Journal of Archaeological Science 38 (2011) 2897e2907

Contents lists available at ScienceDirect

Journal of Archaeological Science

journal homepage: http://www.elsevier.com/locate/jas

Geologic constraints on rain-fed Qocha reservoir agricultural infrastructure, northern lake Titicaca Basin, Peru
Nathan Craig a, *, Mark S. Aldenderfer b, Catherine A. Rigsby c, Paul A. Baker d, Luis Flores Blanco e
a

Department of Anthropology, 409 Carpenter Bldg., Pennsylvania State University, University Park, PA 16802, USA University of California Merced, School of Social Sciences, Humanities and Arts, Merced, CA 95343, USA c Department of Geology, East Carolina University, Greenville, NC 27858, USA d Nicholas School Division of Earth and Ocean Sciences, 301 Old Chemistry, Duke University, Durham, NC 27708, USA e Universidad Nacional Mayor de San Marcos, Lima, Peru
b

a r t i c l e i n f o
Article history: Received 3 December 2010 Received in revised form 4 May 2011 Accepted 13 May 2011 Keywords: Andes Agriculture Remote sensing GIS

a b s t r a c t
This paper reports new data on qocha ponds from the Rio PucaraeAzngaro interuvial zone, northern Lake Titicaca Basin, Peru. Qocha are a little known form of Andean agriculture that developed around 800e500 B.C. and remain in use today. Prior estimates suggested that in the study area, there were more than 25,000 qocha. While most Andean sunken beds are excavated to reach groundwater, qocha are rainfed ponds. How these rain-fed ponds functioned has been an open question, but one that is answered in part by research presented in this paper. We suggest that a thick impermeable stratum of clay that was possibly deposited by paleolake Minchin created a perched water table that makes rain-fed qocha reservoir agriculture possible. Field geology shows that within the study area, this stratum only exists under Terrace E. Based on this model, we hypothesized that persistently used qocha should only be found on Terrace E. To test this hypothesis we used remotely sensed data to inventory qocha and to determine their distribution by each terrace present. We identied 11,737 qocha. By area 93.77% and by count 94.33% of the qocha are located on Terrace E. These results strongly supported our hypothesis. This case study illustrates that the long term viability of this form of agriculture is made possible by a physical context that is beyond human control. 2011 Elsevier Ltd. All rights reserved.

1. Introduction This paper reports new data from the northern Lake Titicaca Basin regarding a type of poorly understood precolumbian agricultural infrastructure that remains in use todaydthe rain-fed qocha reservoir. Cultural processes such as the development and long term use of agricultural infrastructure must be considered in light of how they are permitted and constrained by spatio-temporally variable physical contexts that are beyond human control. We provide an example of this relationship through a case study from the PucaraeAzangaro interuvial area of the northern Lake Titicaca Basin. We hypothesize that in this area, lacustrine clays formed by paleolake Minchin create a perched water table, and this geologic phenomenon makes rain-fed qocha pond agriculture hydrologically possible. Given this model, we hypothesize that persistently used

qocha are restricted to those areas where the paleolake clays are extant. We test this hypothesis by using remote sensing to inventory qocha and evaluate their distribution in relation to the extent of the paleolake clays. First we introduce relevant background information that highlights the anthropological importance of the region under study and summarize the relevant prior knowledge regarding qocha. We then continue by presenting our geologic model and specifying the hypothesis that derives from it. Following this we present our methods and results. These data are then brought to bear on the hypothesis that derives from our model. 2. Background With respect to geography, high resolution climate reconstructions, the origins of agriculture, and the development of sociopolitical complexity, the Lake Titicaca Basindand the northern basin in particulardis an extremely important region. By volume of water, Lake Titicaca is the largest lake in South America (WWDR 2003). Laminated sediments from the main basin of Lake Titicaca provide a high resolution, long term, multi-proxy climate record

* Corresponding author. Tel.: 1 814 863 4388. E-mail addresses: ncraig@psu.edu (N. Craig), maldenderfer@ucmerced.edu (M.S. Aldenderfer), rigsbyc@mail.ecu.edu (C.A. Rigsby), pbaker@duke.edu (P.A. Baker), Lores78@gmail.com (L.F. Blanco). 0305-4403/$ e see front matter 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.jas.2011.05.005

2898

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907

(Baker et al., 2001a,b). Five major rivers drain into Lake Titicaca. Located in the northern basin, and formed at the conuence of the Ros Pucara and Azngro, the Ro Ramis has the greatest discharge of any of these ve drainages (Fig. 1). Recent studies indicate that the northern basin is a probable location for the domestication of potatoes and chenopods (Bruno, 2006, 2008; Rumold, 2010; Spooner et al., 2005). Considered as a whole, the Lake Titicaca Basin is a largely autonomous area of ancient state formation (Stanish, 2001, 2003). Within this region, the earliest evidence for sustained socio-political complexity comes from the archaeological site of Pukara (Klarich, 2005, 2009), located in the northern basin to the west of the Ro Pucara. Immediately to the south of the site of Pukara, in the area between the Ros Pucara and Azangaro, lies the system of intensied agriculture that consists of reservoirs or ponds that are known as qochas. Qochas from this area were initially described as occurring in the mesopotamia or chapwimayu of the Ros Pucara and Azngaro (Flores Ochoa and Paz Flores, 1983a: 69, 72, 1984: 68, 88, 93). In Quechua, chapwi means middle, center, or between and mayu means river. This chapwimayu is host to the single largest concentration of qochas known anywhere in the Andes (Erickson, 2000: 341), and this concentration of qochas has been linked to the rise of the Pukara polity. This interuvial zone, from the conuence of the Ros Pucara and Azngro northward past the town of Pucara, forms our study area (Fig. 2). Many physical and social facets of qocha agriculture are documented in the pioneering work of Flores Ochoa and Paz Flores. Yet,

in 1999, qocha were still described as poorly studied infrastructure infraestructura muy poco estudiada (Valdivia et al., 1999: 147; authors translation), and it was argued that qochas were without doubt the least understood agricultural infrastructure of the Titicaca altiplano es sin duda la infraestructura agraria menos estudiada en el altiplano del Titicaca (Valdivia et al., 1999: 149; authors translation). To the best of our knowledge, since 1999, no new empirical studies of qocha agriculture in the northern Lake Titicaca Basin have been conducted or published. As of the year 2000, the origin and history of qochas remained largely unknown, and a comprehensive inventory of qochas in the PucaraeAzngaro interuvial did not yet exist (Erickson, 2000: 341). In this paper, we report an inventory of qochas that considers their distribution in relation to landscape evolution. In doing so, we provide a rened understanding of how natural and cultural processes interact with respect to this agricultural technology. First we briey review relevant prior research on qocha. This review both highlights earlier results that our study renes and identies the fact that how qocha ponds are capable of holding water remains, up until now, an unanswered question. This question is answered by our new geologically oriented model for qocha distribution. 2.1. Temporal context It has been suggested that within the Lake Titicaca Basin, associated, temporally diagnostic artifacts indicate that all of the major

