Andean Past

Volume 9 Article 15

11-1-2009 Climate, Agricultural Strategies, and Sustainability in the Precolumbian Charles Ortloff [email protected]

Michael E. Moseley University of Florida, [email protected]

Follow this and additional works at: https://digitalcommons.library.umaine.edu/andean_past Part of the Archaeological Anthropology Commons, Natural Resource Economics Commons, Natural Resources Management and Policy Commons, Sustainability Commons, and the Water Resource Management Commons

Recommended Citation Ortloff, Charles and Moseley, Michael E. (2009) "Climate, Agricultural Strategies, and Sustainability in the Precolumbian Andes," Andean Past: Vol. 9 , Article 15. Available at: https://digitalcommons.library.umaine.edu/andean_past/vol9/iss1/15

This Article is brought to you for free and open access by DigitalCommons@UMaine. It has been accepted for inclusion in Andean Past by an authorized administrator of DigitalCommons@UMaine. For more information, please contact [email protected]. CLIMATE, AGRICULTURAL STATEGIES, AND SUSTAINABILITY IN THE PRECOLUMBIAN ANDES

CHARLES R. ORTLOFF University of Chicago and MICHAEL E. MOSELEY University of Florida

INTRODUCTION allowed each society to design and manage complex water supply networks and to adapt Throughout ancient South America, mil- them as climate changed. While shifts to marine lions of hectares of abandoned farmland attest resources, pastoralism, and trade may have that much more terrain was cultivated in mitigated declines in agricultural production, precolumbian times than at present. For damage to the sustainability of the main agricul- alone, the millions of hectares of abandoned tural system often led to societal changes and/or agricultural land show that in some regions 30 additional modifications to those systems. to 100 percent more terrain was cultivated in precolumbian times than at present (Clement To achieve agricultural sustainability, An- and Moseley 1991:425). While many cultural dean administrators needed to record changes in explanations for agrarian collapse can be formu- climate patterns, weather events, and natural lated, the most compelling reason for the loss of disasters, then conduct analyses to plan modifi- cultivatable land is changing climate, including cations allowing agricultural systems to function shifting rainfall patterns and amounts. Agricul- in the face of changing water supplies. Modifica- ture was expanded many times in many places tions took the form of physical alteration of when conditions favorable to land reclamation existing water delivery systems, and the devel- were perceived by past populations. When opment of new agro-systems suitable to new climatic trends led to diminished water supplies, hydrological conditions. When climate deterio- temporary or permanent agrarian regression rated beyond a system’s ability to make modifi- ensued, with consequences for social structure. cations to maintain sustainability, field system abandonment was an inevitable outcome. A Ancient Andean utilized a wide sustainable agricultural base, on the other hand, diversity of agricultural techniques in different can lead to overall population growth and ecological zones, and developed agriculural patterns of population concentration and/or strategies consistent with local climate patterns, dispersal, with specialized labor to work and hydrological characteristics, soil and crop types, manage the system. Such a division of the work- and local labor supply. The strategies chosen force underlies urban centers that controlled depended upon a society’s hydraulic engineer- and administered adjacent agricultural zones ing, surveying, and civil engineering skills com- and may have exerted centralized control of bined with its perception of ecological and labor. hydrological conditions. Taken together, these

ANDEAN PAST 9 (2009): 277-304. ANDEAN PAST 9 (2009) - 278

Even if water supplies are adequate for high groundwater levels on the altiplano sustainable agriculture, inappropriate strategies throughout wet and dry seasons. Drainage from of agro-engineering and labor management can western cordillera rainfall is mostly directed to interrupt the development of otherwise well- coastal river valleys (Figures 1, 2) with outflow functioning societies. Agricultural sustainability to the Pacific Ocean with the exception of the requires administrative skills to guide adaptive intermontane source of the Santa River. technical innovations in the design and manage- ment of an agricultural system to maintain high Biotic diversity is pronounced in the many yields in spite of changing weather and climate. highly varied ecological zones of Peru. For A degree of flexibility to modify an agricultural example, with 35 of the world’s life zones, Peru system should, therefore, be part of the original contains the largest number of ecological zones design of a system if knowledge of prior weather of any country on earth (Perú, ONERN 1976; and climate patterns, and their consequences for Tosi 1960). However, diversity is asymmetrically sustainability, exists. As an integral part of distributed by altitude, latitude, and longitude. system design and the potential for innovative As in all mountain ranges, ecological zones are and adaptive design change, an understanding stratified by altitude and far fewer species of of the dynamics of water flow from original plants live at high elevations than at low ones. highland rainfall sources to lowland and coastal regions must be in place. The effects of exces- The Andean mountain ranges form South sive rainfall or drought are disproportionately America’s continental divide. Normally, all felt in the highlands compared to the run-off rainfall in the eastern cordillera comes from the dependent coastal field systems. The differences Atlantic Ocean with a longitudinal gradient in arise from altitude-dependent soil types and precipitation. Fronting the Amazon Basin, the their water infiltration and water retention high eastern Andean escarpment receives characteristics, soil saturation levels, and poros- abundant precipitation, creating a rain shadow ity, as well as transport, evaporation, and seep- to the west. Consequently, bio-diversity is age loss rates from water source to final destina- greatest along the lower eastern flanks of the tion for agricultural use. Based upon such con- eastern cordillera. The eastern escarpment is siderations, reconstructions of interactive, exceptionally steep and therefore difficult to climate-related, and societal-dependent struc- farm. Because the eastern watershed reaches tural factors have a hydrological component and deep into the intermontane sierra, it receives are thus key to understanding highland-lowland and discharges approximately 90% of all mois- interactional dynamics and their possible rela- ture in the range. Sierran basins have relatively tion to Andean development. modest slopes amenable to rainfall and runoff farming. Cultivation, in conjunction with the THE SETTING: REGIONAL CLIMATE NORMS use of high altitude grasslands for herding, sustains agro-pastoralism and was the basis for The central Andes consists of parallel east- large sierran populations in prehispanic times. ern and western cordilleras. In the north-central Andes the higher eastern range and the lower DROUGHT EVENTS western range enclose intermontane uplands that drain mostly into the Amazon and its Analysis of the ice cores from the southern tributaries. In the southern altiplano region region Quelccaya peak (Thompson et al. 1985, rainfall drainage is mostly into 1986, 1994) and from the north Andean Huas- with a high degree of infiltration that maintains carán mountain (Thompson et al. 1995a), and 279 - Ortloff & Moseley: Agricultural strategies

