p. 338 -- Figure 1. Southern Peru
This paper is presented exactly as it appeared in the 1987 publication with the exception that the graphics have been improved, and a few spelling errors have been corrected. The page numbers of the original publication are listed at the top of the page.
Vast expanses of abandoned terrace systems dominate the extreme southern Peruvian drainages. The Moquegua valley is one of a number of major drainages located in this extremely arid environment. This paper (1) a reconstruction of the patterns of agrarian land use in one small valley of the Moquegua Basin, from the latest phases of the Tiwanaku occupation to the Spanish Colonial period, and (2) a testable model of agricultural contraction and micro-regional patterns of land use through time. A continuing and varied set of pressures on the agricultural system are isolated as the key elements in this contraction process. These stresses include demographic growth, labor minimization, increased productive demands, and possibly a reduced level of available hydrological resources.
The productive base of the labor intensive agrarian economies of prehispanic Andean America was irrigation agriculture. The implicit or explicit recognition of irrigation agriculture as central to Andean society has guided a long and successful history of research into agrarian technology (Day 1970; Denevan 1966; Farrington 1974, 1977; Farrington and Park 1978; Kosok 1940, 1965, Kus 1972; Nuñez 1974; Mitchell 1976; Pozorski and Pozorski 1982; Regal 1970; Rowe 1969; Sherbondy 1979).
The last decade as seen a shift in the research focus from the empirical description of agricultural technologies to one concerned with the long-term land use dynamics and sociological concomitants of regional agrarian systems. The work of Moseley and associates in the Moche Valley has provided a methodological and theoretical framework for addressing diachronic changes in land use systems (Moseley et al. 1983; Moseley 1974, 1978; Ortloff et al. 1982, 1983). Regional patterns of field location, canal use dynamics, and long term tectonic landscape alterations figure prominently in their reconstruction and explanation of the evolution of Moche Valley agriculture. The ensuing debate over interpretations of the Moche data (Pozorski and Pozorski 1982; Farrington 1974, 1977) has served to expand this framework to include technological variables of canal construction and engineering successes and failures.
The Otora Valley of the Moquegua Drainage of southern Peru (Figs. 1,2) is located in the heavily dissected and arid upper sierras. This landscape is dominated by abandoned agricultural terraces. Throughout the Peruvian south and Chilean north, such abandoned
terrace systems are found associated with prehispanic settlement in virtually every irrigable drainage as yet investigated. Topographically, ecologically, and agriculturally, the Otora Valley typifies the majority of other south central Andean drainages. The empirical patterns of human land use documented in Otora should therefore be representative of the southern drainages as a whole.
The Otora research sought to investigate the prehispanic agricultural systems within the broad framework developed in the Moche Valley. An intensive surface reconnaissance located 17 prehistoric and early Spanish Colonial sites in the valley. Air photo and ground survey defined the area of abandoned terrace systems. These terrace tracts were the associated with their respective settlements through simple geographical proximity and other canal-site-terrace relationships. A six period cultural sequence (Fig. 3) was developed beginning in the latest phase of the Moquegua Tiwanaku occupation (ca. A.D. 1200). The diachronic patterns of land use were incorporated into this historical sequence and are presented below.
As with the Moche Valley, the Otora data demonstrate a progressive agricultural land use shift through time: terraced land was abandoned in a context of demographic growth and agricultural intensification. Unlike the Moche patterns, however, the Otora Valley settlements and terraced lands contracted up-river toward the canal sources, a pattern not previously reported in the central Andes. As opposed to the Moche coastal floodplain, the Otora data suggest that the differing topographical and ecological characteristics of the south central sierras promote a differing set of agrarian strategies.
A fundamental and initially surprising result of the Otora research was that the engineering features of modern agriculture show remarkable similarities to the prehispanic systems. In the past, as today, all plant agriculture and animal husbandry was possible only with canal-watered terraces below approximately 3,200 meters above sea level (Dolan 1985). The vast majority of agricultural land is fed by permanent rivers, whose sources are found in the extensive subterranean aquifers and snow melts above 4,000 m. Occasionally, small areas of land are watered by permanent springs called pukios. Although these pukios are particularly common around 3,200-3,500 m where aquifers break (F. Stevenson, pers. comm.), they do not provide a significant percentage of the total agricultural water utilized in the Otora Valley, an observation which holds true for the Moquegua drainage as a whole.
The basic principle behind south central Andean agriculture is to construct an earth banked canal from a permanent water source to terraceable land. Virtually all agricultural land below 3,000 m in the Moquegua drainage is terraced, even that immediately adjacent to the rivers. This is due to the steep landscape gradients, a product of severe tectonic uplift and downcutting of the watershed (M. Moseley, pers. comm.). In such an ecological context, the contrast between irrigated and non-irrigated areas is dramatically evident with a dark green vegetation strip marking canals and irrigated fields
against a virtually plantless, brown landscape (see Tosi 1960 for a detailed description of the ecological characteristics of this area).
