INTRODUCTION: This still-unpublished monograph chapter is intended as a supplement to course lectures about Andean herding and agricultural ecology. The chapter itself is still incomplete, a work in progress, and should not be thought of as definitive.




This chapter seeks to provide a conceptual framework for the interpretation of the archaeological data to be presented in Chapter 6. A few aspects of this effort were anticipated in Chapter 2. We rely primarily on ethnographic and ethnohistoric studies from the southern half of the Central Andean highlands. We are not searching for highly specific analogs between historic and prehispanic herding and agriculture. Our purpose is to highlight general patterns of production, zonal complementarity, and resource redistribution inherent for preindustrial agricultural and pastoral economies set in a natural and cultural matrix that poses a finite range of constraints and offers a finite range of possibilities. We assume that historic patterns of agriculture and herding have direct links, however uncertain, to those of the prehispanic past.

Because economy and polity are so closely inter-related, we begin this chapter with a general overview of our current understanding of prehispanic Central Andean cultural development. This should help provide a basis for understanding which aspects of the historic-period patterns are most pertinent to the interpretation of our own archaeological data. Because our fieldwork contributes nothing to the Archaic period, we extend our view back only as far as the beginning of the Formative era (ca. 2000 B.C., Table 5.02). Because archaeological knowledge of the ceja de montaña and montaña zones remains limited, we will not discuss the prehistory of these regions.


A Synopsis of Prehispanic Cultural Development

Both agriculture and camelid herding were fully established in much of the Central Andes by the beginning of Formative times (Pearsall 1992; Wing and Wheeler 1988). The Junín puna was an early center of camelid domestication, while remains of the full spectrum of Andean cultigens have been found in Late Archaic and Early Formative archaeological deposits in several kichwa valleys


Early and Middle Formative Chiefdoms (Initial Period and Early Horizon)

Prior to the end of the third millennium B.C. there is little indication of hierarchical organization anywhere in the Central Andes. The most complex South American societies of the third millennium B.C. were concentrated in the far northwest, between Panama and Ecuador (Damp 1984; DeBoer 1996; Holm 1987; Lathrap and Marcos 1977; Meggers et al. 1965; Reichel-Dolmatoff 1965, 1985; Schwartz and Raymond 1996; Zeidler 1991). By ca. 2000 B.C. there is clear evidence in several parts of Peru for public architecture; for increases in the size and density of sedentary population; for the intensification of agriculture and herding; and for the intensification and expansion of inter-regional exchange of both materials and information (e.g., Fung 1969, 1988; Grieder et al. 1988; Matsuza 1978; Moseley and Willey 1973; S. Pozorski and T. Pozorski 1979; Quilter 1985).

By later Early Formative times (ca. 1500-900 B.C.) the central and northern Peruvian coasts had emerged as the cores of a cultural florescence that included formal public architecture and an impressive development of formal ritual iconography (e.g., Burger 1992; Burger and Salazar-Burger 1991; Donnan 1985; Patterson 1985; S. Pozorski and T. Pozorski 1986, 1987, 1988, 1992; T. Pozorski 1975; T. Pozorski and S. Pozorski 1995; Ravines and Isbell 1975; Rossello 1997). Some adjacent highland kichwa valleys participated in this florescence (e.g., Burger 1992; Fung 1975; Izumi et al. 1972; Izumi and Sono 1963; Izumi and Terada 1972; Kaulike 1975, 1976; Onuki 1985; Rosas and Shady 1970, 1974; Shady 1992), but many others did not, and the entire southern half of the Central Andes -- both coastal and highland sectors -- remained a developmental backwater until the Middle Formative in the first millennium B.C..

We view these Early Formative developments as the formation of chiefdom polities, rooted in Late Archaic Big Man systems that may have flourished most notably in the larger coastal valley deltas of central and northern Peru. Inter-zonal exchange, across the full ecological spectrum from Pacific coast through the highlands and into the upper Amazon Basin, is attested by the occurrences of small quantities of exotic marine, sierra, and montaña materials in several archaeological sites (e.g., [citations], maybe Burger 1992, with specific page nos.; maybe Matos on Ancon). Such exchanges, in the context of prestige-building activities carried out by competitive Big Men and Chiefs, may have initially introduced cultigens and agriculture into the Pacific coastal valleys. It is for this same time that Pearsall (1980a, 1980b) reports evidence from archaeological sites in the Junín puna for an intensification of puna-kichwa plant exchange; comparable intensification is seen to be underway at several other excavated highland sites (e.g., Lynch 1980; MacNeish 1979, 1981).

By the beginning of the first millennium B.C. the pace of these developments had accelerated throughout central and northern Peru, and had got underway as well farther south. The Chavin stylistic complex that spread selectively throughout much of northern Peru is one example (e.g., Burger 1988, 1992; Lumbreras 1971), and the first presence of substantial sedentary settlements in a well-defined site-size hierarchy in the Titicaca Basin is another (Albarracin-Jordan 1996; Hastorf (ed.) 1998; Mathews 1995; McAndrews et al. 1997; Stanish et al. 1997). Settlement pattern studies on the Peruvian north coast (Feldman 1983; Willey 1953; Wilson 1988, 1995) reveal the construction of impressive complexes of monumental public buildings, the use of formal ritual iconography, and the development of canal irrigation and significant population growth -- although such agricultural intensification and population growth were still modest by the standards of subsequent centuries.

The trends toward growing cultural complexity that had been set in motion during the antecedent Early Formative accelerated and expanded during the Middle Formative. Just as impressive is the apparent absence of such development in many places, such as our own survey area. There are many other regions where Early and Middle Formative occupation is known to be absent or very modest (e.g., Meddens 1984, 1991; Schjellerup 1997; Schreiber 1987; Stanish 1992; Valdez and Vivanco 1994).

The underlying causes of this Middle Formative variability remain unclear. Florescent developments appear to have occurred either in areas that are unusually productive in terms of Formative agriculture (e.g., several larger coastal valleys with permanent rivers and large, irrigable floodplains), or where there is an unusual potential for the direct combination of highly productive agriculture and highly productive herding (e.g., the circum-lakeshore region of the Titicaca Basin). Miller and Burger (1995) have found an increased importance of camelid meat in the diet of Middle Formative kichwa agriculturalists in north-central Peru -- suggestive of intensified interaction between puna herders and kichwa cultivators.

The Early and Middle Formative florescence in northern Peru and southern Ecuador occurs where the spacing between coast, sierra, and montaña zones is narrowest, and where the intervening mountain passes are lowest. Such settings may have been the best places for inter-zonal interaction between chiefly elites seeking prestige-building "foreign" exotica, but mainly dependent for their access to such exotica upon spatially restricted, kin-based networks. The Middle Formative inhabitants of the Tarama-Chinchaycocha Region were apparently only modestly involved in such inter-zonal networking (e.g., Matos 1971).


Small States during the Late Formative/Early Intermediate Period.

Late in the first millennium B.C. there is archaeological evidence that more complex polities were developing. These developments are most clearly apparent for the Mochica culture on the Peruvian north coast (e.g., Brennan 1980; Donnan 1976; Hocquenghem 1987; Shimada 1994; Uceda and Mujica 1994; Wilson 1983, 1987, 1988, 1995), for the Lima culture on the Peruvian central coast (e.g., Earle 1972), for the Nasca culture on the Peruvian south-central coast (e.g., Silverman 1992, 1993a, 1993b, 1996), for the early stages of Wari's development in the Peruvian south-central highlands (Isbell 1985, 1987; Lumbreras 1974b), and for the early stages of Tiwanaku's development in the Titicaca Basin (Albarracin-Jordan 1996; Albarracin-Jordan and Mathews 1990; Berman 1994; K. Chavez 1988; S. Chavez 1988; Kolata 1993; Stanish et al. 1997).

All these developments were characterized by substantial population growth, impressive monumental public architecture, signs of agricultural intensification, and formal ritual iconography. There is a clear and pervasive stylistic linkage between decorative motifs and vessel forms employed in coastal Nasca and highland Wari ceramics (Knobloch 1991; Menzel 1964; Paulson 1983; Silverman 1988b). This hints at new forms of coastal-highland interaction in south-central Peru that may have included inter-zonal exchange based on something (e.g., tribute) other than seeking out of "foreign" exotica through networks of kin-based reciprocity for purposes of chiefly prestige-building (Spencer [1982] develops an analogous hypothesis for the early development of the Monte Alban state in southern Mexico). The growing importance of domestic camelid meat in the diet of north-coastal Mochica populations may be another sign of new forms of coast-highland interaction (S. Pozorski 1979).

Just as in the antecedent Early and Middle Formative, these Late Formative/EIP developments were distinctly uneven over space. Peruvian coastal valleys south of Nasca remained sparsely occupied, without major sites or monumental public architecture, as did many highland valleys and virtually all the puna except for the circum-lake portion of the Titicaca Basin. Just as in the Early and Middle Formative, coastal polities appear to have been notably more complex than their highland contemporaries in terms of the monumentality of their public architecture, the size of their settlements and regional populations, and overall agricultural productivity founded in increasingly large-scale canal irrigation (e.g., Brennan 1980; Moseley and Deeds 1982). With the possible exception of the Nasca-Wari linkage noted above, the influence of most of these coastal EIP states apparently did not extend significantly outside their coastal heartlands (e.g., Dillehay 1976, 1979; Schadel 1985; J. Topic and T. Topic 1983).

Archaeologists are not yet in a position to resolve the question of why some areas were in the "mainstream" of earliest Andean state development while others were not. The difficulties are compounded by the fact that the EIP is approximately 1000 years long -- a great span of time that cannot always be subdivided into chronological phases short enough to perceive key developmental processes. Additional problems arise in terms of conceptualizing such overall variability: e.g., if states developed in some areas and not in others, how were changes in the developmentally peripheral areas affected by the presence of new forms of political and economic organization in the core regions? These issues have begun to be addressed in central (Dillehay 1976, 1979) and northern Peru (e.g., Onuki 1985; Schadel 1985; Shimada 1985; T. Topic and T. Topic 1983), but the archaeological data are simply not yet up to the task of providing good answers.


Large States in the Middle Horizon.

The predominance of the larger Peruvian coastal valleys as the main foci of Andean chiefdom and state development ended definitively during the Middle Horizon. It was then that two highland centers, Wari and Tiwanaku, emerged as the focal points of large polities that extended their influence across major coastal and highland zones in ways that earlier, coastal-centered states apparently had not. Marcahuamachuco in the Peruvian north highlands was another major MH center, whose regional influence is still less well understood than either Wari's or Tiwanaku's (J. Topic and T. Topic 1986; T. Topic and J. Topic 1987). The now-prevailing view is that during the Middle Horizon Wari and Tiwanaku were centers of proto-imperial polities whose expansive character and hierarchical organization foreshadowed the better-known Inka (e.g., Browman 1985; Feldman 1989; Goldstein 1987; Isbell and Mcewan 1991; Kolata 1993; Moseley et al. 1991; Raymond 1992; Schreiber 1987).

These scholars, and others, have marshaled much evidence in support of the proto-imperial model. This includes the widespread distribution of distinctive ceramic and architectural styles: e.g., the Wari-like architectonic configurations in far-flung highland sites like Pikillaqta in the south (Mcewan 1987, 1989, 1991, 1996; Sanders 1973), Jincamocco in the south-central region (Schreiber 1992), and Viracochapampa in the far north (J. Topic 1991; T. Topic 1991) ([ref. to map showing these sites]). Several other sites (e.g., Wari Willka in the Mantaro Valley near Huancayo [Matos 1968; Shea 1969] and Honcopampa in Peru's north-central highlands [Isbell 1989]) also have strong ceramic and architectural linkages to Wari, suggestive of different forms of "influence" that might include the expansion of religious ideology (Keatinge 1981). In other cases, far-flung Wari contact is expressed primarily in the form of decorated ceramics (e.g., Thatcher 1974, 1977; Topic and Topic 1986).

Shady (1988, 1989) has challenged this proto-imperial model, arguing that Wari (and, by extension, Tiwanaku as well) was but one important component in the context of numerous large, interactive Middle Horizon state polities. Shady's position is appealing in light of the substantial Peruvian central- and north-coastal Middle Horizon polities that appear to have developed and flourished in the absence of any significant influence from Wari (Bawden 1982, 1983; Shady 1982; Shimada 1978, 1981, 1985, 1994; Shimada and Cavallaro 1985; Willey 1953; Wilson 1988). Further support of this view comes from indications of the apparently autonomous development from local EIP antecedents of the major Middle Horizon center at Marcahuamachuco, in the Peruvian north highlands, with little or no direct ceramic or architectural influence from Wari (J. Topic and T. Topic 1985).

The nature of the largest Middle Horizon states remains to be adequately comprehended. Nevertheless, there seems little doubt that there was a major difference between the smaller, spatially more restricted EIP polities, on the one hand, and the larger, spatially and ecologically more extended Middle Horizon systems, on the other. For the first time the two largest regional centers -- Wari and Tiwanaku -- were situated in the highlands of the southern half of the Central Andes, a region that had previously been distinctly secondary in pan-Andean developmental terms.

From this point onward in time, the north Peruvian highlands, a region that had seen the development of some major Formative and EIP centers, was increasingly secondary in terms of the growth and expansion of the largest polities -- the latter were invariably centered farther south, in or near the heartland of Central Andean camelid pastoralism. Domestic camelids existed in the Peruvian north highlands, and even into Ecuador (Norton and Stahl 1987; Stahl 1988), but this northern region has always been peripheral to the core zone of Central Andean camelid pastoralism in the central and southern highlands (Troll 1958, 1968).

A glance at Figure 2.02 reveals that both Wari (the largest known Middle Horizon center in Peru) and Cuzco (the Inka capital during the Late Horizon) are at the junctures of major herding and agricultural zones. This suggests that the integration of large-scale agricultural and herding economies lay at the foundation of both centers' dominance. As noted above, Tiwanaku's location is uniquely strategic in terms of the close proximity of unusually productive lakeshore agriculture and camelid herding; there is also easy accessibility to ecologically complementary zones to the west and east of the Titicaca Basin, in a setting where regional socio-economic integration would have been facilitated by water-borne transportation and communication (Parsons 1968).

During the Middle Horizon, for the first time in Andean prehistory, the full integration of more specialized agricultural and herding economies was apparently beginning to be important in some particularly favored and strategic highland regions of central and southern Peru. Arguably, only these regions would have been capable of attaining the higher levels of overall productivity necessary for underwriting the costs of increased socio-political centralization.

From the Middle Horizon onward, coastal polities appear to have become increasingly peripheral to the "mainstream" of Central Andean socio-political development. This may be another indication of the growing importance of fully integrated specialized agricultural and herding economies. Some level of local camelid breeding and herding may have developed in later prehispanic periods on the Peruvian coasts (e.g., Shimada and Shimada 1985). However, the coastal valleys have always been peripheral to camelid pastoralism simply because of environmental factors -- e.g., harmful bacteria thrive in the warm climate, and coastal vegetation does not include the plants necessary for proper camelid diet.