Fig. 1. Overview showing the location of the study area and areas that are mentioned in the text as places where qochas are reported. The schematic hypothesized extent of paleolake Minchin clays is based on the presence of a ca. 40,000 year old lacustrine clay strata that is found 150 m above the modern level of Lake Titicaca (Farabaugh and Rigsby, 2005: 25). This idealized representation of paleolake Minchin was estimated by drawing a contour line 150 m higher than the modern level of Lake Titicaca. This schematic does not take into account the processes of tectonic tilting or isostatic rebound.

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907

2899

Fig. 2. Map showing the distribution of use zones as dened by Flores Ochoa and Paz Flores (Flores Ochoa, 1987: 283; Flores Ochoa and Paz Flores 1983a: 69e70, Map 3; 1983b: 134, 151, Fig. 14, 1984: 89 Mapa 3; 1986: 97; 1987: 87, 89e90; 1992: 262; Flores Ochoa et al., 1996: 254, Fig. 122; Valdivia et al., 1999: 147, 148, Fig. 1).

agricultural techniques, including qocha, appear to have developed by as early as 800 B.C., and probably before (Stanish, 2007: 130, 2006: 390). Qocha agriculture was rst formally reported in the early 1980s by Flores Ochoa and Paz Flores, (1983a,b). In the area where they examined qochas, they found few pieces of pottery but those recovered were generally diagnostic Pukara type fragments. Near Llallahua, where there is a major concentration of qochas (Fig. 2), researchers have reported the presence of a Pukara style carved monolith located on top of Mount Kojra (Flores Ochoa and Paz Flores, 1986: 103; Rowe, 1942: 72). Flores Ochoa and Paz Flores suggested that all the qocha in the PucaraeAzngaro interuvial are located within a day to a day and a half walk from the site of Pukara (Flores Ochoa, 1987: 286; Flores Ochoa and Paz Flores, 1983a: 75, 1983b: 139, 1984: 95, 1986: 103, 1987: 96; Washington and Alvarez, 1991: 170). The close spatial association of the qochas to Pukara, the presence of Pukara style polychrome pottery in direct association with the qochas, and the location of a Pukara style carved monolith near the qochas led these investigators to

suggest that qochas helped to supply the growing needs of the newly emerging urban center at Pukara (Flores Ochoa, 1987: 286e287; Flores Ochoa and Paz Flores, 1983a: 74e75, 1983b: 139, 1984: 95, 1986: 103e104, 1987: 95e96, 1992: 264; Flores Ochoa et al., 1996: 256; Washington and Alvarez, 1991: 171). Recent pedestrian survey in the PucaraeAzngaro interuvial sheds new light on associations between temporally diagnostic archaeological materials and qocha reservoirs. Aldenderfer and Flores Blanco (2008) identied a carved stone monolith that is located in the middle of a large qocha that is named Mara Huancane Qocha that is located adjacent to the Quechua community of Chiqchipani. The carvings on the monolith depict a stylized goggle eyed feline catsh water cat creature that is typical of QaluyuePukara transitional (ca. 750e200 B.C.) early Middle Formative motifs. Additionally, pedestrian survey of the Pucarae Azngaro interuvial zone (Aldenderfer and Flores Blanco, 2008) revealed that Qaluyu (ca. 800e500 B.C.; Browman, 1980; Chvez, 1977), QaluyuePukara transitional (ca. 750e200 B.C.; Chvez

2900

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907

Justo, 2008a: Fig. 14) and Pukara (ca. 800 B.C.eA.D. 200; Stanish 2003: 85e89; Klarich, 2005: 242) styles are the earliest types of pottery that are found in direct association with qochas. These three styles span the Middle Formative period (Stanish 2003: 90, Fig. 5.2). Limited sub-surface testing was undertaken at two sites that are directly associated with qochas (Aldenderfer and Flores Blanco, 2008): Tulani (RM621e622) and Laroqocha (RM1190, 1192, 1194e1195). Tulani is a 1 ha dense surface scatter that contains pottery, chipped stone debitage, obsidian, and stone hoe fragments. Following the typology of Chvez Justo (2008a,b), the surface pottery assemblage from the site consists of Formative (ca. 800 B.C.eA.D. 400), Early Huaa (ca. A.D. 100e600), Altiplano (A.D. 1100e1475), and Inca (A.D. 1475e1532) styles. Three 1 1 m sub-surface test pits were excavated into Tulani. These excavations revealed the presence of stone hoe fragments, burned camelid bone, obsidian projectile ponts, temporally diagnostic pottery, and numerous small ash lled pits. Following the typology of Klink and Aldenderfer (2005), the projectile points are Type 5B and 5D forms. Excavations recovered a single Qaluyu sherd; QaluyuePukara transitional and Pukara were the dominant pottery styles (Table 1). Early Huaa and Altiplano style fragments were present, but in much lower abundance than either QaluyuePukara or Pukara materials. Laroqocha is a 3.2 ha scatter of artifacts that is composed of pottery, chipped stone debris, and stone hoe fragments (Fig. 3). Surface pottery consists of MiddleeUpper Formative (ca. 800 B.C.eA.D. 200), Early Huaa, Altiplano, and Inca styles. Three 1 1 m sub-surface test units were excavated into Laroqocha. This sub-surface testing revealed the presence of fragments of burned camelid bone, concentrations of burned clay, chipped stone debitage, hoe fragments, and pieces of temporally diagnositc pottery. Excavations recovered eight Qaluyu sherds but the pottery assemblage is again dominated by QaluyuePukara and Pukara sherds; Huaa and Altiplano style materials were much less abundant (Table 1). Excavations also exposed an earth oven in association with Pukara pottery. Together, Tulani and Laroqocha reveal that signicant amounts of Middle Formative pottery and agricultural implements are found in direct association with qocha ponds (Fig. 3). Both Tulani and Laroqocha are located on river Terrace E (see below). While stylistically diagnostic pottery reveals associations between QaluyuePukara transitional (ca. 800e500 B.C.) and the qochas of the PucaraeAzngaro interuvial, somewhat later Pukara (ca. 800 B.C.eA.D. 200) style pottery is far more abundant. The presence of Qaluyu and QaluyuePukara pottery, both earlier Middle Formative styles, in association with these qochas indicates that in the PucaraeAzngaro interuvial this agricultural technology likely predates the apogee of the Pukara polity. However, the greater abundance of Pukara pottery in association with these qochas indicates that the ponds were probably used more intensively during the height of the Middle Formative Period (ca. 500 B.C.). Material from later traditions including Huaa, Altiplano, and Inca are also found in direct association with the PucaraeAzangaro qochas. These associations with later temporally diagnostic styles indicates that qocha infrastructure remained in use well after the decline of the Pukara tradition. However, the lower abundance of post-Pukara forms suggests that after the decline of Pukara the intensity of qocha use probably declined.
Table 1 Diagnostic pottery types recovered from excavations at Tulani and Laroqocha. Site Tulani Laroqocha Qaluyu 1 8 QaluyuePukara 215 376 Pukara 549 601 Huaa 5 11 Altiplano 17 6