A typology of canal construction, based primarily upon topographical placement and function has been developed for the Otora prehispanic canals. Three types of canals were (and are) utilized in the valley. A "contour canal" is one that draws off a water source and follows a hillside at less than a 1° slope. To the eye, it appears to be an even contour. A "ridgetop canal" is one that draws off a water source and then follows a descending ridgetop, usually straddling the crest in a sinuous pattern allowing access to either side of the hill. A "vertical canal" is defined as any canal that is placed more or less perpendicular to the contours of the hill and is designed to feed individual terraces or terrace tracts off of a contour or ridgetop canal.
In many cases, both prehistoric and modern, natural erosional channels were used as vertical "canals" to drop water from a contour or ridgetop canal down a gully to terraces constructed in the lower, colluvial fans. There is no evidence of any stone lining of the natural erosion points although there was probably a certain reworking of the water flows to the terraced fields.
Contour canals are particularly prone to seepage losses due to the generally severe slopes (>25-30°) and the occasional rocky terrain onto which they must be constructed. In cases of severe seepage loss in concentrated areas, modern farmers oftentimes construct small terraces which collect the otherwise lost water (Dolan 1985). These "opportunistic terraces" are placed to compensate for severe topographical conditions. Similar small terraced areas can be seen along now-abandoned systems.
Contour canals are approximately 50-80 cm wide and, with a few modern exceptions, are not stone paved. Ridgetop canals tend to be slightly wider due to the absence of gradient constraints on construction. The sides of both ridgetop and contour canals have been continually reinforced and raised with sediment cleaned out of the channel. The effect of such a practice is to produce a 50 cm wide canal, approximately 50 cm deep from surface level, with either one (contour) or two (ridgetop) sediment-raised walls ranging from 20 to 100 cm high.
On rare occasions, vertical canals in the terrace areas themselves were stone sided, but were never paved (this differs for some modern cases). Vertical canals and steep gradient ridgetop canals were often flanked on one side by a pirca wall. These walls were approximately 1 m high (using modern walls as analogies) and served at least two functions. First, today they mark off property lines and keep animals from grazing in other fields. Second, the walls are almost always on the down-slope side of the canal which prevents lateral movement of the canal itself. This protects against field and canal erosion and deterioration of the water delivery system to individual terraces (see Dolan 1985). In a geophysical context of a highly active tectonic landscape in which ground movements are the rule, and not the exceptional occurrence, such an engineering feature is extremely useful (see Moseley et al. 1983).
Figure 4. Prehispanic reservoir excavation in Otora
---------------------p. 344---------------------
Terrace wall height and distance between individual terraces is purely function of slope with the largest and most closely packed terraces on the steepest slopes while low gradients permit the construction of shorter terrace walls. On low slopes (2-10°) terrace heights are as low as 50 cm, while on the steepest slopes (over 25°) wall heights approach 1-2 m. Distance between the terraces ranges from a minimum of 1 m up to several meters apart.
All ancient terraces walls were constructed with field stones and mud mortar, virtually identical to residential structure walls. There is no indication of any rigid selection for stone size nor placement of particular stone types within the wall. In some abandoned, prehispanic terraces, repairs were made which are clearly detectable as mortarless sections in the wall.
Unlike terraces in rainfall areas, such as the Titicaca Basin, there are no houses or non-agricultural structures built directly in the terraced areas. Rather, all structures are built above the contour canal outside of the agricultural area in order to conserve cultivable land and avoid having the structure periodically inundated by irrigation waters. Any case in which domestic structures are found built over terraces indicates that the agricultural fields were out of use at the time of domestic construction.
Reservoirs and aqueducts were also vital engineering components of the prehispanic agricultural systems. At least five reservoirs have been located in the Otora Valley, all currently out of use and associated with abandoned canals. These water-holding structures were built with stone sided, earthen filled walls approximately 1 m wide. They were placed directly on the main canal lines. Excavations into several of these discovered a pattern of laminated silty sediments expected in such a depositional context (Fig. 4).
Three prehistoric aqueducts were located in the Otora Valley. These were modest constructions designed to maintain water flow over unusually low swales. The side walls were constructed identically to the walls of the reservoirs with a wide brick wall filled with earth.
These ecological factors and engineering responses provided a system characterized by a long canal off of a river source to its principal fields, with a series of small agricultural areas along its route where seepage was severe or where natural gullies allowed terraceable land. The "principal fields" were chosen as the areas of minimal slope close to the water sources. In the case of major, multi-contour canal systems however, sociological and economic factors into the field location calculus.
In sum, the Otora Valley survey has identified four technological components of prehispanic agriculture: earth-banked canals, terraces, reservoirs, and aqueducts. With these four construction techniques, water can be efficiently delivered to level planting surfaces. Even today, peasant agricultural land is engineered under similar techniques (see Dolan 1985 for an extended discussion of contemporary agricultural practices).
Figure 5. Otora Valley archaeological sites.
---------------------p. 346---------------------
Seventeen sites were located in an intensive surface survey of the Otora Valley (Fig. 5). These are designated P-1 - P-17 ("P" = Porobaya). The sites were dated principally by comparing ceramics form existing chronologies and styles from the Titicaca Basin (Lumbreras and Amat 1968; Rydén 1947; Tschopik 1946; Arellano 1975; Julien 1978), the Moquegua and Ilo Valley (Ghersi 1956; Belán 1981), and Northern Chile (Dauelsberg 1969, 1972; Foccaci 1969, 1981). Domestic architecture and funerary styles proved to be chronologically sensitive as well.