The Middle Horizon was clearly a time of radical cultural change. We still understand little of what this change actually entailed or of how it was achieved. There may have been significant ideological changes accompanying the implementation of new forms of socio-political hierarchy, new forms of surplus generation and tribute extraction, and new forms of supra-household and supra-community and inter-regional interaction. Hints of such change include the widespread distribution of Wari- and Tiwanaku-linked architectural and ceramic forms that may have been associated with public ritual, analogous perhaps to the Inka emphasis on ritual feasting in their provincial centers (e.g., Morris 1982). Archaeological evidence for ritual feasting at administrative centers occurs at least as early as late EIP times in Peru's north-central highlands (Gero 1986, 1990).

J. Topic and T. Topic (1986) and T. Topic (1991) have suggested that Marcahuamachuco in Peru's north highlands became a major regional center during the Middle Horizon because of its new role in mediating coast-highland exchanges and interaction at a time of increased regional and inter-regional tensions. The radical architectonic innovations at Marcahuamachuco may reflect new forms of ritual and settlement pattern associated with changing forms of inter-regional linkage. There is a definite Wari "influence" in the Marcahuamachuco region, but this appears to be more related to considerations of local prestige building than of Wari dominance. The Topics' model may have some applicability as well to the changed roles of Wari and Tiwanaku as these larger centers emerged as new regional and inter-regional centers during the Middle Horizon. As noted earlier, Wari probably began to be important as a node in sierra-coastal interaction during the later EIP.

We have noted that distinctly Wari-related architecture and ceramics are not found everywhere, even in some areas relatively close to Wari itself. Furthermore, much of the existing archaeological evidence for Wari influence comes from excavated data, that our surface survey cannot provide. We must thus be prepared to think about other potential archaeological indicators (e.g., shifts in regional settlement patterns) that have only seldom been considered by others in highland contexts (Albarracin-Jordan 1996; Schreiber 1987; Stanish et al. 1997).


State Collapse and Rebuilding during the Late Intermediate Period.

This period is usually seen as a time of socio-political instability, de-centralization, and intense warfare in the aftermath of the collapse of the large Middle Horizon polities centered at Wari and Tiwanaku (e.g., Parsons and Hastings 1988). The causes and processes of this collapse remain obscure, although scenarios emphasizing military conflict have been proposed (e.g., Amat 1978). The main known exception to this pattern of center collapse and regional instability is the expansive LIP Chimu polity centered at Chan Chan on the north Peruvian coast (e.g., Moseley and Day 1982; Moseley and Cordy-Collins 1990; Rowe 1948).

Somewhat surprisingly, several regional surveys on Peru's north coast (Keatinge 1982; Willey 1953; Wilson 1988, 1995) have revealed a consistent pattern of population decline after the Middle Horizon. On the other hand, less systematic reconnaissance in the far north coastal Lambayeque region (Shimada 1981) suggests substantial LIP population growth. Highland surveys (Albarracin-Jordan 1996; Albaracin-Jordan and Mathews 1990; Earle et al. 1980; D. Julien 1994; Krzanowski 1985; Meddens 1984; Parsons and Hastings 1977; Stanish et al. 1997; Valdez and Vivanco 1994) consistently indicate that the LIP was a period of significant population growth in the sierra.

[check w/Brian Bauer re Cuzco LIP patterns]

Perhaps more than for any other period in Central Andean prehistory, the LIP is a time for which archaeologists have their greatest difficulty in conceptualizing socio-political forms. By the beginning of the LIP at ca. A.D. 1000, state polities had existed for approximately a millennium. There were probably few areas between western Bolivia and southern Ecuador that had not been significantly affected by state polities in varying stages of development, florescence, or collapse. How do we comprehend the societies that remained in the wake of the collapse of large states following a millennium-long period of state presence? Such discombobulated systems may no longer be centralized states, but the standard "chiefdom" model -- with its emphasis on kinship-based authority, generalized administration, and limited secular power -- also seems inadequate, although some highland LIP polities have been labeled as chiefdoms (e.g., Costin 1986; D'Altroy 1992; Earle et al. 1980, 1987; Hastorf 1993; Hastorf et al. 1989).

The complicated and confusing socio-political landscape of nineteenth century West Africa (e.g., Forde and Kaberry 1967) may offer a better analogy, with its ever-shifting alignments and re-alignments of agriculturalists, herders, artisans, traders, war-lords, and powerful outsiders seeking to extend and consolidate their influence from relatively stable core regions to unstable peripheries and interstitial zones where they found themselves in competition with other forces structurally comparable to themselves. This West African scene, like highland Peru during the LIP, defies easy typological classification, but it is probably generally comparable to what archaeologists must deal with in the Central Andes during the three or four centuries after ca. A.D. 1000.

Although the early stages of the development of the imperial Inka capital at Cuzco remain unclear, there are now good indications that the Cuzco Region was precocious in terms of population growth and settlement nucleation during the LIP (Bauer 1991, 1992b; Dwyer 1971; Kendall 1976; Parsons and Hastings 1988). Furthermore, a recent synthesis of several dozen radiocarbon dates (D'Altroy, Williams, and Bauer 1998) suggests that some far-flung Inka-related developments, previously regarded as chronologically Late Horizon, are actually of Late Intermediate Period age.

The LIP was clearly a time of significant change. Part of this change may have been related to the fracturing of long-established regional and inter-regional linkages following the collapse of Wari and Tiwanaku (and perhaps of other major centers as well) at the end of the Middle Horizon. The development of new organizational forms that grew out of these fractured networks ultimately gave rise to the Inka empire centered at Cuzco. Earle et al. (1980, 1987) have demonstrated major organizational differences between earlier and later phases of the LIP in the Wanka Region in Peru's central highlands; such a developmental dichotomy was probably widespread. The all-too-common difficulty of being unable to subdivide the 400-year long LIP often precludes us from defining the processes of change within this period.

The fact that the Late Horizon pan-Andean imperial capital emerged at Cuzco, situated at the juncture of major kichwa and puna zones, and not at Chan Chan or elsewhere in the Chimu domain on the far Peruvian north coast, suggests that the integration of agriculture and camelid herding continued to play a key developmental role during the LIP. The apparent failure of the coastal Chimu polity to expand significantly into the adjacent north Peruvian highlands suggests that major LIP coastal polities continued to be economically based on large-scale canal irrigation -- an adaptation that apparently did not prove to be fully "competitive" in terms of Late Horizon pan-Andean geo-politics.


The Inka Empire during the Late Horizon.

Understanding of the Inka empire has come largely from documentary sources compiled during the sixteenth and seventeenth centuries (e.g., C. Julien 1983, 1988, 1991; D. LaLone 1982; M. LaLone and D. LaLone 1987; Murra 1958, 1980; Parssinen 1992; Rowe 1946; Wachtel 1973; Zuidema 1964; Zuidema and Poole 1982). Only since the late 1960s has archaeology made a significant contribution (e.g., Bauer 1991, 1992a, 1992b; D'Altroy 1992; Hyslop 1984, 1985, 1990; Morris 1967; Morris and Thompson 1970, 1985; Morris and Von Hagen 1993). The Inka imperial structure continues to be seen as expansive, multi-tiered, and highly bureaucratized, with a strong component of secular force and a command economy that included massive terracing and irrigation systems and craft workshops in many strategic localities (e.g., M. LaLone and D. LaLone 1987; Levine 1987). A network of formal roads, provincial centers, storage facilities, and way-stations (tambos) provided an extensive transportation-communication infrastructure (Hyslop 1984, 1985, 1990; LePaige 1993; Lynch 1996; Stehberg and Carvajal 1986).

Nevertheless, we now know that there was considerable variation within the empire in terms of local administration and linkage to Cuzco (e.g., Barcena 1992; Berberián 1991; D'Altroy 1992; Gonzales 1983; Hayashida 1995; D. Julien 1994; Levine 1985, 1982; Lorandi 1983; Lorandi and Cremonte 1991; Mallpass (ed.) 1993; Morris and Thompson 1985; Niemeyer 1986; Niemeyer and Rivera 1983; Plaza 1976; Raffino 1982, 1993; Salomon 1986a, 1986b, 1987; Schjellerup 1984; Stehberg 1992; Williams 1994; Williams and Lorandi 1986). The empire is now seen as far less monolithic than was formerly believed, and there is a growing awareness that the Inka imperial effort (including the organization of labor for agriculture, herding, craft production, and infrastructure maintenance) was constrained and shaped by the varied character of local polity, economy, and population size and density. Although Inka "conquest" could produce major transformations of local settlement and production (e.g., Hastorf 1990; Hastorf and Johannessen 1993), in other cases there appear to have been strong continuities with the pre-Inka past, especially in terms of the operation of local-level (as opposed to imperial-level) economies (e.g., Conrad 1977; Menzel 1959; Parsons 1998).


The Legacies of Historic Colonialism and Capitalism

In Chapter 2 we outlined the specific impacts of historic-period transformations in the Peruvian central highlands. These included the disastrous population declines resulting from the deadly combination of introduced diseases and onerous forced labor in mines and textile workshops during the sixteenth and seventeenth centuries. By the later sixteenth century Spanish policies of encomienda (Indian labor service to Spanish and Creole landlords) and reducción (population resettlement) had begun to be fully implemented, resulting in the abandonment of many older settlements and the formation of new villages and towns modeled on Spanish concepts of urban planning.

This was paralleled by a strong dedication on the part of Spanish authorities to the extirpation of indigenous religious ideology. By the time of independence from Spain in the 1820s, the impact of world systems capitalism began to be strongly felt in highland Peru, especially in mining and wool production. Remaining communally controlled resources, including agricultural lands and pastures, were increasingly alienated from indigenous people, many of whom became landless peasants and laborers. Even Indians who retained some control over productive resources became increasingly involved in the production of wool (both sheep and alpaca) and foodstuffs for sale in local or regional commercial markets, including those in new mining centers.


Summary and Conclusions

Our overview of some 4000 years of Central Andean cultural development indicates six main developmental stages:

(1) The growth of simple chiefdoms during the Early Formative (ca. 2000-900 B.C.), with florescent developments in the deltas of major valleys along Peru's central and northern coasts.

(2) The appearance of larger, more complex chiefdoms during the Middle Formative (ca. 900-400 B.C.). Although the lower valleys of Peru's central and north coasts remained at the core of this development, some parts of the adjacent highlands also participated, and similar societies began to develop in the unusually productive and strategically situated Titicaca Basin in the southern highlands. This development was characterized by the construction of formal public buildings, formal ritual iconography, population growth and agricultural intensification through canal irrigation in the coastal valleys. Inter-zonal and inter-regional exchange expanded and intensified. Some coastal and highland areas remained peripheral, particularly in the southern half of the Central Andes.

(3) The appearance of small states during the Late Formative/Early Intermediate Period (ca. 400 B.C.-A.D. 600). The best known and most florescent of these polities (Mochica, Lima, and Nasca) were centered in the major valleys of Peru's northern, central, and south-central coasts, and in the Titicaca Basin. Nonetheless, some highland areas in central and northern Peru also witnessed more rapid change, particularly in the Wari region of the south-central highlands where a strong stylistic linkage with decorated coastal Nasca ceramics may indicate new forms of inter-zonal interaction. The coastal state heartlands witnessed rapid population growth and a major expansion of canal irrigation systems. The first presence of state-like polities in highland settings (Wari and Tiwanaku) may indicate the beginning of new forms of interaction between more specialized herding and agricultural groups.

(4) The development of large, expansive states during the Middle Horizon (ca. A.D. 600-1000), with the largest regional centers (Wari and Tiwanaku) situated in the southern half of the Central Andean highlands. From this point on northern Peru became increasingly secondary -- even the LIP north-coastal Chimu polity had only modest impact in the adjacent sierra. These geo-political shifts suggest the full integration of more specialized puna herding and kichwa agricultural economies, and the emergence of strategic highland regions as the loci of dominant states or proto-imperial systems. It is during the Middle Horizon when we might expect to find the first effective mechanisms for maintaining order and security that did not depend on kin-based alliances.

(5) An era of state collapse and rebuilding, comprising the Late Intermediate Period and the Late Horizon (ca. A.D. 1000-1532), culminating in the Inka empire. Inka state building was deeply rooted in the LIP, and archaeologists are just beginning to understand more about this development. The Inka empire itself is now seen as internally varied, with imperial administration and organization of production constrained by the highly varied nature of local population, polity, and economy. The location of the Inka capital, Cuzco, at the kichwa-puna juncture indicates the continued importance of fully integrated, large-scale agricultural and herding economies.

(6) The massive transformation of the Colonial and Modern periods after A.D. 1532 in which involuntary labor obligations and resettlement were compounded by the great loss of native population resulting from introduced diseases and overwork. The impact of capitalistic world systems after the early nineteenth century produced further dislocations of indigenous people from resources as new forms of land tenure and commercialized production of mineral ores, foodstuffs, and wool became dominant.


The Cultural Ecology of Historic Central Andean Herders and Cultivators

We are interested in the links between pastoral and agricultural economies. We particularly want to discern the expectable relationships between the size, composition, location, and boundary definition of interactive units of herders and cultivators. Most of the relevant ethnographic and ethnohistoric studies have been carried out in places where the impact of modern Peruvian and Bolivian states is generally limited, and where local production is not strongly integrated into the commercial economy. Consequently, these studies may be most applicable to prehispanic contexts where strong centralized organization was weak or absent. Our discussion may thus be least applicable to the Late Horizon, when our study area appears to have been fully incorporated into the Inka empire, and perhaps also to the Middle Horizon, when Wari may have exercised some degree of supra-regional control or hegemony.

Nevertheless, virtually all of these historic-period studies derive from contexts where both a state and an over-arching commercial economy do exist, however weak or under-developed. State institutions and commercial markets do have some significant impact on the behavior of individuals, even in more remote communities: children often attend state schools; births, marriages, and deaths are registered by bureaucrats employed by municipal, departmental, or national agencies to whom taxes (including military service) are sometimes paid; laws prohibiting theft and violence and other state-defined transgressions are typically well known and sometimes enforced by outside authorities; outside buyers often purchase crops, wool, and handicrafts; access to land is usually structured by externally imposed rules concerned with ownership and transference of property; money minted and printed by the state circulates, and every household needs some cash to purchase a few necessities; opportunities for wage employment lure young people away to work in distant places; most rural Indians find themselves in the lowest stratum of an imposed socio-economic hierarchy that severely constrains upward mobility; and many local officials are partly empowered and legitimized through their roles as functionaries at the bottom rungs of state-imposed administrative structures. In most cases these external forces have been in play for centuries.