analysis of the Lake Titicaca sediment cores ies we review the record of cultural change (Abbott et al. 1997; Binford et al. 1997; Ortloff through time using the Uhle-Rowe chronologi- and Kolata 1993; Seltzer 1991) reveals dramatic cal sequence. The sequence begins with Forma- climate shifts. The Quelccaya ice cores indicate tive and Preceramic Periods of long duration (c. periods of wet and dry climate, as well as dust 9500-1800 B.C.E.), and continues with the maxima, over a 1500 year span of time. The Initial Period (IP; 1800-900 B.C.E.), the Early Huascarán ice cores show similar climate varia- Horizon (EH; 900-200 B.C.E.), the Early Inter- tions with dust concentration events character- mediate Period (EIP; 200-600 C.E.), and the izing dry periods. The Lake Titicaca sediment Middle Horizon (MH; 600-1000 C.E.). This last cores present limnological data corroborating division is followed by the Late Intermediate the major wet and dry period climate shifts Period (LIP; 1000-1476 C.E.), then climaxed by found in the Quelccaya and Huascarán ice core the Late Horizon (LH; 1476-1534 C.E.). data (Ortloff and Kolata 1993:200). Initial analysis of the cores documents a 25 to 30 Climate data show that early and middle percent decline in precipitation between 563 phases of the EIP climate were sufficiently stable and 594 C.E. (Shimada et al. 1991:261). This to provide adequate water resources for the drought is notable for both its rapid onset and development of canal based agricul- exceptional severity (Schaaf 1988; Shimada et ture by the Peruvian north coast Gallinazo and al. 1991:248, 261-262). A protracted precipita- Moche polities, as well as by the south coast tion downturn between 1100 and 1500 C.E. Nasca and central coast Lima polities. Highland occurred when rainfall was, on average, 5 to 15 Wari and early phase also flourished percent below previous norms before precipita- during this time, reinforcing the conclusion that tion returned to long term averages around water supplies and runoff were adequate in both 1700 C.E. the coastal and highland zones, although agri- cultural techniques varied greatly from locale to The limnological cores from Lake Titicaca locale. have provided a 3500 year record of precipita- tion induced lake level variation. These cores Towards the end of the EIP, a dry period show early aridity, mid-Holocene lake (Thompson et al. 1985:973) apparently played filling around 1400 B.C.E., drought-induced some role in the decline of the Moche lake level low stands at about 900-800 B.C.E. around 640 C.E. (Shimada et al. 1991), as well and 400-200 B.C.E., as well as at 1-300 C.E. and as in the collapse of Recuay and Lima polities, 1100-1450 C.E. (Abbott et al. 1997; Binford et and that of the south coast Nasca polity. During al. 1997). the MH there was a dramatic expansion and rise in influence of the highland Tiwanaku and Wari Huascarán ice cores from northern Peru states with their highland-adapted agricultural reveal a glacial record of climatic conditions strategies and plentiful water supplies. By con- extending back to late times trast, in coastal areas there was a decline in the (Thompson et al. 1995). Evidence of the area of irrigated land and highland states ex- drought beginning around 1100 C.E. is also panded into coastal regions (e.g., Tiwanaku found in dust maxima and elevated temperature colonies in the Valley and Wari variations seen in the Huascarán cores. influence on southern coastal regions, at Cerro Baúl in the Moquegua Valley, and at Beringa in To reconstruct the effects of changing the Majes Valley; Owen 2007:287-289; 291- climate on Andean highland and coastal societ- 292, 305-316, 321-325; Tung 2007:254-255). ANDEAN PAST 9 (2009) - 280

Towards the end of the Middle Horizon the Rainfall farming is more efficient than decline of the Tiwanaku and Wari states (c. canalized runoff farming (per unit of water 1000-1200 C.E.) may have been due to a pro- input) because of the evaporation and seepage longed drought. Chimu, Chancay, and Ica- associated with rivers and canals. In the arid Chincha societies continued to sustain their sierra at elevations around 2250 m, the Moque- agricultural bases through efficient use of limited gua River’s flow forfeiture reaches 4 percent per irrigation water supplies, and dependence upon kilometer (Williams 1997). Mountain runoff is marine resources. greatly diminished by the time it reaches the lower coastal valleys. Drought, therefore, always Late kingdoms in the altiplano (the Lupaqa, has a more severe effect in coastal desert zones Colla, and minor local groups) arose concur- than in mountain headwater zones. On the rently with the fragmentation of the dominant other hand, when water is adequate, irrigated Tiwanaku state. Towards the end of the LIP, a farming produces far higher yields on average decline in agricultural productivity in the Chimu than rainfall farming. Thus, there is substantial region is associated with sierra investment in irrigation reclamation during rainfall decline as shown by sequential canal protracted episodes of normal or above normal cross-section area decreases and flow rates precipitation. However, growth is not sustain- (Ortloff et al. 1985:78, 86-89, 94, 96-97). High- able when long term precipitation rates decline land polities (Wanka, Chanca, and early ) on the order of 5 percent or more from average arise in this period, but are of lesser regional because runoff drops disproportionately. Conse- influence compared to the Chimu state. A shift quently, over the millennia, populations de- to a wetter period during the start of the Late pendent on irrigation agriculture have repeat- Horizon (Thompson et al. 1985:973, 1986: 364, edly pulsed outward over arid landscapes in wet 1994:85) was followed by political dominance of periods and defensively reconfigured in times of the highland Inca state over coastal, highland, rainfall and runoff decline. This process is and altiplano regions extending from present- reflected in ruins of vast agrarian works that day to mid-. blanket the arid Andean landscape.