Immediately to the east of the present study area, in the HuancanePutina river valley, and to the south in the JuliePomata region, pedestrian survey indicates that qochas in these other areas were also in use by the Middle Formative (Stanish and Steadman, 1994; Stanish et al., 1997; Stanish, 1994, 2006: 384). Qochas located on the Huatta Pampa to the north (Erickson, 2000: 341) and in the Tiwanaku Valley to the south (Albarracn-Jordan, 1996) are reportedly associated with multicomponent occupations that span from the Middle Formative to the present. These multiple associations with temporally diagnostic artifact styles indicate that prior to Spanish contact, qochas were used for a long period of time. Unlike other prehispanic agricultural technologies that were abandoned after contact (i.e. raised elds), the modern indigenous inhabitants of the region continue to use qocha. This continued use of the technology is further indication that in some contexts qocha infrastructure can be used over the long term. However, we recognize that over time the nature and intensity of qocha use may have shifted. 2.2. Qocha water Sources It has been suggested that in the PucaraeAzngaro interuvial region, the groundwater is too deep for qocha to function as wells (Flores Ochoa and Paz Flores, 1983a: 57; 1983b: 130, 1984: 79; 1986: 93; 1987: 82; Kendall and Rodrguez, 2002: 242; Valdivia et al., 1999: 152; Washington and Alvarez, 1991: 166). Flores Ochoa and Paz Flores (1987): 277 states that, [t]he substantial depth at which the subterranean water table is encountered eliminates the possibility of utilizing water from the water table for agriculture (authorstranslation). For those few wells that have been excavated in the region, the groundwater is found at a depth of 4e15 m. The fact that qochas do not function as wells stands in contrast to the majority of archaeological and contemporary sunken bed agricultural elds from Peru (Soldi 1982: 17; Denevan, 1980: 631e632). These sunken beds are located along the coast of Peru, and nearly all known examples of these are excavated to groundwater or the phreatic zone. Examples of Peruvian groundwater wells include: hoyas, mahames, sachaques, totorales, balsares, melgas, canchones, and occasionally puquios, which are actually subterranean irrigation systems (Denevan, 1980, 2001; Moseley, 1969; Parsons and Psuty, 1974, 1975; Rowe, 1969; Schreiber and Rojas, 1988, 2003; Soldi 1982). Rather than basins designed to tap into groundwater, the use of qochas is based on rain water which is stored in and transferred between the reservoirs or ponds through open channels (Flores Ochoa, 1987: 277; Flores Ochoa and Paz Flores, 1983a: 57e58; 1983b: 130; 1984: 79; 1986: 93; 1987: 82; 1992: 255; Flores Ochoa et al., 1996: 247; Rozas 1987: 110; Valdivia et al., 1999: 152; Washington and Alvarez, 1991: 166; Denevan, 2001: 163). Water can be stored in a qocha for months after a rain. How it is possible to store water in these ponds for extended periods of time without loss of water through percolation remains unspecied by previous descriptions of qochas. Our geology-based model provides an explanation as to why water can be contained in the qocha ponds for extended periods of time. This model is based on the premise that the distribution of rain-fed qochas is controlled by the presence of a stratum of a buried, impermeable sediment. Before turning to our model we discuss the results of prior attempts to count the total number of qochas and to estimate their density within the study area. 2.3. Estimates of the count and density of Qocha Flores Ochoa and Paz Flores attempted to count qochas using a combined approach of eld visits, aerial photographs taken in 1961 by Servicio Aereofotogrco Nacional (SAN), and 1:100,000

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907

2901

Fig. 3. Map showing the location of Laroqocha (RM1190 and RM1192) and a concentration of qocha. The boundary of Laroqocha is shown with dashed red lines. The background imagery is a 1 m spatial resolution IKNOS-2 satellite scene that was captured on 20 June 2001. Qocha reservoirs are visible throughout the image as round features. Some of the qocha contain water while other are dry. Also visible are small canals that connect various qocha reservoirs. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