Eight of the 17 sites were small sherd scatters or proved to be minor ancillary settlements in a larger system (P9, P11 - P17). These were surface collected. The remaining 9 major sites were either excavated by complete room units or were intensively surface collected.
A six-period chronology was defined (Fig. 3). The initial occupation was a very late Tiwanaku hamlet (P5) dated to "Moquegua Tiwanaku VI" or post-Tiwanaku Expansive (see Goldstein 1985). P5 was the only site in the valley during this period. Although the settlement was small, with no more than four domestic units, excavations revealed a permanently occupied and intensively utilized site. The site was culturally linked to the larger Moquegua valley Tiwanaku settlement system and represented a pioneering occupation in an area of prime agricultural potential.
The subsequent Otora period in the Moquegua Valley represents a time of profound socio-political and economic restructurings after the collapse of Tiwanaku hegemony. In the Otora Valley, at least five sites have been located that date to this period. Two of these sites (P4 and P8) are best interpreted as coastal and circum-Titicaca basin colonies respectively, while a third (P7) is most likely a local population developing out of the Tiwanaku hamlet. The small site of P9 and an associated corral (P12) are also dated to this period.
The exact chronological position of the Otora period sites relative to one another remains problematic although there is little doubt that all of these are bracketed between the Tiwanaku and later Estuquiña periods. P7, P9, and P12 would appear to be earlier, based upon ceramic criteria, while P4 and P8 share distinctive proto-chulpas (Stanish 1985a) and have later decorated ceramics (Chiribaya and a contemporary Tricolor del Sur variant, probably Collao black-on-red--see Lumbreras 1974; Tschopik 1946; and Rydén 1947).
Cultural hallmarks of the succeeding Estuquiña (P1, P2, P3) and Estuquiña-Inka (P1, P3, P6) periods include the development of fortified architecture, intensification of interregional exchange, the formation of a nascent elite, and the development of chulpa funerary styles. During this period P4, P7, P8, P9, and P12 were abandoned. The multi-ethnic character of the Otora Valley gave way to an indigenous, regional sierra culture (Estuquiña) of which P1, P2, and P3 are but three of the many other examples in the Moquegua and southern Tambo drainages.
Figure 6. Moquegua Tiwanaku VI Period land use.
---------------------p. 348---------------------
The dominant exotic ceramic styles on the Estuquiña-Inka sites are Sillustani (black-on-red, black-and-white-on-red, polychrome) and Imperial Inka (Cuzco polychrome A and B) (see Rowe 1944; Julien 1978; Tschopik 1946). As a percentage of total diagnostic ceramics however, these are very rare. The local decorated ceramic, a later variant of the widespread Tricolor del Sur (Tricolor Porobaya) is found in larger numbers. In contrast, the small valley bottom hamlets of P11, P13, and P15 contain considerably larger quantities of Imperial Inka and Chuquito (the latter not found on the Estuquiña-Inka sites). These sites are later in date and constitute the actual Inka occupation of the Otora valley.
Spanish Colonial settlements (P11, P14, P15) approximate the Inka period pattern with small, dispersed hamlets along the river bottom.
Moquegua Tiwanaku VI
In the Otora drainage, there are 8 ancient and/or modern canals designated C-1 through C-8. The first agricultural settlement in the Otora Valley is the very small Tiwanaku site (P5) located at the distal end of the C-1 canal, the original canal constructed in the valley. This contour canal draws off of the Río Porobaya and is located on the northern side of the Cuajone Ridge. There are approximately 15 ha of terraced land associated with the site located in five discrete areas (Fig. 6). The average slopes in the main fields of this settlement are the lowest in the sequence (Table 1).
A small reservoir was constructed out of local field stones directly in the main grouping of terraces. This tank was partially destroyed by the construction of the later corral site (P9). The construction further increased water use efficiency by permitting the capture of a lowered water flow over an extended period of time. As today, the water is stored until it is needed for the fields. This is particularly critical during the dry season when there is insufficient natural water flow to adequately cover a field within a scheduled watering cycle.
There is absolutely no evidence for earlier occupations in the valley, although the possibility that a light density, pre-Moquegua Tiwanaku VI population farmed the immediate sides of the river remains plausible. It is certainly logical to presume an initial settlement pattern such as this which took advantage of small areas of easily irrigable land along the river banks. Such an agricultural practice could also have been employed to provide sufficient food for the missed planting cycle when C-1 was constructed. The fact that P5 is undefended also lends credence to this hypothesis in that the need for a nucleated, protected settlement was not evident in the Tiwanaku periods. Given the fact that these sites would now be covered by terrace agriculture, testing of this proposition would require an intensive reconnaissance methodology far beyond the capacities of the project reported here.
Table 1. Land use dynamics in Otora
Otora Period
The site of Cuesta Alta (P7) was founded as P5 was abandoned. The population of P7 utilized the same canal as P5 (C-1) but opened up 45 ha of terraced fields on slopes averaging 17° (Fig. 7). Contemporary with this resettlement up-canal was the construction of a very large reservoir at the swale where C-1 enters the Paralaque drainage. A smaller tank was constructed north of the first one on the canal segment which waters the fields directly below the residential area of P7.