I.e., the ethnographic and historic data we look to for guidance do not come from tribal or chiefdom societies where supra-household behavior is structured in large part by competitive prestige-building activities of Big Men and Chiefs operating within localized networks of shifting kinship alliances. Kinship alliances do exist in the historic period, of course, as do non-commercial production and redistribution, and sometimes these are locally of primary importance. However, any attempt to understand historic-period cultural ecology at the local level must remember that individuals, households, settlements, and communities are not un-modified remnants of prehispanic societies, nor do they behave according to purely local, or even purely regional, constraints and possibilities. From this perspective, our insights from historic-period data are probably least applicable to the Early and Middle Formative, a time before the existence of state organization anywhere in the Central Andes.

Since the later sixteenth century, cattle and sheep have increasingly replaced domestic camelids in much of the Central Andes, including the Titicaca Basin and some of the most desirable pastures in southern Peru, particularly at elevations below 4000 m asl (Flores 1980). Similarly, Orlove (1977a) has noted the major increase of commercialized sheep and alpaca production over the past 150 years -- with llamas relegated to a secondary role in most places. In some cases, such herd expansion has encroached upon prime agricultural lands. Even in remote valleys in Peru's southern highlands the sale for cash of alpaca wool is sometimes the dominant factor determining local settlement patterning (Webster 1971, 1973).

As we noted in Chapter 2, these changes have accompanied the general reduction of local community access to the most productive agricultural and herding lands in the face of expanding private land-holdings by Spanish, creole, mestizo, and foreign elites seeking to generate wealth from certain types of agriculture and from wool and mining production. This process began in the later sixteenth century as Spanish policies of resettlement and concentration (reducción) of Indian populations got fully underway (Gade and Escobar 1982; Molinié-Fioravanti 1986), and it accelerated after independence from Spain in the early nineteenth century when western industrial capitalism extended its reach into the Central Andes (Orlove 1977a). This has had the effect of increasing competition between local Indian communities over remaining fields and pastures. Yamamoto (1981:95), for example, describes the periodic ritualized checking and marking of community territorial borders -- occasions that sometimes produce violent confrontations between neighboring groups that accuse each other of land encroachment.

Because many of the best agricultural and herding lands have long ago been overtaken by commercial operations of little relevance for our needs, we lack good ethnographic analogies for the prehispanic use of these most productive sectors of the Central Andean highland economy.


The degree to which agriculture and herding are combined within household and local community economies varies a great deal: some herders practice no cultivation at all, while others are committed about equally to both; in other cases herding is dominant while cultivation is secondary; at the other end of the spectrum are predominantly agricultural households with only a few domestic animals.

Not surprisingly, people who live permanently above ca. 4200 m asl do little or no cultivation, especially if they live relatively far away (more than a few hours travel) from productive agricultural land. Households in the lower puna and upper kichwa are likely to combine some mixture of herding and agriculture, particularly if they are not too far away from both pasture and agricultural land of reasonable quality.

Some modern households and communities in the lower puna may be differentially involved in herding and agriculture because of historic factors. For example, in their discussion of the settlement nucleation caused by colonial reducción policies, Gade and Escobar (1983:438) note that "mindful that pastoralism would undermine the efforts to control and acculturate the natives, the early colonial administrators tried to ensure that farming would be the primary mode of livelihood in the reducciones." Moreover, as had been accomplished a few years earlier in colonial Mexico, the sixteenth century Spaniards in Peru wanted to monopolize commercial livestock production, and so Indian villagers were further discouraged from the pursuit of herding divorced from agriculture. Although many reducción villages had begun to break apart by the end of colonial rule, nevertheless some degree of modern agro-pastoralism may be a legacy of these colonial policies.

Gade and Escobar (1982:441-442) estimate that in the Peruvian southern highlands a typical nuclear-family agro-pastoral household, about equally dependent on both agriculture and herding and working with simple hand technology, can cultivate a maximum of approximately six hectares of land (a considerable portion of which is fallow at any given time) while maintaining a mixed flock of approximately 25 sheep, 4 llamas, 3 cows, and a horse; actual numbers of animals and size of cultivated plots vary widely. For inhabitants of the lower puna who are primarily sheep-llama herders, but who also have a secondary involvement in potato cultivation, Flannery et al. (1989:104-105) find that household llama flocks of 25 animals are about average, while flocks of 15 animals are considered "small," and flocks of 35-40 animals are "excellent" (the numbers of sheep are unspecified). [see if there is any comparable data on households who are primarily agriculturalists but with secondary herding; also check Goland).

Yamamoto (1985:91, 1988:144) makes the important point that where agriculture and herding are combined today, the traditional agro-pastoral economy is almost always based on camelids and tubers. Cereal cultivation, especially in the case of maize, is much less complementary with herding because of maize's needs for fertile soil, warmer climate, and a labor intensity that conflicts overmuch with herding activities.

A fairly typical situation for agro-pastoral communities in Peru's central and southern highlands is described by Fonseca and Mayer (1988). Individual households in each of these communities directly control agricultural fields and pastures situated between 2000-4000 m asl; these fields and pastures supply most of the plant and animal products consumed by each household. At the same time, higher terrain above 4000 m asl in the same general region is occupied by dispersed hamlets of full-time herders, who themselves do no agriculture, and who exchange their wool, textiles, dried meat, dung, and other animal products for potatoes, maize, a variety of vegetables, coca, and craft products produced or acquired by their agro-pastoral neighbors at lower elevations (Mayer 1971).

Several writers (e.g., Camino 1980; Camino et al. 1981; Fujii and Tomoeda 1981; McCorkle 1987; Orlove 1977b; Webster 1971, 1973; and Yamamoto 1981, 1985) describe agro-pastoral societies in which a community's major nucleated settlement is strategically placed at an intermediate elevation (typically between 3100 and 3700 m asl), while many people spend much of their time in smaller hamlets and camps at higher and lower elevations (between 2500-4500 m asl) engaged in different kinds of herding and agricultural tasks. At any given time, segments of the entire community population (several hundred people grouped in scores of households) can be found widely dispersed throughout the entire community territory. In some cases the centrally located major settlement is used only for occasional residence by people cultivating crops in the vicinity or, on a few occasions for a few days each year, by the entire community for major ceremonies; each member household maintains its own house in the main settlement. Murra (1972) describes a similar mid-sixteenth century community system in the north-central highlands of Peru.

In another variant of modern Central Andean agro-pastoralism, most members of a community spend virtually all their time at agricultural tasks in their kichwa settlements, while a few households (or even hired specialists) care full-time for all the community's flocks that are herded together in neighboring puna pastures (e.g., Flores 1985:262; Fonseca and Mayer 1988; Guillet 1987).

More complex interactions between larger groups of socio-economically integrated herders and agriculturalists are described by Bastien (1978a, 1978b), Harris (1982, 1985), and Platt (1982, 1986) in the highlands of western Bolivia. Here, specialized kichwa agriculturalists and puna pastoralists reside in separate settlements at different elevation zones between ca. 2500 m asl and 4500 m asl. These settlements are integrated within multi-settlement "ethnic" or "ayllu" groupings whose territories extend over the full elevation range. The distances between herder and agricultural settlements are considerable, up to several dozen kilometers. Complementary agricultural and herding products move between these settlements through an intricate web of inter-household linkages. These linkages are un-centralized, maintained through well defined rules of zonal exogamy and ritual focused on inter-zonal integration.

These examples illustrate the great diversity of modern agro-pastoral adaptations in the Central Andes.


The Complementarity of Herding and Cultivating Economies

The degree to which agriculture and herding were combined or separated in prehispanic times is a key question for archaeologists. Because the Central Andes lies within tropical latitudes, ecological zones succeed each other rapidly with change in elevation (Rhoades and Thompson 1975). Unlike highland zones in temperate regions, there is often little or no spatial separation between herding and agricultural zones, and these zones often overlap and interdigitate. Central Andean herders can use most pastures during the entire annual cycle, unlike herders at temperate latitudes for whom many pastures are seasonally unavailable because of cold or aridity (e.g, Barth 1961; Vincze 1980). Consequently, Central Andean herders typically have much less need than their temperate-latitude counterparts for long-distance transhumant or nomadic mobility (Inamura 1986).

As we have just noted, some local communities and even individual households effectively combine herding and cultivation. Nevertheless, spatial separation between more specialized herders, on the one hand, and cultivators, on the other, is well documented. No camelid pastoralists are known to be involved in the production of grain or grass for their animals - unlike European pastoralists who must store up fodder for the winter season (McCorkle 1987; Vincze 1980:392).

Access to both herding products and cultivated plants is essential for highland populations (e.g., Flores 1979; Harris 1982, 1985; Nachtigall 1975; Thomas 1976; Tomoeda 1985; Winterhalder and Thomas 1978; Yamamoto 1985). Without herding, the entire puna zone would remain marginalized and peripheral. Without access to dependable supplies of tubers and cereals produced in the kichwa, the diets of most puna herders would remain calorically and nutritionally incomplete (Table 4.01), especially for herders who do not consume the milk or blood of their living animals (Yamamoto 1988:144). Furthermore, puna herders have historically relied on kichwa-produced maize for the chicha (maize beer) and maize-paste figurines commonly used in their rituals (e.g., Tomoeda 1985:295; Yamamoto 1985:94).





Table 4.01. Common foods for puna camelid pastoralists in southern Peru,

(after Flores 1979:42)

Food (most are soups)



boiled water with ulluco, isaño, and chuño



ground chuño with water, salt, and a little meat


wheat and barley ground and mixed with alpaca blood


flour and water, plus eggs and milk, boiled in oil

yawal longani

blood sausage, plus eggs and milk, boiled in oil


soup made by boiling alpaca feet and lower legs in water with toasted maize


roast camelid or sheep meat


ground kañiwa, water, lime, and salt -- molded into cakes for use in trading expeditions


toasted wheat or maize flour


Conversely, without access to the wool, textiles, pack-animal transport, meat, skins, and dung (for fertilizer and fuel) produced in great quantities only by puna herders, kichwa cultivators would have lacked adequate supplies of some of their most basic raw materials for tools and clothing, and would have found their diets much less varied and appealing. They would also have been hard pressed to transport their harvested crops from field to storehouse, or to move heavy loads of animal dung for fertilizer at planting time. In the ritual realm, kichwa agriculturalists commonly make use of aborted camelid fetuses and live animals as offerings and sacrifices in ceremonies (Bastien 1978a:37).

The overall carrying capacity of the Central Andean highlands increases in direct proportion to the degree to which kichwa agriculture and puna herding are fully integrated and complementary. Nevertheless, we should remember the findings of Mazess and Baker (1964) who found that only 2.5-3.5 percent of the diet of an isolated puna group in southern Peru consisted of food not produced above 4000 m asl. Similarly, Yamamoto (1985:94) notes that for one lower puna community in southern Peru, potatoes comprise ca. 80 percent of the diet.


Herding Ecology

The Central Andean puna provides large expanses of good pasture but is only marginally suitable for plant cultivation (especially at elevations above ca. 4200 m asl); conversely, large numbers of camelids are rare in the intensively cultivated kichwa valleys below ca. 3700 m asl (Troll 1958, 1968; Winterhalder and Thomas 1978). Generally speaking, where puna and kichwa zones interdigitate closely and local vertical relief is relatively high, there is a greater tendency for herding and cultivating to be fully combined at the household and local settlement levels (e.g., Brush 1976; Fujii and Tomoeda 1981; Gade 1975; Inamura 1986; Orlove 1977b; Valle 1970; Webster 1971, 1973; Yamamoto 1981, 1985). Such an ecological configuration can occur anywhere, but it is most predominant in the northern third of the Central Andean highlands (Figure 2.02) -- an area where herding and agriculture may always have been integrated at the local settlement level (McGreevy 1989).

Most reports of modern kichwa agriculturalists who also keep domestic herbivores indicate that a high proportion of these animals are introduced species -- donkeys, horses, cattle, goats, and (possibly to a lesser extent) sheep (Custred 1977b:77). These species may be more compatible with a relatively sedentary mode of life, and perhaps less dependent than domestic camelids on puna plants for their dietary needs -- i.e., they supplement, rather than complement, agricultural production.

Where the puna and kichwa zones are separated by greater distances, and where local vertical relief is relatively low, herders in the puna and cultivators in the kichwa are more likely to be specialized and separated (e.g., Browman 1974; Flannery et al. 1989; Flores 1979; Fuji and Tomoeda 1981; Harris 1985; Platt 1982; Yamamoto 1985). Such ecological configurations are most predominant in southern Peru and adjacent Bolivia (Figure 2.02). Such specialization does not have to be complete. For example, some herders in the lower puna may devote most of their attention to caring for their animals, but also raise some hardy varieties of bitter potatoes as a secondary supplement to their household economy (e.g., Flannery et al. 1989:50-51). Part of this zonal specialization has to do with the high labor demands of agricultural and herding tasks that often conflict with each other in terms of both scheduling and the adequacy of local labor pools (McCorkle 1987).

Of special interest to us is the degree to which inhabitants of the lower puna and upper kichwa combine herding and agriculture and function as intermediaries between more fully specialized herders in the upper puna and more fully specialized agriculturalists in the lower kichwa (e.g., Flores 1985; Inamura 1981, 1986; Tomoeda 1985).

Caring for Llamas vs. Alpacas. Recent ecological, skeletal, fiber, and DNA studies indicate that the two species of modern domestic camelids (llamas and alpacas) are descended from two similar species of prehispanic animals (Stanley et al. 1992; Wheeler 1995; Wheeler et al. 1995). These studies suggest that the modern domestic camelids are the products of centuries of hybridization between llamas and alpacas in the face of the post-contact breakdown of the more exacting breeding practices that had existed in late prehispanic times.

Today alpacas are often much more numerous than llamas. This is because alpaca wool has such high commercial value, while coarser llama wool has very little. Consequently, llamas are used mainly as pack animals, and as a limited source of wool, meat, and hides for household use and inter-household and inter-community exchange. Casaverde's (1977:174) 1970 census in a highland district of southern Peru showed totals of 10,395 alpacas, 3,580 llamas, and 3,886 sheep (58 percent, 20 percent, and 22 percent, respectively). Similarly, Inamura (1986:149) found that in another puna locality that 70-80 percent of domestic animals were female alpacas, while the remaining 20-30 percent consisted of male alpacas, llamas (of both sexes), and sheep. Inamura (1986:151) has also noted the extent to which pastures reserved for alpacas have been improved through canal irrigation, while llamas are restricted to unimproved grazing land. This emphasis on alpacas is probably a product of modern commerce, and seems unlikely for pre-modern or prehispanic herders. Flannery et al. (1989), for example, describe isolated herders in the Peruvian south-central highlands who combine llamas and sheep, with no alpacas.

