Cultivating Diversity: Field Scattering as Agricultural Risk Management in Cuyo Cuyo, Department of Puno, Peru

By Carol Goland, 1993.

Chapter 10 - Agricultural Practice, Production Zone, and Risk Management in Cuyo Cuyo

(page 355)

SUMMARY OF FINDINGS

The residents of Cuyo Cuyo practice a complex, highly organized system of agricultural production. Patterns of land tenure, cropping choices, fallow schedules, fertilizer regimes, and labor inputs are differentiated in space and time. They provide precise definition of communally created and maintained production zones. The agricultural lands of the two study communities extend from 2700 to 4100 meters in elevation. The organization of production on these lands is in part differentiated by placement on the altitudinal gradient. Higher elevation lands are subject to greater degrees of communal control and longer periods of fallow.

Landholdings of individual households are quite small when compared to other Andean communities. In Puna Ayllu, households cultivate an average of 0.35 hectare of land each year; in Ura Ayllu, the average is slightly greater, 0.45 hectare. In both communities these diminutive landholdings are divided into numerous, spatially separated parcels: between 9 and 25 plots are planted by each household in Puna Ayllu, and between 11 and 26 in Ura Ayllu. Most fields are obtained through inheritance from both husbands' and wives' parents. The elevation distribution of the fields in the two communities overlaps only slightly. Fields in Puna Ayllu are located between 4100 m and 3470 m, while the highest fields in Ura Ayllu are found at 3820 m and the lowest at 2700 m.

The manda, or the sectorial fallowing system, is primary in the conceptual and practical organization of agriculture. In Puna Ayllu, all agricultural land falls within one of two mandas. The higher elevation Awi Awi manda is organized around seven geographical sectors. Two years of cropping--potatoes in the first year, oca in the second--are followed by five years of fallow. The lower elevation estancia manda is divided into six sectors. Four years are used for cropping (potatoes, oca, habas and tubers, and finally monocropped habas) and the fallow period lasts two years. The estancia manda of Ura Ayllu contains the highest elevation agricultural lands in this community. Like its Puna Ayllu counterpart, the Ura Ayllu estancia manda is used for four years of cropping and then left to fallow for two years. During the cropping years, the Ura Ayllu rotation proceeds from potatoes, to oca, to a repeat crop of oca, and finally to habas.

Other lands utilized by Ura Ayllu households are located at lower elevations. Fields in the Paqhchani zone are only loosely organized as a manda system. Three years of cropping are followed by four years of fallow. Families use these lands primarily for planting potatoes, especially when their landholdings in the estancia manda are insufficient. This zone can be used for planting early-maturing varieties of potatoes, and because of this two crops per year can be obtained.

Ura Aylleños also grow potatoes in several other zones: they rent lands in the nearby community of Ñacoreque; they own lands in the manda system of their anexo Aripo; and they

(page 356)

sequentially intercrop potatoes and corn in their maize fields. In all of these locales, production of both the early crop (papa milli) and the second crop (papa panq'o) are made possible by lower elevation and more temperate climate.

In their lowest elevation zones Ura Aylleños plant maize. These fields, maizales, are not subject to the community controls and coordination that typify the manda system; there is no set sequence, or inter-household synchrony, of cropping and fallow periods. Households plant fields until they feel that the soil has become exhausted and requires rest. Maize is the primary crop. It may be planted in association with squash, common beans, yacon, and racacha.

Analysis of landholdings for the two agricultural cycles studied shows that households face continual reconfiguration of the distribution of available their lands and crops across space. One year they may have abundant land in the potato manda, the following year they may have little. Changes in the quantities of land available for each crop each year affect decisions about how the harvest is allocated between consumption and seed. One year the potato manda may be located at relatively high elevations, and the following year it will be located in lower zones. Changes in the distribution of lands by elevation re-order labor and its scheduling each year. These decisions are interdependent and must be coordinated for each manda, plot, etc., and always adjusted to household needs.