maps (Flores Ochoa, 1987: 282e283; Flores Ochoa and Paz Flores, 1983a: 69; 1983b: 134; 1984: 88e89; 1986: 97; 1987: 87; Flores Ochoa et al., 1996: 254). They estimate that the study region has a total area of 528 km2, and that about 384 km2 of this is suitable for cultivation and pasture. According to their classication scheme: larger qocha are >200 m dia and >31,400 m2 in area; the smaller and more numerous qocha have a mean area of 6500 m2. Based on a 0.5 km2 sample, Flores Ochoa and Paz Flores estimated the total number of qochas present in the chapwimayu of the Ros Pucara and Azngaro (Flores Ochoa, 1987: 283e284; Flores Ochoa and Paz Flores, 1983a: 71; 1983b: 135; 1984: 91; 1986: 98; 1987: 89; 1992: 262; Flores Ochoa et al., 1996: 254). They counted close to twenty qochas in 0.5 km2, and from this count they estimated that there are 80 qochas in 1 km2. They assumed that qochas are circular, and modeled qocha size as 60 m dia. This gave each qocha an area of approximately 2863 m2 (We note that the correct solution is 2827 m2. However, we also observe that given a rounding operation that these authors later applied to the estimated total number of qocha, the difference in area is insignicant). Based on this estimate, there are 0.226 km2 of qocha area per km2. They assumed that the study area was 256 km2, and from their 0.5 km2 sample they calculated a regional total of 57.88 km2 of qocha area. Dividing this by the estimated area of a single qocha (2863 m2) produced an estimated 20,215 qochas. The authors rounded this gure up to 25,000, and suggested it served as a conservative

estimate of the total number of qochas in the study area (Flores Ochoa, 1987: 284; Flores Ochoa and Paz Flores, 1983a: 71; 1983b: 135; 1984: 91; 1986: 98; 1987: 89; 1992: 263; Flores Ochoa et al., 1996). Based on estimates of the density of qochas per km2, Flores Ochoa and Paz Flores dened three different zones (Fig. 2; Flores Ochoa, 1987: 283; Flores Ochoa and Paz Flores, 1983a: 69e70, Map 3; 1983b: 134, 151, Fig. 14, 1984: 89 Mapa 3; 1986: 97; 1987: 87, 89e90; 1992: 262; Flores Ochoa et al., 1996: 254, Fig. 122; Valdivia et al., 1999: 147, 148, Fig. 1): 1) Disused or Damaged 2) Spaced 3) Numerous. The Disused or Damaged zone is located to the west of the study area, and they suggest that this area has an extent of about 128 km2. Interpretation of aerial photographs led them to suggest that in this zone, salinization may have led to the deterioration of the qochas. They also note that several haciendas are present in this area including Qoa (Ccoa), Pichacani, and Cojempati. This led Flores Ochoa and Paz Flores to offer the alternative hypothesis that agricultural intensication on the part of the haciendas and the use of mechanized agriculture may have contributed to the deterioration of qochas in this zone. By attening and leveling the ground surface,

2902

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907

mechanized agriculture can physically damage qochas. The Spaced zone is dened by areas where there are <100 qochas per km2. They estimate that the Spaced zone has an extent of approximately 96 km2. It is located on the eastern side of the Ro Azngaro in the areas known as Tuturco, Juyacache, Huayrapata, and the exhacienda Corpa in front of the town of Achaya and in the general area of Calapuja. The Numerous zone has a density of >100 qochas per km2, and this zone covers an estimated 160 km2. The Numerous Zone is found around the communities of Iquilo, Mataro Grande, Llallahua, the former hacienda of Sullata, and in the Laro and Charamicaya pampas. Having reviewed prior attempts to count and estimate the distribution of qocha we now turn to relevant aspects of the geologic setting of the region. 3. Geologic setting and expectations for Qocha location Fluvial geomorphological research in the interuvial or chapwimayu of the Ros Pucara and Azngaro indicated the presence of ve river terraces and three major facies associations (Farabaugh and Rigsby, 2005). Termed AeE, A is the lowest terrace above the active oodplain and E is the highest (Fig. 4). The entire modern surface of the area between these two rivers is covered by a combination of braided and meandering uvial sediments that form two of the regions three major facies associations. The deposition and erosion of these two uvial deposits is strongly correlated with periods of high and low precipitation respectively (Baker et al., 2001a). The third facies association is only found underlying Terrace E, and it is always found directly below the aforementioned uvial deposits. This third facies association is composed of a meter-thick accumulation of thinly laminated gray to blueegray clays that bear gastropods, ostracods, and abundant organic material (Fig. 5; Farabaugh and Rigsby, 2005: 19). These clays were deposited by the settling of ne particles suspended in a lacustrine environment (Farabaugh and Rigsby, 2005: 20). Carbon from this clay stratum dates to 40,170 430 cal B P. Although near the limit of nite radiocarbon dating, we trust that this single radiocarbon date corresponds to a wetter period from 46,000e36,000 cal yr B P (Clapperton, 1993; Clapperton et al., 1997; Fritz et al., 2004) during which palaeolake Minchin extended across vast areas of the Andean altiplano (Baker et al., 2001a; Fritz et al., 2004). The blueegray lacustrine clay stratum underlying Terrace E is today found 40 m above the current level of the Ro

Ramis and about 150 m above the modern level of Lake Titicaca (Farabaugh and Rigsby, 2005: 25). This implies that the water level of paleolake Minchin was at least 150 m above the modern level of Lake Titicaca (Fig. 1). We believe that subsequent tectonism, tilting due to folding and faulting, differentially elevated the northern shores of the Lake Titicaca basin since the original deposition of the lacustrine clay. We hypothesize that the extant lacustrine clays deposited in paleolake Minchin are responsible for creating a perched water table, and it is this geologic phenomenon that makes rain-fed qocha pond agriculture hydrologically possible. If our model is correct, then the large majority of qochas should be located on Terrace E where the paleolake Minchin clay is present. Experimentation and attempts to expand this agricultural infrastructure may have resulted in the presence of a few qochas on lower terraces. 4. Methods for inventorying Qocha and evaluating geologic expectations for their location To determine 1) the total number and density of qochas and 2) whether or not they are restricted spatially to Terrace E, the entire lower interuvial zone of the Ros Pucara and Azangaro was examined by inspecting satellite and aerial remote sensing imagery. In any remote sensing survey, the ability to detect features is a function of the size of the targets of interest and the spatial resolution of the imagery (Craig and Chagnon, 2006; Jensen, 1996; McGwire et al., 1996: 102e105, Fig. 6.2; Sabins, 1987). In addition to prior publications, eld survey and GPS data that we collected from the interuvial region indicated that qochas were generally larger than 30 m. We had access to 0.5 m spatial resolution aerial photographs that covered the entire study area. Thus, the size of the features and the resolution of the imagery rendered qocha reservoirs well suited for mapping with satellite and aerial remote sensing imagery. Advanced Spaceborne Thermal Emission and Reection Radiometer (ASTER) level 3 orthorectied imagery covering the study area was obtained from the Land Process Distributed Active Archive Center (LP DAAC). An image captured on 15 August 2007 covered the entire study area in a single cloud-free scene. From this multispectral dataset, a VNR 1,2,3N band false color composite image was created and inspected. Declassied CORONA imagery was obtained from the USGS as a lm positive. The CORONA image

Fig. 4. Schematic diagram illustrating the terrace sequence of the Rio Ramis valley. Figure adapted from Farabaugh and Rigsby (2005): Figs. 6 and 12.