A more detailed breakdown of herd composition is shown in Table 4.02. Although these latter data derive from the arid South Andean puna, the high proportion of adult female animals and the low numbers of sexed adult males stand out consistently.

Although they are closely related biologically, llamas and alpacas have some very different needs and require somewhat different management. Relative to alpacas, llamas are larger, stronger, and hardier, with more flexible dietary needs, a greater capacity to cope with predators, and a capacity for foraging and carrying burdens at elevations between sea level and 5000 m asl. Although llama flocks are routinely left unattended and un-corralled for considerable periods of time in northern Chile and northwestern Argentina (Gundermann 1988:107), the common modern practice in the Central Andes, for both llamas and alpacas, is to corral them every night (e.g., Flannery et al. 1989). Night-time corralling of camelids is done for three primary reasons: (1) to provide protection from the cold and wet; (2) to protect flocks from their natural predators (mainly mountain lions, foxes, and condors); and (3) to discourage human rustlers.

McCorkle (1987:71) reports that camelids are more capable than sheep of caring for themselves because they generally "will not stray far, will keep together in a group, and will often return home at night of their own volition ... and [they possess] a strong dominance hierarchy and a tendency to recognize only one or two human masters, and the existence of a [camelid] band leader, and [consequently] they are more difficult to rustle." Nonetheless, the consequences for inadequate care of camelid flocks can be very severe. Palacios (1977b:61), for example, notes that prolonged droughts and serious snowstorms can produce losses of 50-60 percent.

Alpacas are much more delicate and demanding in terms of care and diet, and they are never used to carry burdens. They are more susceptible than llamas to diseases and infections, especially new-born and very young animals. Flores (1979:90, 95), for example, reports that young alpacas are seriously affected by mites and diarrhea, with a mortality rate of 50-90 percent during the first three months of life; Custred 1977b:66 reports annual mortality rates of 25-40 percent. Llamas are also affected by mite-linked diseases, but are more resistant to them than alpacas. The severity of mites amongst modern camelid flocks may relate to the introduction of some parasites along with sheep in the mid sixteenth century; consequently, prehispanic camelid herds may have been significantly less affected (Flannery et al. 1989:102-104). Nevertheless, several sixteenth century sources refer to a serious, mange-like disease (carache) that affected camelid herds (Dedenbach-Salazar 1990:121-122).

Because alpacas must consume herbaceous plants that grow only at high elevations (Palacios 1977b:20; Webster 1971:177), and because they are so sensitive to micro-organisms that flourish at lower elevations, alpacas are rarely found below 4000 m asl. The optimal elevation range for alpacas is between 4370-4800 m asl, and the only region below 4100 m asl where alpacas have been successfully raised in historic times is the Titicaca Basin, at 3800-4000 m asl (Custred 1977b:67). Llamas, on the other hand, thrive on a much more diverse diet, and can survive in marginal pastures that could not suppport alpacas. Palacios (1977b:20), for example, notes that llamas can even make do with certain cactus species (e.g., Opuntia sp.) and the rough tola bush (Lepydophylum quadrangulare) -- fare that would be completely unpalatable for alpacas.

Because all domestic camelids tend to be gregarious and imitative of the behavior of a herd's dominant animal, one experienced shepherd can usually manage to care for several hundred animals (Palacios 1977b:42). Most herds today are in the range of a few score to a few hundred animals. This size range has considerable time depth -- e.g., a seventeenth century source from the puna of western Bolivia (Dedenbach-Salazar 1990:145-146) indicates that individual camelid herds were separated by sex and comprised 200-400 animals; other seventeenth century sources indicate the importance of dogs in herding.

The grazing habits of llamas and alpacas are different enough so as to require different kinds of shepherding care. Palacios (1977b:39) notes that alpacas tend to cluster tightly in relatively small groups within a restricted area, while llamas move much more widely and continuously while feeding. Male animals of both species are typically taken to the most marginal and distant pastures, while females with their young are pastured much closer to their corrals and in areas with better vegetation and more water.

Table 4.02. Herd composition in two puna communities in northern Chile

(adapted from Gundermann 1988:104)

Category Parinacota Community

4400 m asl

August 1981 Sitani Community

3800 m asl

Jan. 1977 Totals


No. % No. % No. %




Females 1240 58.6 108 57.7 1348 58.8


Sexed Males 161 7.6 17 9.1 178 7.8


Castrated Males 251 11.9 21 11.3 272 11.9


"Maltones" 103 4.8 23 9.6 126 5.5


Young 351 17.1 18 9.5 369 16.1



Total % 2106 100.0

43.8 187 100.0

7.1 2293 100.0





Females 1183 64.2 875 57.6 2058 61.3


Sexed Males 132 7.2 28 1.8 160 4.8


Castrated Males 141 7.7 164 10.8 305 9.1


"Maltones" 121 6.6 307 20.2 428 12.7


Young 265 14.3 144 9.5 409 12.2



Total % 1842 100.0

38.4 1518 100.0

57.8 3360 100.0





Females 529 61.8 742 80.2 1271 71.4


Sexed Males 59 6.9 25 2.7 84 4.7


Castrated Males 20 2.3 29 3.1 49 2.8


"Maltones" 48 5.6 123 13.3 171 9.6


Young 199 23.2 6 0.6 205 11.5



Total % 855 100.0

17.8 925 100.0

35.3 1780 100.0



OVERALL TOTAL 4803 100.0 2630 100.0 7433 100.0



Adult males and female camelids are usually separated while grazing and while corralled in order to avoid two problems: (1) copulations that would produce "out of season" births during the dry season when colder temperatures and more limited pasture make it more difficult to care for very young animals; and (2) male sexual dis-interest that can be caused by continuous, daily contact with females. Palacios (1977b:36) observes that male llamas lose their sexual interest in females if they are herded together continuously for more than approximately eight days; successful mating can only occur by maintaining a degree of separation between un-castrated males and sexually receptive females.

Alpacas are raised primarily for their fine wool for sale in commercial markets. Some adult animals are shorn annually, usually in December and January, and yield about 1.5 kilos of wool; animals that are shorn only every two years yield about 3 kilos of wool (Flores 1979:93). Because alpaca wool is so valuable commercially, alpaca herders devote great energy and attention to the breeding and reproduction of their animals. This involves not only maximizing successful pregnancies and births, but also the maintenance of desirable wool quality (length, thickness, and color). Since white wool is the commercially most desirable, some effort is made to maintain herds of pure white animals (Flores 1979:91). Flannery et al. (1989:112) note the abundant ethnohistoric evidence that in late prehispanic times camelids of different colors and color combinations were preferred for different ritual activities. This may also have required additional care to prevent the indiscriminate intermixing of male and female animals of different colors.

Achieving these goals of alpaca production requires a great deal of watching, separation of animals of different sexes, ages, maturation stages, and reproductive status, and penning different categories of animals in separate corrals (e.g., Custred 1977b:67-68). Specific tasks include: (1) mate selection to maintain desired wool color and vigor; (2) placing appropriate males and sexually receptive females together in special corrals for mating at intervals during Dec.-Feb.; (3) the separation of adult males from pregnant females and un-weaned young animals (in order to prevent assaults by the males), while unusually disruptive or unruly males are often castrated; and (4) providing special care for adult females and new-born animals in special facilities (usually small corral and hut attachments to larger corrals) during the birth period (Dec.-Feb.) when the very young animals are particularly susceptible to diseases and predators. Inamura (1986:154) notes that sometimes young animals are periodically rotated to unoccupied corrals where mud, damaging to the animals' health, has not built up.

Flannery et al. (1989:96-97) report that llama herders, although they have little concern with the color of their animals, still need to separate receptive adult females from aggressive males. Llama herders accomplish this by castrating most males in their herds at 3-4 years of age, leaving only a few sexually active adult males, usually selected for their more docile dispositions. Most castrated male llamas become pack animals.

Because alpacas are such valuable and vulnerable animals, they must also be continuously protected by their human guardians from the wind, cold, and wet, and predators (e.g., Flannery et al. 1989; Webster 1971:177). This requires daily overseeing by human escorts who live close to their herds. Early each morning these guardians take the animals from their nearby corrals (where they are protected overnight) out to graze in nearby pastures, and return them to their corrals at the end of the day. This daily protection is especially critical and intense for nursing females and their young (who are weaned at about eight months of age) (Flores 1979:91).

The nightly penning of camelids in corrals also facilitates dung collection (for use as fuel and fertilizer), ear-marking for identification, and checking on females' pregnancy status. At elevations below 4200 m asl, dung-rich corrals are commonly used as agricultural plots every few years (Orlove 1977b:95).

Special care for pregnant females and very young animals is also critical because of the low overall fertility of domestic camelids. Flannery et al. (1989:99) report, for example, that on average llamas give birth only once every other year, with no more than 4-6 offspring per individual animal during her lifetime; Flores (1979:90) indicates that female alpacas typically have 5-7 lifetime pregnancies. Palacios (1977b:34, 36) reports a fertility rate of about 85 percent for llamas vs. only about 50 percent for alpacas.

Flores (1986:147) notes that domestic camelids in southern Peru are presently classified in 53 named categories, according to variations in sex, color, age, reproductive status, and species type. Some 20,000 categories are theoretically available if all classificatory variables were to be fully employed. This complex classification terminology reflects herders' great concern with maximizing livestock productivity and with maintaining the physical integrity of different categories of their animals. Inamura (1981:69) reports a 4-category classificatory scheme that may be more typical of modern subsistence-oriented camelid herders: (1) female alpacas and their immature offspring; (2) male camelids; (3) female llamas and their immature offspring; and (4) cargo llamas (castrated males).

Camelid Meat and Hides. Adult llamas weigh between 75-125 kilos (Palacios 1977b:34). Domestic camelids are usually killed for their meat and skins when they are too old to be useful anymore for breeding purposes or as cargo animals, or (in the case of alpacas) when they cease to yield sufficient wool. For llamas this occurs at no more than about 15 years of age (Flannery et al. 1989:99); alpacas are usually slaughtered by the time they are eight years old (Palacios 1977b:36). Such unproductive camelids are annually culled from the herd at the beginning of the dry, cold season.

Palacios (1977a:157-159) reports that on average about 15 percent percent of a camelid herd is butchered annually. Flores (1979:41) has found that a "relatively poor" herder family consumes the meat of 3-4 animals per year, while "the well-to-do are accustomed to slaughtering up to one animal a month" for their household needs. Some camelid meat is dried and salted, for both household use and for sale or exchange, and this activity is best done at the coldest, driest time of the year (June and July). Such freeze-dried meat (charki) can be stored for up to "a number" of years (Orlove 1977b:92-93).

The herders' annual round (adapted from Custred 1977b; Dedenbach-Salazar 1990:121-122; Flores 1979; Palacios 1977b:42-48; and Tomoeda 1985. This summary represents an "average," "typical" round, one whose specific details may vary somewhat over time and space. Dedenback-Salazar (1990:122) points out that some of this variance relates to herders who occupy the upper kichwa vs. those who reside in the puna.

(1) Birth, breeding, and shearing, December, January and February. Females about to give birth are removed to better pastures and drier quarters with fewer parasites and with fewer opportunities for very young animals to drown in streams or ponds or swamps. For alpacas, human assistance is often needed during the birth process itself; llamas are apparently more self sufficient in this regard. Special vigilance over new-born animals is desirable during the first few weeks of life, especially to prevent their straying and being seized by predators. Adult female animals are sexually receptive within about 24 hours after giving birth; the gestation period is 11.5 months.

Intact males are mixed with receptive females for several days. Females are checked to see if they are pregnant or not; pregnant animals are identified with special colored cords, and are then kept separated from the males. Adult virgin females, not naturally receptive to males, are mixed with males for impregnation in special ceremonies. Shearing is usually done during this same period, when night-time temperatures are relatively high and the newly shorn animals are thus exposed to a minimum of extreme cold.

(2) Census of new lambs, curing of sick animals, and ownership marking of new animals: These activities are carried out in the context of ceremonies during March and April.

(3) Taking animals to dry-season pastures. These movements begin in early May. Adult males and females are kept separated. Older animals are slaughtered during May, and their meat is dried, to be exchanged later in the year for agricultural products; slaughtering is best done by two-person teams.

(4) Routine dry-season pasturing care, May-November. This is the period when the highest pastures are used, when there is the least chance of snow cover. At least three separate herds must be identified and maintained by herders who possess both llamas and alpacas: (a) adult females and their young (alpaca and llama females can be herded together); (b) adult male alpacas; and (c) adult male llamas [I'm still unclear why males of the two species cannot be herded together]. A fourth group must also be separated after the young animals are weaned at an age of about 8 months (usually beginning in Sept.), in order to prevent rough treatment by their mothers and a loss of sexual interest in the young, maturing males if they remain with adult females. These still immature animals cannot be put with the herd of adult males because the latter are too aggressive and rough. Each of these four groups requires its own shepherd. Overly aggressive adult males, or males which do not appear to be good breeding studs, are castrated from time to time; such castrated males may be herded together with adult females.

(5) Trading expeditions occur throughout the year, although most of them are in the dry-season (beginning as early as March and April, and extending at intervals through Sept. and October), in order to acquire agricultural products and other goods from kichwa settlements (see below).

Camelids per household. For specialized puna herders, primarily dependent on domestic animals for their livelihood, Custred (1977:69-70) finds that the minimal number of animals for a herder household to be economically viable is 30 llamas, 15 alpacas, and 10 sheep. "Wealthy" herding households in the same region have flocks containing upwards of 200 llamas, 180 alpacas, and 20 sheep; maximal household herds are up to 500 alpacas and 200-300 llamas. "Poor" herding families, who usually work part-time as seasonal laborers for "rich" households, may have small flocks of 3-5 llamas and 3-5 alpacas. Yamamoto (1981:130) notes that herder households in southern Peru, with no involvement in agriculture, have average camelid flocks of 200-300 animals. Other puna households in the region who combine camelid herding with some cultivation of bitter potatoes, have flocks averaging 100-200 animals.

Palacios (1977b:63) finds that a herd of 250 camelids is regarded as the preferred size for adequate maintenance of a primarily herding household; 150 animals/household is said to be the minimal size for household maintenance; a household can no longer be primarily dependent on herding when its flock fall below 50 animals. The ideal household camelid herd composition (Table 4.03) yields a maximum of 160 new animals/year.


Table 4.03. Ideal household camelid-herd composition, according to Palacios (1977b:63)









 For the much drier puna of southwestern Bolivia, in a region averaging ca. 3700 m asl that provides inadequate pasture for alpacas, and where herders also cultivate a few potatoes and some quinoa, T. West (1988) reports that the size of household llama-sheep flocks varies as shown in Table 4.04.