Significantly, the Titicaca lake cores, Quel- ADAPTATION TO PROTRACTED DROUGHT ccaya ice cores, and Huascarán dust maxima are concordant in their documentation of a long- The recurrence of protracted drought raises term decline in rainfall levels beginning around the probability that indigenous populations 1100 C.E. This decline appears to be an Andean reacted to episodic dessication in patterned ways expression of the worldwide perturbations in based on prior experiences, and that some of rainfall and temperature known as the Medieval these responses are evident from the archaeo- Warm Period. The long duration of this dry logical record. A very high degree of subsistence period allowed coping strategies to be developed mobility characterizes highland agro-pastoral over many centuries. Drought defensive re- adaptations because they are based upon the sponses that can be inferred from the archaeo- exploitation of multiple, dispersed ecological logical record can be viewed as a measure of a zones stratified by altitude (Murra 1972). An- society’s accomplishments in technical innova- nual hazards associated with short growing tion to reconfigure agro-systems towards greater seasons and poorly developed mountain soils sustainability under climate stress. include topsoil erosion, erratic precipitation, temperature fluctuations, saturated soils, frost, and hail. Mediation requires rapid transmission 281 - Ortloff & Moseley: Agricultural strategies of information so that agricultural and pastoral and placement of commoners’ houses. Erected activities can be reprogrammed on short notice in rows, hundreds, and in some cases, thousands (Earls 1996:302, 304-305). It also requires of qollka were strategically positioned on high preserving, storing, and stockpiling food reserves hills and could be seen from great distances. because poor harvests are frequent (Orlove and Although prominently displayed for reassurance Guillet 1985:10). Drought exacerbates many that the state provided contingency food sup- negative factors affecting human adaptations in plies, their locations on mountainsides also the central Andes and contributes to declines in provided cold air currents to help preserve food productivity, botanical variability, and increas- quality by removal of heat that serves as a ing distances between valued commodities. catalyst for the spoilage of organic material (Morris 1992; Rowe 1946). Although the stores Large, dense population areas are also af- were generally used for state purposes, they were fected by drought because these had long tradi- also used to mitigate food shortages among the tions of complex social organization, culminat- common people (Rowe 1946:266-267). ing with the LH Inca imperium. Inca political formation was a slow process that began shortly In contrast, the coastal Chimu polity re- after 1000 C.E. with the gradual consolidation sponded to low rainfall periods by contracting its of local ethnic groups (Bauer 1992:1, 40-48, 72, agricultural base commensurately with its lower 90-94, 109-123, 124-139, 149-147). Thus, the water supplies. While canals were infilled to nascent polity was formulated and grew during create smaller channels during drought, no the long periods characterized by low average evidence of the reverse process is evident for the rainfall. After 1400 C.E., as average rainfall late MH to LH times when water supplies levels increased, the Inca adapted corporate increased. This may be attributed to the domi- styles of art, architecture, and construction on a nation of the north coast by the Inca and the monumental scale both in the capital region and disassembly of the Chimu state’s agricultural in the provinces. Corvée labor was employed for multi-valley domains to suit Inca political goals large-scale agrarian reclamation of land that was for the region. not farmed, or that was underutilized. Initially, much of the reclaimed terrain was at high In EIP times, it appears that some shift from elevations along the eastern Andean escarp- the Mochica capital in the Moche Valley center ment, utilizing terracing in high rainfall zones, occurred to incorporate, through conquest, although such terracing was a drought response larger northern valleys, the Jequetepeque and at the folk level. Later, as rainfall levels in- Lambayeque in particular, with rivers less sub- creased even more above normal, corvée labor ject to flow rate intermittency and large land was used to reopen farming in lower, warmer areas suitable for agriculture. In the late LIP, a elevations where conquered communities were similar expansion into northern valleys by the often resettled. Therefore, certain Inca corvée Chimu, who were centered at , in policies over time may be considered as adapta- the Moche Valley, had, as its goal, an increase tions to both drought and to increased rainfall. in agricultural sustainability to support an expanding population. This expansion was likely Inca food storing and stockpiling are unsur- driven by similar drought effects that challenged passed in the annals of South American civiliza- agricultural sustainability in the Moche Valley tions (LeVine 1992:15). The monumental with its small land area and intermittent river construction and prominent display of ware- water supplies. Territorial expansion into large, houses (qollka) frequently surpassed the quality ANDEAN PAST 9 (2009) - 282 irrigable valleys is also a coping strategy, albeit a higher fields located near high water table zones, last resort when innovation is lacking. and occupy small rural settlements (Alberracín- Jordán and Mathews 1990:146-148). HIGHLAND AGRICULTURAL STRESS AND RESPONSE While creation of small sunken gardens as a drought response was tenable in limited regions Recent work has postulated that the 1100 of the land-locked Titicaca Basin, where water C.E. drought played a role in the collapse of the was not far below the ground surface, this was Tiwanaku polity centered around Lake Titicaca not an option in most sierra basins with steep (Kolata and Ortloff 1996:110, 151; Ortloff and drainages. Along the western Andean escarp- Kolata 1993). This is premised upon the fact ment there were few means to compensate for that the agricultural base of Tiwanaku was food loss in the dry sierra below 2000 m. Here anchored in some 80 square kilometers of raised slopes are steep, ground- water is deep below the fields in the extensive low-lying areas along lake land surface, runoff is limited, and natural margins north of Tiwanaku (Binford et al. 1997: vegetation is sparse. In this region of the 235, 243, 245). The raised field mounded ridges Moquegua Basin, Tiwanaku colonies and later are elevated 1-1.5m above the water table and post-Tiwanaku Chirabaya populations depend- have planting surfaces 2-10 m wide that range ent upon rainfall and irrigation agriculture from 10 m to 200 m long (Kolata and Ortloff declined significantly during the post-1100 C.E. 1996:118). Built in parallel row segments, each dry period (Ortloff 1989:472-475, 477). ridge is separated from the next by a depressed trough of similar dimensions that held slow Unlike the coastal valleys, where moving spring-supplied or standing ground tectonically-induced river down-cutting forced water which is essential to the high productivity farmers to shift their river canal inlets to lower of ridged field systems (Kolata 1996:118-120). altitudes downstream to channel water into canals (Ortloff et al. 1985:77-78, 85, 90-91, 96- Warmed during the day by solar radiation 97), sierra farmers shifted to higher elevations in absorbed by dark, decomposing organic material pursuit of higher rainfall rates and pre-drought in the water troughs, heat is released during cold quantities of subsurface water supplying moist nights into the soil mounds to maintain internal pasture lands at altitudes about 100 to 400 m or mound temperatures near the freezing point, but more above the normal level of cultivation. insufficiently cold to cause a phase change to There were many constraints on the uphill ice, thus limiting damage to root crop biomass. pursuit of rainfall and soil moisture. Fewer types The mound phreatic zones are also effective in of crops can be grown at higher elevations collecting and storing solar radiative heat due to because soil quality decreases and the frequency the high specific heat of moist soils (Kolata and of frost, hail, and erosion increases. While Ortloff 1989: 252, 256-260). terracing can provide planting surface stability, the labor investment is high and time consum- As lake and runoff levels declined after 1100 ing to implement and thus does not constitute C.E., the water table subsided, desiccating raised a short-term solution to climate variation. In the field systems and reducing their thermal storage central highlands of Peru, by about 1300 C.E. potential. By the time the lake fell to its -12m colder, dryer climate conditions forced a down- low-stand, more than 50,000 hectares of raised ward shift of as much as 150 m in the altitude fields had been abandoned. The population of distribution of natural vegetation zones relative Tiwanaku’s urban core dispersed to utilize 283 - Ortloff & Moseley: Agricultural strategies to today’s levels (Seltzer and Hastorf 1990:402, innovation applied to these fixed systems was 405-408, 410). the only option for maintaining sustainability.

Thus, as highland people moved agriculture To augment agrarian tax revenues, the Inca and pastoralism into higher, wetter altitudes, the imperium often forcibly resettled conquered elevations at which plants, pasture, and crops high altitude communities on lower terrain could grow diminished, and drought-response made more productive by increased rainfall, soil farming moved from gentle to steep inclines. To moisture, and runoff. Although post-LH precipi- reclaim upland mountain slopes, agriculturalists tation rose above long term normal levels, constructed terraces over a span of many gener- demographic decimation in the wake of Euro- ations to control soil loss and regulate moisture pean pandemics left Spanish overlords with few levels for specific crops. Along the Pacific water- people to farm large expanses of arable land. shed and sierra zones, terracing was combined Because above normal rainfall and runoff per- with canal irrigation to capture high elevation sisted until about 1700 C.E., remnants of the runoff streams. In addition to permitting crop indigenous population could still be forcibly cultivation at higher altitudes, terracing was relocated to even lower elevations. This facili- widely used to move farming eastward into the tated political control and religious conversion Atlantic watershed. Even during drought this and imposed cultivation of cultigens region was better watered than the rest of the intolerant of extreme altitudes. If the drought central Cordillera. Hence terracing allowed had not broken neither Inca nor Castilian farming to expand into less extreme elevations resettlement policies would have been tenable. where more types of crops can grow. Thus, during the last millennium, farming, Over the course of many drought-influenced and the millions it supports, have shifted over centuries in the latter part of the LIP, millions of the elevated slopes of the Andean range in terraces were built to reclaim vast areas of the concordance with long term changes in rainfall Andean uplands in the eastern escarpment. and runoff, with political boundaries set as Whereas both agrarian productivity and popula- constraints. This story of climate change and tions declined along the lower Pacific water- human response over the mountain landscape is shed, the drop in rainfall was a major catalyst for shown in ubiquitous terraces, fields, and ruins of economic and demographic radiation into the past agrarian endeavors during EIP to LH times. upper and eastern highlands, culminating by about 1400 C.E. in large populations at high REGIONAL ADAPTIVE AGRICULTURAL altitudes. Because normal sierra runoff farming STRATEGIES produces higher yields than sierra rainfall farm- ing, as the drought mitigated during the Little The adaptive strategies used to defend agro- Ice Age, farmers reverted to lower, warmer production in the face of drought, excessive El settings better for plant growth. Thus, where Niño rainfall, or the presence of above-average mobility was possible, population concentrations rainfall over time, are summarized in Table 1 shifted to maintain agricultural sustainability. and Figure 3. Specific to different geographic For cases for which large fixed investment in a sectors and cultural periods, these strategies specialized agricultural method was tied inti- represent sustainability programs devised to mately to the landscape (altiplano raised fields protect agricultural fields and water supply and coastal irrigation networks, for example), systems. The defensive measures give direct evidence of cultural memory of responses to past ANDEAN PAST 9 (2009) - 284

catastrophic events and constitute evidence of the importance of such events in shaping the technological response pattern relevant to specific areas and polities.

