein 1993; Stanish and de la Vega n.d.).

     Pottery and other objects in the distinctive Tiwanaku-style are widely distributed throughout the South- Central Andes, from the southern coastal valleys of Peru and Chile to the lowland eastern slopes of the Andes. The distribution of Tiwanaku-style artifacts exhibits a great deal of variation, with different types and quantities of Tiwanaku-style materials occurring in different regions. The mechanisms that lead to this distribution of Tiwanaku-style artifacts, and the wide regional variation, have long been the subject of intense debate (Serracino 1980; Browman 1980; Mu¤oz 1983; Oakland 1985; Bermann 1990; Stanish 1992).

     Four general explanations have been proposed in the literature to account for the distribution of Tiwanaku-style materials in the South Central Andes. The first explanation holds the distribution of artifacts to be the result of Tiwanaku imperial expansion outside the Titicaca Basin with colonies and conquest aimed at lowland resource extraction (Ponce 1972; Moseley et al. 1991; Cespedes 1992, pers.comm.). A second scenario proposes the growth of an archipelago system where discontinuous territorial niches were exploited through placed colonies (Mujica 1985; Kolata 1987; Berenguer and Dauelsberg 1989; Goldstein 1989, 1990). A third scenario is that Tiwanaku-style materials spreaded spatially through trading networks headed by the Tiwanaku center (Nu¤ez and Dillehay 1979; Browman 1980, 1984; Lynch 1983, 1988). Finally, a fourth scenario views Tiwanaku expansion as purely ideological and/or ritual in nature devoid of political control or colonization (Browman 1978; Wallace 1989; Kolata 1992). Scholars (Berenguer et al. 1980; Byrne 1984; Browman 1985; Mujica 1988:101-107) have also proposed combinations of parallel and/or sequential mechanisms used by Tiwanaku in distinct regions.

     Tiwanaku-style materials have long been known to be present in the Cochabamba region (Bennett 1936; Byrne 1984; Ibarra 1944). Two of the above scenarios have been advanced specifically to interpret the relationship between the Tiwanaku polity and the Cochabamba region: (1) the Cochabamba region interacted with the Tiwanaku system through long-distance networks, and Tiwanaku materials flowed into the Cochabamba Valleys through pre-existing trade networks (Browman 1980); and (2) the Tiwanaku polity established colonies under the verticality mode in the Cochabamba region deliberately seeking staple resources, maize and coca, from the yunga zones (Goldstein 1989; Kolata 1992:80). These interaction hypothesis have been based solely on casual comparisons of Cochabamba Tiwanaku-style artifacts with assemblages from what are considered Tiwanaku colonies in Moquegua, Peru (Goldstein 1990). Neither of these two scenarios for Cochabamba have been tested archaeologically.

     For the purposes of this research, I will assume that differences between Tiwanaku-style artifact assemblages correlated with changes in settlement locations or densities in the two areas of research reflect different forms of interaction with Tiwanaku. Traditional approaches to interpreting Tiwanaku materials outside the Tiwanaku nuclear area have generally been concerned with studying only Tiwanaku-style artifact assemblages in order to reconstruct Tiwanaku state strategies (Stanish 1992). In contrast, mine is one of the few projects to focus on settlement and subsistence patterns to discern Tiwanaku complementarity.


     The Cochabamba Valleys, situated 400 km southeast of Tiwanaku, are located on the subtropical eastern slopes of the Bolivian Andes at an average elevation of 2700 m.s.l. They extend from west to east, over an area of approximately 2500 kmý (Figure 1).

     This region provides the ideal setting in which to address issues relating to Tiwanaku provincial interaction for three reasons: (1) ethnohistoric sources show that the region was actively exploited for maize production by highland polities, such as the Aymara kingdoms, and the Inca empire, in late prehispanic times (Murra 1975; 1985a, 1985b; Dillehay 1979; Wachtel 1982); (2) it is an area where a highly complex polity (Tiwanaku) would have been interacting with less complex peripheral societies; and (3) it has long been assumed based on ethnohistoric analogy that Cochabamba was incorporated, in one form or another, in the Tiwanaku politico-economic system.