Table 4.04. Size of household llama-sheep flocks in southwestern Bolivia,

according to T. West (1988:192)

Household Wealth

No. Llamas

No. Sheep











The importance of camelid dung as a source of household fuel in the almost-treeless puna cannot be over-emphasized. Winterhalder et al. (1974:[get page nos.]), for example, have calculated that a typical herder household requires access to the dung output of 75 animals in order to obtain 30 kilos of fuel per day for their domestic cooking needs. Today, and in the historic past, dung has been supplemented by peat and (more recently) kerosene for household fuel.

Murra's (1965:192) examination of sixteenth century documentary sources pertaining to the Titicaca Basin shows household camelid herds in the 1560s ranging from 2-3 up to 500-1700 animals per household, with modal sizes between 10-20 and 50-100 animals; some form of communal grazing of flocks totaling several hundred, or even several thousand animals, appears to have been common at that time, with specific ownership of individual animals denoted by distinctive ear-markings (Flannery et al. 1989:111). It seems unlikely that the very large size of some documented sixteenth century communal herds in the Titicaca Basin (up to 16,846 animals in one case) would ever have been duplicated elsewhere: pastures in other parts of the Central Andean puna are simply not extensive enough to accommodate such large numbers of animals. Nevertheless, the Junín puna, with over 3.5 million hectares of potential pasture (Matos 1994:45), would probably have been capable of sustaining individual herds of several thousand animals. We still know very little about the necessarily complex management of such large herds.

Settlement patterns. Several writers describe modern puna herder communities (e.g., Flannery et al. 1989, Flores 1975, 1979, Fonseca and Mayer 1988, Gade and Escobar 1982, and Inamura 1981, 1986). Such communities typically (although not always) comprise several score herder households configured in two types of settlements: (1) a single large, centrally situated nucleated village or small town, in which each household maintains its own infrequently used residence; and (2) numerous small hamlets (estancias), spaced several kms apart, usually comprised of one extended family or several related households (Inamura 1986:156-157).

Most puna pasture controlled by Indian herder communities is communally owned, with some areas reserved for common grazing while others are permanently ceded for the exclusive use of particular extended family groups (e.g., Ossio 1983:50-52). This form of land tenure has sometimes resulted in a strong pattern of local-group endogamy in order to conserve traditional land titles. A common pattern over the past century has been for sons to inherit their fathers' rights to herds, pastures, and herding tools, while daughters inherit their mothers' rights to agricultural lands and equipment (Ossio 1983:57). There is also another, exogamous tendency that functions to link puna herding households and communities more securely to kichwa agriculturalists through kinship ties (Bastien 1978b; Flores 1975:14). The interplay between endogamous and exogamous tendencies produces considerable variability in local social structure.

The central village/town is a focal point for interaction between herders and higher levels of regional and national administrative hierarchy; it is also a source for certain necessities (e.g., radios, batteries, metal tools, medicines, and some items of clothing) that cannot be otherwise obtained.

The hamlets are occupied for periods ranging from a few weeks to a few months, according to the frequency with which flocks must be moved to fresh pastures; they are often supplemented by outlying camps used as temporary shelters for shorter periods. Inamura (1981:69-70, 1986:153-154) describes a pattern of household transhumance in which, during the dry season, each family moves back and forth between its main hamlet residence (estancia) and one or more secondary residences (astanas or cabañas) located about a half-hour's walk apart. During the wet season a family often occupies its secondary residence(s), normally situated in more marginal pastures that are inadequate during the dry season. Dedenback-Salar (1990:145) notes sixteenth century references to long-term residence of youthful herders living in small huts in remote pastures, well removed from their families' main households.

Puna herding settlements tend to be much less nucleated than kichwa settlements inhabited by agriculturalists (e.g., Flores 1975:15; Gade and Escobar 1982:446-447). This difference relates to the herders' greater need for daily and seasonal movement between pastures, and to the agriculturalists' dependence upon intensive, irrigation-based agriculture requiring community administration and cooperative maintenance. The expanding network of improved roads since ca. 1950 has radically transformed the spatial configuration of regional hierarchies as strategic road-side location has become a primary factor in the placement of central socio-political and economic functions.

During the wet season (Dec.-March) suitable pasture is potentially available virtually everywhere in the Central Andean puna (Custred 1977a:159). Herders must periodically move their flocks as pasture in any given locality becomes inadequate for one of two reasons: (1) because less vegetation is available during the dry season (May-Sept.); or (2) because flocks exhaust the available grass within a few kms from the corrals to which they return every night. Seasonal dryness is less critical in the wetter Peruvian central highlands than for the drier puna of southern Peru and western Bolivia. Nevertheless, even in the central highlands the dry season does lower both the quantity and quality of most pasture. Consequently, relatively well-watered pastures -- along stream and lakeshores, around the edges of swampy ground (bofedales) -- are always at a premium. Such localities are invariably sought after by herders because they provide the richest, most dependable, and most continuously available sources of feed for their animals.

Flores (1984), Palacios (1977a), and Orlove (1977b:95-96) point out that bofedal pastures are often extended through artificial irrigation and ponding. Palacios (1977a:166) describes two community pasture-irrigation systems in the southern Peruvian puna that provide water to an area totaling ca. 2200 hectares through two main canals of 17 and 16 kms in length. Efforts to control such improved pastures produce conflicts that often must be resolved through the mediation of outside authorities (Custred 1977a:168-169). Comparable competition over improved pastures, is documented for the sixteenth and seventeenth century (Dedenbach-Salazar 1990:153-156).

A further complexity results from the different pasture needs of llamas vs. alpacas: only llamas can make do with the low-quality vegetation available in more marginal pastures. Alpaca herders seek out areas where succulent young growth is available (Custred 1977b:68-69). Alpacas must usually be re-located more frequently than llamas, simply because of their different dietary requirements. In the wetter puna of central Peru, the movement of herds is typically less frequent and over shorter distances than in the drier puna of southern Peru and western Bolivia. Animals are often moved between higher vs. lower elevations and between swampy vs. non-swampy zones where different quantities and qualities of vegetation are seasonally available. Herders reside alternatively in larger vs. smaller settlements according to the interplay between their animals' needs and pasture availability.

Herders' hamlets range between 3-10 households (Custred 1977b:71). The largest and most-occupied settlements are invariably near the best watered pastures, often strung out along stream or river margins (Inamura 1981:68); the smallest and least frequently occupied settlements usually are found in more marginal pastures. The larger settlements typically contain the largest, most numerous, and most substantial human dwellings grouped somewhat apart from the camelid corrals. Houses are often rectangular in plan, with mortared stone walls, and separate cooking and sleeping areas, and they often include space for the long-term storage of cooking and eating implements, food, wool, textiles, dried meat, skins, and tools (Flores 1979:46-48; Inamura 1986:154). These rectangular dwellings are variable in size; generally they range from 2-3 m on a side up to ca. 5.5 x 3 m (Flores 1979:48-49); the details for one precisely measured structure are presented in Table 4.05. These larger settlements, situated as they are near the best pastures, are almost always occupied during the dry season, when more marginal pastures are least productive (Custred 1977a:159; cf. Flores 1979:92).


Table 4.05. Exterior dimensions of a typical herder's house,

after Arangueren (1975:123-124)

Section of house


rear length


front length


left-side width


right-side width



Smaller herders' settlements typically have more rustic dwellings that are usually occupied by less than a full complement of household members. These houses are often circular and less carefully built, with lower stone walls, and are usually too small for anything beside sleeping and casual cooking; they are often constructed immediately adjacent to small corrals (Custred 1977a:159, 1977b:71; Flannery et al. 1989:43-45). These settlements are usually supplemented by temporary camps where one or two individuals remain for short periods caring for animals in pastures too far from the hamlets for daily commuting. Such camps often comprise little more than a small stone-walled hut and a single small corral (e.g., Inamura 1986:153).

Camelid corrals are quite variable in size. Most have stone walls standing 0.8-1.0 m high, measuring 0.2-0.3 m thick, and enclosing areas between 80-1000 m²; most corrals are 100-250 m² in area; some corrals in the lower puna may be built of rectangular sod blocks (Orlove 1977b:95). Corral area is related to both flock size and to function -- e.g., some of the smallest corrals are set aside for giving birth and caring for very young animals; some corrals are used mainly for ceremonial purposes. Corrals in larger settlements tend to be relatively large and rectangular, while those in smaller settlements are more likely to be smaller and circular (e.g., Flannery et al. 1989:43-57).

In the relatively moist puna of Peru's central highlands, we might expect to find less differentiation between wet-season and dry-season herders' settlements, because herders are less constrained in where they can pasture their animals during the dry season. Nevertheless, Dedenbach-Salazar (1990:127) notes a 1534 documentary source that refers to seasonal transhumance of full-time pastoralists in the Junín puna. Land tenure considerations that affect access to pastures can also have a significant impact on the timing and duration of transhumant movements.

Houses in modern herders' settlements usually shelter a nuclear family. Herders' hamlets typically comprise clusters of households headed by closely related men (e.g., Flannery et al. 1989:50). These households pool their labor for tasks associated with herd management, house and corral building and repair, caring for aged relatives and children, and trading expeditions. Groupings of a few dozen hamlets typically comprise a local "district" community -- whose existence is often recognized as the lowest rung of the state-imposed administrative hierarchy. Community solidarity and cohesion is also achieved through integrative rituals that involve all members in petitions or thanks-giving to the supernatural forces that control the well-being of themselves and their herds (see below).

Over the annual cycle, individual herder households move back and forth between their residences in different-sized hamlets and camps, alternatively breaking apart and coalescing according to the dictates of their herds' pasture needs and to whatever constraints may be imposed by land tenure arrangements. It is difficult to quantify this general pattern because precise information on the camelid carrying capacity of different puna environments is scarce. Palacios (1977a:159, 1977b:60) indicates that in the puna of southern Peru, humid bofedal pasture can sustain three camelids/ha during the dry season. This compares with 1 camelid/ha on non-bofedal pasture during the wet season

As we noted in Chapter 2, Matos (1994) has reported recent studies of annual pasture requirements for sheep and camelids in the Junín puna (Table 4.06).


Table 4.06. Annual Pasture Requirements in the Junín puna,

after Matos (1994:45, based on unpublished 1984 ONERN Studies)

one sheep

5-10 ha of "improved" pasture

one camelid

10-17 ha of "un-improved" pasture

one camelid

20-25 ha of "least favorable" pasture


[try to get more carrying capacity data]

[check with Ramiro re meaning of "improved" pasture]


For the upper kichwa of the main Mantaro Valley in the Peruvian central highlands, Browman (1974:195) has estimated an overall average camelid carrying capacity of 50-125 animals/km² (.5-1.25 animals/ha, or .8-2.0 ha/animal); he suggests these figures might even be doubled. These figures greatly exceed all other estimates available to us, and we suspect they cannot be realistically applied to puna contexts.

It is difficult to generalize about camelid carrying capacity in the Central Andean puna from these diverse figures. The most reasonable and comprehensive general estimate is probably that of Matos (1994:45) who calculates that some 3,850,000 ha (38,500 km²) [check figures] of available pasture in the Junín puna (an area much larger than our survey area) is presently capable of sustaining 150,000-200,000 camelids -- an average of ca. .04-.05 animals/ha, or 19-26 ha/camelid on a long-term basis. Matos suggests that the prehispanic carrying capacity of this same area may have been significantly higher, owing to modern environmental deterioration.

Herders' cosmology and ritual. Herders desire the continued and adequate supply of healthy animals. This end is served by a series of formal rituals that propitiate and thank the supernatural forces that control animal life. In common with kichwa agriculturalists, puna herders share a belief in a primary life-giving force (a remote being, usually referred to as Pacha Mama), and a hierarchy of more approachable regional and local spirits, usually referred to as Apu or Wamani (e.g., Flannery et al. 1989:182; Flores 1977:229; Gow and Gow 1975:148-149; Matos 1994:39; Zuidema and Quispe 1967). The wamani, who are considered to be the ultimate "owners" of camelid flocks, are usually associated with prominent hills, but also with lakes, cliffs, quebradas, springs, and other natural features (e.g., Duviols 1984:208-209; Rowe 1980), and it is commonly to these places that rituals are directed, or where they are performed.

Several descriptions of contemporary Andean cosmology (e.g., Bastien 1978a) indicate that landscapes are traditionally visualized in anthropomorphic terms, as living organisms comprised of different components -- including natural features, animals, and humans (both living and dead) that must interact appropriately in order for the whole organism to survive and function. Ritual performances (including pilgrimages, dances, and fighting), offerings, and feasting often occur at, or near, shrines placed at locations deemed to be particularly important within the anthropomorphized landscapes. Often such places are at ecological junctures (e.g., between puna and kichwa zones), or along major quebradas that separate socio-political territories. Rituals are organized on household, community, regional, and inter-regional levels.

The herders' shrines themselves are diverse: they may consist of subterranean chambers (Quispe 1984; Valderrama and Escalante 1988:142-140); miniature house-like structures (Gow and Gow 1975:150); small chapels (Paz 1988; Poole 1982; Quispe 1984:610); boulder cairns (Flannery et al. 1989:172)); special corrals (Flannery et al. 1989:172; Flores 1977:233-234; Quispe 1984:611-612); cemeteries or isolated tombs (Bastien 1978a:182; Duviols 1976a:287); free-standing boulders (Duviols 1984:208-209); or simple natural openings into the earth's surface, or holes dug into the ground surface and covered with rocks (Bastien 1978a:57). Quispe (1984) indicates that these shrines often comprise a complex of related features, widely dispersed over the landscape, to which people from related households and hamlets periodically come in ritual processions to deposit offerings.

Herders' rituals take several different forms, but almost always include feasting and the deposition of offerings. Offerings include small stones (sometimes bezoar stones, removed from camelid intestines); zoomorphic stone or ceramic figures; small figurines made of maize paste or grass; marine shells; coca leaves; camelid fetuses; sacrificed adult camelids or parts of these animals; cloth; flowers; maize beer (chicha), or other forms of alcohol (e.g., Arangueren 1975; Nachtigall 1975). These offerings symbolize the essential generative life force (enqa) that is necessary for successful animal reproduction, growth, and general well-being. The life force of the material objects slowly seeps away and dissipates, and so must be periodically renewed through ritual performances associated with the deposition of new offerings (Flores 1975, 1977). If these ritual performances are not carried out properly, the camelid herds "on loan" from their wamani owners will desert their human caretakers.

In her study of two early seventeenth century documents that describe herders' rituals in the Junín puna, Dedenbach-Salazar (1990:220-221) notes the special role of Lake Junín, and neighboring smaller lakes, as the places of origin and residence of wamani spirits. Participants in ritual dances used slings to hurl special stones, or anthropomorphic figures made of straw, into the lakes.