Table 1 - Regional and Adaptive Agro-engineering Technological Strategies

EH EIP MH LIP LH

North Sierra

high rainfall – – – – –

drought 11 – 19 11, 19 11

Southern Altiplano

high rainfall – 5, 8 2, 5, 8, 12, 14 6, 14 6

drought – 8, 12 3, 5, 8, 12, 13, 19 4, 6, 7, 8, 13, 14 7, 19

North Coast

high rainfall 2 2 1 1, 5, 16 –

drought 4, 11, 17, 18 3, 4, 11, 17, 18 3, 4, 12, 15 3, 4, 7, 10, 15, 16, 18 –

South Coast small valleys

high rainfall 17 – – 1, 2 –

drought 18 9, 18 4, 9, 10 4, 9, 10 9

Key: A dash indicates lack of available information. means of reactivating dessicating intra-valley canals. They redirected water in excess of that 1. Canal hydraulic controls: canal inlet blockage required for available arable land, directing flow from rivers to regulate canal intake flow rate; from large rivers to adjacent valleys of political use of canal overflow weirs (downhill water importance to increase the latter’s agricultural spillage from canals) triggered by El Niño-in- footprints and increase local sources of food. duced canal flow rates exceeding design capac- Examples present in Chimu Chicama-Moche ity, i.e., excessive rainfall-induced canal flows Inter-valley, Lambayeque-Supe-Leche Inter- activate supercritical chokes causing side over- valley, and Chillón-Rimac-Lurin (Lima) Inter- fall weirs to activate to release excess flows; valley complexes (Kosok 1965: map page 24, stream-wise alteration of canal slope, wall map page 86 figure 8 page 90, map page 146; roughness, and cross-section shaping to alter Ortloff 1993: 345, 347-351, 356, 2009; Ortloff flow height and velocity; dividing canal flow et al. 1982:581, 583-588, 591-593, Ortloff et al. into two separate streams at different slopes and 1985). velocities, then recombining to induce a hydrau- lic jump in a single channel used as an energy 2. Flood diversion channels: used mainly in the dissipation method to slow flow velocity; Tiwanaku and Lukurmata areas of the Bolivian coupling of intra- to inter-valley canals as a altiplano to intercept and shunt excessive 285 - Ortloff & Moseley: Agricultural strategies

rainfall runoff from adjacent hill slopes directly mountainous terrain (Ortloff 2009) diverted into Lake Titicaca (Kolata and Ortloff 1996: into quebradas to limit inflow damage to major 115, 121, 147, 149-50; Ortloff 1996) in order to canals. modulate ground water level with respect to planting surfaces. Some application to Moque- 6. Terrace agriculture: post-Tiwanaku V, high- gua Valley mountain region terraces (in Wari land Wari, Inca polities, planting surfaces and Tiwanaku colonies) to divert excessive moved to higher elevations during low average rainfall runoff in terrace supply canals into rainfall periods. Use of high elevation canals downhill spillage channels draining into supplied by snowmelt channeled water to irri- quebradas. Use of Pre-Moche or Moche Great gate terrace systems (Ortloff 2009). Trenches for flood water diversion (Ortloff 2009). 7. Sunken gardens (cochas): used by late and post-Tiwanaku V altiplano cultures to supple- 3. Ground water recharging: north coast Moche ment pasturalism-derived food supplies. Chimu and Chimu sunken gardens (wachaques) and coastal wachaques near Chan Chan (Moseley Chan Chan compound wells (Ortloff 1993) and Deeds 1982:31, 33, 35). activated by canal water seepage from field systems; Tiwanaku and Pampa Koani use of 8. Lake Titicaca raised field agricultural zone shifts spring-fed canals to deliver water to local raised to incorporate the optimum raised field moisture field water troughs to alter local water table levels for subsiding or increasing lake and rainfall height and chemical nutrient composition level: optimization method applied to field (Ortloff 2009). systems on Pampa Koani (Ortloff 2009).

4. Springs, wells, and minor sunken gardens: 9. Underground galleries and channels collecting Moche Valley pukio (spring) systems; north groundwater for surface field agriculture:(south coast Chimu wells in Chan Chan (Ortloff 1993: coast Nasca galleries (Schreiber and Lancho 343, 356, 364) and use by earlier north and 1995; 2006). south coast cultures; minor sunken garden systems (cochas)in late to post-Tiwanaku (Kola- 10. Canal and river seepage utilization: north coast ta and Ortloff 1996:134). Use of perpetual Chimu (Chan Chan wells, aquifer recharge from pukio-sourced canals for valley bottom agricul- field system seepage; Ortloff 1993:356, 363); ture at and other sites in the Supe Valley Moquegua Valley Chirabaya coastal ground seep (Ortloff 2009). agriculture (Clement and Moseley 1991:430, 434-435, 441); north coast valley mouth agricul- 5. Runoff interception and river canal shunts to ture at Casma, Virú, and Moche Valleys. modulate groundwater levels: Tiwanaku and Pampa Koani raised field systems (Ortloff 1996: 11. High and mid-sierra reservoir and lagoon water 156, 157, 159, 166; Kolata and Ortloff 1996: storage and transport to coastal valleys: Chimu, 128, 134-137, 148-151) laced with main canals lower Jequetepeque and Lambayeque Valley having elevated weir structures that activate at systems; also Moche Nepeña and San José high water to drain excessive runoff water reservoirs and use in mid-sierra farming. Proba- directly to Lake Titicaca; possible Chiripa ble water delivery by geological fault (or canals) antecedents; Chimu canal systems near Farfán to the Supe Valley to maintain high water table (Jequetepeque Valley) with trenches uphill of and springs to support Caral agricultural base canal systems to intercept rainfall runoff from (Villafana 1986; Ortloff 2009). ANDEAN PAST 9 (2009) - 286