4.1 Cochabamba Prehistory

     There has been little systematic archaeological research in Cochabamba. Several local ceramic styles have been identified for the Formative period (ca. 2000 B.C.-A.D. 400; Ryd‚n 1952, 1961; Byrne 1964; Brockington et al. 1985, 1986, n.d.; Brockington and Sanzetenea 1989; see Table 1) but the chronological and geographic relationships between these styles and subsequent traditions remain unclear. For the following period, archaeologists have defined the Tupuraya tradition in the Central Valley (Ryd‚n 1959; Ibarra 1971), the Mojocoya tradition in the Mizque valley, the Omereque Polychrome tradition in the Aiquile area (Anderson, pers.comm. 1992) and two local Tiwanaku phases, Illataco and Pi¤ami (phases Tiwanaku IV and V of the altiplano, respectively; Cespedes, pers.comm. 1992). Tiwanaku-style objects have been documented on the surface of sites, in tomb contexts (Ponce 1972: fig.10; Byrne 1984; Tapia 1984) and recovered from stratigraphic excavations (Bennett 1936; Brockington et al. 1985). However, the relationships of these styles (e.g., such as hypothesis of Tupuraya preceding and influencing Tiwanaku in the region; Mujica, pers.comm. 1993) still need to be clarified. Virtually nothing is known of the sociopolitical organization in the Valleys in the Formative and subsequent Middle Horizon-Tiwanaku periods. The lack of known large centers or evidence of political integration suggests that Cochabamba populations were organized into simpler societies -- egalitarian or simple chiefdoms-- in comparison to highland populations.

4.2 Areas of research

     The Cochabamba Valleys comprise two main ecologically distinct zones: (1) a montane moist forest biome on higher elevations (2600 m.s.l. and up); and (2) montane moist forest and montane thorn steppe biomes on the valley plains (below 2600 m.s.l). The research in these two ecologically contrasting zones will provide comparative databases essential to identifying important ecological/productive-related variables of interaction between the local Formative period populations and the Tiwanaku polity in Cochabamba. The distinct environmental features of the two areas will permit a better control of key variables (e.g., agricultural potential based on soil types, elevations, distance to water, etc.) in exploring whether differential interaction of the Tiwanaku polity with each area was due to agricultural productivity of staples such as maize.

     I have chosen two research areas: one in each ecological zone. Each research area measures 200 kmý roughly corresponds to a natural drainage, but has been arbitrarily bounded. Visual inspection of these areas in my 1992 pilot season yielded evidence of prehistoric occupation from the Formative and Tiwanaku periods, so each has the elements necessary to test the correlation between agricultural productivity, staple production, and degree of Tiwanaku interaction with the area. The first research area is located immediately north of the town of Independencia, 180 km northwest of the city of Cochabamba. It is delimited by two narrow alluvial valleys at an elevation range of 2800-3400 m. with a transitional ecology combining mesothermal and altiplano cultigens (e.g. grains and tubers, and in the lower reaches, maize). The Independencia research area consists of narrow interandean drainages, with relatively cold temperatures (means in the range of 6-13C), seasonal frost and dryness, and seasonal precipitation (December through March) of 300 to 500 mm. The area is formed by steep and narrow water cuts and valleys with seasonal springs, with alluvial terraces in the bottom. Soils are normally deep with medium to low permeability with shallow organic sediments. The relative humidity of soils is suitable for crops such as potato, oca, ullucu, and quinua in the rich class II soils in the alluvial plain (Fig.2). The steep slopes, lacking irrigation, are used today for goat pasturage and wood production.