The second canal (C-2) in the sequence was constructed on Cuesta Cuajone and is associated with the intrusive circum-Titicaca Basin period site of P8 (Cuajone). This ridgetop canal drew its water immediately to the south of Otora. In effect, by drawing water from a neighboring drainage, C-2 was technically an "intervalley" canal which diverted the water resources of the Paralaque river over the Cuajone ridge into Otora.
C-2 is 6.3 km from its source to the furthest fields associated with P8. This is the only Otora period site located on the Cuajone ridge. The canal is constructed similarly to C-1 (earth-banked) and follows a sinuous pattern across the ridgetop. The entire distal end near P8 has been abandoned but the more easterly sections of the canal are currently used to water a small amount of pasture land. Along the northern side of the Cuajone ridge (Otora drainage side) vertical canals were opened up to natural gullies and a series of small, terraced tracts are found following the canal line (Fig. 7). We can therefore surmise that C-2 was probably used in the following Estuquiña period to water the terraces found above the C-1 canal line.
Porobaya Chica (P4) is associated with an extension of the in-use valley bottom canal of C-4 (C-4a). The date of this canal segment is also difficult to determine. C-4 parallels C-1, but is lower by approximately 30 m. C-4 may only be associated with the later Inka or Spanish Colonial period occupations and therefore intrusive on P4. A section of C-4a actually cuts through an aqueduct on C-1 as well as [through] domestic structures on Porobaya (P1). Therefore, at least in its latter phase, the canal was intrusive on P1 and later than the Estuquiña-Inka period. On the other hand, the earliest phases of the canal may have been utilized by the colonial (Chiribaya) settlement at P4 either after the abandonment of P7, or while the latter site was still occupied, as an attempt to maintain independent control over irrigation waters by ethnically distinct groups.
Estuquiña Period
The principal sites of this period in Otora are P1, P2, and P3 (Fig. 8). Associated with the founding of P2 on the Colana ridge was the construction of the third major canal in the valley (C-3). The 139 ha Colana ridge agricultural system was fed by this single contour and ridgetop canal 3.0 km from the intake to the first terrace. Included in this system were two reservoirs and two low aqueducts used to shunt water over a swale and around the residential area of the settlement.
Figure 9. Estuquiña-Inka Period land use
---------------------p. 354---------------------
Figure 10. Inka Period land use
Both C-3 and P2 were abandoned before the Estuquiña-Inka
period. This is clearly documented by the lack of any Inka or
Inka-influenced ceramics on the surface or from room excavations. P2
is therefore the only single occupation Estuquiña period site
without an Inka influence phase. The fact that the agricultural
system associated with this site represented the most sophisticated
constructions in the valley emphasizes the indigenous origin of
terrace agriculture. That is, the expansive terraced hillsides and
extensive water delivery systems were not Inka imperial impositions,
as is suggested by early historical accounts.
Both Sajena (P3) and Porobaya (P1) have extensive Late Horizon components but do not represent Inka control. In its earlier, Estuquiña period, P1 probably utilized the C1 and/or C4 canals which were limited to watering a small area of fields near the sites. P3 opened up a large terrace system on the Río Sajena (canal C-7) around 3000 m. The principal occupations at both P1 and P3 were during the later Inka influenced phase. The degree to which these latter agricultural patterns reflect the earlier Estuquiña period ones remains unresolved for both P3 and P1.
Estuquiña-Inka and Inka Periods
With the abandonment of C-3, the valley bottom canal of C-5 was constructed and C-4 was probably expanded down to its present location west of fork of the Ríos Sajena and Porobaya (Fig. 9). The water conserved from the abandonment of C-3 was diverted to these lower canals and to some small areas of fields near P4 and P1. It is undoubtedly not a coincidence that the significance of the lower canals increased when Inka influence first appeared. Such canals were considerably more water efficient although they forced lineation of agricultural land and residential settlements. While such a settlement system may have been possible in a cultural context without intermittent warfare, such as that found in the Late Horizon, it would have been difficult to maintain in the pre-Inka periods where defensibility was a major settlement determinant.
The terraces and canals associated with the settlement of P1 are difficult to assess due to the site’s proximity to P4, the Otora period site. Compounding this problem, it is virtually impossible to determine the extent of the Estuquiña period occupation at P1, although a few pre-Estuquiña ceramics cautiously suggest an early founding of the site. It would appear that those terraced quebradas near the site as well as the C-1 terraces were in use during the Estuquiña and Estuquiña-Inka periods. Certainly these terraces are some of the steepest in the valley and represent an intensive labor investment in construction and maintenance.
The major change in the Inka period (Fig. 10) was the construction of C-5, as indicated by the founding of P11 on the north side of the Río Otora. P11 is the first site on this side of the river. It was constructed off of the Río Sajena and fed valley bottom terraces associated exclusively with low density, Inka period settlements alongside the river. Similarly, another valley bottom canal (C-8) was constructed below Otora in Inogoya (Fig. 2). C-7 was not abandoned but contracted several hundred meters up river to the
site of P13. All Inka sites in Otora are associated exclusively with valley bottom canals.