Ethnohistoric perspectives on herd management. Dedenbach-Salazar's (1990) ethnohistoric analysis extends Murra's (1965) earlier study and offers new insights into camelid herd management at the time of European-contact. These sixteenth and seventeenth century sources reflect the powerful impact of Inka imperial administration. Particularly interesting is the extensive Quechua-language terminology that makes clear distinctions between different categories of herds, herders, and pastures (Table 4.07). [work Murra 1965 into this]

The ethnohistoric sources indicate that herding was a pillar of highland Andean economy, from the local community to the over-arching Inka empire. Herders, animals, and pastures had different statuses according to how they related to different organizational levels. Differential wealth and prestige, for example, were defined by different degrees of access to camelid resources. Some herders were members of local communities who contributed their shepherding labor as state tribute; others were more closely linked to Inka state institutions as herding specialists. Some animals and pastures were identified with local communities, while others were more directly linked to the state -- some flocks and pastures were directly granted to deserving individuals for different kinds of service to the state; others were associated with different kinds of state secular or religious institutions.


Table 4.07. Sixteenth century Quechua terms relating to camelid herding,

(after Dedenbach-Salazar 1990:123-151; note: our translations to English).

Quechua Terms Meaning


llama sapa

herd of camelids

michi-q or

llama michi-q

"la persona que guarda y pastorea al ganado"

the person who guards and pastures the flocks

llama kamayuq

herder serving the state


"ganadero que tiene ganado"

institutional or individual herder who owns animals

llama chakraruna

"el ganadero en ganadaos y no en chacras"

herder without agricultural fields

llama kama-yuq

"ovejero diestro en criar las ovejas de la tierra"

herder skilled in raising camelids

qhapaq llama

"los rebaños del estado," or "dueños ricos"

state herds, or rich owners

wakcha llama

non-state herds or non-wealthy owners


herders who care for flocks linked to Inka religious cults

huk kancha llama

llama kancha

kancha kancha

different kinds of camelid corrals




chunku or chunku chunku-lla


apa apa-ntin

terms having to do with managing cargo llamas, or herding groups of camelids in general


"enviar el ganado de día al pasto para volver a la noche"

take animals out to pasture by day and return at night


"echar al pasto muchos días"

take out to pasture for many days, without returning to main base

[check p. 142-143]

returning animals to corral at nightfall

[check p. 142-143]

grouping animals in corral

[check p. 142-143]

careless intermixing of camelids belonging to different owners

[check p. 143-144]

leading and controlling animals with cords




common pasture


communal pasture

     [recheck these in notes]


Agricultural Ecology

In Table 2.24-A we summarized the altitudinal and productive ranges of indigenous cultigens. Although these cereals and tubers are adapted to a wide range of climatic variability, only a few tuber species are important in the lower puna; above 4200 m asl agricultural production is insignificant because of cold temperatures. Highly productive agriculture is confined to the kichwa, below 3850 m asl.

Because of the uncertainties of rainfall, water control in the forms of canal irrigation and swamp drainage is often necessary to achieve high agricultural productivity. Uncertainties about water supply are compounded at elevations above 3000 m asl by the unpredictable onset and termination of frosts. High agricultural productivity can only be maintained through the application of fertilizers (e.g., camelid dung, ashes, household wastes, and grasses) and the maintenance of fallowing regimes. The high proportion of sloping terrain means that terracing must often be employed in order to achieve full agricultural productivity.

Because there are so many different altitude-defined crop niches, and because there is such great altitude variation within short horizontal distances, agriculturalists face a complicated schedule of planting, weeding, and harvesting. Different variants of cereals and tubers have different planting and harvesting times. This means that many different crops are available over a large part of an annual cycle, and that poor harvests in one niche may be offset by good harvests in another. This complicated scheduling is also demanding in terms of administration and travel time between fields at different altitudes.

Agriculturalists must be concerned with the demands of field preparation, fertilizing, planting, weeding, caring for growing crops, guarding and harvesting mature crops, and maintaining an infrastructure of terraces, irrigation canals, and field walls (Camino 1980; Camino et al. 1981; J. Matos et al. 1958). Many of these tasks must be performed at the same times as some of the most demanding activities related to herding. Growing and standing crops can be seriously menaced by the predations of domestic animals. Consequently, where herding and cultivation are locally combined, cultivated fields must be strongly fenced and/or the daily movements of local flocks must be carefully controlled (McCorkle 1987).

Traditional agriculture employs a variety of sectorial fallowing regimes that maintain long-term soil fertility and minimize the risk of poor harvests for individual households (e.g., Camino 1980; Camino et al. 1981; Goland 1992; Guillet 1981; Mayer 1985; Orlove and Godoy 1986). These regimes demand considerable investment in administration, definition and protection of community territory, and walking back and forth between scattered fields. Additional walking and prolonged camping time must be expended if some members of agricultural households need to devote themselves to the care of domestic animals in higher pastures well away from their main residences.

Agricultural tasks. Table 4.08 summarizes the principal agricultural tasks -- these are in addition to the maintenance of irrigation, drainage, and terrace facilities.

Much agricultural land in the Andean highlands is terraced. Some terraces, especially where irrigation is used, are well-made, stone-faced features. Less elaborate terraces are probably more typical: e.g., the "andenes rusticas" (rustic terraces) described by Fonseca and Mayer (1988:70). The latter features do not have stone retention walls, but rather take advantage of naturally occurring


Table 4.08. Agricultural tasks, (adapted from Camino et al. 1981:176)

Name of Task



Fertilizing, with animal dung, grass, or ashes1

Barbecho or Chakmay

Plowing, usually with a foot plow




Mounding earth up around plant stalks





1 -- Small quantities of commercial fertilizers are now also used.


breaks in slope whose natural level surfaces are reinforced by rows of bushes, cactus, or small trees that stabilize the terrain. This type of agricultural field system is similar to what R. West (1968) has called "semi terracing" in highland Mexico -- the main difference being that in the Andes the cultivated hillslopes tend to be steeper.

Traditionally, fertilizer is applied in three main ways: (1) by setting up temporary corrals atop the terraced fields and placing livestock there for a few days; (2) by hauling animal dung from pastures to fields on pack animals; and (3) by covering the fields with cut grass (both local, and brought in from the puna), which is allowed to dry and then turned under during the barbecho. Ashes from household cooking hearths are also used as fertilizer, and since the mid 1970s there has been minor use of commercial fertilizers.

There are few detailed studies of traditional vegetative fertilizers in contemporary Central Andean agriculture [check Goland; maybe ask Winterhalder or Goland]. This is an important topic that merits more investigation. King (1911:202-212), for example, provides a richly detailed account of how important "green manure" (grasses, weeds, stalks, leaves) was in the intensive agriculture of eastern Asia in the early 20th century, before the advent of chemical fertilizers in that region, and in a context where animal dung (including human) was insufficient relative to the overall need to maintain continuously high agricultural productivity. King's study suggests that green manure could have been important in precolumbian Central Andean agriculture.

Land preparation prior to planting is usually done with the traditional Andean foot plow (chaquitaclla). Typically this hard work is performed by a husband-wife team: the man thrusting the iron-tipped tool into the earth to bring up the clods of fresh earth, and the woman overturning the clods to fully expose the newly turned soil. Planting is similarly done in a field that has been plowed some weeks or months earlier, with the woman inserting the seed into the turned (or re-turned) earth. [check on this]

The hilling-up of potatoes and other crops is generally undertaken about a month or so after initial planting (Yanamoto 1981:126). Here the foot-plow is employed to open a shallow ditch (surco), about 30 cms wide and 20 cms deep, alongside each row of plants. The earth from this ditch is then thrown up around the sides of the growing plants with a J-shaped wooden mattock tipped with an iron blade (Orlove 1977b:93-94). This operation is apparently necessary in order to reduce the number of plant stems and to prevent the buildup of excessive humidity around the growing plant [check on this].

Mature tubers are dug out of the earth with a mattock. Some are taken directly to the dwellings for consumption and storage, but many are placed in specially prepared grass-lined and brush-covered pits dug in or near the field. Layers of tubers, usually separated by grass layers according to variety, are placed within such pits, from which they are removed as needed. These pits typically measure 1.2 m in diameter and 1.5 m deep (Fujii and Tomeada 1981:47-48).

Mature stands of wheat, barley, and quinoa are cut with an iron sickle, and are commonly threshed and winnowed in or near the fields. Threshing is commonly accomplished by dumping the cut grain onto stone floors where they be trod upon by oxen or beaten with wooden rods. Winnowing is typically performed by tossing the threshed grain and adhering chaff into the air with wooden forks so that the wind removes the lighter chaff, while the heavier grains sink back to the earth. Mature maize ears are left to dry on the stalks in the fields, then carried to the permanent residences where they are stored until used (ca. 70 percent for chicha) (Fujii and Tomoeda 1981:45).


[maybe add fuller descriptions of other tasks and tools; quantify amt. of land cultivable per household, maybe get photos or illustrations of these tools in use -- maybe Michael Brown's slides can be made into B&W prints, or maybe there are illustrations in the UNAM Andean technology book; are there any time and motion studies -- maybe ask Enrique Mayer, Goland, Winterhalder]


Agricultural zones. Camino (1980) and Camino et al.'s (1981) five-division classification of agricultural zones in the southern Peruvian highlands appears to have broad applicability. The main outlines of this classification are summarized in Table 4.09. Ideally, each household has direct access to fields in each of the five zones within its community territory. Zone C, the primary and most accessible niche, is by far the most important in economic terms. Zone A, not listed in Table 4.09, is uncultivated herding land above 4100 m asl.




Table 4.09. Agricultural zones in highland southern Peru,

(adapted from Camino et al 1981:176-187)



(m asl)


Cultivation Practices

B, "Luki Manda"


mainly bitter potatoes plus oca and izaño

6 rotating sectors; 1 in cultivation and 5 in fallow

C, "Uray Manda"


potatoes1, izaño, oca, olluco, habas

6-sector rotational sequence (see Table 4.10)

D, "Anexo"2


2 types of potatoes

minor maize

6-sector rotational sequence

E, "Tierras de Maiz"


mainly maize3; some potatoes; minor beans, squash, yacón, and racacha 1 yr. of potatoes, followed by several years of maize interplanted with minor crops. Use of ox-drawn plows.


Below 2000

coffee, coca for cash sales

Far from community; commercial use only

   1 -- Up to 30 varieties of potatoes are planted.

2 -- This sector is 12 kms distant, and therefore less intensively used than "C".

3 -- Several varieties of maize are planted and kept separated.


Sectorial fallowing regimes. Orlove and Godoy (1986:171, 185) summarize the general characteristics of 51 ethnographically known sectorial fallowing regimes in the kichwa and lower puna between central Peru and western Bolivia:

(1) Each of the sectorial fallowing systems is a land-use system which consists of a set of lands associated with a set of households.

(2) The set of lands are divided into a number (n) of sectors. The lands which make up each sector are contiguous.

(3) All households own plots in most or all sectors....

(4) There is a sequence of n-year-long uses for the lands. Some of these uses consist of the planting of a specific annual crop or small number of annual crops. The set of crops may different among the successive cropping uses out of the total set of uses. The other uses are fallowing, combined with grazing. All the fallowing uses occur after all the planting uses.

(5) All n sectors pass through the same sequence of n-year-long uses. In any given year, one and only one sector will have each cropping use, and the number of sectors which are fallowed is equal to the number of fallow uses in the n-year cycle.

(6) When a sector is used for planting, each household has access to usufruct rights to its plot or plots in that sector.

(7) When a sector is used for fallowing and grazing, all households have access and grazing rights to the entire sector.

(8) This land use is maintained and enforced by institutional means.


The average number of households participating in each of these community-regulated fallowing systems is 350. The origin and time depth of these systems are unknown, although their absence in the northern and southern Andean highlands (and even in northern Peru) suggests they are mainly adaptive in the southern half of the Central Andes. In most cases where irrigated land exists today, it tends to be privately owned and privately controlled (Fonseca and Mayer 1988:78-80; Guillet 1981:142); community-administered sectorial fallowing regimes occur predominantly where rainfall-based cultivation predominates. Areas where commercial agriculture has made substantial inroads have witnessed a considerable deterioration of traditional sectorial fallowing systems (e.g., Guillet 1987:86-87; Mayer 1979).

Guillet (1981:145) argues that community oversight is necessary in order to maintain the long-term integrity and viability of sectorial fallowing regimes: the adherence of individual households to rules about fallowing, planting sequences, and grazing rights affects the well-being of the entire community. This is reinforced by the fact that individual households are often dependent upon infrastructural facilities (e.g., terraces, canals, roadways, fences) that can only be adequately maintained through community effort.

The communities studied by Camino et al. (1981) employ a six-sector fallowing system. This means that communal lands are subdivided into six different sectors. Each designated sector follows a specific crop-rotational sequence (Table 4.10), and at any given time each of the other five sectors in the zone is at a different point in its rotational cycle. This deliberately produces a maximum diversification of crops over time and space -- a result which buffers against crop failures caused by unpredictable and highly localized climatic variations. What this means for each household is that all available cultigens are growing somewhere within its usufruct holdings at any given time, while at the same time some of its lands are always lying fallow.

Another detailed description of a comparable sectorial fallowing system comes from the upper Cañete drainage on the Pacific slope in central Peru (Fonseca and Mayer 1988:72-74). This is an 8-sector system, and in any given year there are three sectors in use and five in fallow. Individual households maintain use rights in each sector. Each year a new sector is opened to a 6-year rotational cycle, and each year a sector that was cultivated the previous year goes into a 3-year fallow period, whose conclusion will bring the full 6-year cycle to a close. Community management is crucial to viability of the enterprise. The great variability of sectorial fallowing regimes in the upper Cañete and in other parts of the Peruvian central highlands (e.g., Mayer 1971) reflects the flexibility of these systems as they are periodically readjusted to take into account changes in community and household demography.


Table 4.10. Rotational sequence in 6-year cycle in Zone C. After Year 6, the cycle begins again with Year 1. (Adapted from Camino et al. 1981:179-182). 





potatoes interplanted with izaño

2 aporques (1st in Oct., 2nd in Nov.-Dec.)


mainly oca; sometimes olluco or izaño

no barbecho; no fertilizer applied


oca, olluco, izaño; sometimes fava beans

no fertilizer applied


fava beans

some fertilizer applied



grazed by livestock



barbecho and main application of fertilizer

[check: barbecho in Years 3, 4?]