     The second research area, 150 km southeast of Cochabamba, in located in the hinterland and south of the town of Aiquile, at an elevation range of 2400-2700 m. in a drainage that leads to lowland areas. Its fully mesothermal ecology permits the optimal production of important non-altiplano products including maize, legumes, fruits and coca leaves. The Aiquile research area is located in a valley-bottom setting (mean temperatures in the range of 13-18C) with seasonal differences in dry climate and frost period from April to October, and precipitation (500-1000 mm) from November to March. It is a suitable area for intensive non- irrigation agriculture. In the dry period only frost resistent crops are cultivated with irrigation. The plains are generally flat and filled with strata of clay and sand sediments. The soils have good retention of water of rainfall, reducing risk during maturation of crops. Arid xerofitic vegetation is combined with patches of humid and sub-spinous vegetation.


     The main objective of this research is to evaluate whether the two discrete research areas differ in terms of interaction with Tiwanaku. If so, I will explore whether this differential interaction can be explained in relation to maize production potential, or to pre-existing patterns of regional interaction. As noted earlier, if the Kolata hypothesis is correct, I would expect to see differential interaction with Tiwanaku in these two research area: with a greater density of Tiwanaku-style artifacts at sites in the maize area (i.e. Independencia), Tiwanaku items at sites most closely associated with terraces, or changes in settlement during the Tiwanaku period to optimal maize production locations. Specific research questions include the following:

(1) What is the spatial relationship of Formative Period (Pre-Tiwanaku) settlements to ecological features, soil types, and agricultural terracing in each area?;

(2) What types of Tiwanaku-style goods appear in each of the research areas, and how are they distributed between sites and productive zones?

(3) What change(s) in settlement locations, site size, and productive activities followed the appearance of the Tiwanaku-style materials?

     Although I do not expect the results of the research to be as simple and clearcut as any of those presented below, I will briefly delineate several broad consequences of Tiwanaku attempts at maize acquisition, and their archaeological correlates.

     This project is designed to address the Kolata hypothesis on two levels: (1) comparing the archaeological record of two ecologically dissimilar research areas; and (2) examining the potential settlement shifts within each area. As noted previously, we would expect, given a Tiwanaku interest in maize and coca, to find greater evidence of Tiwanaku interaction with the better maize growing area (Aiquile), even though this area is significantly farther from Tiwanaku. However, if there is no evidence of differential interaction between Tiwanaku and the two research areas, we cannot reject the Kolata hypothesis. Instead we turn to examining changes through time in settlement within each area.

     If the Kolata hypothesis is correct, we might expect to find sites with relatively high proportions of Tiwanaku-style materials to be spatially associated with optimal agricultural loci (based among other things, on soil types, elevation, temperature regime, and rainfall). Similarly, based on analogy from elsewhere in South America, we might expect intensified maize production to be associated with a shift to settlement locations adjacent to --but not on-- prime agricultural land (Schreiber 1987; Drennan and Quattrin 1993).

     Alternatively, if the Kolata hypothesis is not correct, we might expect the spatial distribution of Tiwanaku-style materials to bear little relationship to maize production potential. Similarly, we might expect to find in neither area any evidence of intensified maize production in the form of settlement shifts. If this were the case, we would expect to see no changes from Formative Period land-use patterns, with settlements optimally located to exploit a variety of resource zones rather that the best maize production lands.

     The situation will be complicated by any pre-existing differences in settlement and land-use between the two areas. For instance, pre-Tiwanaku Period settlement in Independencia may already have been arranged to maximize maize production. In this case, interaction with Tiwanaku would not produce any significant shifts in settlement locations. Finally, settlement locations are shaped as much by sociopolitical considerations as environmental variables and subsistence needs. Thus recognizing the effects of Tiwanaku interaction on local settlement, or prehispanic attempts to maximize maize production, will be a difficult process, requiring careful weighing of settlement and artifactual patterns at the regional, intra-regional, and community level.


     The major fieldwork will consist of a stratified sampling settlement survey. This will be complemented by a random sampling surface collection, and limited test pits. Settlement survey will have two major objectives: (1) to record the spatial distribution of settlements to particular resource zones and agricultural productive potential (Steponaitis 1981); and (2) to document a portion of the range of settlements distributed in each area. The survey is not designed or intended to find low uncommon site types, such as regional centers or provide information on the range of settlement in each area. Instead, this strategy will provide a solid grounding for assessing the relationship between agricultural potential and the most common site types -- probably villages and homesteads.