Per unit of land, labor intensities in Cuyo Cuyo are extremely high relative to data reported in comparable Andean studies. These same studies have reported household landholdings considerably larger than those found in Cuyo Cuyo. Reckoned as a percentage of days of the year, Cuyo Cuyo farmers spend about equal the amount of time engaged in agricultural production as farmers elsewhere in the Andes: 27% of their days in Ura Ayllu, and 21% of days in Puna Ayllu. But although equal in overall production time, Cuyo Cuyo farmers prepare fields and husband each meter of growing crop with greater attention and effort. The estimates provided by the data gathered and analyzed for this study are nearly identical to independent time allocation data collected by the PSE project (Winterhalder et al. n.d.).

Differences in labor investments, spatial location, degree of communal control, and crop use define production zones. Potatoes require the greatest labor inputs of any crop grown in Cuyo Cuyo. Being the first crop of the rotation, part of this relatively high labor cost is due to the extra preparations required--especially the plowing--when a field is taken out of fallow. When field preparations are excluded from analysis of potato labor inputs, labor for the oca crop is found to be greater. Average labor per hectare for both of the tuber crops is comparatively high. In contrast, the production of broad beans requires substantially less labor per unit of land area. Weeding consumes the greatest proportion of all of the labor time spent on the crops in the latter part of the rotation cycle, an observation which is also true for the non-manda maize crop.

Labor inputs appear to be related to distance from the household. They are highest for those fields nearest the communities, i.e., in the estancia mandas, and lowest for those located at the greatest distance (at least for those crops which can be compared across production zones). This distance effect is consistent across the various tasks. It is striking that nearby fields are weeded twice during the growing season, while distant fields are weeded only once.

Puna Ayllu and Ura Ayllu share many similarities with respect to labor allocation and the patterns that differentiate near and far fields. Fertilization is heavy, and in both communities, fertilizer practices are differentiated by crop and distance to fields. Cropping cycles are at the same time fertilization cycles. The first crop in the rotation cycle (potatoes) is always fertilized, usually with dung; after this crop, little or no additional fertilizer is applied.

(page 357)

There are several general similarities between the communities in terms of fertilizer use. In both communities the potato crop is fertilized ubiquitously and heavily, often with several types of fertilizer. By weight, the majority of the fertilizer is provided by dung from herd animals. After the potato crop, subsequent years of the rotation cycle are less likely to be fertilized. Less than half the fields of the second year of the estancia rotation, and slightly more than half the fields in the third and fourth year of the rotation, are fertilized. The predominant fertilizer in these subsequent years is ash collected from kitchen hearths.

Differences between the two communities are most apparent in the kinds of fertilizers used in potato production. Puna Aylleños use camelid dung, while Ura Aylleños use sheep dung. This difference parallels the actual species composition of herds in the two communities. Ura Ayllu is further distinguished by its widespread use of the wanuna, a transportable pen that is used to collect dung by corralling sheep overnight on fields. The wanuna is used only in the estancia manda.

In both of the manda systems within Puna Ayllu, camelid dung predominates, though there are several important differences. Ashes, guinea pig dung, and the wanuna are used only in the estancia manda. Differences in fertilizer practices between the production zones of Ura Ayllu are also apparent. Except for the wanuna, all sources of fertilizer, especially ashes, are used much more frequently in the non-estancia production zones. Like potato fields, the maize fields of Ura Ayllu are almost always fertilized. In stark contrast to the potato crop, however, the predominant source of fertilizer is guano de isla, with cow dung second in importance.

Analyses of agricultural soils in Cuyo Cuyo (Bennett pers. comm.; Sandor and Eash n.d.) indicate the soils on cultivated landforms are generally adequate for crop growth. Comparison of cultivated and non-cultivated soils suggests that this is the product of careful soil management practices. Inputs of macronutrients through fertilization maintain soil fertility sufficient for crop growth. The growing environment is further enhanced by terracing, which levels slope, reduces frost, decreases run-off, retards erosion, and conserves water.

Yields for crops in Cuyo Cuyo are comparable to those reported elsewhere in the Andes. On average, potatoes produce 8551 kg/ha, oca 8269 kg/ha, illaco 5836 kg/ha, isaño 6892 kg/ha, and habas yield 1162 kg/ha. However, the variance in yield from one field to another is great. Differences between the two communities in average yields are slight. Within each community, differences between production zones are in some cases pronounced.