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907

2903

Fig. 5. Photograph showing the thick stratum of laminated gray to blueegray lacustrine clays. This photograph is from the Rio Iquilo B measured section reported by Farabaugh and Rigsby (2005): Fig. 3. (For interpretation of the references to color in this gure legend, the reader is referred to the web version of this article.)

was captured 30 June 1979 (mission number 1215-5, Ops designation 00473). The image was taken with the KH-9 lm camera, which has a reported spatial resolution of 20e30 ft (6e9 m). The lm was scanned at 3000 dpi and georeferenced to the orthorectied ASTER false color composite. We purchased 1 m spatial resolution IKONOS-2 satellite image that was captured on 20 June 2001 (Fig. 3). Printed paper aerial photographs covering the study area from August of 1979 that are available through Programa Especial de Titulacin Tierras (PETT) were scanned at 3000 dpi and georeferenced to the ASTER and CORONA images. Examination of the scanned PETT imagery indicated that the aerial photographs had a spatial resolution of 0.5 m. This resolution is more than

sufcient to detect qochas (e.g. McGwire et al., 1996: 104). Field checking showed that anomalies interpreted in these air photos were in fact qochas. The georeferenced IKONOS-2 scene and the PETT images were visually inspected and all detectable qochas were digitized into a geographic information system (GIS) as polygon features. In the imagery, qochas were identied based on two primary criteria: 1) the anomaly appeared to be a basin and 2) canals were visible either entering or exiting the basin-like anomaly. In most cases, the basin-like anomaly bore evidence of recent cultivation. In some cases, the basin-like anomalies appeared to be no longer under cultivation. These anomalies were interpreted as qocha which are

2904

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907

no longer in use. Some of these basin-like anomalies were extremely faint, and a conclusive identication could not be made. Field checking conrmed that these features likely represent abandoned and badly eroded qocha. These extremely faint basinlike anomalies were not inventoried. River terrace breaks were identied and dened through eld geology (Farabaugh and Rigsby, 2005). These terraces are large features that are easily observed in the satellite and aerial imagery. The terraces were digitized from inspection of the ASTER, CORONA, IKONOS-2, and PETT images. From the GIS polygon coverages we determined: qocha count, qocha area, terrace extent, and terrace area. This allowed us to calculate: qocha count by terrace, qocha area by terrace, and mean qocha area. These calculations provide a test of the hypothesis that persistently used qochas should only be located on Terrace E where the paleolake Minchin clay is present. We also provide revised density calculations that are based on both qocha count and qocha area.

5. Results of the inventory Within the survey area, a total of 11,737 qochas were remotely identied and digitized as polygons (Fig. 6 and Tables 1 and 2). In terms of the area of these polygons, 99.5% of the qochas fall within two standard deviations of the mean. 57 or 0.49% of the qochas have areas that are larger than two standard deviations of the mean. We characterize these as exceptionally large qocha (area >2s). In terms of either counts or proportions of the total survey area, it is useful to retain the exceptionally large qochas in the sample. Outlier removal helps to reduce skew, and this operation is important for equations that assume a normal distribution. We present both transformed and untransformed data since they provide distinct insights into qocha size. Removing outliers, the mean declines by 178.43 m and it shifts towards the median (Table 2). Outlier removal decreases the standard deviation by 4903.92 m; the untransformed value is 3.45 times this gure. Outlier removal reduces the skewness from 81.99

Fig. 6. Map showing the limit of the study area, the ve river terraces, the qochas, and places mentioned in the text.

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907 Table 2 Summary statistics of area for all qochas with and without outliers included. Sample N Min Max Sum Mean Std. error Variance Stand. dev Median 25 prcntil 75 prcntil Skewness Kurtosis Geom. mean Outliers included 11,737 67.9318 682,474 33,456,500 2850.52 63.7408 47,686,100 6905.52 2095.67 917.221 4641.52 81.9947 7998.25 2186.08 Outliers excluded 11,680 67.9318 16586.2 31,210,100 2672.09 18.5206 4,006,390 2001.6 2089.08 3243.4 1403.91 2.41332 8.25228 2159.55 Table 4 Density values based on counts and areas. Qocha Qocha Qocha Qocha Qocha Qocha Qocha Count/Survey Area Count/Terrace C Count/Terrace D Count/Terrace E Count/Terrace CeE Area Area/Survey Area Area/Terrace CeE 22/km2 0.6/km2 26/km2 62/km2 44/km2 6% 13%

2905

to 2.41. Even with outlier removal, the distribution remains positively or right-skewed. All of the qochas are located on Terraces CeE (Table 3). Together these terraces constitute half (50.6%) of the total survey area. Terrace E is a little more than a third (34.7%) of the total survey area, but the limits of this terrace contain 94.33% of the total qocha count and 93.76% of the total qocha area. Thus, 35% of the study area contains 94% of the qochas. Considering qocha counts by terrace area, densities ranged from 0.6e62 qochas/km2. Qocha density is highest on Terrace E (Table 4). With respect to their area, qocha constitute 6% of the study area. In terms of their area, qocha comprise >1% of Terrace C, 8% of Terrace D, and 17% of Terrace E. Exceptionally large qocha (area>2s) are found on each of these three terraces. 6. Discussion of the results The scanned PETT imagery that we used to detect the qochas has a spatial resolution of 0.5 m, thus each pixel covers an area of 0.25 m2. Once the exceptionally large outliers are removed, qochas have a mean area of 2672 m2. This indicates that typically a qocha is represented by about 10,000 pixels. Given that the distribution is still positive or right-skewed even after removing outliers, most qochas are probably represented by more than 10,000 pixels. At 0.5 m spatial resolution, qochas are easy to spot and dene. While, we do not claim to have dened the absolute total number of qocha, we believe that our count is an extremely close approximation of the actual number of qocha present in the study area. Prior estimates suggested that there were over 25,000 qochas in the PucaraeAzngaro interuvial zone. Our remote sensing survey of this area produced 11,737 qochas, or roughly half the prior estimate. Previously, it was estimated that >100 qocha/km2 were present in the Numerous Zone. The highest density of qochas