12. Canal collection and diversion of runoff to tional protein source (Moseley 1992b:5, 7, 10- modify ground water profiles: Pampa Koani 12, 16, 22, 33). Tiwanaku systems area on the Taraco peninsula (Ortloff 1996). 19. Adaptions toward sierra pastoralism: As a drought response, the shift from land agriculture 13. Snow-melt water collection channels directed to to high sierra animal herding can provide addi- mountainside terraces: Upper highlands Moque- tional food resources. gua Valley drainage, post-Tiwanaku V, Estuqueña and Wari terraces–technique used in The adaptive strategies set out in Table 1 elevated temperature periods to provide water indicate conscious efforts to control water in the highlands. supplies by a variety of technologies specific to different geographical areas and agricultural 14. Shift to terrace agriculture when Lake Titicaca systems. While many of these strategies are completely covers raised fields:Late post-Tiwana- related to observation of long-term climate ku, possible reuse of early Tiwanaku terraces in trends (3, 4, 6, 7, 8, 9, 11, 13, 14, 17, 18, 19) Inca times. others provided specific defenses against short- term El Niño related flooding (1, 2, 3, 5, 12, 14), 15. Multi-valley transport/distribution canals: and drought events (1, 3, 4, 6, 7, 8, 9, 10, 11, Chicama-Moche Inter-valley canal, Motupe- 12, 13, 14, 15, 16, 17, 18, 19). Table 1 indicates Leche-Lambayeque inter-valley canals, Lima a library of agro-engineering responses to both complex (Chillón-Rimac-Lurín inter-valley long and short term climate variations over canals) used as possible drought remediation many time periods and demonstrates that obser- measure based on redistribution of excess water vation of climate over time and related techno- beyond that needed for valley agriculture, to logical innovations and controls were important large land areas in adjacent valleys with small, factors in system design and operation. intermittent rivers (Kosok 1965: map page 34, figure 8, map page 146). MODELING DROUGHT STRESS: KEY COMPO- NENT FACTORS IN HIGHLAND/LOWLAND 16. Hydraulic efficiency improvements in canal RUN-OFF RATIO (R) design by cross-section, slope, and wall roughness changes: (Ortloff 1993:345, 347-351, 356, 2009; During drought, coastal farmland that Ortloff et al. 1982, 1985); canal hydraulic design normally supports an abundance of crop types changes to maximize low canal flow rates during can sustain only a reduced number of types with droughts. Improvements also include canal low water demands. Crop yields are reduced in spatial relocations related to river down-cutting response to lower water supplies. Sierra range- and inlet stranding. land, in contrast, may support pastoralism and some form of agriculture due to elevated rainfall 17. Lomas farming: fog condensation agriculture amounts at higher altitudes (but which, during in coastal, natural surface field–yields amplified drought periods, are lower than normal). One during wet periods (Moseley 1992a:41). reason for the difference between coastal and highland agriculture under drought is related to 18. Adaptions toward a marine resource base: For the soil types that influence rainfall infiltration, drought affected land regions with access to runoff, and transport rates, given highland marine resources this shift can provide an addi- rainfall sources. Highland soils retain water within their porous structure until saturation is 287 - Ortloff & Moseley: Agricultural strategies reached. Past this level runoff occurs. Chan- 10 to 15 percent result in runoff reductions of neled runoff is further subject to evaporation, 25 to 40 percent or more. Significantly, the seepage, and subsurface porous medium reten- asymmetric relationship between rainfall and tion effects through a different set of nonlinear runoff also works in reverse. Increased precipita- relationships than the storage/saturation effect, tion rapidly saturates the soil which then dis- resulting in imbalances in the rainfall/runoff cards water and amplifies runoff. This effect was delivery rate. prevalent in the first two centuries C.E. when precipitation rose by 20 to 25 percent and the To illustrate the effect of drought stress and runoff by 72 to 90 percent. the hydrological relationships between rainfall, soils, runoff, and flow losses, an illustration from Drought stress is exacerbated by the fact the Moquegua Basin is useful. The Moquegua that once rainfall saturates the soil and excess Basin lies on the Pacific watershed to the west of water is released, some surface runoff is lost to Lake Titicaca. This river system is 139 km in evaporation and seepage. Due to these factors, length, with headwaters reaching slightly above the Moquegua River loses about 4 percent of its 5,100 masl. Along the Moquegua coast precipi- flow per kilometer in the arid sierra at elevations tation is negligible, but it increases gradually in around 2250 masl. Other than during spring the interior with altitude. However, the quantity floods, the river channel does not normally carry of rainwater only exceeds (saturated) retention surface flow at elevations below 1200 m. Farm- values in the 19 percent of the basin that is ing in the coastal section of the drainage de- above 3900 masl. Between 3900 and 4500 m, pends on springs fed by subsurface groundwater average rainfall is about 360 mm/yr, of which flows originating high in the river basin. The 260 mm is absorbed at saturation and 100 mm relationship between highland rainfall and is available as runoff (Perú, ONERN 1976). In coastal spring flow is highly asymmetrical be- this zone, a 10 percent, or 36 mm, decline in cause subsurface water flows through porous rainfall to 324 mm decreases runoff by 36 per- geological strata. Similar to soils, porous deposits cent from 100 mm to 64 mm. Given a specific have different hydraulic conductivity and satu- soil type with a given retention capability, the ration values. Although these values are poorly soil always retains the same amount of water, so known, there are indirect indications that a 15 percent decrease in rainfall results in a 54 coastal spring flow may have dropped by 80 percent decrease in runoff. In the elevation zone percent during the 1100 C.E. drought (Ortloff between 4500 and 4900 masl, rainfall averages 1989:457-477). These calculations are approxi- 480 mm/yr and a 10 percent or 15 percent mations for the Moquegua Basin. Other soil rainfall reduction results in runoff reductions of adsorption and precipitation values characterize 21.8 percent and 32.7 percent respectively. other drainages. Nonetheless, relationships Thus the asymmetric disparity between rainfall between rainfall and runoff are always nonlin- and runoff reduction diminishes as precipitation ear, so drought always exerts asymmetrically increases. Comprising less than 3 percent of the greater stress on runoff farming than on rainfall Moquegua Basin, precipitation in the zone of farming. alpine tundra above 4900 masl is principally in the form of snow and ice (Perú, ONERN 1976). The runoff not directly channeled into rivers The runoff contribution is not known because is diminished en route to coastal zones by fur- an unknown amount of this moisture is retained ther infiltration into increasingly porous soils in glaciers and snow-fields. Nonetheless, for the (adding to the local water table profile) as well upper river basin as a whole, rainfall declines of as by evaporation losses. The resulting coastal ANDEAN PAST 9 (2009) - 288 river hydrographs track the availability of coast- rainfall because elaborate field drainage systems al irrigation water over time. Of course, the shunt water directly into the lake, thus limiting details will vary between coastal valleys as infiltration into the water table (Table 1). functions of local soil geomorphology, topogra- Because the collection basin rainfall rates and phy, evapotranspiration, agricultural productiv- Lake Titicaca height vary with seasonal and ity potential, and temperature and humidity climate-related rainfall/runoff fluctuations history. The net effect is one of a nonlinear, but (Binford and Kolata 1996:37-38), the raised generally similar, relationship between unit field water table height sufficient to maintain amounts of input rainwater at different altitudes agriculture shifts vertically and laterally within and times, and net deliverable water to coastal the extensive lacustrine field systems (Ortloff irrigation systems with time lags. From this 1996; Kolata et al. 1996:205), and productive discussion, it may be concluded that in dry farming zones can likewise be shifted. This periods rain at high altitudes may sustain some indicates that not all of the raised fields in the form of agriculture in these zones, but coastal Lukurmata area north of Tiwanaku were farmed agriculture derived from runoff into rivers from simultaneously, but only those areas with water the same watershed will experience a severe trough zones supplied by spring water and deficit of irrigation water. In terms of quantify- elevated groundwater profiles at the correct ing the runoff effect, the Runoff Ratio (R) is height for agriculture were utilized, with remain- defined as the average net runoff rate from an ing areas allowed to fallow. area divided by the average rainfall delivery rate to the same area. Here R=0 denotes zero runoff Prolonged drought over many years can and R=1 denotes that all delivered rainfall to destroy the special heat storage features that an area converts to runoff, implying a total provided frost damage protection under diurnal saturation condition. and seasonal temperature variations (Kolata and Ortloff 1996:130) and decrease the height of the VULNERABILITY INDEX (1-V) water table in raised field troughs necessary to sustain agriculture. The raised field systems can, Figure 4 shows a plot of the main agricul- nevertheless, be optimized to highland climate tural strategies practiced by Andean civiliza- conditions and cycles to produce high crop tions. A vulnerability index is defined relative to yields through interventions shown in Table 1, available rainfall levels. The first entry, raised and as demonstrated by modern resurrection field agriculture (Index 1), was widely practiced and use of these systems (Kolata 1996:203, 206- by MH Tiwanaku III-V around the southern 207, 226, 228-230). and western periphery of Lake Titicaca. Tiwa- naku groundwater based agricultural systems are The next least vulnerable agricultural system largely invulnerable to short-term drought due to rainfall fluctuations is a variant of raised field to the continuous arrival of subsurface water systems–sunken gardens (Index 2). These from earlier rainfall in the immense collection systems are pits excavated to the phreatic zone zone around the lake. Because groundwater and are mostly found as a last resort drought transport velocities are on the order of a few response system used when the water table has centimeters per month, groundwater from declined out of reach of plant root systems. distant collection basins may have originated as While sunken gardens are common in the 1100 infiltrated rain that occurred many years earlier. C.E. post-collapse settlements around Tiwa- The groundwater based systems are likewise naku, similar wachaques are also found in Peru- relatively invulnerable to seasonal excessive vian north coastal valleys in response to the 289 - Ortloff & Moseley: Agricultural strategies