Fonseca and Mayer (1988:75) point out that the most basic factors that underlie and sustain the traditional sectorial fallowing systems are (1) low soil fertility, owing to the slow decomposition and buildup of organic material in soils of cold climates; (2) the scarcity and expense of fertilizer; and (3) the danger of erosion if sloping lands are cultivated or grazed too intensively. These circumstances are somewhat alleviated on irrigated land, and this is a principal reason why such land is often not included in sectorial fallowing regimes. Locally elected officials (camayoq) enforce rules about planting, fallowing, crop rotation, and grazing by making frequent tours of inspection (Fujii and Tomoeda 1981:54). As recompense for their services, these officials are provided with small huts for living quarters in areas where their tours take them, together with a small share of each household's crop.

The annual round. Figure 4.01 graphically depicts a typical annual round of agricultural tasks in the south Peruvian region studied by Camino el al. (1981:Fig. 6) (excluding infrastructural maintenance tasks, such as cleaning irrigation canals or repairing terraces). This illustrates several key points.

(1) Agricultural work is virtually continuous throughout the annual cycle, although there is comparatively more activity during two periods: March-June, and Sept.-November. The least-busy time, late June through early August, is the time when chuño (freeze-dried potatoes) and ccalla (dried oca) are prepared.

(2) Tuber harvesting is nearly continuous during the year, excepting only October and November. Within any given sector and for any given variant, tuber harvesting begins earliest at lower elevations, and extends to fields at higher elevations through the year.

(3) Two main agricultural-task seasons can be distinguished (especially if one focuses on the economically most important terrain between 3200-3800 m asl): (a) Feb.-July, a time of land preparation and harvesting; and (b) August-December, a time of planting and weeding. Harvesting invariably proceeds from lower to higher elevation for given sectors and crop variants; land preparation, planting, and weeding for given sectors and crop variants begin at higher elevation and proceed to lower elevations through the year. This dichotomy in the timing of agricultural tasks reflects the additional time required for plants to reach maturity in the colder, less fertile soils at higher elevations.


Figure 4.01. The annual round of agricultural tasks in highland southern Peru

(adapted from Camino et al. 1981:188-189, Figure 6)


[insert Fig. 4.01 about here]


Note: In Fig. 4.01, each activity unit is shown as a discrete lineal segment that pertains to a specific crop in a specific sector rotation cycle. Ideally, this diagram illustrates the tasks engaged in by each household within a sectorial fallowing regime. No information is available about the timing of barbecho below 3200 m asl.


Additional agricultural annual-round data from the Peruvian central highlands are summarized in Table 4.11.


Table 4.11. The annual round from two kichwa districts in central Peru,

(after Matos et al. 1958:56-58, and Soler 1954:108-109)













Oca, olluco, mishua








Barley, wheat, fava beans





Another perspective on the agricultural round can be obtained from a month-by-month compilation of activities from a kichwa community in central Peru (Table 4.12). This is a mestizo community, more involved in commercial production than those we have discussed above; nevertheless, the data are indicative of general modern practices.

Settlement patterns. For most households, many different agricultural tasks are likely to be going on at the same time, and often in scattered fields at some distance from each other; this is compounded by the fact that many households also have some involvement in herding activities above 4100 m asl. The household thus faces a major challenge in its efforts to effectively apportion and direct its labor force toward activities that often overlap and compete. McCorkle (1987:59) has suggested that the complexities of labor allocation in these contexts may select for extended-family household structures, in which more people can cooperate in the performance of essential tasks that are dispersed in space and time. The Catholic ritual calendar, with its numerous saint's days and fiestas distributed over the year, is an important reference for beginning and ending different agricultural activities (Camino 1980:29).


Table 4.12. Month-by-month agricultural activities in a mestizo kichwa community at 3400 m asl in central Peru, (adapted from Escobar 1973:64-66)




Harvest of "new" potatoes planted in July and August.

Cultivation of potatoes, wheat, and barley.


Harvest of "new" potatoes.

Harvest of arvejas verdes.

Weeding wheat and potatoes.

Hilling-up potatoes.


Harvest of potatoes, arvejas, and quinoa.



Harvest of green maize, potatoes, arvejas, and ocas.

Plowing of fallow fields (with ox-drawn plow).


Harvest of maize and mashua.

Beginning of barley harvest.


Harvest of barley and wheat in non-irrigated fields.

Harvest of maize, potatoes, fava beans, wheat, and alfalfa in low fields.


Clearing fields before planting.

End of wheat and barley harvest.


Pasturing livestock on fallow fields.

Planting maize (Cuzco variant) in low fields.

Planting "new" potatoes in low fields.


Planting fava beans, oca, and mashua in low fields.


Planting maize, wheat, and potatoes in rainfall-watered fields.

Planting fava beans, arvejas, ocas, ollucos, and mashua in low fields.


Planting barley in rainfall-watered fields.

Planting quinoa in low fields.

Planting maize.

First harvest of arvejas verdes in low fields.


Planting quinoa in low fields.

Weeding and hilling-up maize.

Final harvest of barley.

 [check whether low fields are irrigated]


It is usually necessary for several members of each household to spend part of their time each year living in small hamlets and/or with relatives in localities well outside the main settlement. Yamamoto (1981:119-120) notes that, in the Cuzco region, small huts for temporary residence typically measure ca. 4 x 2 m in area, with stone walls and a thatched roof, containing a cooking hearth and grinding stone; new roofing grass and portable kitchen utensils are brought to the structure for use during the time of its occupation.

Seed potatoes are usually stored in stone bins or atop straw mats at or near dwellings in the lower puna, where low temperatures prevent premature sprouting and rotting (Yamamoto 1981:124). At the appropriate time, these seed potatoes, together with loads of animal dung for fertilizer, are transported on pack animals to fields at lower elevations. Because domestic camelids defecate in a single locality in both corrals and pastures, it is particularly convenient to collect their dung.

Occasionally labor demands exceed a household's capacity, and temporary outside workers must be hired, or reciprocal labor exchanges made with other households. Another response, for households that also have small herds of llamas or sheep, is to bring these animals to the fields during the busiest harvest periods: the animals can be cared for at the harvest site, and immediately put out to graze and defecate in the stubble of the harvested fields (Yamamoto 1981:123-124). In some cases, kichwa households who combine herding and agriculture find it convenient to combine all their flocks under the care of a few shepherds during periods when agricultural tasks are particularly demanding (McCorkle 1987:65).

The scheduling of labor demands produces a dispersed regional settlement pattern, typically with a large, centrally situated settlement of a few hundred people situated at 3200-3700 m asl in the heart of the most important tuber-producing zone, and numerous small hamlets (estancias) scattered widely at higher and lower elevations. In addition to the fields outside the major settlements, many households have intensively cultivated kitchen gardens close to their houses inside these large villages (McCorkle 1987:62). Most people spend most of their time at the main settlements, but there is usually some permanent occupation of the smaller hamlets as well, although often not by complete households; and, as previously noted, some central settlements are actually mostly vacant most of the time (Webster 1971, 1973).


Carrying capacity. [any data?; any regional pop. density figures?; compare pop. densities of herding vs. agr. zones]


Water management. Irrigation has greatly expanded agricultural production in the kichwa and lower puna; in many cases irrigation canals are directly incorporated into extensive systems of terraced fields (e.g., Fonseca and Mayer 1988; Mitchell and Guillet (eds.) 1993; Treacy 1989). Unlike the desert coast (e.g., Netherly 1984), highland irrigation systems are usually small scale, with multiple sources (often springs or small puna lakes). They are usually managed at the local community level, with annual communal canal-cleaning and repair operations overseen by locally elected officials (Fujii and Tomoeda 1981:56-57; J. Matos et al. 1958:63). Skar (1982:150-151) has noted that "more than 10 households sharing the same irrigation ditch tends to lead to trouble." Nevertheless, highland irrigation networks often provide the economic foundations for important Andean social units, and the canal networks themselves have often served to physically delineate and symbolize these units on the ground (B. Isbell 1974; Ossio 1978; Sherbondy 1982, 1986; Zuidema 1986).

Fonseca and Mayer (1988:78-80) delineate two very different types of irrigation systems: one intensive (watara), and the other extensive (vichka). Watara land, which typically comprises 10-20 percent of the community's total irrigated terrain, is cultivated every year. The irrigated fields are on well-built terraces, close to the largest settlements, and are heavily fertilized with household wastes and animal dung. Multi-household work groups usually cultivate these watara fields cooperatively, and also work together to maintain the irrigation canals. Vichka land, on the other hand, is farther away from the settlements, with casually built terraces, less irrigation, little or no fertilization, and with fields worked by single household units.

A lesser known form of Andean water management is the Qocha system, described by Flores (1984) for the lower puna of the northern Titicaca Basin. These are shallow depressions (usually circular, averaging about 100 m in diameter) that collect rainwater runoff. It is uncertain whether these features are natural or artificial. The ponded water soaks into the underlying earth and periodically provides sufficient moisture for crops. Today, in an area just northwest of Lake Titicaca, some 256 km² of these depressions (approximately 20,000 individual qochas) remain in use, planted with rotational sequences of potatoes, quinoa, kañiwa, wheat, and barley; an additional 128 km² are found nearby in an abandoned state (Flores 1984:89). Qochas also seasonally provide household water, and they are prized as pastures during fallow years. Flores (1984:94) notes how easy it is to miss seeing remains of abandoned qochas.

Qochas appear to be quite distinct from the Qochawiña discussed by Orlove (1977b:95-96). The latter average only 7-10 m² in area and 1-2 cms deep. Like Qocha, they function to collect rainfall run-off. The ponded moisture improves the growth and quality of grass. Commonly "several" are found in an area of 1-2 hectares in the lower puna. The Qochawiña are mainly a form of improved pasture, although they apparently also function as places where tubers are periodically grown.

Swampy zones in several parts of the Central Andean highlands contain numerous prehispanic ridged fields (Erickson 1987, 1988; Hastorf and Earle 1985; Kolata 1986). These fields, analogous to the Mexican chinampas (Armillas 1971; Parsons 1985), apparently did not continue to be used in historic times.

[check Mitchell 1976 -- Am. Anthrop. article, plus Mitchell and Guillet (eds.) 1993]


Agriculturalists' cosmology and ritual. Like herders, agriculturalists are concerned with propitiating and thanking a hierarchy of wamani spirits who are associated with neighboring hills, lakes, and other places, and who are ultimately responsible for the supply and well being of crops and domestic animals. As for herders, Christian beliefs and ceremonies have become part of agriculturalists' ritual performances.

B. Isbell's (1974) study of the Chuschi community in the Peruvian south-central highlands provides a good example of the kichwa agriculturalists' vertically structured local universe. There are four zones: (1) the main settlement itself (pueblo), situated at 3150 m asl; (2) the quichwa, agricultural land where maize can be grown, immediately surrounding the village at elevations between 2600-3300 m asl; (3) the mayo patan, a zone of low-lying land, below 2600 m asl, along the river; and (4) the sallqa, more distant terrain between 3300-4000 m asl, where maize cannot be grown. Each of these four principal zones is, in turn, sub-divided into two sectors, referred to as "upper" (hanay) and "lower" (uray).

Variations on this vertical zonation are widely known throughout the Central Andean highlands. In many cases these local universes also contain a puna herding niche, at elevations above the upper limits of effective agriculture (e.g., Bastien 1978a, Harris 1978, 1982, 1985; Platt 1982, 1986; Skar 1982; Valderrama and Escalante 1988; Valle 1970).

The borders of these ecological divisions and sub-divisions are often physically marked with shrines (often small chapels) where formal rituals are performed. Zonal subdivisions are typically complementary in their functions -- e.g., the upper and lower sectors of the main village contain, respectively, the architectural centers of religious (the church) and secular (the district administrative building) authority. Integrative ritual performances focus on the saints images and crosses associated with the main church and many of the border-defining chapels. At intervals these images of "minor" saints and crosses are brought in ritual processions from the small chapels to the main church where they are briefly reunited with their "mother" images.

Many of these community rituals are associated with communally performed tasks, such as cleaning irrigation canals. Other components of community ritual focus on water sources (springs and lakes) to which offerings (including coca and chicha) are carried in formal processions whose routes are defined by the small chapels, where additional offerings are made and where the saints' images are replaced after their stays in the main church. Processions to and from the chapels typically begin and end at the main settlement, where additional offerings and feasting often occur. Ritual significance is often attached to points where major irrigation canals converge, and community cemeteries may be placed at such locations in recognition of this significance.


Exchange between Puna Pastoralists and Kichwa agriculturalists

Today herders and agriculturalists exchange their complementary products and services in three main ways: (1) through impersonal buying and selling at commercial marketplaces in towns and larger villages; (2) through non-commercial trading expeditions that link established trading partners along traditional routes traversed by llama pack trains; and (3) labor and product exchanges between close relatives living in different zones (Bastien 1978a; Flores 1985; Fonseca and Mayer 1988; Harris 1982, 1985; Mayer 1971; Platt 1982, 1986).

Before the advent of motorized transport, large annual or semi-annual regional fairs were held in strategic places (usually in puna localities with adequate pasture for hundreds of pack animals for a week or two) in order to redistribute non-local products for which there was only infrequent demand (Wrigley 1919). Commercial markets and fairs need no further discussion here; both seem inextricably associated with nineteenth and twentieth century commerce. Trading expeditions and inter-household exchange, however, may be more relevant to our concerns.

Trading expeditions. It is puna herders who invariably organize and undertake trading expeditions. Prior to motorized transport, pack llamas were the only effective way to move quantities of material through rugged highland landscapes. Even today the high cost of gasoline and the absence of adequate roads in many areas means that motorized vehicles cannot always move effectively. Mules and horses can carry heavy burdens overland through rugged terrain. However, because of the high costs for their feed and breeding, horses and (especially) mules are economically viable as freight carriers only in commercial enterprises.

Only puna camelid herders possess large numbers of cargo llamas. Furthermore, these herders frequently are situated between ecologically complementary zones (kichwa, montaña, and desert coast). Consequently, puna camelid herders naturally occupy a strategic position vis a vis the inter-zonal movement of complementary products. These herders continue to be important as transporters of goods between agricultural communities.

Pack llamas are capable of carrying burdens of 25-55 kgs (depending on length of journey). These animals begin to be trained for transport work at about two years of age, and are usually ready for full service when they are three years old. Groups of related men in herding communities organize trading expeditions, usually comprised of 20-30 pack animals (up to 50 can be managed by a single experienced man), covering 10-30 kms/day in daily marches of about 8-10 hours. Pack llamas can apparently go for 4-5 days without food on such journeys. Pack-train management is greatly facilitated by the presence of well-trained lead animals who are easily managed and who help keep subordinate animals in order. Caravan routes are typically marked by the presence of stone-walled corrals, placed at intervals of a day's travel, intended for overnight encampments (Concha 1975:87; Flannery et al. 1989:106; Flores 1979:94-95; Inamura 1981:69).

Dedenbach-Salazar (1990:168) reports that Inka imperial caravans of the early sixteenth century comprised 400-800 pack llamas. Obviously, the management and infrastructural support of such large caravans would have been much more complex than for the comparatively small undertakings of modern times.