Table 3 Total survey area, individual terrace areas, qocha count, qocha area, and qocha counts and areas by terrace. Feature Survey zone Terrace A Terrace B Terrace C Terrace D Terrace E Qocha All Qocha by C Qocha by D Qocha by E Area (km2) 519.486 4.478 10.207 61.357 21.159 180.405 33.457 0.111 1.705 31.369 Number % Total area 100 0.862 1.965 11.811 4.073 34.728 100 0.332 5.095 93.762 % Total count

(62 qocha/km2) is located on Terrace E. Our inventory indicates that qocha density by count is considerably lower than prior estimates. Even on Terrace E, the area of greatest qocha density, the ponds comprise less than 15% of the space. In light of this revised inventory, let us now return to our geologic model and the expectation regarding qocha distribution that derives from it. In the remote past, palaeolake Minchin apparently extended across a large portion of the Altiplano including the interuvial of the Ros Pucara and Azngaro (Baker et al., 2001a; Fritz et al., 2004). Within the PucaraeAzangaro interuvial, palaeoake Minchin left a thick deposit of laminar blueegray lacustrine clays (Farabaugh and Rigsby, 2005). This deposit now underlies the highest Terrace E (Farabaugh and Rigsby, 2005: 24e26, Fig. 12). Elsewhere in the interuvial region, within the mapped boundaries of the surfaces of Terraces DeA, river downcutting and erosion has removed this lacustrine unit. With respect to our present study area, we believe that the impermeable nature of the lacustrine clay deposit is responsible for creating a perched water table that makes qocha agriculture possible. If this hypothesis is correct, then qochas should be located on Terrace E. Experimentation by early agropastoralists may have led to the expansion of qochas into the lower Terraces DeA, but our hypothesis dictates that the large majority of the qochas should be located on Terrace E. The distribution of qochas conforms well to our hypothesis. Whether considered by count or by area, more than 90% of the qochas are located on Terrace E. The existence of 553 qocha on Terrace D and 39 qocha on Terrace C might appear to stand in contrast to our expectation that qochas should be located on Terrace E. However, eld observations and remote imagery indicate that the qochas on Terraces D and C are largely abandoned. The extremely faint un-denable basin-like anomalies described earlier were located within the limits of Terraces D and C. Notably, these two terrace are within the Disused or Damaged Zone that was dened by Flores Ochoa and Paz Flores (compare Figs. 2 and 6). In light of this, we suspect that qochas were created on Terraces D and C during a period of agricultural expansion. Following this period, qochas within this zone were abandoned. Periodically, people may re-attempt use of qochas on these terraces. Yet compared to Terrace E, use of qochas on Terraces D and C was limited in both scope and duration. Further investigation of this zone is required to more fully understand what we believe represents a uctuating use of Terraces D and C for qocha cultivation. 7. Conclusions Qaluyu (ca. 800e500 B.C.; Browman, 1980; Chvez , 1977), QaluyuePukara transitional (ca. 750e200 B.C.; Chvez Justo, 2008a: Fig. 14), and Pukara (ca. 800 B.C.eA.D. 200; Stanish, 2006: 85e89; Klarich, 2005: 242) are the earliest pottery styles found in association with qocha ponds that are located on Terrace E. Near Llallahua, a Pukara style caved monolith was is found in association with qochas (Flores Ochoa and Paz Flores, 1986: 103; Rowe, 1942: 72). Recent pedestrian surface survey located a QaluyuePukara