pan-Andean LIP drought. Being primarily a 471-472). Such systems rely on groundwater drought remediation measure, these systems seepage to coastal bluffs over long underground have no defense against excessive rainfall and distances, and thus are only marginally produc- groundwater level rises, because simple drainage tive compared to other delivery systems. paths usually do not exist. Survival of the more vulnerable agricultural At the next level of vulnerability are the systems is questionable past a critical level (line terraces widely used by Inca, Wari, and post- DD, Figure 4) and extinction of highly vulnera- Tiwanaku highland civilizations (Index 3). ble systems is inevitable whenever reconfigura- Terraces are mostly supplied by rainfall and tion to lower vulnerability systems is impossible. provide a well-drained agricultural system that In the presence of yearly rainfall and runoff is effective during rainy seasons. Other variants variations, high vulnerability systems must have are supplied by snow-melt and channel water superior technology, innovation, and modifica- during periods of low rainfall and elevated tion features to maintain agricultural produc- temperature (particularly in the upland tion. Figure 4 plots a Vulnerability Index (1-V) Moquegua area). As rainfall diminishes, these such that the largest values of the index denote systems generally become marginal for produc- the least vulnerable agricultural systems. tion unless supplied by channeled water. TECHNOLOGY AND TECHNOLOGICAL Next in increasing order of vulnerability CHANGE UNDER CLIMATIC DURESS (Index 4) is canal-fed irrigation as practiced primarily by north and south coast civilizations. Andean agricultural systems have a long Figure 4 indicates that such systems are only history of evolution and improvement over time. viable in the presence of highland rainfall ex- A sample of agricultural strategies (Figure 3) is ceeding saturation conditions. As such, if coast- shown for modified and unmodified water and al agriculture flourishes, then highland agricul- land surface variables. An initial choice of an ture has an excess of water supplies due to the agricultural system fitting local ecological condi- nonlinear input/delivery R relationship. The tions is made by early inhabitants. System evolu- highlands appear to always have demonstrated tion proceeds through observation of agro- less vulnerability to agricultural stress regardless production changes in response to field system of the level of rainfall, provided the technology design changes. Key requirements are the pres- to use the available water is adequately devel- ervation of the system and its efficient function- oped in each ecological zone, and drainage ing under seasonal weather fluctuations, as well technology to control the water table height is as those arising from large-scale climate fluctua- in place. tions. Therefore, the ability to modify an agri- cultural system to maintain sustainability in Generally, long canals are more vulnerable anticipation of climate-related changes in water (Index 5) than short canals due to greater supply is part of the original concept of the seepage and evaporation losses, tectonic/seismic system as shown by the interventions listed in distortions, and higher technological demands Table 1. The agro-system modifications must be for design, low-angle surveying (Ortloff 1995:60, performed more rapidly than the climate varia- 70-71 ), and construction. Yet more vulnerable tion effect unfolds, e.g., system modifications are (Index 6) are the coastal seeps that supply effective when long-term climate changes are agriculture, mainly in northern Chile and in the initially observed and system modifications are Ilo area of the Moquegua Valley (Ortloff 1989: carried out in anticipation of a long-term trend. ANDEAN PAST 9 (2009) - 290

Therefore, the Vulnerability Index (1-V) of an irrigation. The term, P’=P/Pmx is defined as the agricultural system depends upon the sustain- population density ratio where Pmx is the maxi- ability of an initial design choice in the face of mum population sustainable by the in-place weather and climate variations, as well as the agrosystem. If P’=1, then P’(2-P’)=1 at the ability to modify technology (T) in time (t), maximum population level balanced with the denoted as (dT/dt), faster than the climate- food supply. If P'=0, then P’(2 - P’) = 0, indi- induced creation/evolution rate of a climate cating that a very small population exists (such related disaster (dD/dt). as may occur after a natural or man-made disaster). Thus: AGRICULTURAL SUSTAINABILITY MODEL 0

Several key factors influencing agricultural As before, (1-V) is the agricultural Vulnera- stainability have been discussed from the view- bility Index (0 < V < 1) for the agricultural point of agricultural systems strategies. These system involved as shown in Figure 4. For the factors, and others, are next combined to pro- remaining terms, 0< dT/dt<1 represents the duce a trend equation where increases or de- time rate of technology (T) change to surmount creases in each term imply a net increase or a long term climate effect (excessive rainfall, decrease in the agricultural sustainability (Q) of drought) on agricultural production. Here the a society. The quantities in the equation (Q, R, maximum dT/dt value is assumed to be unity to S, P, V, Y, dT/dt and dD/dt) are non-dimen- represent a technology growth rate typical of sional values normalized to the maximum refer- most advanced agriculture based societies. Here ence state for each variable. A large value of Q dT/dt can be large due to technical innovations connotes agricultural sustainability, while a listed in Table 1. The dD/dt maximum value small value denotes the opposite. From the may be typically very large for rapidly evolving foundation of the above discussion, a simplified disasters such as El Niño events (reducing Q model equation, based upon agricultural parame- dramatically in a short period). The ters only (i.e., excluding implied or induced 1

ized to Smx where S=0 represents zero crop Niño flood event occurs beyond the defense storage and S=1 represents total storage of all mechanisms’ ability to protect, then unconsumed crops. The quantity Y•A•R repre- dT/dD÷dD/dt is a small number indicating no sents the main comestible crop yield per land contribution to agricultural sustainability, Q. If, unit area times the total land area times the however, a climate related disaster evolved at available water supply to the area normalized so the same rate as a defensive technology, then that 0

creases. If P’.1 in the presence of a declining one society exerts overarching influence over agricultural supply, the agricultural resources are vast territories. During intermediate periods, inadequate to feed a large workforce over time dominant regional states may exert control to ensure rapid dT/dt changes to increase Q. primarily through branching government struc- Thus, only a population balanced with agricul- tures capable of integrating adjacent territories tural supply (including storage) promotes large into the same ideological and political template. sustainability Q values. The relative value of Q Although some revision regarding Formative (0.2