     Regional settlement survey has long been recognized as a powerful tool for investigating how agricultural peoples exploited surrounding territories. And, with the recent development of computer-based techniques that readily analyze the interrelationships of site locations, social distances, and ecological variables, information from regional survey data has been used to gain increasingly sophisticated understandings of the determinants of settlement locations. In particular, approaches such as GIS have been of great value in elucidating the motives underlying particular settlement patterns and modeling the intentions of prehistoric peoples in deciding where to live.

     GIS techniques will be critical to: (1) identifying prime agricultural land; and (2) determining the optimal locations for settlements to exploit these lands. These predictive patterns can then be compared to the archaeological patterns to: (3) assess whether a determining factor in settlement was maximization of maize production, and (4) measure other factors potentially important in structuring settlement (for example, maintenance of catchment zones; proximity to diverse ecological zones; proximity to valley passes, distance to water); and (5) explore whether settlement locations in the Formative, Tiwanaku, and Post-Tiwanaku Periods were structures, or can be "explained", by the same factors, or inhabitant preferences.

     It must be recognized that the survey strategy will provide some kinds of information but not other kinds. As Kintigh has noted, an issue in survey is often whether to sample a large area, or fully cover a smaller one (1990:239). Given my need to examine man-land relationships, the environmental context of sites (Kowalewski and Fish 1990:20), and artifact distributions at the regional level, I have opted for a stratified random sampling survey that will allow me to obtain a large and robust database with extensive area coverage (Whalen 1990). Clearly, data obtained by this strategy will not allow me to reconstruct settlement hierarchies (Johnson 1981; Crumley 1979; Evans and Gould 1982), but information on hierarchies is not critical to the questions posed earlier. The information that is critical deals with the spatial relationship between sites and variables of agricultural production, and survey strategy will provide the data sets needed for this.

     Through probabilistic sampling I hope to obtain a dataset adequate for making general inferences about a larger universe using through statistical measures. The large number of "observations" provided by this strategy (vs. one sample unit in a 100% survey of a smaller area) will help to narrow the error range of the estimations (and interpretations) based on the sample population.

     In addition to recording size, elevation, location and surface artifact density for each site I will look at the productive associations of each site with reference to particular soil types (characterized by nutrients, water retention, and hence different levels of productivity), agricultural features such as terraces and canals, proximity to other soil types, distance of fields to water, etc. Using the GRASS GIS package, these features will be indexed against variables related to agricultural activities such as rainfall patterns and vegetation zones. Estimations of maize productivity per hectare will be based partly on modern land use figures but adjusted in order to estimate prehistoric productivity (cf. Kirkby 1973).

     Settlement analysis for each period in each area will entail: (1) measurement of the total size/area of occupation as a relative measure of population; (2) recording the presence and absence, and relative quantity of administrative architecture; (3) recording the distribution of imported pottery and other prestige goods and association with local items (e.g., "provincial" pottery styles; Smith 1987; Costin and Earle 1989; Hodge and Minc 1990); and, (4) recording the location of sites and features of agricultural technology as terracing and canals in relation to ecological, soil productivity and topographic characteristics. These measures will allow me to monitor the development of land-use patterns and as well as relative and general demographic shifts (based on area of occupation; Sanders et al. 1979; Kowalewski et al. 1989). Catchment analysis will also be used in addressing the relationship between the productive capacities of land plots and individual settlements (Vita-Finzi and Higgs 1970; Brumfiel 1976; Drennan and Quattrin 1993).