11,737 39 553 11,071

100 0.332 4.712 94.326

2906

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907 Smith, B.D. (Eds.), Documenting Domestication: New Genetic and Archaeological Paradigms. University of California Press, Berkeley, pp. 32e45. Bruno, M.C., 2008. Waranq waranqa: ethnobotanical perspectives on agricultural intensication in the lake Titicaca Basin (Taraco Peninsula, Bolivia). Ph.D., Dissertation, Washington University. Chvez, K.L.M., 1977. Marcavalle: the ceramics from an early horizon site in the valley of Cusco, Peru, and implications for south highland socio-economic interaction. Ph.D., Dissertation, University of Pennsylvania. Chvez Justo, C., 2008a. Anlisis de la Cermica del Sector Medio y Bajo de la SubCuenca del Ro Huancan (Puno-Per). Chvez Justo, C., 2008b. Anlisis de la Cermica de los Sitios RM-348, RM-621, y RM-1192 del Proyecto Arqueolgico Ramis 2007. Clapperton, C.M., 1993. Glacier readvances in the Andes at 12,500e10,000 yr BP: implications for mechanisms of late glacial climatic change. Journal of Quaternary Science 8, 197e225. Clapperton, C.M., Clayton, J.D., Benn, D.I., Marden, C.J., Argollo, J., 1997. Late quaternary glacier advances and palaeolake highstands in the bolivian altiplano. Quaternary International 38e39, 49e59. Craig, N., Chagnon, N., 2006. Horticultural investment, transportation, and site spacing: remote sensing of Yanomam landscape features as a test case. In: Sellet, F., Greaves, R., Yu, P.L. (Eds.), Archaeology and Ethnoarchaeology of Mobility. University of Florida Press, Miami, pp. 44e74. Denevan, W., 1980. Tipologia de Conguraciones Agricolas Prehispanicas. America Indigena 40 (4), 619e652. Denevan, W., 2001. Cultivated Landscapes of Native Amazonia and the Andes. Oxford University Press, Oxford. Erickson, C.L., 2000. The lake Titicaca Basin: a precolumbian built landscape. In: Lentz, D. (Ed.), Imperfect Balance: Landscape Transformations in the PreColumbian Americas. Columbia University Press, New York, pp. 311e356. Farabaugh, R.L., Rigsby, C.A., 2005. Complex response to holocene climate and baselevel changes in the Rio Ramis valley, Peru. Journal of Sedimentary Research 75, 12e28. Flores Ochoa, J., 1987. Cultivation in the qocha of the south Andean Puna. In: Browman, D. (Ed.), Arid Land Use Strategies and Risk Management in the Andes: A Regional Anthropological Perspective. Westview Press, Colorado Springs, pp. 271e296. Flores Ochoa, J.A., Paz Flores, P., 1983a. El Cultivo en Qocha en la Puna Sur Andina. In: Fries, A.M. (Ed.), Evolucion y Tecnologia de la Agricultura Andina. Instituto Indigenista Interamericano, Cuzco, pp. 45e79. Flores Ochoa, J.A., Paz Flores, P., 1983b. La Agricultura en Lagunas del Altiplano. awpa Pacha 21, 85e106. Flores Ochoa, J.A., Paz Flores, P., 1984. El Cultivo en Qocha en la Puna Sur Andina. In: Masuda, S. (Ed.), Evolucion y Tecnologia de la Agricultura Andina. Universidad de Tokyo, Tokyo, pp. 59e100. Flores Ochoa, J.A., Paz Flores, P., 1986. La Agricultura en Lagunas (Qocha). In: de la Torre, C., Burga, M. (Eds.), Andenes y Camellones en el Per Andino: Historia, Presente, y Futuro. Consejo Nacional de Ciencia y Tecnologia, Lima, pp. 85e106. Flores Ochoa, J.A., Paz Flores, P., 1987. Agricultura en Lagunas (Qocha). In: Universo, E. (Ed.), Cultura y Tecnologa Altoandina, Corporacin de Fomento y Promocin y Exonmia de Puno, pp. 73e97. Lima. Flores Ochoa, J.A., Paz Flores, P., 1992. Une (Re)dcouverte Rcente: La Culture en tangs Temporaires (Qocha), sur LAltiplano. In: Morlon, P. (Ed.), Comprendre Lagriculture Paysanne dans les Andes Centrales: Perou-Bolivie. INRA, Paris, pp. 255e265. Flores Ochoa, J.A., Paz Flores, P., Rozas, J.W., 1996. Un (Re-)Descubrimiento Reciente: La Agricultura en Lagunas Temporales (Qocha) en el Altiplano. In: Morlon, P. (Ed.), Comprender la Agricultura Campesina en los Andes Centrales: PerBolivia. Institut Franais dtudes Andines y Centro de Estudios Regionales Andinos Bartolom de las Casas, Lima, pp. 247e256. Fritz, S.C., Baker, P.A., Lowenstein, T.K., Seltzer, G.O., Rigsby, C.A., Dwyer, G.S., Tapia, P.M., Arnold, K.K., Ku, T.-L., Luo, S., 2004. Hydrologic variation during the last 170,000 years in the southern hemisphere tropics of south America. Quaternary Research 61 (1), 95e104. Jensen, J.R., 1996. Introductory Digital Image Processing: A Remote Sensing Perspective. Prentice Hall Series in Geographic Information Science. Prentice Hall, Upper Saddle River. Kendall, A., Rodrguez, A., 2002. Las Qochas Andinas: Una Solucin para Mitigar el Riesgo Agropercuario y Domstico en la Sierra del Per. In: Viqueira, J.P. (Ed.), Antologa Sobre Qequeo Riego. : Sistemas de Riego no Convencionales, vol. III. Colegio de Postgraduados en Ciencias Agrcolas, Texcoco, MX, pp. 241e255. Klarich, E., 2005. From the monumental to the mundane: dening early leadership strategies at late formative Pukara, Peru. Dissertation, University of California, Santa Barbara. Klarich, E., 2009. Pukara: Investigaciones de la Temporada 2001 y un Nuevo Modelo para el Desarrollo del Sitio. In: Zi1kowski, M.S., Jennings, J., Augusto Belan Franco, L., Drusini, A. (Eds.), Arqueologa del rea Centro Sur Andina: Actas del Simposio Internacional 30 de Junio-2 de Julio de 2005 Arequipa, Peru,, vol. 7. Instituto Francs de Estudios Andinos, Lima, pp. 283e304. Klink, C., Aldenderfer, M.S., 2005. A projectile point chronology for the southcentral andean highlands. In: Stanish, C., Cohen, A.B., Aldenderfer, M.S. (Eds.), Advances in Titicaca Basin Archaeology, vol. 1. Costen Institute of Archaeology, Los Angeles, pp. 25e54. McGwire, K., Chagnon, N.A., Carias, C.B., 1996. Empirical and methodological problems in developing a GIS database for Yanomam Tribesmen located in remote areas. In: Aldenderfer, M.S., Maschner, H. (Eds.), Anthropology, Space,