societies with different agricultural systems that Binford, Michael W., Alan L. Kolata, Mark Brenner, John are subject to different climatic conditions, W. Janusek, Matthew T. Seddon, Mark Abbott, and Jason H. Curtis provide a partial basis for underlying factors that 1997 Climate Variation and the Rise and Fall of an lead to changes in cultural patterns and agricul- Andean Civilization. Quaternary Research 47(2): tural sustainability. 235-248. Clement, Christopher Ohm and Michael E. Moseley Because only fragmentary details of Andean 1991 The Spring-Fed Irrigation System of Carrizal, Peru: A Case Study of the Hypothesis of Agrar- climate cycles are known from ice and lake core ian Collapse. Journal of Field 18(4): data, only an approximate hypothesis can be 425-443. offered at present to explain the effects of Earls, John changing climate upon socio-political structure 1996 Rotative Rank Hierarchy and Recursive Organi- and sustainability. For the present, however, zation: The Andean Peasant Community as a Viable System. Journal of the Steward Anthropo- some factors underlying the agricultural basis of logical Society 24(1, 2):297-320. societies have been discussed and preliminary Kolata, Alan L., editor arguments have been advanced which relate 1996 Tiwanaku and Its Hinterland: Archaeology and climate change to observed agricultural pattern Palaeoecology of an Andean Civilization. Volume changes. The coupling and feedback of these 1, Smithsonian Series in Archaeological Inquiry. Washington, D.C.: Smithsonian Institution effects as they relate to the political, economic, Press. religious, governmental, and social responses of Kolata, Alan L. and Charles R. Ortloff societies remains an area for investiga- 1989 Thermal Analysis of Tiwanaku Raised Field tions. Systems in the Lake Titicaca Basin of . Journal of Archaeological Science 16:233-263. 1996 Tiwanaku Raised-field Agriculture in the Lake EDITORS’ ACKNOWLEDGEMENT Titicaca Basin of Bolivia. In Tiwanaku and its Hinterland: Archaeology and Palaeoecology of an We thank William P. Mitchell for assistance in the editing Andean Civilization, Volume 1, Agroecology, of this paper. edited by Alan L. Kolata, pp. 109-151. Volume 1, Smithsonian Series in Archaeological Inquiry. REFERENCES CITED Washington, D.C.: Smithsonian Institution Press. Abbott, Mark B., Michael W. Binford, Mark Brenner, and Kolata, Alan L., Oswaldo Rivera, Juan Carlos Ramírez, Kerry R. Kelts and Evelyn Bemio 1997 A 3500 14C Yr. High-Resolution Record of Wa- 1996 Rehabilitating Raised-Field Agriculture in the ter-Level Changes in Lake Titicaca, Bolivia/ Southern Lake Titicaca Basin of Bolivia: Theory, Peru. Quaternary Research 47(2):169-80. Practice, and Results. In Tiwanaku and Its Hinter- Albarracín-Jordán, Juan, and James Edward Mathews land: Archaeology and Palaeoecology of an Andean 1990 Asentamientos Prehispánicos del Valle de Tiwanaku, Civilization, Volume 1, pp. 203-230. Smithsonian Volume 1. : Producciones CIMA. Series in Archaeological Inquiry. Washington, Bauer, Brian S. D.C.: Smithsonian Institution Press. 1992 The Development of the Inca State. Austin: Uni- Kosok, Paul versity of Texas Press. 1965 Life, Land and Water in Ancient Peru. New York: Binford, Michael W. and Alan L. Kolata Long Island University Press. 1996 The Natural and Human Setting. In Tiwanaku LeVine, Terry Y. and Its Hinterland: Archaeology and Palaeoecology 1992 The Study of Storage Systems. In Inka Storage of an Andean Civilization. Volume 1, pp. 23-56. Systems, edited by Terry Y. LeVine, pp. 3-28. Smithsonian Series in Archaeological Inquiry. Norman: University of Oklahoma Press. Washington, D.C.: Smithsonian Institution Morris, Craig Press. 1992 The Technology of Highland Inka Food Storage in Inka Storage Systems, edited by Terry Y. Le Vine, pp. 237-258. Norman: University of Okla- homa Press. ANDEAN PAST 9 (2009) - 298

Moseley, Michael E. and Peter Scarr. Oxford: British Archaeological 1992a The Incas and their Ancestors: The Archaeology of Reports International Series 545(ii):457-477. Peru. London and New York: Thames and 1993 Chimu Hydraulics Technology and Statecraft on Hudson. the North Coast of Peru. In Research in Economic 1992b Maritime Foundations and Multilinear Evolu- Anthropology, Supplement 7, edited by Vernon L. tion: Retrospect and Prospect. Andean Past 3:5- Scarborough and Barry L. Isaac, pp. 327-367. 42. Greenwich, Connecticut: JAI Press. Moseley, Michael E. and Eric E. Deeds 1995 Surveying and Hydraulic Engineering of the 1982 The Land in Front of Chan Chan: Agrarian Pre-Columbian Chimu State: AD 900-1450. Expansion, Reform, and Collapse in the Moche Cambridge Archaeological Journal 5(1):55-74. Valley. In Chan Chan: Andean Desert , edited 1996 Engineering Aspects of Tiwanaku Groundwater- by Michael E. Moseley and Kent C. Day, pp. Controlled Agriculture. In Tiwanaku and its 25-53. Albuquerque, New : University of Hinterland: Paleoecological and Archaeological New Mexico Press. Investigations in the Titicaca Basin of Bolivia, Moseley, Michael E., Jorge E. Tapia, Dennis R. Satterlee, Volume 1, Agroecology, edited by Alan L. Kolata, and James B. Richardson III pp. 153-167. Washington, D.C.: The Smithso- 1992a Flood Events, El Niño Events, and Tectonic nian Press. Events, In Paleo-ENSO Records, International 2009 Water Engineering in the Ancient World: Archaeo- Symposium, Extended Abstracts, edited by L. logical and Climate Perspectives on Societies of Ortlieb and J. Machare, pp. 207-212. Lima: Ancient South America, the , and ORSTOM. South-East . Chapter 1, South American Moseley, Michael E., David. Wagner, and James B. Water Engineering. Oxford: Oxford University Richardson III Press. 1992b Space Shuttle Imagery of Recent Catastrophic Ortloff, Charles R., Robert A. Feldman, and Michael E. Change Along the Arid Andean Coast. In Paleo- Moseley shorelines and : An Investigation of 1985 Hydraulic Engineering and Historical Aspects of Method, edited by Lucille Lewis Johnson and the Pre-Columbian Intravalley Canal Systems of Melanie Stright, pp. 215-35. Boca Raton, Flor- the Moche Valley, Peru. Journal of Field Archae- ida: CRC Press. ology 12:77-98. Murra, John Victor Ortloff, Charles R. and Alan Kolata 1972 El “control vertical” de un máximo de pisos 1993 Climate and Collapse: Agro-Ecological Perspec- ecológicos en la economía de las sociedades tives on the Decline of the Tiwanaku State. andinas. In Ortiz de Zúñiga, Iñigo, Visita de la Journal of Archaeological Science 20(2):195-221. provincia de León de Huánuco en 1562, edited by Ortloff, Charles R., Michael E. Moseley, and Robert A. John V. Murra, Volume 2, Visita de los Yacha y Feldman mitmaqkuna cuzqueños encomendados en Juan San- 1982 Hydraulic Engineering Aspects of the Chimu- chez Falcon, transcribed by Felipe Márquez Aban- Chicama-Moche Intervalley Canal. American to, pp. 428-476. Huánuco, Perú: Universidad Na- Antiquity 47(3):572-595. cional Hermilio Valdizán, Facultad de Letras y Owen, Bruce Educación, Series Documentos para la Historia y 1993a A Model of Multiethnicity: State Collapse, Competi- Etnología de Huánuco y la Selva Central. tion, and Social Complexity from Tiwanaku to Orlove, Benjamin S. and David W. Guillet Chiribaya in the Osmore Valley, Peru. Ph.D. 1985 Theoretical and Methodological Considerations dissertation, University of California, Los on the Study of Mountain Peoples: Reflections Angeles. on the Idea of Subsistence Type and the Role of 1993b Early Ceramic Settlement in the Coastal Osmore History in Human Ecology. Mountain Research Valley: Preliminary Report. Paper presented at and Development 5(l):3-18. Convergence and the Institute of Andean Studies 1993 Annual Difference in Mountain Economics and Societies: A Meeting. Comparison of the Andes and Himalayas. http://bruceowen.com/research/researchperu/ Ortloff, Charles R. htm (Consulted 5 December 2008). 1989 A Mathematical Model of the Dynamics of 2007 Rural Wari Far from the Heartland: Huamanga Hydraulic Societies in Ancient Peru. In Ecology, Ceramics from Beringa, Majes Valley, Peru Settlement, and History in the Osmore Drainage, Andean Past 8:287-373. edited by Don Stephen Rice, Charles Stanish, http://bruceowen.com/research/researchperu/ htm (Consulted 5 December 2008). 299 - Ortloff & Moseley: Agricultural strategies