6.1 Fieldwork

     The initial step of settlement survey will be stratifying the two 200 kmý areas according to ecological/topographic zones (e.g. grassland higher slopes; shrub-like medium slopes, and riverine alluvial plain). This procedure guarantees that measures related to topographical and ecological features associated with the site or sites within the quadrat may be documented (Read 1986). Based on my preliminary assessments of site sizes and densities, I will divide the strata in square quadrats (4 ha in size), and select a random sample of these quadrats to be surveyed. A sample size of 200 quadrats to be drawn per each strata (an overall sample fraction of .12) is a sample (N=600; ca. 24 kmý) larger than the sample required to speak with a high degree of precision and confidence (95% confidence level) about site locations. Sites/artifacts encountered outside the survey area will be recorded but will not be considered in the inferences made from the cluster samples of artifacts. Survey members will walk in parallel transects of 33 m (that will ensure a 95% probability of finding sites larger than 30 sq.m; Sundstrom 1993). I have allotted three teams of three persons each to cover 3 quadrats per team/day) during a 80-85 day period of work in each research area. This estimate includes the execution of the surface collection as well. In the process of surveying the drawed units in the field I expect to encounter occasional problems for recording, such as in private lands; the sampled population will be accordingly replaced each of these cases.

     Prior to survey I will: (1) define in the maps the definitive boundaries on the drainage areas to survey; (2) define the ecological/ topographic strata of the research areas; (3) superpose a north-south grid (4 ha quadrats) to the survey areas; (4) draw randomly the sample of quadrats to survey; and (5) prepare plates based on topographic maps (1:10,000 scale) for the recording of the spatial limits of the artifact scatters and a precise location of sherd collection lots. The extension of a site will be defined by a surface scatter of sherds and/or architectural features; however, sites are not expected to conform to preconceived categories such as hamlets since they may well be representative of single structures, representing isolated homesteads in a dispersed settlement pattern (Drennan 1985).

     Site data forms and sketch maps will be produced for each site. Instrument maps will be produced for large sites, or those with visible architectural features. The degree of disturbance generated by modern plowing, looting, slope erosion and other factors affecting the visibility of each site will also be recorded for later consideration of possible biases in the survey results.

     Random surface artifact collections will be made at each site. Transect collection units, 10 m wide, randomly oriented, will provide a sample of the density and types of the artifacts on surface (Redman 1974; Plog 1986). Surface collections of pottery from surveyed sites will allow me to (1) map the spatial distribution of pottery-style preferences, and possible function specialization within each site; (2) compare intersite distribution of ceramic assemblages within and between the research areas; (3) monitor stylistic changes in local assemblages; and (4) estimate the area of occupation at different time periods at each site.

     A limited number of judgementally placed test pits (5 test pits in each research area) will be excavated in selected sites. These pits will be important in: (1) helping to establish more precise chronological controls; and, (2) analyzing surface-subsurface artifact relations in order to evaluate the reliability of surface-gathered information. The 1x1 m test pits will be excavated in natural stratigraphic layers with control levels of 10 cm. Soils will be sieved in a 1/4" mesh. Stratified middens will be preferred contexts for test excavation.

     Laboratory analysis in Cochabamba will concentrate on gathering the qualitative, quantitative, and locational data of artifacts, calculating diversity and proportions indexes, and refining the Cochabamba ceramic sequence. I will examine both technological and stylistic attributes in distinguishing Formative and Tiwanaku pottery, and local vs. highland pottery. Samples of Tiwanaku pottery from habitational areas of the Tiwanaku site (Rivera, comm.pers.) and in the Valley of Moquegua (Goldstein 1990) will allow comparison with highland pottery. Laboratory analysis in Pittsburgh will concentrate on elaboration of maps of settlement distributions, and analysis of the spatial distribution of artifacts in relation to ecological and topographical features using GIS and statistical packages.


     This research constitutes a critical problem-oriented first step in addressing the issue of Tiwanaku- Cochabamba interaction and Tiwanaku imperial expansion. It will provide an excellent opportunity to address archaeologically many aspects of the current model of Tiwanaku political economy and Andean verticality.