style stone monolith that was placed inside a large qocha named Mara Huancane Qocha which is located near the community of Chiqchipani. Excavations at Tulani (RM621e622) and Laroqocha (RM1190, 1192, 1194e1195) produced an abundance of Middle Formative ceramics and stone hoes. Both of these sites are situated on Terrace E and are found in very close association with a dense concentration of qochas. Together, these associations suggest that on Terrace E, use of qochas began by at least 500 B.C. and probably earlier. Within this same area, pottery diagnostic to all major time periods is found in association with qochas (Aldenderfer and Flores Blanco, 2008). Based on pottery styles, Pukara appears to represent the most intense use of qochas. Pottery styles that post-date the Middle Formative are much less abundant. This suggests that qocha were used less intensively after the decline of Pukara. Yet, the qocha on Terrace E are still used today by modern Quechua agropastoralists. On Terrace E, the multiple associations with ancient and modern material culture indicate that qocha have been used, perhaps in different ways and probably with varying degrees of intensity, over a period of 2500 years. Yet the capacity for the prolonged use of this agricultural technology is dictated by geologic conditions that are well beyond human control. Despite attempts to expand qocha cultivation into Terraces D and C, in these contexts, the viability of this agricultural technology appears limited. The success of long term qocha use is due in part to the presence of an ancient lake that left a relatively impermeable strata of clay, and this strata creates a perched water table that makes qocha agriculture possible. Qochas are known for other parts of the Andes (Fig. 1). Many of these qocha groupings are located in relatively at areas near modern lakes. The techniques presented in this paper provide a method for inventorying qochas in other parts of the Andes. If our guiding perched water table hypothesis is correct, then other persistently used qocha may be found in areas that are underlain by ancient lake clays or other relatively impermeable deposits. Acknowledgments The authors would like to thank Honorato Tacca, Albino Pilco Quispe, and Silvia Roman for their help with eld work. We would also like to thank the Quechua communities of the PucaraeAzngro interuvial for their gracious hospitality. We are also grateful to Matthew Riggle for help with digitizing the qochas. Field work was performed under the authority of the National Institute of Culture, Resolucin Suprema No 004-2000-ED. Research was made possible by support from NSF grants BCS-0737793 awarded to Aldenderfer, EAR-0227999 awarded to Rigsby, and EAR-0227550 awarded to Baker. All computer processing took place at the Pennsylvania State University Anthropology Department GIS and Analytical Cartography Laboratory. Any errors or omissions are the sole responsibility of the authors. References
Albarracn-Jordan, J., 1996. Tiwanaku: Arqueologa Regional y Dinmica Segmentaria. Editores Plural, La Paz. Aldenderfer, M. S., Flores Blanco, L., 2008. Informe Final: Reconocimiento Arqueolgico de la Cuenca del Rio Ramis, Sub-regin de Azangaro, Regin Puno. Report submitted to Instituto Nacional de Culture. Baker, P.A., Rigsby, C., Seltzer, G.O., Fritz, S.C., Lowenstein, T.K., Miklas Bacher, P., Veliz, C., 2001a. Tropical climate changes at millennial and orbital timescales on the bolivian altiplano. Nature 409, 698e701. Baker, P.A., Seltzer, G.O., Fritz, S.C., Dunbar, R.B., Grove, M.J., Tapia, P.M., Cross, Scott L., Rowe, H.D., Broda, J.P., 2001b. The history of south american tropical precipitation for the past 25,000 years. Science 291, 640e643. Browman, D.L., 1980. Tiwanaku expansion and altiplano economic patterns. Estudios Arqueolgicos 5, 107e120. Bruno, M.C., 2006. A morphological approach to documenting the domestication of Chenopodium in the Andes. In: Zeder, M.A., Bradley, D.G., Emshwiller, E.,

N. Craig et al. / Journal of Archaeological Science 38 (2011) 2897e2907 and Geographic Information Systems, Spatial Information Series. Oxford University Press, New York, pp. 97e106. Moseley, M.E., 1969. Assessing the archaeological signicance of mahamaes. American Antiquity 34, 48e50. Parsons, J.R., Psuty, N., 1974. Agricultura de Chacras Hundidas en el Antiguo Peru. Revista del Museo Nacional 40, 31e54. Parsons, J.R., Psuty, N., 1975. Sunken elds and prehistoric subsistence on the Peruvian coast. American Antiquity 40, 259e282. Rowe, J., 1942. Sitios Histricos en la Regin de Pucara, Puno. Revista del Instituto Arqueolgico 6 (10e11), 66e75. Rowe, J., 1969. The sunken gardens of the Peruvian coast. American Antiquity 34, 320e325. Rozas, J.W., 1987. El Sistema ed Cultivo en Qocha. In: de la Torre, C., Burga, M. (Eds.), Andenes y Camellones en el Per Andino: Historia, Presente, y Futuro. Consejo Nacional de Ciencia y Tecnologia, Lima, pp. 107e126. Rumold, C., 2010. Illuminating womens work and the advent of plant cultivation in the highland Titicaca Basin of south America: new evidence from grinding tool and starch grain analyses. Ph.D., Dissertation, University of California, Santa Barbara. Sabins, F.F., 1987. Remote Sensing Principles and Interpretation, second ed. W. H. Freeman and Company, New York. Schreiber, K., Rojas, J.L., 1988. Los Puquios de Nasca: Un Sistema de Galeras Filtrantes. Boletn de Lima 10 (59), 51e62. Schreiber, K., Rojas, J.L., 2003. Irrigation and Society in the Peruvian Desert: The Puquios of Nasca. Lexington Books, Lanham. Soldi, A.M., 1982. La Agricultura Tradicional en Hoyas. Pontica Universidad Catolica de Peru, Lima.

2907

Spooner, D.M., McLean, K., Ramsay, G., Waugh, R., Bryan, G.J., 2005. A single domestication for potato based on multilocus amplied fragment length polymorphism genotyping. Proceedings of the National Academy of Sciences 102 (41), 14694e14699. Stanish, C., 1994. The hydraulic hypothesis revisited: lake Titicaca Basin raised elds in theoretical perspective. Latin American Antiquity 5 (4), 312e332. Stanish, C., 2001. The origin of State Societies in south America. Annual Review of Anthropology 30, 41e64. Stanish, C., 2003. Ancient Titicaca: the Evolution of Complex Society in Southern Peru and northern Bolivia. University of California Press, Berkeley. Stanish, C., 2006. Prehispanic agricultural strategies of intensication in the Titicaca Basin of Peru and Bolivia. In: Marcus, J., Stanish, C. (Eds.), Agricultural Strategies. Cotsen institute of Archaeology, Los Angeles, pp. 364e400. Stanish, C., 2007. Agricultural intensication in the Titicaca Basin. In: Thurston, T., Fisher, C. (Eds.), Seeking a Richer Harvest. Springer, New York, pp. 125e139. Stanish, C., Steadman, L., 1994. Archaeological survey in the Juli-Desaguadero area, Lake Titicaca basin, Peru Fieldiana Anthropologica No 23. Field Museum, Chicago. Stanish, C., Steadman, L., Justo, C.C., Frye, K., Mamani, L.O., Seddon, M.T., Chuquimia, P.C., 1997. Archaeological Survey in the Juli-Desaguadero Region of Lake Titicaca Basin, Southern Peru. Fieldiana p. 29. Valdivia, R.F., Reinoso, J.R., Elas, B.M., 1999. Descripcin y Evaluacin de un Systema de Qochas en la Cuenca Norte del Titicaca. Gaceta Arqueolgica Andina 25, 147e166. Washington, J., Alvarez, R., 1991. Experiencias y Alternativas en la Tecnologia en Qocha. Revista del Museo e Instituto de Arqueologia 24, 155e175. WWDR, 2003. Lake Titicaca basin, Bolivia and Peru. In: Water for People Water for Life: The United Nations World Water Development Report. UNESCO Publishing, Paris, pp. 466e480.