Perú, ONERN [Oficina Nacional de Evaluación de Thompson, Lonnie G., Mary E. Davis, and Ellen Recursos Naturales] Mosley-Thompson 1976 Mapa ecológico del Perú: Guía explicativa. Lima. 1994 Glacial Records of Global Climate: A 1500-Year Rowe, John Howland Tropical Ice Core Record of Climate. Human 1946 Inca Culture at the Time of the Spanish Con- Ecology 22(l):83-95. quest. In Handbook of South American Indians, Thompson, Lonnie G., Ellen Mosley-Thompson, J. F. volume 2, The Andean Civilizations, edited by Bolzan, and B. R. Koci Julian H. Steward, pp. 183-330. Washington, 1985 A 1500-Year Record of Tropical Precipitation in D.C.: U.S. Government Printing Office. Ice Cores from the Quelccaya Ice Cap, Peru. Satterlee, Dennis E. Science 229(4717):971-973. 1993 Impact of a Fourteenth Century El Niño Flood on Thompson, Lonnie G., Ellen Mosley-Thompson, W. an Indigenous Population Near Ilo, Peru. Ph.D. Dansgaard, and P. M. Grootes Dissertation, University of Florida. 1986 The Little Ice Age as Recorded in the Stratigra- Satterlee, Dennis E., Michael E. Moseley, David K. phy of the Tropical Quelccaya Ice Cap. Science Keefer, and Jorge E. Tapia A. 234(4774):361-364. 2000/01 The Miraflores El Niño Disaster: Convergent Thompson, Lonnie G., Ellen Mosley-Thompson, M. E. Catastrophes and Prehistoric Agrarian Change Davis, P.-N. Lin, K. A. Henderson, J. Cole-Dai, J. F. in Southern Peru. Andean Past 6:95-116. Bolzan, and K-b. Liu Schaaf, Crystal Barker 1995 Late Glacial Stage and Holocene Tropical Ice 1988 Establishment and Demise of Moche V: Assessment Core Records from Huascarán, Peru. Science of the Climatic Impact. M.A. Thesis, Harvard 269:46-50. University Extension School, Cambridge, Massa- Tosi, Joseph A., Jr. chusetts. 1960 Zonas de vida natural en el Perú: Memoria expli- Schreiber, Katharina J. and Josüé Lancho Rojas cativa sobre el mapa ecológico del Perú. Lima: 1995 The of . Latin American Antiquity Instituto Interamericano de Ciencias Agrícolas 6(3):229-254. de la OEA, Zona Andina. Proyecto 32, Programa 2006 Aguas en el desierto: Los puquios de Nasca. Lima: de Cooperación Técnica. Fondo Editorial de la Pontificia Universidad Tung, Tiffiny Católica del Perú. 2007 The Village of Beringa at the Periphery of the Seltzer, Geoffrey O. : A Site Overview and New Radio- 1991 Glacial History and Climatic Change in the Peru- carbon Dates. Andean Past 8:253-286. vian-Bolivian Andes. Ph.D. Dissertation, Univer- Villafana Avila, Juan Fernando sity of Minnesota, Minneapolis. 1986 Sistemas hidráulicos incas. Lima: Lluvia Editores. Seltzer, Geoffrey O. and Christine A. Hastorf Willey, Gordon 1990 Climatic Change and its Effect on Prehispanic 1953 Prehistoric Settlement Patterns in the Virú Valley, Agriculture in the Central Peruvian Andes. Perú. Washington, D.C.: Smithsonian Institu- Journal of Field Archaeology 17(4):397-414. tion, Bureau of American Ethnology Bulletin 155. Shimada, Izumi, Crystal Barker Schaaf, Lonnie G. Thomp- Williams, Patrick Ryan son, and Ellen Mosley-Thompson 1997 Disaster in the Development of Agriculture and the 1991 Cultural Impacts of Severe Droughts in the Evolution of Social Complexity in the South- Central Prehistoric Andes: Application of a 1,500-Year Andean Sierra. Ph.D. Dissertation, University of Ice Core Precipitation Record: Archaeology and Florida. Arid Environments. World Archaeology 22(3): Wilson, David J. 247-270. 1988 Prehispanic Settlement Patterns in the Lower Santa Valley Peru: A Regional Perspective on the Origins and Development of Complex North Coast Society. Washington, D.C.: Smithsonian Institution Press. ANDEAN PAST 9 (2009) - 300

Figure 1: Map of Northern and Central Peru showing some of the geographic features (in italic type) and sites (in uncial type) mentioned in the text. The names of modern are written in all capitals. 301 - Ortloff & Moseley: Agricultural strategies

Figure 2: Map of Southern Peru and the Lake Titicaca Region of Bolivia showing some of the geographic features (in italic type) and sites (in uncial type) mentioned in the text. The names of modern cities are written in all capitals. ANDEAN PAST 9 (2009) - 302

Figure 3. Agricultural strategies for artificial (modified) and natural planting surfaces for surface and subsurface water sources. 303 - Ortloff & Moseley: Agricultural strategies

Figure 4. Vulnerability Index (1 - V) for different types of agricultural systems and water supply methodologies (rainfall intercepted and runoff) indexed by numbers 1-6. ANDEAN PAST 9 (2009) - 304

Figure 5:(best-estimate) application of the Q (agricultural sustainability) equation for Moche, Tiwa- naku, and Chimu societies in their heartland areas. The D1 notation indicates the time of drought- related collapse of the (EIP) Moche V society in the Moche Valley area; the D2 notation denotes loss of agricultural sustainability due to a later drought period during the 12th Century CE affecting (MH) Tiwanaku and (LIP) Chimu societies. (The Chimu curve shown applies only to the Moche Valley capital area not for expansion period valley sites to the north.) The curves indicate that the drought- related decline in agricultural productivity most probably underwrote subsequent changes in societal structure and agricultural strategies of these societies as previously noted.