In sum, there were eight separate canals built in the Otora drainage throughout a half millennium of human occupation. The technological nature of this agricultural base remained fundamentally unchanged, even though the ethnic composition of the valley fluctuated. Unlike domestic architecture, funerary practices, pottery, and other material cultural characteristics, canal and terrace technology was not ethnically sensitive. Agricultural constructions responded exclusively to the demands of the local ecological environment (Stanish 1985b).
Throughout the half millennium of human, agriculturally based occupation in the Otora Valley, there were four major, non-contemporary periods of land use in geographically discrete areas. This pattern of land use was characterized by progressive settlement and agricultural land shifts up the river source until the Estuquiña-Inka period when there was a major shift to valley bottom canals. Incredibly, agricultural land was abandoned in a context of net agricultural expansion and demographic growth.
Give the empirical patterns indicated above, the question that immediately arises is why would the initial Tiwanaku population of no more than ten nuclear families opt to build a 7.5 km long canal instead of choosing a shorter canal near the source, a strategy used by later occupations?
First, it should be noted that the construction of such a long canal is not beyond the capacities of the population represented at P5. It is necessary to view this settlement in the political context of a hierarchical Tiwanaku state in which a much larger body of labor and overarching political system could be organized. Based upon preliminary work already completed by members of Programa Contisuyu it would appear that in the latest phases of the Tiwanaku period there was an expansion of population from the Moquegua Valley into the mid and upper sierra. Causes of this population movement are as yet unknown, but it is clear that P5 represents one of the highest settlements in this expansion process designed to exploit unoccupied, small drainages having agricultural potential. As such, the small group of people at P5 were part of a larger socio-political nexus which assisted in the construction of the canal and provided subsistence for a missed planting period. It should be noted that the nearest contemporary Tiwanaku site is only a mere 4 hours walk away on the flanks of Cerro Baul. The construction of a 7.5 km long canal is therefore not beyond the organizational capacity of a settlement such as P5, as long as the larger socio-political context of a state (or possible imperial state) is taken into account.
The Otora data suggest that two factors are primarily responsible for initial settlement location in the valley: (1) progressively lower hillslopes away from the river, and (2) labor minimization in agricultural constructions. That is, in terms of labor requirements
it is much more efficient to build long canals to cultivate flat land with low terraces than it is to build short canals to steep land which require high and closely spaced terraces. In the deeply incised upper and mid sierras, where valley gradients average 6° (ONERN 1974), there is a gross relationship between distance from the river and lower slope gradients.
Topographically, the cultivated land associated with P5 is the closest low gradient land (<10°) that could be reasonably reached from a canal intake on the Río Porobaya. In the fields of P5 the average gradient is 9°, which contrasts sharply with later field systems where terraces sometimes were constructed on slopes of 30° of greater. There is a progressive increase in average field slope through time (Table 1).
In support of this principle of initial field choice, it is important to emphasize that the canal trenches are not, in themselves, labor intensive constructions. Apart from bedrock obstructions, which require cutting of solid rock, the earth-banked canals are not nearly as labor intensive as terrace construction. Furthermore, bedrock obstructions are virtually absent in the C-1 canal path. It should be noted that canal technology was well developed in the agricultural fields at other large Tiwanaku sites in the Moquegua Valley, but terrace construction is rare, restricted to very low gradients (M. Moseley and R. Feldman, pers. comm.).
The settlement shift in the Otora period to an area of considerably greater gradients and therefore labor requirements needs explanation, particularly since there was abundant land near the fields of P5 for expansion. The fact that the earlier P5 fields were actually abandoned in a context of demographic growth serves to emphasize the "dramatic" nature of this resettlement.
Throughout the historical sequence in Otora (at least until the Inka period), there is indicated a continual effort to respond to agricultural stresses. This is the only possible explanation of the progressive, patterned and "trans-ethnic" settlement shifts through time. Once a canal and terrace system was in place, demographic increases, intensification of production, and possibly reduced hydrological resources combined to stress its productive capacities. Under these circumstances, the options available to the population were (1) reduce demographic levels, (2) open up new water sources, (3) change site location to hydrologically more efficient areas, (4) adopt more efficient agro-engineering constructions and/or (5) develop ancillary, non-agricultural economic institutions. All of these options were utilized at various times in the Otora Valley.
The shift from P5 to P7 represents the utilization of the option n which the settlement and agricultural fields were moved tom ore efficient locations. The principle behind this strategy involves essentially the shortening of canal lengths which reduce evaporation and particularly seepage losses, a strategy isolated and described in detail by Moseley et al. (1983) for the Moche Valley. Modern studies of earth-banked canals indicate that more than 50% of the initial intake water is lost to such factors (USDA 1955). Evaporation rates in the Peruvian south are some of the highest in the world (F.
Stevenson, pers. comm.). One of the simplest and most efficient ways to conserve water in such environmental and technological contexts is to shorten the length of the canal itself. This simple strategy proportionally cuts the absolute rate of water loss.