Today these expeditions vary in length between a few days and a few weeks, and typically involve groups of 3-10 related men and their animals. They move along traditional routes between their established trading partners at lower elevations from whom they acquire maize, quinoa, potatoes, fruits, coca, and pottery in exchange for dried meat, wool textiles, salt, dung, and harvest-transport services. Boys often accompany their fathers or uncles in order to learn the routes and procedures, and to meet the trading partners and their families, with whom they eventually will come to have personal ties themselves. The typical pattern is to make several short expeditions, of a few days duration, to nearby upper-kichwa areas, and a few longer expeditions, up to several weeks in length, to more distant places at lower elevations in the kichwa and even the Pacific coast and ceja de montaña (e.g., Bandelier 1910: ;Casaverde 1977; Concha 1975; Flannery et al. 1989:108; Flores 1979; Inamura 1981 Tomoeda 1985 [Scott sponsor for mid 19th cent.]).

Salt is commonly moved in trading expeditions between its puna source areas and consumers in the kichwa (e.g., Matos 1994:35-37). Some puna herders spend significant amounts of time in extracting and packaging salt. Concha (1975:74-76) and Inamura (1981:73), for example, note that some puna herders make special expeditions of 2-8 days duration to salt sources they control in order to acquire quantities sufficient for their subsequent trading expeditions. Because salt sources occur primarily in the puna, and because only puna herders control large numbers of pack llamas, in traditional Andean economies only puna herders can acquire and move salt in any quantity.

Table 4.13 illustrates the trading activities of a typical herding community.


Table 4.13. Trading expeditions from the Chalhuanca community in 1970,

(adapted from Casaverde 1977:179)

Areas Costa Kichwa

(Collca Quebrada) Upper Kichwa



Communities Visited Sihuas Camaná Cabanaconde

Tapay Huanca


Lluta Cororaque Paucartambo


Months of trip March-April Feb.-Apr. May-July May-July July-August Aug.-Oct.


Duration 15 days 8 days 10 days 12 days 8 days 40-60 days


Mode of travel llamas truck llamas llamas llamas llamas



Goods offered meat meat, money meat, wool, cloth meat, wool, cloth meat, wool, cloth, coca, sugar, figs, aji, meat, cloth, llamas


Goods obtained figs, maize, wheat, figs, aji, rice maize, fruit maize, wheat, barley barley, wheat, beans chuño



Form of exchange barter and purchase purchase barter barter barter barter and purchase



Two patterns emerge from these data. First, most trading expeditions occur during the dry season, especially for those of greater than one week's duration. The only wet-season expeditions undertaken by the Chalhuanca herders are to the rainless Pacific coast. Second, although much of what these herder-traders carry for exchange are what they produce themselves (e.g., meat, cloth, llamas), they also supply their trade partners with products from other zones (e.g., coca, sugar, fruit).

Concha (1975:82) notes that the puna herders who organize and undertake trading expeditions usually lack good information about (1) the precise time of the main crop harvests in specific kichwa agricultural settlements, and (2) the relative success or failure or availability of specific crops in specific settlements. Analogous uncertainties also exist for the kichwa agriculturalists at the other end of the exchange, who are dependent upon the puna-based traders for key goods and services (including transporting harvested crops from field to storehouse): e.g., it is difficult for them to plan the precise timing of their harvests if they remain uncertain about when the herder-traders will actually arrive to assist them in this essential task (Inamura 1981:70-72). These uncertainties produce great concern and stress on the part of those involved.

Other forms of labor and product exchange between herders and cultivators. When herders and cultivators live in relatively close proximity (i.e., within less than a day's travel time), it becomes possible for them to exchange complementary goods and services in more regular and predictable ways. As in the case of the long-distance trading expedition described above, these relationships are often defined and validated in ritual terms, and occasionally through affinal ties between inter-married households. In such cases, specific groups of puna herders supply specific groups of kichwa agriculturalists with such things as dung and animal transport at planting or harvesting times when agriculturalists most urgently require them for fertilizing their fields and carrying their crops from field to storehouse.

Such visits are also occasions when herders bring dried meat, textiles, etc. to their agriculturalist neighbors. In return, the visiting herders are supplied by their kichwa hosts with lodging and meals, and with supplies of recently harvested tubers and cereals to take back home with them. Harvest times are particularly appropriate for this kind of herder-cultivator interaction. It is at this time when the stubble and weeds of recently harvested fields serve as excellent fodder, when the harvested fields are in need of the manuring provided by a few days of animal grazing, and when the agriculturalists are most likely to have surplus supplies of the tubers and cereals needed by herders (McCorkle 1987:66-67).

[add material from Harris, Platt, Bastien,]


Integration of Herding and Cultivating Economies through Ritual

The extent of inter-marriage between puna herders and kichwa agriculturalists varies greatly today. Tomoeda and Fujii (1985:301), for example, report that no more than 1 percent of all marriages in one district they studied are of this type. In other nearby places, however, the inter-marriage rate is much higher, up to 60 percent in some cases (Inamura 1981:76). Inter-marriage between herders and agriculturalists obviously facilitates access to complementary products and resources that become available through the normal channels of kin-based sharing and reciprocity. On the other hand, community exogamy can disperse rights to land, fields, and herds -- basic resources that often can remain fully concentrated and consolidated only through community endogamy.

The very real difficulties of establishing bonds of inter-marriage between puna herders and kichwa agriculturalists are often overcome through building other kinds of linkages, typically defined and validated through formal ritual, between these complementary groups. We will now consider several inter-related aspects of such ritual.

Anthropomorphic Landscapes. Ethnographic studies of the isolated Qollahuaya in Bolivia (Bastien 1978) describe local groups (ayllus), typically comprising many hundreds of individuals, that occupy well-defined territories extending over areas measuring several dozen kilometers in length and width. The Qollahuaya are particularly well described in this regard, but comparable anthropomorphized landscapes in other parts of the central Andes have also been described (e.g., Allen [1982], Gow and Gow [1975]), and Vokral [1991]). Classen (1993) and Zuidema (1983) suggest that a living landscape metaphor is deeply rooted in Andean cosmology and polity.

These ayllu groups incorporate dispersed settlements of specialized puna herders and kichwa cultivators into integrated socioeconomic units. Ayllu borders are defined by major quebradas where ritual performances linking adjacent ayllus occur.

Qollahuaya ayllus maintain their territorial integrity and socioeconomic integration through belief systems that emphasize the role of human beings as components of anthropomorphized landscapes. Both living and dead human beings are key agents in maintaining the well being of these landscapes. The living accomplish this essential task by adhering to rules about marriage, residence, exchange, burial, and ritual. People, places, and products are all endowed with complementary qualities that derive in large part from their origin in different elevation zones, and all components of the living landscapes are "nourished" by their associations over time and space with people, products, and places possessing ecologically complementary qualities.

It is by means of ritual offerings at ayllu shrines that the living landscape is nourished. Each such ritual offering involves products from all of the main ecological zones, as well as the presence and participation in the ritual of people who are associated with each of these zones. The largest shrines and the most important rituals typically occur at intermediate locations on, or near, the puna-kichwa border.

Burial Ritual. The importance of deceased ancestors and their mummified or interred remains in household, community, and imperial ritual is well documented (Allen 1982; Bandelier 1904; C. Mcewan and Van der Guchte 1992; Salomon 1995; Zuidema 1973, 1977a, 1990). These sources describe the public veneration of ancestor mummies, public feasting, and public renewal of burial offerings.

Regional Pilgrimages. Several studies (e.g., Poole 1982, 1991; Reinhard 1985; Sallnow 1987, 1991; Zuidema and Poole 1982) emphasize the importance of major ecological junctures (often where principal puna and kichwa zones border one another) as the loci of regional pilgrimage shrines. These pilgrimages typically incorporate multicommunity groupings of puna herders and kichwa cultivators who interact at these strategic loci to insure continued productivity and interaction through feasting, dancing, material exchanges, and ritual offerings.

Ritual Fighting. An extensive literature (e.g., Gorbak et al. 1962; Hopkins 1982; Orlove 1994; Platt 1986; Skar 1982) describes practices, extending back at least 200 years, of regularly scheduled intra-community and inter-community ritualized conflicts (tinku). These feature confrontations between groups of young men armed with slings, bolas, whips, or clubs, who are supported by groups of women and older men who perform dances, prepare feasts, and encourage the fighters. The total number of participants varies from several dozen to several hundred. Serious injuries and even deaths occur, and there is some indication that these, and the flow of human blood in general, are esteemed as signs of future good harvests or successful animal breeding.

These ritual battles are often competitions between moieties of dually structured communities. When intra-village moieties are involved, the individual settlement itself becomes the "stage" for ritual fighting, and the centrally located cemetery and church situated on the border between the two divisions are the loci for fighting, feasting, or avoidance. There are also reports (Hopkins 1982; Platt 1986; Skar 1982) of ritual conflicts between more distantly linked, multivillage groupings who share common sociopolitical borders along which ritual fighting occurs.

Duviols (1973) notes a deep-seated distinction in the seventeenth century between adjacent groups of puna herders ("los lacuaz") and kichwa cultivators ("los huari") in the Peruvian central highlands. This dual opposition was manifested in several ways, including occasional physical violence; marked distinctions in language, costume, and association with different supernatural forces and different sacred places; exchange of complementary products; and bi-ethnic communities.

Of particular interest is Duviols' (1973:175) mention of the "Danza de Guari-Libiac" ("Danza de Guerra"), an important ceremony in the integration of potentially hostile groups of herders and cultivators. This ritual-fighting performance appealed to the long-standing differences between puna herders and kichwa cultivators, emphasizing their traditional enmity while providing a ritualized basis for enduring interaction. The Danza de Guari-Libiac might be an ancestral form of modern ritual fighting, and perhaps a descendant of prehispanic forms of ritualized herder-cultivator interaction.

Structures of duality and tripartition. There is an extensive literature on Andean structural duality and tripartition at all organizational levels (e.g., Gow 1978; Harris 1985; Izko 1992; Murra 1968; Netherly 1990, 1993; Paerregaard 1992; Palomino 1971; Rostworowski 1983; Sallnow 1991; Wachtel 1973). We have already noted the significance of structural duality in ritual fighting and in the performance of other integrative Andean ritual. We have also referred to the differentiation between puna herders and kichwa cultivators in the conceptualization and definition of this duality. Any formal distinction between two sectors can create a formally defined border that may function as a third division. In some cases this becomes the location where public ritual linking the two sectors is performed at cemeteries, churches, shrines, processional routes, or feasting rooms.




(1) Herding and agriculture are basic, complementary components of central Andean highland economy. Neither is complete or sufficient in isolation.

(2) Herding and agriculture may be combined within a single household or settlement. However, several factors encourage specialization and physical separation between puna herders and kichwa cultivators.

(3) Ideology and ritual play a major role in the integration of specialized puna herders and kichwa cultivators. Ecological variability, structural duality and tripartition, and the metaphor of living landscapes provide the conceptual foundation for this integrative ritual. The fundamental purposes of ritual performances are (a) to define and maintain social borders, and (b) to insure continued productivity of the bordered units and on-going interaction between them. Moiety borders and the junctures of puna and kichwa zones stand out as the loci where these rituals are performed. Rituals include (1) making simple material offerings at modest shrines, (2) feasting and dancing in association with ancestral tombs and/or mummies, and (3) engaging in elaborate ritual battles and pilgrimages.


Overall Summary and Conclusions

[good quote for use in this section -- Guillet 1987:84]

The implications of incorporating higher and lower zones into an overall [household or local community] production strategy are that it increases the difficulties of scheduling labor. For example, in areas where two agricultural zones are exploited, variation in rates of plant growth and length of rainfall in relation to altitude produces a separate agricultural calendar for each vegetation zone. The problems of integrating an agricultural with a pastoral calendar are even greater. As a result, conflicting demands on household time and labor and the necessity for continuous vertical movement create an extremely complex scheduling problem. These disjunctions become a serious obstacle to the intensification of agro-pastoral production [in the context of household or local community management].


[Another useful quote for this section, McCorkle 1987:72-73], for communities in southern Peru that combine kichwa agriculture and puna herding

access to labor may be more problematic than access to land. Grazing lands [in the puna] are communally held and freely available. However, their location vis-a-vis agricultural workplaces often engenders spatial and ... labor disjunctions between cultivation and herding for the peasant household.

Some of these disjunctions can be offset by choice of pasturing regime, at least partially or seasonally. Beyond this, certain socio-organizational strategies may be called into play. Predictably enough, most strategies involve aggregating animals and dividing up daily herding duties across two or more households. In other words ... households must forego their 'ideal' of productive autarchy and seek assistance beyond the 'domestic mode of production' ... in the ... 'supra-household' sphere.

Although villagers' small-scale pastoral associations are designed to combat dialectical disjunctions, they do so only imperfectly, even in Usinos' [name of local community] eyes....They are usually quite costly, not only materially, but also socially....they open the way to serious intercine conflict over animals and their care -- due to the informal, dyadic nature of these contracts. Indeed, village pastoral associations not infrequently collapse from such conflict, leaving bitter feelings that may ripple distrust and factionalism through the entire community, and may last for generations.

[suggestive of need to supra-community level of resolution -- leads directly into my argument for Junin]


[Another good quote, for this section or maybe elsewhere: Morris 1978:20] M. states that his discussion of exchange "brings us at once to what may well be the single most interesting and difficult question in Andean studies: What constitutes the community?"

[JRP: Answer=there are many communities, at many different levels; e.g., ayllu/calpulli -- confusion in terminology comes from fact that these terms refer to "communities" at different organizational levels]


[Think about: how does inter-zonal, inter-regional operate in absence of over-arching system that maintains some degree of security and peace? I.e., pre-state, or non-state contexts]





Table 4.01. Common foods for puna herders in southern Peru.

Table 4.02. Composition of mixed alpaca-llama-sheep herds in the South Andean puna.

Table 4.03. Ideal household camelid-herd composition.

Table 4.04. Sizes of household llama-sheep herds in southwestern Bolivia.

Table 4.05. Exterior dimensions of a typical herder's house.

Table 4.06. Annual pasture requirements in the Junín puna.

Table 4.07. Sixteenth century Quechua herding terms.

Table 4.08. Agricultural tasks.

Table 4.09. Agricultural zones.

Table 4.10. Crop-fallow rotational sequence.

Table 4.11. The annual round from two kichwa districts of central Peru.

Table 4.12. Month-by-month agricultural activities in a mestizo kichwa community.

Table 4.13. Trading expeditions from Chalhuanca.



Figure 4.01. The annual round of agricultural tasks.


PLATES (adapt several photos of tool use; check slides too)

return to main page

Copyright 1999 Jeffrey R. Parsons. May not be used without written permission.