Within each community, yields of potatoes differ greatly between the near and distant fields. In both Puna Ayllu and Ura Ayllu, potato fields in the estancia mandas produced 4500 kg or more per hectare than fields in other production zones. Analysis of yields in potato production demonstrates that the factors which can account for the greatest proportion of variance are labor (field preparation and weeding), fertilizer, and proximity to the community. Together, differences in these three input factors can explain 29.5% of the variance in potato yields.

A similar analysis of oca yields indicated that two variables--density of seeding and planting date--explain 18% of the variance in crop yield. Increased seed densities, and earlier planting dates were positively correlated with yield. The very low proportion of variance explained by the regression model is unsurprising, given the general failure of this crop in the year for which data were analyzed. The poor performance of the oca crop was largely attributed to pests, which are patchily distributed and have uneven impacts. In contrast, 1986-87 was reported to be a superb year for habas production. For this crop, seed density and use of fertilizer explained 26.4% of variance in yields.

(page 358)

These regression models indicate that a high proportion of the variance in crop yields cannot be explained by reference to management factors (such as year in the cropping rotation, seed density, date of planting, fumigation, labor, and fertilizer inputs) or systematic qualities of the field (e.g., distance, altitude, field area, study year, community affiliation). What cannot be controlled in the analyses are precisely the same factors which are beyond the control or management of households: patchy distributions of rain, uneven soil qualities, pest infestations, crop disease, etc. These non-systematic and uncontrollable factors are of special interest in the analysis of risk.

In summary, descriptive analyses of agricultural practices and crop yields in Cuyo Cuyo draw attention to three conclusions. First, relative to land area, agriculture is input (labor, fertilizer) intensive. Second, the average family's landholding is small. It appears to be difficult for households to satisfy subsistence needs. Third, most of the high degree of variance in productivity cannot be statistically explained by the input factors under the control of households. These findings generate this question: given high labor inputs and total yields marginal to subsistence needs, why would households further increase labor demands and reduce net yields by dispersing fields?

Although scattering fields over the landscape increases travel and transport costs, I have shown that risk minimization can explain this phenomenon. Models suggest that if the fluctuations in yield of scattered plots are not strongly positively correlated, then a harvest that pools the produce of several fields reduces variance relative to what would be experienced from year to year if households relied instead on a single plot.

The risk reduction Z-score model has generated the following rule of thumb: if you can expect to meet your requirement (set by the critical threshold) be variance-averse. In the present analysis this means plant many fields (the low variance option). However, if your requirement is greater than what you can expect on average from production, be variance-prone. In this case, plant fewer fields (the high variance option) on the slight chance one or more will do unusually well.

Analysis of potato yields for each family indicates that planting several dispersed fields reduces variance. As increasing numbers of plots are cultivated, the coefficient of variation of yields for each household steadily decreases, rapidly at first and then at decreasing rates. Maximum possible yields drop (the poorly producing fields dampen the effects of high yielding fields), while the minimum yield obtainable increases. This increase in pooled minimum yield is the mechanism by which field scattering reduces risk, when risk is defined as the probability of falling below a critical threshold set by household subsistence needs.

The analysis of field scattering as a means of reducing risk has focussed on potato production. The potato crop is nutritionally and agronomically the most important of those grown in Cuyo Cuyo. Household disaster levels were set by calculating each family's total caloric needs based on the age-sex structure, adjusted by subtracting the nutritional demands of temporary migrants (for the time they spent outside the community), and the caloric contributions of non-potato crops and purchased foods. Remaining caloric needs must be provided by the potato fields available to each family, and this determines the minimum level of production (yield per hectare) required to satisfy subsistence needs.

The ability to avoid disaster as the number of utilized fields is increased was projected for each family. In 12 of 19 cases, families would have faced a significant probability of disaster had they cultivated an enlarged version of but one of the fields actually used in 1986-87. All twelve reduced this predicted disaster probability to 0% as increasing numbers of fields were added to their array; all twelve had an actual average yield from their many potato fields in excess of their calculated minimum need. In contrast, three families faced an initially

(page 359)

high probability of disaster with one field, and increased the probability of disaster to 100% by cultivating many fields. These three families all had minimum needs greater than the actual average production for all of their potato fields. If they could have known in advance of planting that their average production would fall below their needs, then by the Z-score rule of thumb they should have consolidated potato plantings into a single well-yielding locale. However, Cuyo Cuyo families cannot know prior to planting (and well into the agricultural season) how well fields will produce. Nor do they have the option of quickly expanding the area of each field to sufficient size. Absence of prescience, and constraints in altering field size, means that at planting time families must assume that their average yield will be above requirements. They all must begin with the scattering bias that serves the majority so well each year.