Such a strategy is not without its costs, however, given the topographical characteristics of the Otora Valley. Up-valley movement forces agricultural terrace construction onto gradients originally not considered optimal in terms of labor input. C-1, for instance, was originally constructed in a context of sufficient water resources (maximum of ten nuclear families). Its extensive length reflects the desire to reach low gradient land which permitted low terrace construction. That is, C-1 was optimal in labor efficiency terms in that it balanced both canal length and agricultural terrace height. Once conservation of water resources becomes a critical factor in agro-engineering strategies, however, shortening of the canal forces terrace construction on higher slopes.
In the Otora period, the shift from P5 to P7 decreased canal length by 1.6 km or 21% of the original C-1. Such a move cut water loss and, in gross terms, we can calculate that there was approximately a 10-20% increase in water delivery by this simple strategy. Table 1 illustrates this process throughout the historical sequence. The average canal length per period is cut more than half while the average slope of the fields more than doubles from the Tiwanaku period to the Estuquiña-Inka period. That is, there is a rough inverse proportional relationship between canal length from the source and slope gradient.
The next two sites founded in the Otora Valley were Porobaya Chica (P4) and Cuajone (P8). The canal associated with Cuajone (C-2) draws off of the Quebrada Cueva Quemada from the Paralaque drainage and does not in any way compete for water with the systems that draw off of the Río Porobaya (C-1, and the later C-3 and C-4 canals). In cultural terms this pattern may be understood as an attempt by an ethnically distinct population to control its own irrigation waters. Viewed from the perspective of valley-wide water resource utilization, the construction of C-2 represents the use of option 2 above, in which a new water source is opened.
The placement of C-2 on the ridgetop considerably reduces seepage losses relative to C-1 because the former was not built on steep hillsides, such as C-1. Such a qualitative difference between these two canals helps explain why contraction rates for the former are considerably greater than for the latter (see Table 1).
In the Otora period, three out of the five hypothetical options were utilized by the populations in the valley. The site of P5 was abandoned for more agriculturally efficient fields up canal; two reservoirs and two aqueducts were constructed; and a new water source was opened. There were, on the other hand, non population reductions, at least as evident in the increase of domestic units (Table 1). Furthermore, no intensification of agricultural economic production is evident, at least not on any archaeologically recognizable scale such as that found in the Estuquiña-Inka period.
The Estuquiña period involved a settlement shift of the population into at least four, possibly more sites (Fig. 8). The principal site of P1 continued to use the C-1 canal while other settlements either moved down to lower canals (P6), opened up a new and more efficient canal on the same source as the Porobaya (P2), or opened up more efficient canals on other sources such as P3 on the Río Sajena. C-2 continued to be utilized with very large vertical canals that functioned by opening up the ridgetop canal at natural erosional points and dropping the water to terraces or other canals below. One such canal functions today in which water falls almost 350 m from the top of Cerro Cuajone to the modern C-4 canal where it is collected and serves to augment water from the Río Porobaya. The Otora/Estuquiña period settlement shift on C-1 (P7 to P1) represented a shortening of the canal by approximately 2.2 km or 37% of the Otora period canal length.
Concomitant with the contraction of C-1 was the opening of the Colana ridge agricultural system and the defended settlement of P2. This system had one canal (C-3) drawing off of the Río Porobaya, which in effect served to shift the water saved in the cutting of C-1 to this new system. This indicates that the population had not yet reached absolute water supply limits relative to its productive requirements. That is, there was no need to reduce population nor presumably production levels due to water resource constraints.
It was only after the abandonment of the Colana agricultural system that the Otora Valley reached absolute limits for continued growth. The option of canal shortening and other water-efficiency measures were insufficient to compensate for the demands on the agricultural system. Any subsequent contraction of C-1 would have been so close to the river that there would have been insufficient land to cultivate.
A hallmark of the Estuquiña-Inka period was the expansion of economic exchange networks throughout the region. Extensive coastal as well as altiplano contacts are evident from the quantities of fish bone, shell, ceramics, and camelid bone recovered from the domestic residential middens. The beginning of this regional economic process correlates to the collapse of the Colana system and abandonment of P2, which in turn signifies the absolute maximum expansion of prehispanic agriculture in the Otora Valley.
The Inka occupation remains problematic at this point. As suggested above, there appears to be a distinction between ceramics on P1 versus those of P11 and P15, with the latter sites’ ceramics suggesting greater iconographic affinities to those found at the later site of Torata Alta. The shift to valley bottom canals occurred after Inka influence was felt in the Otora Valley, as determined by these ceramic indicators (Fig. 10). In terms of seepage losses, this strategy is the most efficient because the fields are placed directly below the canal. Any seepage falls onto the fields or into the river where it can be picked up by lower canals. This strategy, however, forces a linear dispersal of settlements along the top of the canal, as is common today in the valley.
The Late Horizon settlement data indicate a considerably reduced population from that in the previous period. A conscious depopulation and resettlement to agriculturally more productive areas of the Moquegua drainage by the Inka state is suggested. While the Late Horizon demographic levels of the Moquegua Valley as a whole are unknown (and may, in fact, indicate no change or even population increases), the Otora data indicate considerable reductions in population sizes.
The data from Otora clearly support the proposition or a progressive land abandonment process beginning with the earliest agricultural occupations to at least the Inka period. As with the Moche valley settlement sequence (Moseley et al. 1983), the proximate cause of this contraction process is a set of continual and severe agricultural stresses. Unlike the Moche valley however, where settlements responded by moving down the source toward the coast, the Otora pattern is one of up-river resettlement toward the water source.