     This research is designed to test the important Kolata hypothesis that Tiwanaku expanded into Cochabamba to acquire non-altiplano agricultural products. As an approach to the general problem of verticality this study adds two important features: (1) a comparison of pre-Tiwanaku and Tiwanaku period patterns; and (2) focus on agricultural production to examine verticality in its most essential feature, that is acquisition of lowland agricultural resources. To date, Tiwanaku's interaction with different populations has only been superficially addressed (Lumbreras 1981; Berenguer 1978; Isbell 1983; Kolata 1983, 1992). Pre- existing local political and economic conditions have only been recently recognized as a cause in the variability in the regional distribution of imperial structures and artifacts in Andean polities (Menzel 1959; Morris 1972; Earle at al. 1987; Schreiber 1987, 1992; Hastorf 1990, 1991; D'Altroy 1987, 1992).

     This research is concerned with the currently popular hypothesis that vertical economies (in the archipelago and other forms) or acquisition of vertical resources, were a common characteristic of prehispanic political formations. The research will provide a preliminary test of the time depth of this assumption by examining Tiwanaku resource acquisition in the historical/diachronic context of the Cochabamba region. We cannot hope in a single season of work to determine the precise nature of interaction between Tiwanaku and the local Cochabamba population, but the research will provide some insights into how direct such interaction may have been, and whether Tiwanaku interaction followed pre-Tiwanaku patterns of highland-yungas interaction. Furthermore, the research will provide important preliminary information on the possible effects on the local population of interaction with Tiwanaku.

     Equally, I cannot hope to determine conclusively whether Tiwanaku maintained colonies in the yungas or simply interacted closely with local elites. However, I can examine the nature and distribution of highland materials in the two areas of research in local and historical context, compare the assemblages in each zone to highland Tiwanaku regions, and thus be able to rule out some explanations for the observed archaeological variability, and suggest promising avenues for exploring other explanations.

     In summary, my proposed research in the Cochabamba Valleys will make the following contributions to South-Central Andean archaeology:

(1) Provide a systematic diachronic investigation of the settlement dynamics in the transition from the Formative to Tiwanaku period in Cochabamba (ca. AD 500-700);

(2) Assess how such interaction is related to agricultural features of Cochabamba, and present a systematic database to begin delineating the consequences of such interests in the local sociopolitical and economic organization;

(3) Generate a specific and useful comparative case study on interregional interaction of a large-scale Andean polity that will enhance our understanding of the most important prehistoric polities of the South-Central Andes;

(4) Produce a long overdue regional understanding of Middle Horizon settlement in Cochabamba: the A.D. 300-1000 period remains the least investigated period of Cochabamba prehistory;

(5) Help refine the Cochabamba ceramic sequence with a systematic ceramic analysis; and

(6) Help connect chronologically the current important research efforts in Cochabamba that focus on the early Formative Period (Brockington) and future research on the Late Intermediate/Late Horizon (Morris).

       Hopefully, future work will build on the evidence provided by this research in order to gain a better understanding of prehispanic imperialism, political economy, and intersocietal interaction.


     The research is planned for a 12-month field season, from September 1993 through August 1994:

1) 15 September-7 October: Cochabamba. Administrative details; examination of aerial photography and preparation of the survey maps;
2) 8 October-11 February: Independencia zone. Survey, surface collections, and mapping of selected sites; preliminary artifact analysis of sherd collections and spatial density distributions; excavation of test pits;
3) 15 February-21 June: Aiquile zone. Same tasks as in previous zone;
4) 25 June-14 September: Cochabamba. Laboratory analysis: artifact processing (weights, counts, drawings, context documentation, and typological analysis of sherd collection lots).

     This research will have support of the Instituto de Investigaciones Antropológicas y Arqueológicas and Museo de Arqueología of the Universidad Mayor de San Simón of Cochabamba. The project was discussed with Cochabamba authorities during the pilot season in 1992. Permission for the research was granted in November 1992. Mr. R. Céspedes, member of the Museum, will act as project co-director and will participate actively in this project. Collections will be permanently stored at the Museum, and available for study to other scholars.

Copyright © 1998-01, Alvaro Higueras. Derechos Reservados/All rights reserved.
Please send comments on content and presentation to
URL of this document: http://www.tiwanakuarcheo.net/1_main/nsf93.html
Revised: 29 May 1999