Three families were found to experience a 100% probability of disaster for all numbers of fields. In these cases needs are so great relative to both average and individual plot yields that no single field or subset of their fields could have saved them from falling below their minimum requirement. Finally, one family with an extremely low requirement and adequate production on all fields maintained a 0% probability of disaster for all numbers of fields.

Field scattering provides important risk reduction benefits in the majority of cases. However, this benefit must be compared against disadvantages, particularly reductions in net yields due to travel and transport to distant fields. The caloric costs of travel to fields were calculated based on distance, number of trips made during the agricultural season and energetic expenditure (adjusted for slope, rate of travel, and load). Net yields in potato fields were then decremented to reflect this cost. Under scattering, the average loss in net yield for families is about 7%. If fields were consolidated, reduction in yield for travel and transport would be less than 4%. This is a significant cost but not a large difference. Given that its costs are quite modest, and given the importance of achieving minimum subsistence needs along with the effectiveness of field scattering in meeting this goal, field scattering appears to be a relatively low-cost and appropriate response to risk.

The very modest increase in travel costs measured here may be atypical. The lower the initial yield, the greater the impact of travel on total net production. Since the 1986-87 potato crop in Cuyo Cuyo was unusually productive, it can be expected that in other years the reduction in net yield due to travel to scattered fields would be greater. This observation is linked to the high level of labor and fertilizer inputs in Cuyo Cuyo agriculture. To whatever degree labor and fertilizer inputs raise yields, this increase helps to achieve yields that maintain family agricultural production farther from disaster, and secondly, the increase elevates yields to the point where the travel costs of field scattering have only a modest impact on the final net yield.

VERTICALITY AND THE DEFINITION OF PRODUCTION ZONES

The Andean natural environment is one of altitudinal organization. Elevation gradients produce vertically layered ecological zones. The ways in which households, communities, and diverse larger polities utilize this natural zonation has been a primary focus of Andean research in the last quarter century.

The importance of verticality for socioeconomic organization was first articulated based on ethnohistoric information (Murra 1972). Study of both prehistoric and contemporary communities has confirmed the long time depth and persistence of vertically-based ecological and economic relations (Mayer 1985; Salomon 1982). Verticality links natural geoecological

(page 360)

zonation, human cognitive patterns, and socioeconomic organization within a single frame of reference.

Research on contemporary variants of prehispanic verticality has produced a typology that recognizes an archipelago variant, an extended form, and a compressed form (Brush 1976). A key underlying factor in these variants is the relative horizontal distance separating vertical zones.

Studies of household- and village-based verticality have relied on the definition of zones as broadly classified by natural zonation and/or emic criteria, often incorporating broad regions. Mayer (1985) argues that these broad definitions of zones have obscured the political and economic aspects of verticality, as it was first formulated by Murra (1972). Mayer (1985) proposes instead that we superimpose the concept of production zone on top of existing natural and emic classifications. The definition of production zone is

A communally managed set of specific productive resources in which crops are grown in distinctive ways. It has a set of infrastructural features, a particular system of rationing resources (such as irrigation water, natural grasses) and the existence of rule-making mechanisms that regulate how these resources are to be used. Complementary to the management of these resources, individual productive units (such as households) hold access rights to specific portions of these resources. They have full rights to all products obtained by them from their labor and they have the right to transmit them to others (Mayer 1985: 51).

The concept of production zones enhances our ability to understand Andean economic and ecological organization by permitting analyses to include dynamic considerations of the role of humans in defining, creating, and responding to their natural environment.

In Cuyo Cuyo, the concept of production zone elucidates how space is used and how activities are organized within it. The broadest contrast is between lands that are organized into the manda (sectorial fallowing) system and those that are not. Manda lands are subject to the highest degree of community control. Community consensus dictates what is to be grown, when and where it is to be grown, and how it is to be grown. But even where lands are not organized into a manda system, production zones are demarcated by the association of distinctive arrays of crops and technologies. Analyses of labor and fertilizer inputs show that the same crop may be produced with very different inputs (both qualitative and quantitative) depending on where the crop is located.