I have suggested that the diachronic geographical patterns of agricultural land use based upon irrigated terraced hillsides can be understood as the interplay between labor minimization and agricultural stresses. Initial, low density occupations seek low gradient land in order to minimize labor inputs for terrace construction. As productive demands increase and/or as water supplies decrease, settlements cut canal length to prevent seepage losses. Such a strategy is hydrologically efficient but forces terrace construction on steeper gradient land requiring increased labor inputs to build high terraces. From the perspective of seepage loss and evaporation, valley bottom canals are the most efficient and characterize Late Horizon through modern agricultural strategies.
For the upper sierras (above 2000 m), therefore, we can predict that in low density regional demographic contexts with a relative abundance of water, settlements will be found on low gradient land away from the deeply entrenched river basins. As agricultural stresses multiply, settlements should move progressively up canal. We predict that valley bottom canals represent the final stages in the local evolution of agricultural land use as populations choose this high labor input, but agronomically efficient option.
Acknowledgments. This paper benefited from the comments and assistance of many people in Chicago, Lima, Moquegua, Arequipa, and Otora. This include Don Rice, Michael Moseley, Robert Feldman, Luis Watanabe, Luis Lumbreras, Joyce Marcus, Robert Mc C. Adams, David Jessup, Charles Ortloff, Victor Barua, Frank Stevenson, José Chávez, Augusto Belán, Romulo Pari, Marc Bermann, Manuel García, Gloria Salinas, Jesús Gordillo, Paul Goldstein, Patricia Dodson, Rita Basuray, Mark Aldenderfer, Antonio Oquiche, Elva Alatrista, Edmundo de la Vega Machicao, members of Programa Contisuyu, and the Archaeology Program of the Universidad Católica de "Santa María" of Arequipa.
A special thanks goes to Barbara Dolan who conducted ethnographic studies in Otora and took time to help me with my own research. Bob
Feldman, Mike Moseley, and Jim Richardson unselfishly loaned me their vehicles. The Southern Peru Copper Corporation provided invaluable logistical support. A special thanks goes to Victor Barua and Nelson Molina of SPCC, who took a personal interest in the success of my project.
Funding was provided by the Henry and Grace Doherty Fellowship in
Latin American Studies, a National Science Foundation Dissertation
Improvement Grant, the Tinker Foundation through the Center for Latin
American Studies at the University of Chicago, Ing. Victor Barua,
Lucy Barua, and Mr. and Mrs. Robert Pritsker. The Field Museum of
Natural History, Chicago, provided a graduate education fellowship.
The School of American Research, Santa Fe, provided a post-doctoral
fellowship which allowed me time to finally complete this article.
The help of all is greatly appreciated.
Arellano Lópes, Jorge C. 1975b La Ceramica de las Tumbas de Iskanwaya. Jornadas (La Paz) 2:103-124. La Paz.
Belan, Augusto Franco. 1981 Chiribaya: Apuntes para el Conocimiento de la Arqueológica Surperuana.Arequipa: Editorial Arqueos.
Dauelsberg H., Percy. 1969 Arqueología de la Zona de Arica: Secuencia Cultural y Cuadro Cronológico. Actas del V Congreso Nacional de Arqueología pp. 15-19. La Serena.
Dauelsberg H., Percy. 1972 La ceramica de Arica y su situación cronológica. Chungara 1-2: 17-24. Arica
Day, Kent C.1970 Walk-in wells and water management at Chanchan, Peru. Ponencia presentada al 39th Congreso Internacional de Americanistas. Lima.
Denevan, William. 1966 The Aboriginal Cultural Geography of the Llanos de Mojos of Bolivia. Ibero Americana, 48, University of California,Berkeley.
Dolan, Barbara. 1985 Agriculture and society. Preliminary report presented to Programa Contisuyu.
Farrington, Ian. 1974 Irrigation and settlement pattern: preliminary research from the north coast of Peru. In: Irrigation's Impact on Society. T. Downing and McG. Gibson (eds.) University of Arizona Press.
Farrington, Ian. 1977 Land use, irrigation and society on the north coast of Peru in the prehispanic era. Zeitschrift für Bewassarungswirtschaft. 151-186.
Farrington, Ian and C. Park. 1978 Hydraulic engineering and irrigation agriculture in the Moche Valley, Peru: c. A.D. 1250-1532. Journal of Archaeological Science 5(3): 255-268.
Foccaci, Guillermo. 1969 Arqueología de Arica: secuencia cultural del periodo agroalfarero -Horizonte Tiahuanaco. Actas del V Congreso Nacional de Arqueologia: 21-26. La Serena.
1981 Descripcion de un cementerio incaico en el valle de Azapa. Chungara 7:212-216. Arica.
Franco Inojosa, Jose M. y Alejandro Gonzalez. 1936 Departmento de Puno. Revista del Museo Nacional. vol. V(2): 157-183. Lima.
Ghersi Barrera, Humberto. 1956 Informe sobre las excavaciones en Chiribaya. Revista del Museo Nacional 25: 89-119. Lima.