Use of the concept of production zone also allows us to understand why relations between altitude and agricultural practice are not always direct and consistent. Much has been written about the relation between altitude and intensification, which is often measured by the shortening of fallow cycles (Godoy 1984; Orlove and Godoy 1986), but is also defined as higher input levels (Guillet 1987), or higher degrees of communal control (Mayer 1985; Yamamoto 1985).

In the Cuyo Cuyo case, there is a clear relationship between altitude and intensity of land use, as measured by the ratio of fallow years to all years in the cycle. The highest elevation lands are those with the highest fallow ratio; the lowest lands have no set fallow cycle, and may be cultivated continuously for a decade or more. Higher elevation lands are subject to the most stringent communal controls, lower lying lands the least.

In contrast, if the definition of intensification is based on inputs (either greater quantities or more costly inputs), then we do not find the expected inverse relation between altitude and intensification. In the Cuyo Cuyo case, it seems clear that it is not elevation but

(page 361)

rather distance that is related to differences in the intensity of inputs. The closest fields (those located in the estancia manda) are those which receive the most intensive inputs of the primary resources under control of households: their labor and fertilizer.

IMPLICATIONS FOR PREHISTORY AND ARCHAEOLOGY

Students of the Andes have long noted the abandonment of prehispanic agricultural structures (terraces, raised fields, irrigation canals, etc.). These abandonments are perplexing, especially given that contemporary population densities are comparable to, if not in excess of, those presumed for prehispanic periods. Several explanations are offered. The loss of the sociopolitical organization required to maintain these infrastructural features is one. Another hypothesis cites climatic change, i.e., fields were once used at higher elevations because past periods had warmer climate. I address myself to this latter claim.

Cardich's (1985) work on abandoned fields and irrigation systems in the Andes uses associated occupation sites to date agricultural features. Presumed dates of use and abandonment are linked by Cardich to global climatic patterns. These are used to argue that a change from warmer to cooler conditions explains abandonment of agricultural features at higher altitudes. His work has been criticized by Seltzer and Hastorf (1990) on two counts. First, global patterns of climate change are simply too generalized to support inferences about climatic events and cultural behaviors in an area as diverse as the Andes; detailed local studies are needed. Second, changes in cultural behavior (abandonment of sites and agricultural features) are used to infer climatic changes, which in turn are used to explain the cultural developments observed. The circularity makes it difficult to sort out cause and effect.

By contrast, Seltzer and Hastorf's (1990) study of local climatic change and prehistoric agriculture in the Mantaro River Valley is based on independent data sets. Climatic change is inferred from late Holocene glacial history; prehistoric agricultural production is reconstructed through analysis of paleoethnobotanical remains from sites spanning a sequence of some 900 years. The authors use contemporary agricultural analogs to establish changes in arable land distributions for a period of presumed climatic cooling beginning about 1290 A.D. and lasting approximately 200 - 300 years. They argue that climatic conditions would have been depressed about 150 m from modern levels, resulting in a loss of 29% and 54% of arable land for potatoes and maize, respectively.

Prior to the period of climatic cooling (during the Middle Horizon and early Late Intermediate Period, A.D. 600 - 1300) sites were located near the valley floors. During the succeeding late Late Intermediate Period (locally termed the Wanka II phase, A.D. 1300 - 1460) settlements shifted markedly. Sites were moved to very high hills overlooking the valley. They became large and highly aggregated. This shift coincided with climatic cooling. Later, sites were relocated to lower elevations with the Inka conquest of the region about A.D. 1460.

Changes in the relative frequencies of potato and maize remains from excavated samples mirror these shifts in settlement. Potato ubiquity was highest and maize ubiquity lowest in samples from the Wanka II phase (i.e, when climate was cooler and settlements were located at higher elevations). From the beginning of the sequence to the end, potato ubiquity remained relatively constant, until the last (Inka) phase, at which time it dropped significantly (from 30% to 19%). In contrast, maize ubiquity was halved when the population moved to higher elevations (from 41% to 20%), but experienced a greater than three-fold increase from Wanka II to the Inka period (from 20% to 70%) when settlements were relocated downslope.