Goldstein, Paul. 1985 The Tiwanaku Occupation in the Moquegua Valley. Unpublished Master's Thesis. Department of Anthropology, University of Chicago.
Julien, Catherine. 1978 Inca Administration in the Titicaca Basin as Reflected at the Provincial Capital of Hatunqolla. Ph.D Dissertation,University Microfilms, Ann Arbor.
Kosok, Paul. 1940 The role of irrigation in ancient Peru. In 8th American Scientific Congress Proceedings. v II. Anthropological Sciences pp. 169-178. Washington D.C. Dept. of State.
1965 Life, Land and Later in Ancient Peru. Long Island University Press. New York.
Kus, James. 1972 Selected Aspects of Irrigated Agriculture in the Chimu Heartland. Ph.D. dissertation, Department of Geography, University of California, Los Angeles.
Lumbreras, Luis. 1974 The Peoples and Cultures of Ancient Peru. Translation by B. Meggers. Smithsonian Institution Press. Washington.
Lumbreras, Luis and Hernan Amat. 1968 Secuencia arqueologica del altiplano occidental del Titicaca. Revista del Museo Nacional. Lima
Mitchell, William P. 1976 Irrigation and community in the central Peruvian highlands. American Anthropologist 78:25-44.
Moseley, Michael. 1969 Assessing the archaeological significance of Mahamaes. American Antiquity. v 34(4): 485-487. Salt Lake City.
Moseley, Michael. 1974 Organizational preadaptation to irrigation: the evolution of early water management systems in coastal Peru. In: Irrigation's Impact on Society. edited by T. Downing and M. Gibson, pp. 77-82. University of Arizona Press, Tuscon.
Moseley, Michael. 1978 An empirical approach to prehistoric agrarian collapse: the case of the Moche Valley, Peru. In: Social and Technological Management in Dry Lands: Past and Present, Indigenous and Imposed, edited by N. L. Gonzalez, pp. 9-43. AAAS Selected Symposium Series. Westview Press. Boulder.
Moseley, M.E., R.A. Feldman, C.R. Ortloff, and A. Narvaez. 1983 Principles of agrarian collapse in the cordillera negra, Peru. Annals of the Carnegie Museum v. 52: 299-327.
Netherly, Patricia. 1984 The management of late Andean irrigation systems on the north coast of Peru. American Antiquity v 49(2): 227-254.
Nuñez, Lautaro. 1974 La Agricultura Prehistorica en los Andes Meridionales. Editorial Orbe, Santiago.
O.N.E.R.N. (Oficina Nacional de Evaluacion de Recursos Naturales). 1974 Cuencas del Rios Locumba, Sama y Moquegua.
Ortloff, C.R., M.E. Moseley and R.A. Feldman. 1982 Hydraulic engineering aspects of the Chimu Chicama-Moche Intervalley canal. American Antiquity 47: 572-595.
Ortloff, C.R., M.E. Moseley and R.A. Feldman. 1983 The Chicama-Moche intervalley canal: social explanations and physical paradigms. American Antiquity 48: 375-389.
Parsons, Jeffrey. 1968 The archaeological significance of Mahamaes cultivation on the coast of Peru. American Antiquity v. 33(1) 80-85.
Parsons, J and Norbert Psuty. 1974 Agricultura de chacras hundidas en el antiguo Peru. Revista del Museo Nacional. v. 40: 31-54. Lima.
Pozorski, Thomas and Sheila Pozorski. 1982 Reassessing the Chicama-Moche Intervalley Canal. American Antiquity 47:851-868.
Regal, Alberto. 1970 Los Trabajos Hidraulicos del Inca en el Antiguo Peru. Lima.
Rowe, John. 1944 An Introduction to the Archaeology of Cuzco. Papers of the Peabody Museum of American Archaeology and Ethnology, vol. 22, No. 2, Cambridge.
Rowe, John. 1969 The sunken gardens of the Peruvian coast. American Antiquity v. 34(3): 320-325.
Rydén, Stig. 1947 Archaeological Researches in the Highlands of Bolivia. Elanders Boktryckeri Aktiebolag. Goteborg.
Sherbondy, Jeanette. 1979 Les reseaux d'irrigation dans la geographie politique de Cuzco. Journal de la societe des Americanistes 66: 45-66.
Stanish, Charles. 1985a. Chulpa distribution and the emergence of elite ideology in the South Central Andes. Paper presented at the 13th Midwest Andeanist meetings, Chicago.
Stanish, Charles. 1985b. Defining Ethnicity: Post-Tiwanaku Domestic Households in the Moquegua Valley, Southern Peru. Paper presented at the 60th Annual meeting of the Society for American Archaeology, Denver.
Tosi, J. 1960 Zonas de Vida Natural en el Peru. Organization of American States. Technical Bulletin #5.
Tschopik, Marion. 1946 Some Notes on the Archaeology of the Department of Puno, Peru. Peabody Museum of American Archaeology and Ethnology. Papers. Cambridge.
USDA. 1955 Water, The Yearbook of Agriculture. A. Stefferud.
Department of Agriculture, Washington, D.C.
Copyright 1987 Charles Stanish (stanish@anthro.ucla.edu)
See Chip's web page for further publications and other information about his work.
